Wcfs2024 Day 1

This document contains the final draft of papers to be
presented on Day 1 of WCFS 2024. The papers in this

document follow the sequence of papers of Day 1
listed in the WCFS 2024 Conference Programme.
This is NOT a publication. It is solely for participants

to view during the conference to facilitate the
discussion. Please do not send the file to other people.
Please delete the file after the conference
On
ly
fo
rR
Life Cycle Structural Integrity Management of Floating Structures
ea
with a focus of floating bridges
din Torgeir Moan
g
Norwegian University of Science and Technology, 7491 Trondheim, Norway
at
Faculty of Maritime and Transportation, Ningbo University, Ningbo, 315211, China
tormo@ntnu.no
W
CF
Abstract. An overview of important developments regarding safety management of floating structures is
S
given. The possible transfer of approaches from the oil and gas industry is briefly described. Based on relevant
20
experiences with accidents and means to control the associated risks, are categorized from a technical-physical
as well as human and organizational point of view. Structural risk relates to extreme environmental and acci-
24
dental events, as well as structural degradation and can be controlled by use of adequate design criteria, espe-
cially to provide damage tolerance, as well as life cycle quality assurance and control. A risk analysis relating
to ship impacts on a floating bridge is briefly presented and possible mitigation of the risk of road traffic
disruption, is commented upon.
Keyword, Life cycle, Floating structures, Integrity Management, Ship impact risk.
1 Introduction
Floating structures are used for different purposes in various sectors, like oil and gas industry, offshore renewable
energy, fish farms as well as infrastructure like floating bridges (Fig. 1). In deep water they are used because the
costs are less than for bottom fixed structures while they are used in more shallow water because of challenges
with possible foundations on soft seabeds or simply because they provide some mobility, i.e. for use in different
locations. While ships and rafts were the earliest floaters, support on discrete pontoons- (semi-submersibles, pon-
toon bridges, ) are commonly used for permanently located structures. They are moored by a connection to land,
piles (in shallow water), or catenary, taut of tensioned mooring lines. Ship-shape facilities in harsh environments
are positioned by a turret arrangement; i. e. the vessel is designed to rotate about a cylinder which is moored to
the seabed by catenary mooring lines, while heading control is ensured by a DP system.
Floating structures in different sectors should in general be fit for their purpose and fulfill socio-political criteria
that address the safety of personnel, aesthetics especially for near shore structures, environmental sustainability,
budgetary and legal constraints. Moreover, they should not represent a hindrance to other activities, e. g. bridges
should not hinder ship traffic unnecessarily.
a) Oil & gas floating. b) Wind turbines. c) Aquaculture cages. d) Coastal infrastructure.
and subsea structures (floating bridge).
Fig.1 Examples of floating structures in various ocean sectors.
The continuous innovation to deal with new serviceability and safety requirements in the oil and gas industry
in the last decades, have led to an industry which has been in forefront of developing design and analysis meth-
odology for safety assessment, including decision under uncertainty through reliability and risk analysis methods
1-3. On the other hand, the cost of structures in this industry are relatively small compared to the income. Hence,
transferring technology from this sector to other sectors should be done with cost reduction in mind.
In the paper the structural integrity management (SIM) in emerging ocean sectors, with emphasis on civil in-
frastructure, is discussed with reference to the comprehensive basis for principles and methods established in the
oil and gas industry, a leading ocean sector.
On
ly
2
fo
2 General requirements
rR
ea
din
The basic concepts of floating structures, initially created in the oil and gas industry, are semisubmersible with
catenary or tension-leg mooring system, spar and shipshape floater/barge 1. Those concepts have served as
g
models or inspiration for floating structures in emerging ocean or coastal sectors. Floating structures should be
at
designed to be serviceable, safe and, increasingly, sustainable, as well as fabrication friendly.
W
2.1 Serviceability
CF
Serviceability requirements refer to behavior that hamper the function as an industrial facility or its function as a
means in a transport system etc. The different types of floating structures considered herein; all have the function
S
of carrying a payload. Some oil and gas platforms and fish cages also need to provide containment of oil and gas
20
and live fish, respectively. Oil and gas platforms and wind turbines support industrial plants which call for limited
24
motions (accelerations) to ensure an efficient operation of the machinery and mooring system. Floating bridges
provide support for the public that should “always” be available, i.e. the disruption of road traffic should be lim-
ited.
2.2 Safety
General. The focus in this paper is on structural integrity management during the life cycle: design, fabrication/
installation and operation. While all floating structures are subjected to hazards associated with payloads,
metocean conditions and ship collisions, the latent energy in hydrocarbons in oil and gas platforms implies a
particular risk potential. Safety requirements are specified to avoid fatalities, environmental damage and property
damage, and are related to the following failure modes:
- overall, rigid body instability (capsizing) under permanent, variable and environmental loads
- failure of (parts of) the structure, foundation or mooring systems, considering permanent, variable, and envi-
ronmental as well as accidental loads
which ultimately can lead to fatalities, pollution or economic loss. Fatalities depend on the number of persons on
the facility – on duty or public, evacuation options etc. Economic losses involve failure and damage expenses as
well as the costs of being partly or completely out of service.
Service experiences. Safety is the absence of failures or accidents. Hence, experiences with accidents is an im-
portant basis for establishing a safety management approach. While accident statistics give an overview of the
situation, more insight about the safety features can be obtained from more detailed investigation of individual
accidents. Accident statistics for oil and gas platforms has been regularly compiled in WOAD 4. Information
about individual accidents or failures of oil and gas platforms, described e.g. in 2, 5-7. Most accidents are due
to human errors occurring at some stage in the lifecycle. Accidents statistics and causes for bridges are presented
e.g. by Zhang et al..8. Design errors, construction mistakes, collisions and overloads make up more than 70% of
the failures. It is noted that there is a global failure among the few existing bridges floating bridges (Hood Canal,
9). Statistics of wind turbine accidents, mainly related to turbines on land maybe found in 10,11. The focus is
on relatively frequent electrical and mechanical failures, due to deterioration, lack of maintenance etc. “Blade
failure” can arise from several possible sources, and often occur in areas with a drastic change of the geometry,
e.g. at the blade root. Tower failure is attributed to lightning strikes, but also poor design, fabrication defects and
lack of proper maintenance (which is difficult to perform) as well as underestimation of wind speeds, e.g. in
typhoon prone areas. Moreover, tower failures have been caused by broken blades hitting the tower. Since there
are very few floating wind turbines, the accident potential for the hull and mooring system needs to be inferred
from that of similar oil and gas platforms with due account of the difference in life cycle integrity management.
In the fish farming-industry the safety is focused on prevention of fish escape which implies environmental dam-
age, and occupational safety 12. Fish escape occurs due to operational errors and damage to the cage containing
the fish associated with well boat operations or initial fabrication defects or deficient maintenance.
A common feature of accidents in the oil and gas industry and bridges is the contribution of human factors at
some stage in the life cycle. The experiences with floating structures in the emerging sectors, is limited but the
trend seems to similar as in the mentioned sectors. Operational errors are the primary cause of accidental events
such as explosions, fires and ship impacts. Fabrication errors might cause a deficiency in strength. Design and
On
ly
3
fo
rR
analysis errors occur. The management of structural integrity to ensure acceptable safety is discussed in Section
2.3 and 2.4.
ea
It is noted that it took several years of operation before deterioration especially in terms of fatigue cracks, was
din
recognized as a hazard to ships and offshore structures.
g
Failure modes. Capsizing is assumed according to current practice, to occur for an intact freely floating structure
at
about an axis where the overturning moment due to wind exceeds the stabilizing hydrostatic moment. It is noted
that the motions induced by waves are not considered despite attempt to develop dynamic stability requirements.
W
While for most floating structures the extreme environmental condition with a return period of 50 or 100 years,
CF
results in the governing overturning moment while for wind turbines it is an operational condition with so called
rated wind speed which is most critical. For a floater with tension-leg mooring the tension provides the stability
S
and slack in the mooring implies instability. For a floating bridge in an operational mode, the anchoring at both
20
ends is intended to resist relevant torsional moments in the girder and prevent “capsizing”.
While stability of the intact floater is an ULS criterion, stability of the floater in damaged condition is an ALS
24
requirement which depends on the damage condition and the overturning wind moment. Damage is typically
caused by ship impact which results in flooding and buoyancy loss, but technical or operational errors associated
with a possible ballast system, could also lead to buoyancy loss. Some standards specify damage conditions in
terms if one or two compartment flooding while others require risk analysis to determine relevant damage at a
certain probability level.
It is interesting to note that capsizing of an intact floating platform or wind turbine is absent in the experience
basis, global instability typically develop due to operational error of the structure causing abnormal ballast condi-
tion or structural damage condition, e. g. due to explosions, ship impacts or other external cause. For instance,
the three catastrophic accidents in Fig.2 resulted in capsizing/sinking after a sequence of events with structural
damage and flooding 13.
Fig. 2. Selected catastrophic accidents in the oil and gas industry (Oil platform disasters – Wikipedia).
Basically, structural failure occurs due to overload or deterioration due to fatigue or corrosion, when the re-
sistance, R is less than the load effect, S or when the cumulative damage exceeds the failure limit. From a human
factors perspective failure can be due to too small safety factors to account for the normal uncertainty and varia-
bility in the resistance or load effects relating to ultimate limit state (ULS) and fatigue limit state (FLS) criteria.
But the main cause of actual structural failures is abnormal resistance or accidental loads due to human errors and
omissions, not least in connection with fatigue which is the most important strength degradation process. Due to
the small size of initial (fabrication) defects, even abnormal cracks (with a sizes > 0.1-0.5 mm depth) are difficult
to detect, and such defects have been the cause of major failures 6.
The failure rate for mooring lines on mobile units in the North Sea was 0.04 per line year of. chains made
before 1985. When design procedures were improved, it was reduced to 0.01, e.g. 7, 14. The high early failure
rate was due to use of oversimplified methods for calculating the line tension as well as due to abnormal strength
such as local bending at fairleads (which is not accounted for in design; fabrication defects, wear etc.) and damage
during handling of the lines, The relatively high prevailing failure rate is partly due to the latter type of causes.
The data for mooring systems for other floating structures is limited, but the failure rate for production platforms
seems to be like that of mobile units 15. Further clarification of the experienced failure rate is needed.
However, to fully explain accident event sequences, it is necessary to interpret them in view of the human and
organizational factors (HOF) of influence, as reviewed in 2-3,5.
On
ly
4
fo
2.3
rR
Structural integrity management for oil and gas platforms
ea
General. In Table 1 the causes of structural failures are categorized and the corresponding measures to control
the accident potential are listed. Accidents with load carrying structures in most sectors are due to human errors
din
that materialize in accidental loads or deficient strength 2-3. A rational integrity management approach should
g
therefore be based on: goal-setting; not prescriptive; probabilistic; not deterministic; first principles; not purely
experiential, integrated total; not separately; and balance of safety elements; not only hardware. In general, the
at
measures include design criteria, quality assurance and control (QA/QC) relating to the engineering process as
W
well as the hardware and operational procedures. With the increased digitalization of design, fabrication and op-
CF
eration of structures, software errors are of particular concern in QA/QC. Structural reliability analysis (SRA) can
be used to establish ULS design criteria by calibration of partial safety factors.to be consistent with a target level
S
for the notional failure probability 16-17. It is noted that the random uncertainties in the ultimate strength com-
monly have limited effect on the reliability compared to that inherent in load effects. On the other hand, the
20
systematic uncertainty, bias, in strength and load effects have the same effect on the reliability measure. Regarding
24
fatigue the most important issue is to recognize this failure mode in design and inspection planning. Fatigue cri-
teria are calibrated also to reflect the effect of inspection 7. The SRA is based on normal variability and uncer-
tainties in load effects and strength, and not the effect of human errors. Since the fatigue endurance depends on
local geometry, which is hard to control, it is susceptible to fabrication “errors”. The accidental loads and abnor-
mal strength caused by human errors, are considered in the Accidental Collapse Limit State (ALS), and quantified
(to the extent possible) by Quantitative Risk Analysis 18.
Table 1. Cause of structural failures and risk reduction measures 2.

Cause Risk Reduction Measure
Less than adequate safety - Increase safety factors or margins in ULS, FLS.
margin to cover “normal” - Improve inspection of the structure (FLS)1)
inherent uncertainties.
Gross error or omission dur- - Reduce probability and consequence of hazards, e.g. by
ing: - Improving skills, competence, self- checking (for d, f, o)
- design (d) - QA/QC of engineering process (for d)
- fabrication (f) - Direct design for damage tolerance (ALS) (for events occurring during f, o)
- operation (o) - Provide measures to detect, control and mitigate hazards at an early stage and initiate
mitigation actions, e.g. by monitoring or Inspection/repair of the structure (for f, o)2)
Unknown phenomena. - Research & Development
1) Measure by Structural Reliability Analysis
2) Measure by Risk Assessment
ALS approach to make structures robust (damage tolerant). Despite the efforts made to avoid error-induced
accidental actions or abnormal resistance, they cannot be eliminated. For this reason, ALS criteria were introduced
to ensure that “small damages, which inevitably occur, do not escalate into disproportionate consequences” 7.
Such criteria were first introduced by the Norwegian Petroleum Directorate (Currently: The Norwegian Ocean
Industry Authority) 19, and later adopted for instance in ISO standards for oil and gas platforms, bridges 20.
The design check is a two-step procedure, 21:
1) estimate accident scenarios for various hazards with an annual exceedance probability of the order 10-2 -10-4.
2) demonstrate survival under permanent and environmental loads, by a ULS design check with partial safety
factors equal to 1.0.
It is noted that the robustness that is provided by the ALS criteria is also needed to get the necessary damage-
tolerance to rely on inspection of e.g. crack before they lead to rupture. Finally, it is noted that an ALS criterion
was introduced to limit the risk associated with “abnormal” wave loading at an exceedance probability of 10-4,
say with a high crest or other unusual shape – which is not a simple “extrapolation” of the 10-2 event, with rele-
vance for potential wave-load-in-deck scenarios.
Quality assurance and control (QA/QC). In view of the service experiences with accidents, QA/QC of the
analysis and design and fabrication process as well as during operation, is crucial. The QA/QC in the design phase
is particularly important for novel concepts and, with the increasing digitalization, also of computational tools and
their use. The QA/QC of the engineering process should address errors and omissions of individuals, inadequate
procedures, software and lack of robustness of the organizations. Lack of adequate training, time, and teamwork
or back-up (disciplined process auditing procedures in the organization) has been the root cause of such errors 2,
5. A main challenge in the design phase is to be aware of possible new phenomena, e.g. new load phenomena,
On
ly
5
fo
rR
dynamic response features and failure modes, especially associated with novel conceptual designs. Laboratory
tests are crucial in this context. “Recently” discovered new phenomena such as ringing and springing of TLPs,
ea
and vortex induced motions of floating platforms, were observed in time before any catastrophic accident oc-
din
curred, and the necessary precautions made. It is also interesting to observe that the failure mode of fatigue for
ships and offshore platforms– was fully recognized in 1980-1990s despite earlier experiences 6. Another issue
g
is that fatigue depends on local structural quality and hence easily be susceptible to fabrication errors. and is hence
at
dependent on adequate damage tolerance. The quality check of the structure during fabrication and operation is
primarily done by visual or NDT inspection. Remote monitoring of structural components is carried out mainly
W
performed to ensure the accuracy of methods used in predicting the behavior, while health monitoring to detect
CF
e.g. crack damage before it is critical, is challenging due to the extensive instrumentation needed to have an ade-
quate representation of possible crack sites in large steel structures, .and is not much in use for platforms. How-
S
ever, monitoring of possible leaks in floating structures is a viable approach to detect flooding before it escalates
20
into more severe consequences.
24
Regulations, standards, guidelines. Standards provide a bridge between the state of knowledge (research front)
and engineering design and operation. Moreover, standardization can help reduce costs since they can limit un-
necessary alternatives and thereby make fabrication and assembly focused on optimal life cycle costs. While
structures traditionally should fulfill serviceability and safety requirements, serviceability requirements are rec-
ommended in standards, but are decided by the owner. The focus in standards is hence of safety. Sustainability
requirements in standards, are still at an early stage, and efforts to establish criteria and methods for assessment
is needed.
The regulatory regimes for floating structures in various sectors, differ. There are international standards like
ISO for oil and gas platforms 20., IEC for floating wind turbines, 22. Provisions for stability and other maritime
issues are normally based on those of IMO and their corresponding national organizations. In addition, many
national authorities and classification societies issue standards and recommended practice.
A main challenge with the ULS design criteria is the occurrence of new phenomena, especially associated with
novel conceptual designs. “Recently” discovered new phenomena such as ringing and springing of TLPs, and
vortex induced motions of floating platforms, were observed in time before any catastrophic accident occurred,
and the necessary precautions made. Other phenomena, such as wave-in-deck load, caused severe damage on
some platforms, before risk mitigation actions were taken 23.
2.4 Structural integrity management of floating structures in the emerging sectors

As mentioned above, knowledge transfer from structural engineering in the oil and gas sector should be made
with due attention to the differences in function (Section 2.1), hazards and consequences and hence the required
safety level as well as the cost of improving the safety level. In the following the focus is on safety aspects.
The offshore oil and gas industry has a large damage potential due to the large energy and pollution potential
inherent in the hydrocarbons. All ocean structures are more or less subjected to ship impact risk and environmental
loads.
The fatality risk varies with the personnel at risk - from none on floating wind turbines, small fish farms and
solar power plants, to some manning on large fish farms, to public exposure on large on floating bridges. The
potential environmental damage caused by failures of these emerging floating structures is limited compared to
that for oil and gas platforms. But escape of fish from fish farms represents an environmental threat of concern.
A very important difference between the emerging sectors and the oil and gas sector is concerned with the
level of costs and profit. While floating wind, solar power plants and fish farms, typically represent an economy
in a smaller scale than oil and gas (production) platforms, floating bridges which are part of the national highway
system might represent a large investment and a potential societal loss in case of failure.
These differences would have an influence on the use of various safety measures
- while in general SLS, ULS and FLS design criteria are applied for all types of structures while the ap-
plicability of, e. g. the ALS criteria, needs to be discussed. ALS requirements are used for floating
bridge 24, to have at least the same safety level as oil and gas platforms.
- QA/QC is important for structures in the emerging sectors due to the novelty of function, layout etc. that
can imply new load conditions and failure modes etc.
- Moreover, differences in loads and structural behavior modes might require different environmental
data and analysis methods. For example, long floating bridges are subjected to inhomogeneous environ-
mental conditions, and the behavior of floating wind turbines depends on hydro- and aerodynamic loads
and active automatic control. Another example is the complex interaction between modular, elastic
floating structures and a multi-line mooring systems.
On
ly
6
fo
-
rR
The limited operational experiences with the floating structures in the emerging ocean sectors, also need
to be taken into account in the design of such facilities.
ea
While the standards for oil and gas platforms have reached some maturity, the international standard, IEC
din
61400-3-2 for floating wind turbines is at an early stage and refer extensively to oil and gas platform practice.
Provisions for stability and other maritime issues are based on those of IMO and their corresponding national
g
organizations, even if consequences in terms of fatalities and environmental damage is very limited. The applica-
at
tion of damage tolerance criteria (ALS) for overall stability, strength of the hull and mooring system.is still open
for discussion. For fish-farms and floating civil infrastructure like floating bridges, terminals etc. there are some
W
national standards, but with quite general guidance and reference to other standards. For floating bridges, some
CF
regulations//standards, like N-400 24 exist, while in practice, a specific design basis for each project is estab-
lished. In due time, when more in service experiences have been collected, they can result in a standard.
S
In principle, application of all risk mitigation approaches in Table 1, should be considered for a balance in
20
safety and lifecycle expenditure.
Public infrastructure is in a particular situation since a (global) failure might cause public fatalities and repre-
24
sent large costs, partly for repair/replacement, but mainly due to the costs of road traffic disruption. The Norwe-
gian Public Road Administration 24, hence, implements all risk mitigation measures in Table 1, i. e. limit state
criteria including ALS. Emphasis is placed on QA/QC (including risk analysis) to identify possible new hazards
(load phenomena) and failure modes as well as measures to control the associated risk. The QA/QC performed
for floating bridges in Norway, identified challenges that led to research projects, including extensive model test-
ing, e. g. 25. Moreover, several engineering design studies were conducted by two independent consortia of
consulting companies. For example, in the Bjørnafjord floating bridge project in Norway the QA/QC process
identified new phenomena, e. g. global dynamic instability of a curved floating bridge, hydrodynamic interaction
between pontoons, the effect of inhomogeneity in environmental conditions, wave-current interaction and the
important contribution of relatively “frequent”, moderate ship impacts to the risk of road traffic disruption. 25.
Moreover, there is a challenge to design the mooring system in view of the disturbingly high failure rate of moor-
ing lines in the offshore industry. Rather than introducing extra safety margins to cover such “unknown” phenom-
ena, it is crucial that sufficient numerical and experimental investigations are carried out during the OA/OC pro-
cess to establish a basis for accounting for the possible effects.
2.5 Structural reliability and risk analysis

The total efforts to ensure safety (design criteria, QA/QC) are supposed to ensure an acceptable safety level. In
principle tools like structural reliability analysis (SRA), e.g. 26 and quantitative risk analysis (QRA) , e. g. 18,
provide means to quantify failure probabilities of different failure consequences, and hence the risk of failure. As
mentioned in Section 2.3. SRA has been applied to calibrate the ULS requirements for floating structures in the
oil and gas industry, but not yet in emerging ocean sectors.
Some regulations and standards for oil and gas platforms require a QRA to estimate the risk of fatalities, envi-
ronmental damage and economic loss, e.g. 18, and its compliance with target levels. The contribution to the risk
due to structural failure have commonly been estimated by assuming that the ALS criteria are fulfilled based on
characteristic accidental loads at an annual probability level on e. g. 10-4 (Section 2.3) estimated by the risk anal-
ysis. This is, however, an approximation which does not give a good estimate of the risk, as discussed in Section
3.
Moreover, risk analysis is crucial at the initial stages of developing new concepts, e.g. floating structures in the
emerging sectors – to identify hazards, failure modes and – consequences. The results can then be used to specify
necessary research and development to cope with the identified challenges.
An illustration of the use of risk analysis is presented in Section 3 to use risk analysis to shed light on existing
design criteria for floating bridges.
3 Risk-informed design of floating bridges against ship impacts
3.1 General
Accidental loads are primarily handled by ALS criteria, which were introduced for offshore structures in 1984 to
ensure more robust structures by a design check in two steps, as mentioned in Section 2.3 19. This approach is
also adopted for bridges (e.g. 24), considering relevant accidental loads. Ship impacts is the dominating acci-
dental load and among the consequences: fatalities, environmental damage and economic loss, the latter and es-
pecially the cost of traffic disruption, is the governing design condition 9.. The accidental load should be
On
ly
7
fo
rR
determined by a risk analysis. As reviewed in 27, some standards for bridges (e. g. ASHTO; Eurocodes) refer
to risk-based design of bridges. However, the existing risk analysis procedures have limitations with respect to
ea
application to floating bridges. To establish a basis for risk-informed of structural integrity relating to ship impacts,
din
a comprehensive risk analysis has been conducted for an early design of the Bjørnafjord floating pontoon bridge.
3.2 Ship impact risk analysis procedure
g
at
The purpose of the present risk analysis is to estimate the ultimate consequence of ship impacts, namely: fatalities
to users of the bridge as well as road traffic disruption caused by the impact. The ship collision risk analysis 9,
27 is carried out by estimating the
W
CF
 Frequency versus impact energy and direction (traffic pattern-types and number of vessels, possible lanes
etc., root causes of operational and technical faults, bridge layout, considering the mitigation measures.
 Consequences
S
20
o Immediate consequences in terms of structural damage, possible buoyancy loss and girder motions
o Ultimate consequences in terms of fatalities/injuries and roads traffic disruption
24
It turns out 9 that the impact on the pontoons makes the largest contribution to the ship impact risk for the
current bridge. It is also found that the contribution to the fatality risk due to ship collisions, is small compared
to the road traffic risk of fatalities on the bridge. Hence, the focus in this paper is on road traffic disruption.
A three – step analysis procedure is necessary to apply to determine the immediate consequences in terms of
internal forces, possible permanent damage and the girder motions (accelerations) 9:
1) nonlinear shell analysis is carried out to determine the impact force-displacement (indentation) relationship in
the impact region (as indicated at the top of Fig. 3).
2) global analysis of the impact is made. As illustrated in Fig. 3, the ship-bridge contact, is modeled by a mass-
nonlinear spring system connected to the bridge, with a stiffness obtained by the force-deformation curve in step1.
The default model of the other parts of the bridge is modelled by a linear elastic model but is replaced by a
nonlinear shell sub-structure if the linear analysis indicates nonlinear material behavior.
3) Based on the results from these first two steps, the damage (indentation, fracture) and possible flooding is
determined. Fig. 4 shows an example of damage after the global analysis of a transverse impact on a pontoon.
Fig. 3 Impact analysis model 9. Fig. 4 Example of damage in a 125 MJ transverse
impact on the pontoon of the tall column 9 .
The repair time - including mobilization of the equipment, i.e. by use of drydock for pontoon repair, was esti-
mated to be between 2 and 20 months 9.
The traffic disruption time, when ULS and SLS criteria are nor fulfilled, was estimated by the annual frequency
of each impact scenario, the immediate consequence (structural damage, flooding) and the time for damage and
repair conditions based on the damage condition and choice of repair method.
3.3 Risk estimation and mitigation
It turned out that the expected annual downtime due to ship impacts was about 7 days in the initial study, with
90 % of the downtime stemming from impacts on the pontoons with an energy up to 50 MJ. This result caused
scrutinizing of the impact frequency analysis, which initially was focused on severe impacts (at an annual fre-
quency of 10-4) for the ALS check. Moreover, further ship traffic control measures to reduce the impact probabil-
ity, are pursued. Another option that was proposed 9, 27, was to introduce a limit state in addition to the ALS to
address the risk associated with frequent impacts, implying strengthening of the bridge girder and columns that
did not fulfill the ULS or SLS criteria in damaged and repair conditions.
On
ly
8
fo
4. Concluding remarks
rR
ea
Lifecycle structural integrity management is discussed in view of accident experiences for floating structures in
din
various sectors. The main cause of accidents is human and organizational errors and omissions that cause acci-
dental loads or abnormal strength. To achieve an acceptable safety level, therefore, requires fulfillment of ULS
g
and especially FLS (and inspection plan) requirements, since the latter are often not recognized in early phases of
at
the developments of new concepts. For high reliability/consequence class structures, fulfillment of damage toler-
ance (ALS) requirements, as well as QA and QC of the engineering process (including identification of hazard
W
and failure modes); inspection, monitoring and repair of the structure, are important. QA and QC tasks (including
CF
assessment of digital tools and their use), are particularly important for novel concepts. Finally, a study of risk-
informed design of a floating bridge under ship impacts, is presented, indicating that (“frequent”) impacts with an
S
energy in the range of 20-50 MJ, contribute more to the road traffic disruption than the impacts with an annual
20
exceedance frequency of 10-4, corresponding to an energy of about 150 MJ used for the ALS check. To ensure an
acceptable risk of traffic disruption, one risk mitigation option that was proposed in 9, 27, was to introduce a
limit state for design against frequent ship impacts in addition to the ALS requirement.
24
References
1. Chakrabarti, S.(ed.): Handbook of Offshore Engineering, Vol.1&2, Elsevier, Amsterdam. (2005).
2. Moan, T. Safety of floating offshore Platforms. Keynote lecture. 9th PRADS Conf. Travemünde-Lübeck, STG (2004).
3. Moan, T.: Safety of Offshore Structures, Keynote lecture, OSRC Conf., API, September 16-18, Houston, USA (2014).
4. WOAD: Worldwide Offshore Accident Databank, Det Norske Veritas, Oslo, updated continuously.
5. Bea, R.G.: Design for Reliability: Human and Organizational Factors, Chapter 12, Handbook of Offshore Eng, S.
Chakrabarti, (ed.), Elsevier, Amsterdam. (2005)
6. Moan, T.: Fatigue Reliability of Marine Structures – from the Alexander Kielland Accident to Life Cycle Assessment of
Safety, ISOPE Keynote lecture, San Francisco, J. ISOPE, 17(1), 1-21 (2007).
7. Moan, T.: Development of accidental collapse limit state criteria for offshore structures. Structural Safety, 31, (2), 124-
135 (2009).
8. Zhang, G., Liu, Y., Lan, S. Yang, J.: Causes and statistical characteristics of bridge failures: A Review. J. Traffic and
Transportation Engineering, 9(3) 388-406 (2022).
9. Moan, T., Jin, Y: Risk Analysis of a floating bridge subjected to ship collisions. Report to NPRA, January 15, (2023).
10. CWIF: Summary of Wind Turbine Accident data to 31. March 2020, www.caithnesswindfarms.co.uk accessed April
24.2020.
11. Santelo, T.N., de Oliveira, C. M. R., Dias Maciel, C., de A. Monteiro, R.B.: Wind Turbine Failures Review and Trends.
Journal of Control, Automation and Electrical Systems, 33:505–521 (2022).
12. Holen, S.M., Yang, X., Utne, I. B., Haugen, S.: Major accidents in Norwegian fish farming. Safety Science, 120: Dec
; 32-43 (2019).
13. OPD (2024) Oil platform disasters - Wikipedia. accessed on 20.08.2024
14. Stiff, J. (2014) How reliable are reliability calculations? (Illustrated with Station keeping Examples). Proc.
Offshore Structural Reliability Conference, September 16-18, Houston, USA.
15. Smedley, P. and Petruska, D.: Comparison of Global Design Requirements and Failure Rates for Industry Long Term
Mooring Systems, Proc. Offshore Structural Reliability Conference, API, September 16-18, Houston, USA (2014).
16. Cornell, C.A.: Structural Reliability- Some contributions to Offshore Technology, Paper No. OTC 7753, Proc. 27th Off-
shore Technology Conference, Houston (1995).
17. Moan, T.: Safety Level Across Different Types of Structural Forms and Material – Implicit in Codes for Offshore Struc-
tures, SINTEF Report STF70 A95210, Trondheim, Prepared for ISO/TC250/SC7 (1995).
18. Vinnem, J.E.: Offshore Risk Assessment, Springer, London (2014).
19. NPD: Regulations for load-carrying structures for extraction or exploitation of petroleum, The Norwegian Petroleum
Directorate, Stavanger (1984).
20. ISO 2394. General Principles on Reliability for Structures”, International Standards Organization, London, UK (2015).
21. NORSOK N-001. Integrity of offshore structures”, Norwegian Technology Standards, Oslo, Norway (2021).
22. IEC 61400-3-2. Design requirements for floating wind turbines, IEC, Geneva, Switzerland (2024).
23. Ma, L, Swan, C.: Wave-in-deck loads: an assessment of present design practice given recent improvements in the de
scription of extreme waves and the nature of the applied loads. Ocean Engineering, 285, 115302 (2023).
24. NPRA: Handbook N400: Design of bridge (in Norwegian). Norwegian Public Road Administration, Oslo (2024).
25. Moan, T., Økland, O.D., Øiseth, O.A.: Analysis and design of floating bridges – recent advances and challenges. In Large
Floating Solutions: Design, Construction and Legality of Offshore Structures and Buoyant Urbanism. Wang, Wang,
Weinert, de Graaf-van Dinther (Editors). (2024).
26. Melchers, R.E. Structural Reliability Analysis and Prediction, John Wiley & Sons, Chichester. UK (1999).
27. Moan, T.: Estimation and mitigation of ship collision risk of a floating bridge. Submitted for publication (2024).
On
ly
fo
rR
Impact of Different Tendon Configurations on the Hydrodynamic
ea
Performance of a TLP Floating Offshore Wind Turbine
din
g
Zhe Tan1, 2 and Ronghua Zhu1, 2
1
at
Ocean College, Zhejiang University, Zhoushan, China
W
2 Yangjiang Offshore Wind Power Laboratory, Yangjiang, China
CF
Zhu.richard@zju.edu.cn
S
20
Abstract. This study investigates the impact of different tendon configurations on the hydrodynamic perfor-
mance of a Tension Leg Platform (TLP) floating offshore wind turbine. Utilizing the NREL 5 MW turbine and
24
the MIT TLP design, three new mooring system configurations were developed, including a mooring layout
with two tendons arranged in an X shape between adjacent vertical tendons. The configurations were tested
under various load cases using OpenFAST. Results indicate that the X layout configurations reduce the surge
and yaw motions compared to the original layout under operating conditions. Eight tendon failure scenarios
with different failed tendons were also analyzed.
Keywords: TLP, Floating Offshore Wind Turbine, Mooring System, Hydrodynamic Performance, Tendon
Failure
1 Introduction
The increasing demand for renewable energy has driven significant interest in offshore wind energy, particularly
floating offshore wind turbines (FOWTs). Among various FOWT floating foundation designs, the tension leg
platform (TLP) has an advantage of stability, including low motion response in pitch [1]. However, the perfor-
mance of TLPs is highly dependent on the configuration of their mooring systems. Recent studies have explored
various aspects of TLP FOWT design and performance, including hydrodynamic behavior, structural integrity,
and mooring system dynamics. Xiaoqian Wang et al. [2] proposed a 5 MW typhoon-resistant fully submerged
TLP FOWT, which adopts a hybrid mooring scheme with tendons and catenary mooring lines. Yajun Ren et al.
[3] studied the dynamic characteristics of a tension-leg floating wind turbine under tendon failures through model
experiments. Zihua Zhang et al. [4] investigated the dynamic response of a TLP floating offshore wind turbine
using vertical carbon fiber composite tendons and circumferential catenary mooring lines for combined mooring.
J. S. Rony et al. [5] studied the coupled dynamic characteristics of a tension leg floating wind turbine-wave energy
hybrid platform under different tendon configurations, including four mooring configurations: four-line, five-line,
eight-line, and nine-line.
Existing studies on TLP floating wind turbine mooring systems mostly use vertical tendons with different
numbers of tendons. In this paper, a mooring layout with two tendons arranged in an X shape between adjacent
vertical tendons is tested. The study employs a 5 MW wind turbine model and an MIT TLP design as the basis
for analysis. Three mooring configurations are developed, including an original vertical mooring layout and an X
layout. The performance of these configurations is evaluated through numerical simulations, and six degrees of
freedom platform motions are analyzed. The simulations are under operational conditions, extreme load condi-
tions, and tendon failure scenarios to provide a comprehensive understanding of the mooring system’s impact on
TLP FOWT performance.
2 Model description
The wind turbine and the tension leg platform are based on the research of NREL, including a 5 MW turbine and
an MIT TLP design [6–9]. The main properties of the wind turbine, TLP platform, and original mooring system
are listed in Table 1. The original mooring system consists of 8 tendons, with two tendons per fairlead. Based on
the original mooring system, three new mooring system configurations were generated, with the total tension on
each fairlead remaining unchanged. The three mooring system configurations have the same total mooring line
volume as the original configuration. This setup is intended to ensure that different mooring configurations have
the same material cost. Under this assumption, if the cross-sectional area of some mooring lines increases, en-
hancing their extensional stiffness, the cross-sectional area of other lines in this mooring system will decrease to
On
ly
2
fo
rR
keep the total volume of the mooring lines constant. This makes different mooring system configurations more
comparable.
ea
din
Table 1. Main properties of the wind turbine, TLP platform, and original mooring system
g
Properties Value
Rated power 5 MW
Cut-in, rated, and cut-out wind speed
at 3, 11.4, and 25 m/s
Platform diameter
W 18 m
Platform draft
CM location below still water level CF 47.89 m
40.61 m
Water displacement
S
12180 m3
Water depth 200 m
20
Radius to fairleads
Depth to fairleads
27 m
47.89 m 24
Number of mooring lines 8
Line diameter 0.127 m
Line mass per unit length 116.03 kg/m
Line extensional stiffness 1500000000 N
Upstretched line length 151.73 m
As shown in the Fig. 1, the three mooring configurations include two layouts. The left layout is the same as the
original layout of the MIT TLP, but each fairlead has three vertically downward tendons, increasing the total
number of tendons to 12. These three tendons are equivalent to the original two tendons, with the same connection
positions, total extensional stiffness, and unstretched length. In the right layout (X layout), each fairlead has one
vertically downward tendon numbered from 1 to 4, and additionally, there are two thinner tendons arranged in an
X layout between adjacent vertical tendons. This layout is feasible because the solver does not account for the
interactions between mooring lines. However, in a real-world scenario, the positions of the fairleads need to be
adjusted to avoid line collisions and crossings.
The properties of the three mooring configurations can be seen in Table 2. The twelve tendons of MC1 have
the same properties. MC2 and MC3 using an X layout have two kinds of tendons, including the thicker tendons
numbered from 1 to 4 and the thinner tendons numbered from 5 to 12 arranged in an X layout. The detailed
parameters of the load cases are shown in Table 3. The wave is generated using the JONSWAP spectrum and the
wind is steady. LC1 and LC2 represent operating conditions and LC3 represents parked conditions. In LC2, the
wind and waves are misaligned. Tendon failure simulations were also performed. Failure scenarios are shown in
Table 4. The tendons in the table were selected to break because they are the ones under the highest tension. And
load case, LC3, was applied. The simulations were run by OPENFAST with a duration of 1800 s, and the data
from the last 1000 seconds was used for analysis.
Fig. 1. Diagram of different layouts with numbered tendons

On
ly
3
fo
rR
Table 2. Properties of three mooring configurations
ea
Mooring configurations Line diameter (m) Line mass per unit length (m) Line extensional stiffness (N)
din
MC1 (Original layout) 0.1037 77.3513 1.00E+09
0.127 116.027 1.50E+09
MC2 (X layout)
0.0884
g 56.2643 7.27E+08
0.147 at
154.703 2.00E+09
MC3 (X layout)
W
CF
0.072 37.5064 4.85E+08
Table 3. Parameters of load cases

S
20
Case name Wave height (m) Peak-spectral period (s) Wave direction (°) Wind speed (m/s) Wind direction (°)
LC1
LC2
7
7
13
13
0
15
11.4
11.4
24 0
0
LC3 14 19 0 42 0
Table 4. Failure scenarios of tendon failure simulations
Failure scenarios Mooring configurations Number of failed tendons Abbreviated as

1 MC1 3 MC1 F3
2 MC1 3 and 7 MC1 F3&7
3 MC1 3, 7, and 11 MC1 F3&7&11
4 MC3 3 MC3 F3
5 MC3 10 MC3 F10
6 MC3 3 and 10 MC3 F3&10
7 MC3 10 and 11 MC3 F10&11
8 MC3 3, 10, and 11 MC3 F3&10&11
3 Results and discussion
3.1 Response analysis of different mooring configurations

The motion data of the TLP platform under different load cases is shown sequentially in Fig. 2 (LC1), Fig. 3
(LC2), Fig. 4 (LC3). The data metrics used include the mean, maximum, minimum, standard deviation (Std), and
root mean square (Rms). In LC1, the X layout (MC2, MC3) shows a better effect in suppressing surge motion.
Their mean, maximum, and root mean square values of surge motion are smaller than those of the original layout
(MC1). In terms of sway, heave, and roll motions, MC1, MC2, and MC3 show little difference. For pitch motion,
the X layout has a slightly larger amplitude. It can be seen that the vertical tendons play a major role in suppressing
pitch motion. In the X layout, each fairlead has only one vertical tendon, resulting in an increased pitch amplitude,
but the maximum value is still within 1°. Both use the X layout, but the vertical tendons of MC3 have a larger
extensional stiffness, resulting in a smaller pitch amplitude than MC2. For yaw motion, the X layout can signifi-
cantly reduce the yaw amplitude, which is only about half of that of MC1. Moreover, under LC2 where the wind
and waves are in different directions, the X layout has a more obvious effect in suppressing yaw motion. In LC3,
due to the large wave height and wind speed, the amplitude of all degrees of freedom increases. The X layout no
longer has an advantage in surge, sway, and heave amplitudes, and the roll and pitch amplitudes are larger than
those of the original layout, but the yaw amplitude is still smaller than that of MC1. Due to the stronger vertical
tendons of MC3, its overall motion performance is better than that of MC2.
On
ly
4
fo
Mean rR
Max Min Std Rms
ea
8
6 1
din
4
2 0
0
g
-2 -1
-4
MC1 MC2 MC3 MC1 MC2 MC3
0.1
at
0.2
W
0
0
CF
-0.1
-0.2
-0.2
S
1 2
0.5
0
20
24
0
-2
Fig. 2. Statistics of six degrees of freedom motion response under LC1
Mean Max Min Std Rms

8 2
6
4 1
2 0
0
-2 -1
-4
0.05 0.2
0
0
-0.05
-0.1 -0.2
-0.15
1 2
0.5
0
0
-2

20 5
0 0
-20 -5
1
0
0
-1
-2 -1
2 2
0 0
-2 -2

On
ly
5
fo
3.2 Tendon failure analysis
rR
ea
In the tendon failure analysis, eight failure scenarios were set under the load case, LC3. However, due to the rapid
capsizing of the platform when tendons 3, 7, and 11 of MC1 failed simultaneously, no data was generated within
din
the 800-1800 second range for analysis. Therefore, only the remaining 7 scenarios are shown in Fig. 5. After
g
tendon failure, the motion amplitudes of sway, heave, and roll did not change significantly compared to the intact
LC3 condition, but the amplitudes of surge, pitch, and yaw motions increased significantly. The maximum surge
at
values in all scenarios exceeded 10 m, and the surge amplitude was greater under MC3 tendon failure, with a
W
maximum value approaching 20 m. As for pitch, the scenarios with smaller amplitudes included MC1 F3, MC1
CF
F3&7, MC3 F10, and MC3 F10&11. However, the maximum pitch value in scenarios where tendon 3 failed
within MC3 reached about 5°, indicating that losing vertical tendons in the X layout has a highly adverse effect
S
on pitch response. MC3 F10 and MC3 F3&10 had significantly larger yaw amplitudes due to the failure of the
tendons disrupting the symmetry of the mooring system, with minimum values approaching -10°. Overall, the two
20
tendon failure scenarios under MC1 performed better in six degrees of freedom motion. However, the platform
24
quickly capsized when tendons 3, 7, and 11 of MC1 failed simultaneously, whereas the platform remained stable
within a certain range of motion when tendons 3, 10, and 11 of MC3 failed simultaneously.

20
Surge (m)
10
-10
MC1 F3 MC1 F3&7 MC3 F3 MC3 F10 MC3 F3&10 MC3 F10&11 MC3 F3&10&11
5
Sway (m)
-5
1
Heave (m)
-1
0.5
Roll (°)
-0.5
6
4
Pitch (°)
2
0
-2
5
0
Yaw (°)
-5
-10
Fig. 5. Statistics of six degrees of freedom motion response under tendon failure scenarios
On
ly
6
fo
4 Conclusions
rR
ea
This study investigates the impact of different tendon configurations on the hydrodynamic performance of a TLP
din
floating offshore wind turbine. The analysis reveals that the X layout mooring configurations (MC2, MC3) effec-
tively suppress surge and yaw motions compared to the original layout (MC1) under operating conditions. In the
g
extreme condition (LC3), the X layout’s advantage diminishes, but MC3’s stronger vertical tendons provide better
at
overall motion performance. Tendon failure scenarios indicate that the platform’s stability is significantly affected
by the loss of vertical tendons. MC3 is better at maintaining stability compared to MC1 under multiple tendon
W
failures. Overall, the tendon configuration in the X layout has certain advantages under different conditions. It is
CF
necessary to optimize the properties of the tendons in the new mooring layout to improve the overall hydrody-
namic performance.
S
References
20
1. 24
Edwards, E.C., Holcombe, A., Brown, S., Ransley, E., Hann, M., Greaves, D.: Evolution of floating offshore wind plat-
forms: A review of at-sea devices. Renewable and Sustainable Energy Reviews. 183, 113416 (2023)
2. Wang, X., Zhang, Z., Zhuge, P.: Transient analysis of a novel full submersible floating offshore wind turbine with CFRP
tendons. Ocean Engineering. 266, 112686 (2022)
3. Ren, Y., Shi, W., Venugopal, V., Zhang, L., Li, X.: Experimental study of tendon failure analysis for a TLP floating
offshore wind turbine. Applied Energy. 358, 122633 (2024)
4. Zhang, Z., Wang, X., Zhang, X., Zhou, C., Wang, X.: Dynamic responses and mooring line failure analysis of the fully
submersible platform for floating wind turbine under typhoon. Engineering Structures. 301, 117334 (2024)
5. Rony, J.S., Karmakar, D.: Coupled dynamic analysis of hybrid STLP-WEC offshore floating wind turbine with different
mooring configurations. J. Ocean Eng. Mar. Energy. 9, 623–651 (2023)
6. Vardaroglu, M., Gao, Z., Avossa, A.M., Ricciardelli, F.: Numerical Modelling of the MIT/NREL TLP Wind Turbine and
Comparison with the Experimental Results. J. Phys.: Conf. Ser. 1669, 012015 (2020)
7. Matha, D., Fischer, T., Kuhn, M., Jonkman, J.: Model Development and Loads Analysis of a Wind Turbine on a Floating
Offshore Tension Leg Platform. (2010)
8. Matha, D.: Model Development and Loads Analysis of an Offshore Wind Turbine on a Tension Leg Platform with a
Comparison to Other Floating Turbine Concepts: April 2009. (2010)
9. Jonkman, J., Butterfield, S., Musial, W., Scott, G.: Definition of a 5-MW Reference Wind Turbine for Offshore System
Development. (2009)
On
ly
fo
rR
Hydroelastic Responses of Submersible Cages under Extreme Waves
ea
din
Mingyuan Ma1, Hong Zhang1,2, Xinyu Zhang2, Chien Ming Wang3,4
g
1Cities Research Institutes, Griffith University, Gold Coast, Queensland 4222, Australia
at
2 School of Engineering and Built Environment, Griffith University, Gold Coast, Queensland 4222, Australia
3 School of Civil Engineering, The University of Queensland, St Lucia, Queensland 4072, Australia
W
4 Blue Economy Cooperative Research Centre, Launceston, Tasmania, Australia
CF
hong.zhang@griffith.edu.au
S
20
Abstract: The accurate prediction of the dynamic responses of fish cages under extreme sea states is of vital
24
importance for the structural reliability of marine aquaculture infrastructure. The present study investigates the
hydroelastic behaviour of flexible fish pens under different extreme wave conditions with specified recurrence
periods. Initially, environmental contours for extreme wave conditions were derived using the direct Inverse
First Order Reliability Method (IFORM), by fitting the generalised extreme value (GEV) distribution to the
declustered storm peaks and extrapolating to conditions with specified recurrence periods. The wave condition
contours are then mapped to the response space of the fish cage according to the Norwegian standard NS9415
using finite element method (FEM) numerical simulations. The research results indicate that significant differ-
ences in the wave responses of the fish cage occur under severe non-severe extreme wave conditions. The
study also examines the reliability and serviceability of the fish cage under these extreme sea states. It is found
that while the fish cage exhibits a substantial hydroelastic response at low frequencies and high energy, this
response is reduced under severe sea conditions. The techniques and findings provide valuable insights for the
reliability analysis of marine aquaculture infrastructure.
Keywords: Extreme sea states; Marine aquaculture infrastructure; Finite element analysis
1 INTRODUCTION
Marine aquaculture is an important component of the global food supply chain. According to DNV [1], global
production of marine aquaculture is expected to increase by about 1.5 times by 2050. In Australia, aquaculture is
a rapidly growing industry known for providing high-quality, safe, and sustainable seafood, with an output value
of A$3.3 billion in 2020 [2]. Storm Bay in Tasmania has been identified by both industry and the Tasmanian
Government as a suitable location for sustainable development of the industry, particularly for offshore fish farm-
ing [3]. However, unlike traditional fish cages located in nearshore, sheltered waters, fish cages exposed to high-
energy environments offshore are at great risk of structural failure, which could result in loss of personnel and
property. Therefore, it is necessary to analyze the dynamic response of such structures, especially under extreme
sea conditions.
Numerical modelling is an effective and flexible tool, as physical tests and field monitoring are limited by
conditions and external factors. Currently, there are extensive models for simulating the dynamic response of fish
cages. Due to the slender and flexible characteristics of fish cages, the hydrodynamic loads can be estimated using
Morison equations [4] or screen models [5], and the dynamic responses are analyzed using the lumped-mass
method [4] or finite element method (FEM) [5].
In addition, forecasting the extreme sea states is an open topic. The common approach is to establish a joint
probability distribution based on buoy measurement data or hindcast data and extrapolate to the sea state under a
specified probability of exceedance. Various extreme modelling methods have been tested and compared in [6].
Based on the aforementioned background and literature review, the present study hopes to investigate the hy-
droelastic response of fish cages under the sea states of a given extreme recurrence period.
2 STUDE AREA AND EMPLOYED DATA
The selected study area is located at a potential fish farming site (Longitude: 147.47°, Latitude: -43.40°, water
depth: 75.5 m) in Storm Bay, Tasmania, Australia. Storm Bay has been identified as a priority area for the potential
On
ly
2
fo
rR
expansion of salmonid marine farming in the Government’s Sustainable Industry Growth Plan for the Salmonid
Industry (refer to Fig. 1).
ea
In the present study, the potential fish farming site analysed is located in the offshore waters on the western
din
side of Bruny Island. The Collaboration for Australian Weather and Climate Research (CAWCR) wave hindcast
dataset [7] from 1990 to 2023, which is simulated using the third-generation spectral wave model Wave Watch
g
III [8], is employed to extrapolate to the extreme sea states reaching up to 50 years.
at
W
CF
S
20
24
Fig. 1. Study site of fish farms in Storm Bay, Tasmania, Australia.
3 METHODOLOGIES
To predict extreme sea conditions with a specific recurrence period, a common practice is to use the entire dataset
for parameter estimation to ensure sufficient samples and thus reduce the variance of the estimated values. How-
ever, probability distribution estimation usually requires the sample data to be independent and identically dis-
tributed. [9] have pointed out that time series of meteorological and ocean variables typically exhibit significant
short-term serial correlation. Therefore, resampling the sample data is necessary. We employ an algorithm to
decluster the sea state series into a series of storm peaks (Fig. 2), ensuring that the interval between adjacent storm
extremes is not less than five days, according to [13], to eliminate the autocorrelation in the series.
The declustered storm peak variables, Xpeak can be fitted by the generalised extreme value (GEV) distribution:
1 𝑥 𝜇 / 𝑥 𝜇 /
⎧ exp 1 𝜉 1 𝜉 , 𝜉 0
𝜎 𝜎 𝜎
𝑓 𝑥; 𝜇, 𝜎, 𝜉 , 1
⎨ 1 𝑥 𝜇 𝑥 𝜇
⎩ exp exp , 𝜉 0
𝜎 𝜎 𝜎
where μ, σ and ξ are the location, scale, and shape parameters, respectively. To describe the tail of the distribution
more accurately, we suppose a two-part model to fit the distribution of the storm peaks (Fig. 3a). The probability
of exceedance P(Xpeak > x) is derived as
⎧ 1 𝑓 𝑥 𝑑𝑥 , 𝑥 𝜃
⎪
𝑃 𝑋 𝑥 /
. 2
⎨ 𝑥 𝜃
⎪𝑃 𝑋 𝜃 1 𝜉 , 𝑥 𝜃
⎩ 𝜎
In Eq. (2), the part of samples over the threshold θ is fitted by the generalised Pareto (GP) distribution:
/
1 𝑥 𝜃
𝑓 𝑥; 𝑘, 𝜀, 𝜃|𝑥 𝜃 1 𝑘 , 3
𝜀 𝜀
where k ≠ 0 and ε are the shape and scale parameters, respectively. The threshold θ is determined from the empir-
ical estimates of recurrence values of variables from 3 to 6 months, and the one with the minimum root mean
square error for the empirical estimates of recurrence values of one year and above is chosen. Therefore, the
recurrence period TR corresponds to a specified probability of exceedance of P (Xpeak > x) is calculated as
On
ly
3
fo
rR 𝑇
1
. 4
ea
𝑀𝑃 𝑋 𝑥
din
where M is the average number of storm peaks per year. In order to determine the joint extreme sea states of the
two variables, namely the significant wave height Hs and the mean zero-up wave period Tz, we used a direct
g
Inverse First-Order Reliability Method to describe the environmental contours (EC) of TR = 50 years. The two
at
variables Hs and Tz are projected on a straight line through the origin at an angle φ, so that the exceedance bound
can be constructed using the univariate estimate of the variable X = Tzcosφ+Hssinφ (as seen in Fig. 3b).
W
CF
S
20
24
Fig. 2. Declustering the wave series into storm peaks; the dashed line represents the minima of the adjacent peaks.
0
10
10-1
10-2
Empirical estimates
10-3 GEV
GP
95% C.I. of GP
10-4
5 10 15
Xpeak (m)
(a) (b)
Fig. 3. Estimating extreme sea states: (a) fitting of the univariate distribution, and (b) constructed EC.
On the other hand, the commercial software AquaSim [10] based on the FEM was used to simulate the dynamic
response of the fish cage, where the floating collar was modelled by beam elements, the fish cage net was modelled
as membrane elements, and the mooring ropes were modelled as truss elements. The hydrodynamic forces dF on
the structural element are estimated by the Morison equation:
1 𝜋𝐷 𝜋𝐷
𝑑𝐹 𝜌𝐶 𝐷𝑑𝐿 𝑢 𝑥 |𝑢 𝑥| 𝜌 1 𝐶 𝑑𝐿𝑢 𝜌𝐶 𝑑𝐿𝑥. 5
2 4 4
where ρ is the water density, u is the wave water velocity, x is the displacement of the structural element, D is the
element cross-section diameter, dL is the length of the element, Cd = 1.2 is the drag coefficient, and Cm = 1 is the
added mass coefficient. In addition, the maximum wave height for a sea state shall be assumed around twice Hs
when using a regular sea for the analysis according to the Norwegian standard NS9415 [11]. In addition, NS9415
suggests that the sea state under a recurrence period of TR = 50 years is considered as the design condition.
4 RESULTS DISCUSSIONS
For fish cages exposed to high-energy environments, submerging them underwater is beneficial in avoiding strong
wave loads, especially during extreme sea conditions. As illustrated in Fig. 4, we consider a fish cage submerged
20 meters below the mean water level and analyse its hydroelastic response under all possible sea states under TR
On
ly
4
fo
rR
= 50 years. The dimensions and material parameters of all structural components of this cage can be found in
Table 1. Additionally, as described in the methodology, twelve representative sea states are selected from the EC
ea
in Fig. 3b for analysis. The specific wave conditions are shown in Table 2. For each wave condition, the wave
din
steepness is defined as Sz = 2πHs/(gTz²).
Currently, only unidirectional monochromatic waves, as shown in Fig. 4, are considered. Based on the FEM
g
simulation results, the most severe failure of the floating collar occurs at the connection point with the mooring
at
lines, as shown by the distribution of von-Mises stress in Fig. 5a. The maximum vertical displacement occurs at
W
the mid-span of the mooring constraints from Fig. 5b. In the direction of the incoming waves, the mooring lines
on the windward side experience greater mooring forces. Additionally, the volume change of the cage is also a
CF
crucial indicator to monitor, as it is related to fish welfare. Therefore, these four indicators will be used in the
S
subsequent analysis. An example of the time series for these four indicators is plotted in Fig. 6, where wave
condition #1 in Table 2 is applied, corresponding to the peak point of the EC contour and can be considered as a
severe condition counted as exceedances.
20
24
(a) (b)
Fig. 4. Sketch of a submersible fish cage below the water surface: (a) isometric view, and (b) plan view.
(a) (b)
Fig. 5. The part of the fish cage floating collar with the most critical response based on FEM simulation: (a) von-Mises
stress, and (b) vertical displacement.
The maximum responses of the four indicators for the fish cage under all possible extreme sea states with a TR
= 50 years are plotted in Fig. 7, associated with the varying wave steepness Sz. For wave conditions #1 to #4, the
fish cage exhibits a significant dynamic response, and the response intensity gradually decreases as the steepness
decreases, although the mooring force and von-Mises stress slightly increase for wave condition #2. For sea states
#5 to 8, the period Tz varies significantly, but the significant wave height Hs is relatively small, indicating a minor
response of the fish cage. The steepness for sea states #9 to #12 increases noticeably, but the dynamic response
of the cage is not as significant as for those of #1 to 4. This implies that low-frequency but high-energy sea states
contribute more to structural failure.
On
ly
5
fo
rR
Table 1. Dimensions and material parameters of structural components of the fish cage.
ea
Structural components Parameter Value
din
Outer/Inner diameter (m) 42/40
Cross-section outer diameter/Pipe thickness (m) 0.45/2.65E-2
g
Floating collar Mass in water (kg/m) 21.06
at
Bending stiffness (Nꞏm2) 2.83E5
Poisson’s ratio 0.4
W
Mask width/Thread diameter (m) 0.03/0.0025
CF
Net chamber Mass in water (kg/m) 3.51E-4
Line stiffness (kN/m) 80.6
S
Diameter/Length (m) 0.064/56.56
Mooring line Mass in water (kg/m) 0.37
Line stiffness (kN/m)
20 153.57
24
Bottom weight Mass (kg/m) 50
Table 2. Analysed wave conditions from the sea state with recurrence period TR = 50 years.
No. Hs (m) Tz (s) Sz No. Hs (m) Tz (s) Sz
#1 7.01 9.27 0.0522 #7 0.25 9.68 0.0017
#2 5.92 11.30 0.0297 #8 0.13 5.92 0.0023
#3 4.82 13.17 0.0178 #9 0.13 1.89 0.0233
#4 3.52 14.65 0.0105 #10 2.01 3.37 0.1134
#5 2 15.19 0.0056 #11 3.52 4.58 0.1075
#6 0.84 13.01 0.0032 #12 5.44 6.86 0.0740
Fig. 6. Time series plots of the four analysed indicators of the fish cage based on FEM simulation
5 CONCLUSIONS
The study presents a case analysis on the response of a submersible fish cage under extreme sea conditions in
Storm Bay, Tasmania, Australia. Using a hindcast dataset, sea states with a 50-year recurrence period were ex-
trapolated, and the dynamic response of the fish cage was simulated using finite element method (FEM) simula-
tions. Representative wave conditions were extracted from the Environmental Conditions (EC), and design wave
conditions were determined in accordance with NS9415. The dynamic response of the fish cage was analyzed to
identify the most critical conditions. The study reveals that significant stress concentrations develop at the moor-
ing connections, accompanied by notable vertical displacement between these connections. The wave conditions
in the EC can be categorized into three regimes, with the fish cage exhibiting substantial dynamic response under
low-frequency, high-energy sea conditions. These findings provide valuable insights for engineers analyzing the
On
ly
6
fo
rR
extreme response behavior of offshore aquaculture structures. Future studies will focus on incorporating realistic
irregular waves to further refine the analysis.
ea
din
g
at
W
CF
S
20
24
Fig. 7. The max response plots of the four analysed indicators of the fish cage under different wave conditions from the EC.
Acknowledgements
The authors acknowledge the financial support of the Blue Economy Cooperative Research Centre, established,
and supported under the Australian Government’s Cooperative Research Centres Program, grant number CRC-
20180101. The authors also would like to thank Aquastructures AS for granting access to their software AquaSim
(https://aquasim.no).
References
1. DNV: Marine Aquaculture Forecast to 2050. Det Norske Veritas Group, Hovik (2021).
2. Chen, D.,Wang, C. M., Zhang, H.: Examination of net volume reduction of gravity type open-net fish cages under sea
currents. Aquacultural Engineering 92, 102128 (2021).
3. Peter & Una Rockliff: ENVIRONMENTAL IMPACT STATEMENT - Draft Storm Bay North Marine Farming Devel-
opment Plan. Department of Natural Resources and Environment Tasmania, Hobart (2017).
4. Zhao, Y. P., Li, Y. C., Dong, G. H., Gui, F. K., Wu, H.: An experimental and numerical study of hydrodynamic charac-
teristics of submerged flexible plane nets in waves. Aquacultural engineering 38(1), 16-25 (2008).
5. Kristiansen, T., Faltinsen, O. M.: Modelling of current loads on aquaculture net cages. Journal of Fluids and Structures
34, 218-235 (2012).
6. Haselsteiner, A. F., Coe, R. G., Manuel, L., Chai, W., Leira, B., Clarindo, G., Guedes Soares, C., Hannesdóttir, Á., Dimi-
trov, N., Sander, A., Ohlendorf, J.-H., Thoben, K.-D., Hauteclocque, G. d., Mackay, E., Jonathan, P., Qiao, C., Myers,
A., Rode, A., Hildebrandt, A., et al.: A benchmarking exercise for environmental contours. Ocean Engineering 236,
109504 (2021).
7. Durrant, T., Hemer, M., Smith, G., Trenham, C., Greenslade, D.: CAWCR Wave Hindcast - Aggregated Collection. v5.
CSIRO. Service Collection (2019), http://hdl.handle.net/102.100.100/137152?index=1, last accessed 2024/08/26.
8. Tolman, H.L.: A Third-Generation Model for Wind Waves on Slowly Varying, Unsteady, and Inhomogeneous Depths
and Currents. Journal of Physical Oceanography 21, 782-97 (1991).
9. Mackay, E., de Hauteclocque, G., Vanem, E., Jonathan, P.: The effect of serial correlation in environmental conditions
on estimates of extreme events. Ocean Engineering 242, 110092 (2021).
10. Aquastructures Homepage, https://aquasim.no/resources/documentation.html, last accessed 2024/08/26.
11. Norge, S.: NS9415 Marine fish farms. Requirements for site survey, risk analyses, design dimensioning, production,
installation and operation. Standards Norway, Lysaker (2021).
On
ly
fo
rR
Hydroelastic Response of Large Floating Offshore Base for Offshore Wind
Farm
ea
din
g
Saika Iwamatsu1, Mako Yamamoto2, Ananda Raiz Pambela2 and Kazuhiro Iijima2
1
at
Joint Research Chair for Offshore Wind System Integration, The University of Osaka, Osaka, Japan
W
2 Department of Naval Architecture and Ocean Engineering, The University of Osaka, Osaka, Japan
iwamatsu.saika.b4k@osaka-u.ac.jp
CF
S
20
Abstract. The offshore wind industry is a rapidly growing sector, and it is estimated that the number of off-
shore Wind Farms (WF) will increase in the future. At present, offshore WF is based at close-by onshore
24
substations or port facilities where power transmission, construction, and operation and maintenance (O&M)
are performed. This is reasonable for offshore area close to coastal area. However, it is no more justified for
far offshore areas such as EEZ. In the case of such offshore WF, the construction of offshore base is required
since it can reduce the trip time between the base port and the offshore WF site and increase the workability.
Therefore, this study examines a very large floating structure with substation and O&M functions by numerical
simulations. Similar projects such as the North Sea Wind Power Hub are reported, but they are assumed to be
deployed at a water depth of less than 50m. In our companion paper, we propose a new concept of an offshore
base which is a kind of a very large floating structure installed at a water depth more than 50 meters in Japan’s
EEZ. The proposed offshore base has a complex geometry and properties. It is semi-submersible and rectan-
gular shape, but with a few cut-outs for its functionalities such as the port function as a base, and with heavy
concentrated-mass objects mounted on its upper deck. In this study, the elastic response of a very large floating
structure with non-uniform mass and with complex geometry, is analyzed by using numerical calculations.
Keywords: hydroelasticity, very large floating structure, offshore wind farm
1 Introduction
In the North Sea, plans are progressing for the construction of artificial islands within the offshore wind farm
(WF) development areas. A notable project is the North Sea Wind Power Hub ([1], see Fig. 1). This artificial
island features a substation function and is designed to interconnect with multiple planned offshore WFs in the
vicinity. Additionally, it incorporates storage via P2X and operational and maintenance (O&M) functionalities,
aiming to enhance its utility as a central hub for offshore WF operations in the North Sea, thereby potentially
reducing costs relative to other options. Generally, for offshore WFs located beyond a 60 km distance from shore,
it is suggested from an O&M perspective that using a Service Operation Vessel (SOV) for several weeks of off-
shore anchorage may lead to cost savings compared to daily Crew Transfer Vessel (CTV) operations[2].
In Japan, the majority of waters, including its Exclusive Economic Zone (EEZ), have depths far exceeding 50
meters, necessitating the essential development of floating wind turbines suitable for deep waters. Similarly, arti-
ficial islands created through land reclamation, serving as bases for offshore WFs, are not well-suited for Japan.
Therefore, Yamamoto [3] has proposed and designed a Semi-Submersible Offshore Base tailored for Japan's ma-
rine environment (see Fig. 2). This concept includes not only substation and O&M functions but also assembly
capabilities. It is semi-submersible and rectangular shape, but with a few cut-outs for its functionalities such as
the port function as a base, and with heavy concentrated-mass objects mounted on its upper deck. Verification is
required using a non-rectangular, non-uniform mass model to ensure its functionalities and safety. Hence, this
study focuses on examining its response by using non-uniform mass massive floating structures with openings.
Refer to the companion paper for detailed concept design of the Semi-Submersible Offshore Base [4].
During assembly and O&M operations, it is concerned that heavy items such as cranes and wind turbine com-
ponents will move on the deck, necessitating consideration of their impact on the structural response of the floater
due to their movement. The cut-out part of the structure may exhibit complex hydroelastic response. Therefore,
the objective of this study is to elucidate the elastic response of a mega floating structure with openings and loaded
with heavy items to non-uniform loads.
On
ly
fo
rR
ea
din
g
Fig. 1. North Sea Wind Power Hub Program. [1] at
Fig. 2. Offshore wind firm float base concept. [3]
W
2 Numerical Method CF
S
20
The numerical analysis theory utilized in this paper employs VODAC (Very Large Floating Structure Oriented
Dynamic Analysis Code), developed by Iijima et al. [5]. Here, an overview of its theory is briefly given.
24
In VODAC, for the fluid domain, the method of singular point distribution and Kagemoto's hydrodynamic
interference theory [6] are combined, while in the structural domain, the concept of Group body is incorporated,
and for the structural domain, the substructure method is applied, enabling handling of even larger structures. As
given in Fig. 3, it is a hierarchical hydroelasticity method both in fluid and structural domains.
Verification of structural responses of ultra-large floating structures using VODAC has been conducted by Iijima
et al. [7] and Suzuki et al. [8], confirming the capability and accuracy of VODAC.
Goo’s method Present method

Response Analysis in Waves Response Analysis in Waves
Structure domain Fluid domain Structure domain Fluid domain
STRUCTURE FLOATING BODIES STRUCTURE GROUP BODIES
PANELS SUB-STRUCTURES FLOATING BODIES

(Singularity Distribution Method)
PANELS
Fig. 3. Hierarchy of the present method. [5]
2.1 Structural design
In this study, we envisage an ultra-large floating structure with dimensions of 1200 m width and 600 m depth,
corresponding to Model Base 3 as proposed by Yamamoto (see Fig. 2) [3]. Base 3 is conceptualized for an offshore
distance of 300 km and water depth of 5000 m, equipped with O&M and assembly functions. The structural
analysis of this model utilized the analytical model of the mega floating structure proposed by Suzuki et al. [9].
This approach clarifies the characteristics of structural responses and serves as an effective tool for guiding deci-
sions in selecting structural forms. For detailed calculation methods, refer to our companion paper [4]. In this
paper, the parameters outlined in the companion paper are presented in Table 1.
Table 1. Key parameters of semisub-large floating structure.
Item Symbol Unit Value

Deck height H m 4.0
Deck thickness t m 0.025
Bending rigidity EI Nm2/m 4.2×1010
Mass per unit area M kg/m2 3000
Distance between columns L m 60.0
Column diameter D m 25.0
Draught d m 22.0
The Dogger Bank Wind Farm substation [10] is referenced for planning the offshore base. It is assumed that it
accommodates an offshore substation capable of 1.2 GW (total weight 7000 tons).
Wind turbine components stored on the deck for assembly and O&M are based on the IEA 15MW wind turbine
[11]. The masses of these components are 65.25 tons for the blades, 860 tons for the tower, and 820 tons for the
nacelle. The mass and moment of inertia of the blades are calculated assuming three stacked blades, using a
simplified assumption of a rectangular prism for the moment of inertia.
On
ly
3
fo
rR
The crane used for assembly is assumed based on existing cranes on catalogues. While Yamamoto [3] concept
includes a gantry crane, ring lift cranes and crawler cranes are also potential options. Referring to the weight and
ea
dimensions of a 3200-ton lifting capacity crawler crane [12] capable of assembling 15MW wind turbines, the
din
assumed weight including the load and main body is 4000 tons. This assumption exceeds the weight of existing
crawler cranes because there is potential to lift floating foundations or wind turbines larger than 15MW in the
g
future.
at
Each port function as a harbor opening is assumed to measure 240 m in width and 300 m in depth. This as-
sumption is based on the 240 m rotor diameter of the IEA 15MW wind turbine [11], and the approximate length
of hydrogen transport ships [13].
W
2.2 Model Case CF
S
Model-1 is depicted in Fig. 4, serving as the rectangular base model for this study. This is a reference model.
20
24
a) Upper deck. b) Y-Z Axis.
c) X-Z Axis.
Fig. 4. Model – 1 of Large Floating Offshore Base.
Model-2 and 3 are shown in Fig. 5 and Fig. 6, respectively. Compared to Model-1, Model-2 and 3 maintain the
width and depth while incorporating harbor and assembly functions through an opening.
Fig. 5. Model - 2 Fig. 6. Model - 3
2.3 Analysis Case

In this study, we investigate the structural response of a large floating structure with openings and non-uniform
mass distribution. To this end, it is necessary to determine the deck payload. The large floating structure in this
study is intended for installation near offshore wind farms, potentially equipped with functions such as power
conversion, O&M, and assembly. Among these functions, assembly is emphasized. When using cranes for assem-
bly necessitating ground bearing capacities of 35 ton/m² on land, local reinforcement immediately beneath the
cranes may be necessary [14]. In this study, we conduct analyses assuming the simultaneous operation of two
cranes on the deck to assemble floating wind turbines. Table 2 presents the x, y coordinates of crane positions for
each analysis case, while Fig. 7 and Fig. 8 depict schematic diagrams of crane positions.
Furthermore, we load all assumed payloads and conduct structural assessments. The payloads assumed include
a substation, two cranes, and ten IEA15MW wind turbines. Fig. 9 illustrates a schematic diagram depicting all
payloads loaded.
Table 2. Crane placement conditions in Model-2.
Crane No. Case - 1 Case - 2 Case – 3 Case - 4 Case - 5

Crane – 1 (150, 0) (150, -150) (150, -270) (150, -270) (150, 0)
Crane – 2 (-150, 0) (-150, -150) (-150, -270) (-150, 0) (-150, -270)
On
ly
4
fo
rR
ea
din
Fig. 7. Case -1, 2, 3 definitions.
g
Fig. 8. Case -4, 5 definitions. Fig. 9. Fully loaded.
at
W
CF
3 Results and Discussion
S
3.1 The effect of openings on the floating structure
Fig. 10 shows the results of static analysis of vertical displacement distribution and angular displacement at Y=0
20
24
m. Compared to Model - 1, 2 and 3 exhibit significant variations at the location of the openings. From Fig. 10 a),
it is observed that in Model 1, except at the ends, the deck vertical displacement distributes within a range from
0.039 to 0.080 m. In Model - 2 and 3, the vertical displacement at the areas with openings are -0.045 m and -0.035
m, respectively.
From Fig. 10 b), it is evident that while the openings in Model - 2 are at X=-120 to 120 m, rapid changes in
angular displacement around Y-axis occur at X=-150 and 150 m, indicating occurrence at the edge of the opening.
This phenomenon is similarly observed in Model - 3, where results of angular displacement are obtained outside
the opening. The presence of openings is presumed to cause non-uniform buoyant forces acting on the floaters,
leading to abrupt changes near the openings.
Vertical displacement distribution [m]
Angular displacement [deg.]
0.5 0.2
0.1
0
-0.1
0 -0.2
-0.3
-0.4
Model - 1 Model - 2 Model - 3
Model - 1 Model - 2 Mode - 3
-0.5 -0.5
-600 -300 0 300 600 -600 -300 0 300 600
X [m] X [m]
a) Vertical displacement distribution. b) Angular displacement.

Fig. 10. Displacements due to openings in the structure (Y = 0).
Fig. 11 shows the RAO (response amplitude operators) of bending stress under waves with varying frequencies.
Fig. 11 a) indicates the point at the center coordinates (x, y) = (0, 0) of the decks for Model - 1 to 3. In large
floating structures, maximum bending stress occurs when the wave frequency is near the characteristic frequency
[9]. The characteristic frequency of this model is 0.362 rad/s [3], and maximum bending stress occurred at the
wave frequencies near the characteristic frequency in all models. This is commonly observed for all models.
Fig. 11 b) shows results from the point at (x, y) = (-150, 0) for Model - 1 and 2. This point is at the base of the
opening in Model - 2, which is presumed to be a vulnerable area prone to damage. From Fig. 11 b), it is observed
that similar to the deck center coordinates, maximum bending stress occurred near the characteristic frequency of
0.362 rad/s. Furthermore, while Fig. 11 a) showed no significant difference in bending stress among all models,
the bending stress at this point is 25 MPa for Model 1 and 37 MPa for Model - 2, indicating that the stress in
Model - 2 is 1.48 times higher. This suggests that thorough validation at the base of the openings is essential.
Comparing Model - 2 and 3 based on Fig. 10 and Fig. 11, similar trends in structural deformation and bending
stress can be inferred from the results of Model - 2 to Model - 3.
50 50
Model - 1 Model - 2 Model - 3 Model - 1 Model - 2
Stress amplitude [MPa]
Stress amplitude [MPa]
40 Characteristic frequency
40
0.362 rad/s
30 30
20 20
10 10
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Wave frequency [rad/s] Wave frequency [rad/s]
a) Stress amplitude (X=0, Y=30). b) Stress amplitude (X=-120, Y=30).
Fig. 11. Stress due to openings in the structure under wave conditions of a significant wave height of 13.0 m and an average
wave period of 14.0 s (Y = 0).
On
ly
5
fo
3.2 Structural response to crane movement on the deck
rR
ea
Fig. 12 shows the results of static analysis for vertical displacement distribution and angular displacement around
Y-axis along the line at Y=0 m, while Fig. 13 depicts the same but at X=30 m.
From Fig. 12 a), it is observed that in Case 4 and 5 where two cranes are positioned near the center of the deck
din
g
and at the tip of the protrusion, the vertical displacement reaches approximately -0.8 m. The displacement is
smallest in Case 1, where both cranes are placed at the center, with a displacement of about -0.05 m. Fig. 13 a)
indicates that in the vertical direction along the Y-axis where the crane is installed, Case 3, positioned at the tip
at
W
of the protrusion, shows a displacement of -2.5 m. This is influenced by the buoyancy supported by surrounding
CF
columns; in the central part of the deck like Case 1, surrounded by columns, vertical displacement is minimized.
However, when the crane is positioned at the tip of a protrusion as in Case 3, fewer columns can support the
S
crane's mass, resulting in a displacement of -2.5 m.
From Fig. 12 b), it is noted that angular displacement around X-axis shows a similar trend across all Cases.
Only Cases 4 and 5 do not exhibit angles near 0 degrees within the range of X = -120 to 120 m, attributed to the
20
24
diagonally offset crane positions. Fig. 13 b) illustrates that the installation of cranes causes the floater to tilt. Even
in Case 3 and 4 where the crane is positioned at the tip of an opening, the inclination angle is approximately -0.4
degrees, suggesting minimal overall structural inclination.

0.5 0.2
0.1
0
0
-0.1
-0.2
-0.5 -0.3
-0.4 Case - 1 Case - 2 Case - 3
Case - 4 Case - 5
Case - 1 Case - 2 Case - 3 Case - 4 Case - 5
-1.0 -0.5
-600 -300 0 300 600 -600 -300 0 300 600
X [m] X [m]

Fig. 12. Displacements due to changing the positions of two cranes installed around the openings in the structure (Y = 0).
0.5 0.2 Case - 1 Case - 2 Case - 3 Case - 4 Case - 5

Case - 1 Case - 2 Case - 3 Case - 4 Case - 5 0.1
0
0
-0.5
-0.1
-1.0 -0.2
-1.5 -0.3
-2.0 -0.4
-2.5 -0.5
-600 -300 0 300 600 -600 -300 0 300 600
Y [m] Y [m]

Fig. 13. Displacements due to changing the positions of two cranes installed around the openings in the structure (X = 30).
3.3 Fully loaded
Fig. 14 shows the vertical displacement distribution from the static analysis. In this numerical simulation, point
loads representing a substation, two cranes, and ten 15 MW wind turbines (nacelles, and blades) are applied. At
the location of the heaviest substation, the displacement is -1.0 m. In contrast, the greatest displacement occurred
at the point near the crane installation location, specifically at X=-150 m, Y=-90 m, where the displacement is -
1.2 m. The displacements at each crane position are both about -1.0 m. The larger displacement around the cranes
compare to the substation is attributed to differences in buoyancy due to the number of surrounding columns.
While multiple columns are present around the substation, the crane areas are open with fewer columns. As a
result, although the substation is heavier, the crane areas exhibited greater displacements.
a) Only self-weight. b) All payloads.

Fig. 14. Vertical deformation diagram of Model-2 when all payloads are input as point loads.
On
ly
6
fo
4 Conclusion
rR
ea
This study investigates the structural response of a mega floating structure with non-uniform mass distribution
din
and with complex geometry, designed for a floating offshore base. The key findings are summarized as follows:
g
 Using models without openings (Model - 1), with one opening (Model - 2), and with two openings (Model -
at
3), the effect of openings on the hydroelastic response was compared. The influence of the openings was evi-
dent in the vertical displacement and inclination angles. When focusing on bending stress, there was no signif-
W
icant difference in trend among the three models at the center of the deck, but at the vicinity of the openings,
CF
bending stress was approximately 1.8 times higher in Model - 1 compared to the others.
 In scenarios where assembly operations involve crane movements on the deck, the structural response was
S
investigated by varying the crane positions along the openings in Model - 2. Comparing vertical displacements
20
at different crane positions—near the center of the deck (Case - 1) and at the tip of the openings (Case - 3)—
revealed displacements of -0.8 m and -2.5 m, respectively. This disparity is attributed to the fewer columns
24
surrounding the crane position at the tip of the openings compared to the center of the deck. The maximum
inclination angle near the crane installation was -0.4 degrees, indicating minimal overall structural inclination
due to the crane's position at the opening's tip.
 Using Model - 2, the structural response was examined while loading equipment equivalent to a 1.2 GW sub-
station, components for 10 units of 15 MW wind turbines, and 2 cranes. The results con-firmed the feasibility
of accommodating all planned equipment on Model - 2.
Countermeasures against the effect of the openings, especially strengthening of local structures may have to be
considered in the actual design.
Acknowledgements
This work was supported by the Sasakawa Scientific Research Grant from The Japan Science Society.
References
1. North Sea Wind Power Hub, https://www.tennet.eu/north-sea-wind-power-hub, last accessed 2024/7/19.
2. CATAPULT Offshore Renewable Energy, https://guidetoanoffshorewindfarm.com/wind-farm-costs, last accessed
2024/7/19.
3. Yamamoto, M.: Study on the Basic Concept of Offshore Bases for EEZ Wind Farms. Osaka University, bachelor thesis.
(2024).
4. Yamamoto, M., Iwamatsu, S., Iijima, K., Shibata, M., Tatsuhara, H.: A proposal of floating offshore base for wind farm
deployed in EEZ, to appear in Proceedings of WCFS2024, Hong Kong, (2024).
5. Iijima, K., Suzuki, H., Yoshida, K.: Structural Analysis of Very Large Semi-submersibles in Waves. Journal of the Society
of Naval Architects of Japan, Vol. 181, pp.281-288. (1997). (in Japanese)
6. Kagemoto, H., Dick, K. P. Y.: Interaction among multiple three-dimensional bodies in water waves: an exact algebraic
method. J. Fluid Mech., Vol. 166, 189–209 (1986).
7. Iijima, K., Yoshida, K., Suzuki, H.: Structural Design Methodology of VLFS from the Viewpoint of Dynamic Response
Characteristics. Proceedings of the Third International Workshop on Very Large Floating Structures (VLFS’ 99), Vol. 1,
pp. 249-258. Honolulu (1999).
8. Suzuki, H., Yoshida, K., Iijima, K., Kobayashi, K.: Response Characteristics of Semisubmersible-Type-Megafloat in
Waves and Accuracy of Hydroelastic Response Analysis Program VODAC. OMAE2002, Vol. 2, pp. 773-781. Oslo,
Norway (2009).
9. Suzuki, H., Yoshida, K.: A Consideration on the Dynamic Behavior and the Structural Design of Large Scale Floating
Structure. Journal of the Society of Naval Architects of Japan, Vol. 178, pp. 473-483 (1995). (in Japanese)
10. aibel, https://aibel.com/news/three-giants-at-the-yard-in-haugesund, last accessed 2024/7/19.
11. Evan, G., Jennifer, R., Latha, S., Frederik, Z., Benjamin, A., Garrett, B., Nikhar, A., Fanzhong, M., Pietro, B., Witold, S.,
George, S., Roland, F., Henrik, B., Katherine, D., Matt, S., Christopher, A., Anthony, V.: Definition of the IEA 15-
Megawatt Offshore Reference Wind. Golden, CO: National Renewable Energy Laboratory. (2020)
12. MIC Co., Ltd. https://www.micjp.com/wp-content/uploads/2020/07/1ca2f8d06b3984d3f5a40d90250c0e80.pdf, last ac-
cessed 2024/7/19.
13. Kawasaki Heavy Industries, Ltd. https://www.khi.co.jp/pressrelease/detail/20220422_1.html, last accessed 2024/7/19.
14. IHI Transport Machinery Co.,Ltd. https://www.iuk.co.jp/crane/offshorewind.html, last accessed 2024/7/19.
On
ly
fo
rR
ea
din
g
at
A transient analysis of the tension leg platform for float-
ing wind turbine
W
Wei Huang1 CF
1 School of Civil Engineering, Guang Zhou, China
S
20
24
Abstract. Tension leg platform is one of classical supporting structures for float-
ing wind turbine. The motion and tension response of the tendon is crucial to be
estimated accurately in the detailed design stage. In this study, the floating struc-
ture model has been established considering several sources related to wind tur-
bine. Thus, the aerodynamic analysis model, hydrodynamic model and numerical
model for mooring tendons have been set up and the coupled effects have been
taken into account. The aerodynamic model is based on the blade element mo-
mentum theory. The panel model of the platform has been established based on
the classical potential flow software i.e. Hydrostar, and the mooring tendons has
been modelled based on the lumped mass method. Several cases are designed
considering the wave loads extracted from the metocean data report. One tendon
sudden failure case is the focus of the present work, the enlargement of the ten-
sion of the adjacent tendon under the severe environment has been discussed
thoroughly. The detailed effect of the failure of the tendon can be seen in the
discussion, which would be interested in the engineering application for the ten-
sion leg floating wind turbine.
Keywords: Tension leg platform; floating wind turbine; transient effect.
1 Introduction
The development of offshore wind energy is advancing rapidly, with technology be-
coming a key focus for researchers. Over the next decade, offshore wind capacity is
expected to increase by over 380GW [1]. As wind farms move into deeper waters,
floating structures are needed to support the turbines. These floating platforms, which
are more cost-effective than fixed foundations, come in four main types: Tension Leg
Platform (TLP), Semisubmersible, Spar platform, and barge. Mooring systems for these
platforms include catenary, taut mooring and tendon supported mooring types. TLP
offers a smaller footprint and better vertical stability, with horizontal motion being
more flexible. While TLP is not widely used for wind turbines yet, they offer significant
potential due to their balance of performance and cost. Yang and Kim [2] Studied an
extended tension leg platform (ETLP) in the Gulf of Mexico, finding that the failure of
one or two tendons alters system stiffness and natural periods. Wu et al. [3] Analyzed
the WindStar TLP at 160m depth with one tendon failure, revealing limited effective-
ness of shutdown strategies in reducing platform responses. Yang et al. [4] investigated
a 10MW FOWT at 110m depth, showing that tendon failures significantly increased
roll and yaw motions and peak tension. Tabeshpour et al. [5] examined the ISSC TLP
On
ly
fo
rR
ea
2
din
g
at
at 230m depth under tendon damage, noting similar response amplitude operators
(RAOs) for tension, pitch, and roll, and critical wave heading effects. Ren et al. [6]
W
Developed a 5MW Tri-Floater with a heave plate for 60m depth, finding that tendon
CF
failures doubled adjacent tendon tension and affected heave, pitch, and roll motions.
They also conducted an experimental study on a scaled model [7]. Besides, the transient
S
effects when mooring failures for other types like semi-submersible or Spar platforms
20
are also studied [8, 9]. Previous research indicates that TLP are well-established for oil
and gas fields but show significant potential for wind turbines. However, there is lim-
24
ited study on the transient response of TLP floating offshore wind turbines (FOWTs)
to sudden tendon failures or damage, especially under varying environmental condi-
tions and for different factors such as maximum tension, yaw angle, or offset. This
paper investigates these issues by designing a nine-tendon system for the OC4 FOWT.
Hydrodynamic analysis using Hydrostar provides key coefficients and forces, while
Orcaflex models the mooring tendons. The study includes transient analysis for single
tendon failure. The paper is structured as follows: Section 2 details the methodology,
including hydrodynamic methods, time-domain motion equations, and lumped mass
methods. Section 3 covers parameters for the OC4 and mooring tendons, environmental
conditions, and design cases. Section 4 presents numerical simulation results on TLP
motions and tendon tensions. Section 5 concludes the study.
2 Methodology
2.1 Hydrodynamics
Normally the floating structure is assumed to be a rigid body without forward speed on
the perfect fluid in some specific position, under the presence of the incidence propa-
gative wave. So the boundary value problem for the velocity potential 𝜙 can be written
as,
𝛻𝜙 0 𝑖𝑛 𝑡ℎ𝑒 𝑓𝑙𝑢𝑖𝑑 𝑑𝑜𝑚𝑎𝑖𝑛 1
𝜙 0 at the sea floor 𝑧 ℎ 2
𝑔𝜙 𝜔 𝜙 0 at the mean free surface 𝑧 0 3
𝜕𝜙
lim √𝑅 i𝑘𝜙 0 in the far field 4
→ 𝜕𝑅
Where ℎ denotes the water depth, 𝑅 is the radial distance 𝑥 𝑦 . The motion equa-
tion in the frequency domain can be expressed as [10],
𝐌 𝐌𝐚 𝜔 𝜔 i𝐁 𝜔 𝜔 𝐂 𝐊 𝐀 𝑥⃗ i𝜔𝜌 𝜙 𝜙 𝑁⃗ d𝑆 5
Where 𝐌 is the mass matrix, 𝐌𝐚 is the added mass matrix, 𝐁 is the radiational damping
matrix, 𝐂 is the hydrostatic stiffness matrix, 𝐊 𝐀 is the stiffness matrix, 𝜌 is the sea wa-
ter density, 𝜔 is the circular frequency, 𝜙 denotes the velocity potential of the incom-
ing wave, and 𝜙 denotes the velocity potential of the wave system diffracted by the
body.
On
ly
fo
rR
ea
din 3
g
2.2 Motion equation in time domain
at
W
Based on the Newton’s second law, the motion equation in the time domain can be
CF
obtained as [11],
𝒕
S
𝐌 𝐌𝐚 ∞ 𝐱 𝑡 𝐊 𝑡 𝜏 𝒙 𝜏 d𝜏 𝐂𝐱 𝑡 𝐅 𝑡 6
𝟎
20
Where 𝐊 is the retardation function, which can be expressed as,
2
24
𝐊 𝑡 𝐁 𝜔 cos 𝜔𝑡 𝑑𝜔 7
𝜋
And 𝐌𝐚 ∞ can be obtained by,
𝟏
𝐌𝐚 ∞ 𝐌𝐚 𝜔 𝑲 𝑡 sin 𝜔𝑡 d𝑡 8
𝝎 𝟎
2.3 Lumped mass method

The mooring line is simulated based on the Software Orcaflex, indeed, the lumped mass
method is employed to establish the line model. One of the main assumptions of this
method is that the line is divided into a finite number of elements constituting of a linear
spring with the mass of the element concentrated at each node, there are two nodes for
one element. The motion equation of the element can be expressed as,
𝐌 𝐒 𝐓 𝐅𝐃 𝐅𝐈 𝐅𝐖 𝐐 9
Where 𝐌 denotes the mass of the element, 𝐒 denotes the acceleration, 𝐓 is the axial
tension, 𝐅𝐃 and 𝐅𝐈 are the fluid induced drag and inertial force, 𝐅𝐖 denotes the weight
of the element in the fluid, 𝐐 is the shear force.
3 TLP wind turbine parameters and designed cases
3.1 OC4 and Tendon parameters
This paper analyzes the OC4 semisubmersible floating offshore wind turbine using Hy-
drostar for hydrodynamic coefficients, including added mass and radiation damping.
The setup features a 200m water depth, 24m diameter columns spaced 50m apart, and
a 20m draft. The platform weighs 1.3473 10 kg with its center of gravity at 13.46m.
Nine tendons are designed for mooring, and panel models are illustrated in Fig. 1.
(a) Panel model (b) TLP FOWT

Fig. 1. The panel model and TLP model for OC4 wind turbine.
On
ly
fo
rR
ea
4
din
g
3.2 Designed cases
at
W
Three designed wave loading cases are selected as the input condition for the transient
CF
analysis. Different wave significant height and peak periods are taken into account, and
one wave heading 45° is chosen, and 𝛾 equals 1, the detailed parameters are all listed
S
in Table. 1.
Table 1. Environmental conditions

20
Case
C1
𝐻 (m)
1.71
𝑇 (s)
5.84
𝛾
1
24
Heading(∘)
45
C2 3.61 8.5 1 45
C3 4.9 10.3 1 45
4 Results and discussions
4.1 Tensions of the tendon
This section presents and discusses the tension and motion results for the OC4 wind
turbine under both intact and transient conditions. Typically, transient analysis for a
single tendon breakage is essential during the design phase. Here, Tendon 7 is assumed
to break at 2700s, and the tension time series for the top of the porch over 500s is
analyzed. Fig. 2(a) shows the tension comparisons for Tendons 2, 5, 7, and 8 between
the intact case and the transient case with Tendon 7 broken. The wave significant height
𝐻 is 1.71m, and the peak period 𝑇 is 5.84s. The tension response varies significantly
when a tendon fails suddenly. Before the break, tensions for all tendons are nearly iden-
tical to the intact case. Tendon 8, adjacent to the broken Tendon 7, shows the largest
increase: the mean tension rises by 48.8% from 2478.24 kN to 3687.04 kN, and the
maximum tension increases by 55.6% from 2792.96 kN to 4346.36 kN. Tendon 2 ex-
periences a mean increase of 5.9% and a maximum increase of 6.6%. Conversely, Ten-
don 5 shows a mean decrease of 7.8% and a maximum decrease of 4.8%.
On
ly
fo
rR
ea
din 5
g
at
W
CF
S
20
24
(a) C1 (b) C2 (c) C3
Fig. 2. Tendon tensions
Fig. 2(b) shows the tension time history for tendons with a significant wave height 𝐻
of 3.61 m and peak period 𝑇 of 8.5 s. Before Tendon 7 breaks, the tension is similar to
the intact case. After Tendon 7 fails, tension changes significantly. The adjacent Ten-
don 8 sees a major increase, with mean tension rising by 48.7% and maximum tension
by 52.6%. The mean tension of Tendon 2 increases by 5.9%, and maximum tension by
8.6%. In contrast, the mean tension of Tendon 5 decreases by 7.8% and maximum ten-
sion by 2.8%. Fig. 2(c) illustrates the tension time history for four tendons with a sig-
nificant wave height of 4.90m and a peak period of 10.3s. After the rupture, Tendon 8,
adjacent to the broken tendon, shows the largest increase: mean tension rises by 48.7%
and maximum tension by 40.2%. The mean tension of Tendon 2 increases by 5.9%,
while its maximum decreases by 6.0%. Tendon 5 experiences a 7.9% drop in mean
tension and a 6.0% drop in maximum tension. This indicates the adjacent tendon un-
dergoes the most significant increase in tension (about 50%).
4.2 Motions of FOWT
Fig. 3 compares pitch responses in the time domain for intact and transient cases when
Tendon 7 broke suddenly at 2700s. The mean pitch angle increased over three times for
all three cases (3.56 for C1, 3.39 for C2, and 3.48 for C3) compared to the intact state.
Maximum pitch errors are 67.8% for C1, 37.0% for C2, and 18.3% for C3. Sudden
tendon failure leads to significant changes in platform rotation, emphasizing the need
for careful monitoring of motion and tension based on operational criteria.
a. C1 b. C2 c. C3
Fig. 3. Pitch motion
On
ly
fo
rR
ea
6
din
g
5 Conclusions
at
W
A TLP supporting the OC4 floating offshore wind turbine features nine tendons
CF
grouped at three corners. A panel model was created, and hydrodynamic analysis was
S
performed using Hydrostar to obtain added mass coefficients, radiation damping coef-
ficients, and second mean drift load for the transient time-domain analysis. Tendon
20
modeling was done with Orcaflex, using the lumped mass method for line structures.
24
The study focused on the transient effects of a sudden break in Tendon 7, using three
wave parameter sets from a 1-year return period. Key findings include, the adjacent
tendon experiences a significant tension increase (40% to 50%) after one tendon fails,
and the mean pitch motion triples, with maximum pitch changes ranging from 18% to
67%, depending on environmental conditions and sea loads. These results underscore
the need for further studies to build a comprehensive database on dynamic properties,
including effects of multiple tendon failures and contributions from different loading
frequencies.
6 References
[1] Global Wind Energy Council. Global offshore wind report 2023. GWEC: Brussels,
Belgium. 2023;19:10-2.
[2] Yang CK, Kim MH. Transient effects of tendon disconnection of a TLP by hull–tendon–
riser coupled dynamic analysis. Ocean Engineering. 2010;37:667-77.
[3] Wu HY, Zhao YS, He YP, Shao YL, Mao WG, Han ZL, et al. Transient response of a TLP-
type floating offshore wind turbine under tendon failure conditions. Ocean Engineering.
2021;220:108486.
[4] Yang Y, Bashir M, Michailides C, Mei X, Wang J, Li C. Coupled analysis of a 10 MW
multi-body floating offshore wind turbine subjected to tendon failures. Renewable Energy.
2021;176:89-105.
[5] Tabeshpour MR, Ahmadi A, Malayjerdi E. Investigation of TLP behavior under tendon
damage. Ocean Engineering. 2018;156:580-95.
[6] Ren YJ, Venugopal V, Shi W. Dynamic analysis of a multi-column TLP floating offshore
wind turbine with tendon failure scenarios. Ocean Engineering. 2022;245:110472.
[7] Ren Y, Shi W, Venugopal V, Zhang L, Li X. Experimental study of tendon failure analysis
for a TLP floating offshore wind turbine. Applied Energy. 2024;358:122633.
[8] Li Y, Zhu Q, Liu L, Tang Y. Transient response of a SPAR-type floating offshore wind
turbine with fractured mooring lines. Renewable Energy. 2018;122:576-88.
[9] Bae Y, Kim M, Kim H. Performance changes of a floating offshore wind turbine with
broken mooring line. Renewable Energy. 2017;101:364-75.
[10] Molin B. Offshore structure hydrodynamics: Cambridge University Press; 2023.
[11] Cummins W. The impulse response function and ship motion. Report 1661, Department of
the Navy, David W Taylor Model Basin, Hydromechanics Laboratory, Research and
Development Report, October 1962. 1962.
On
ly
fo
rR
eaof Pontoon-type Floating Bridges Subjected to
Wavedand
Dynamic Response Analysis
ingTraffic Loads
at
W
Dongqi Jiang1[0000-0002-8548-570X], Bo Wu2, Bin Peng1 and Yuhang Shao1
1
CF
Nanjing University of Science and Technology, Nanjing 210094, China
S
2
China Ship Scientific Research Center, Wuxi 214082, China
20
jiangdongqi@njust.edu.cn
24
Abstract. Floating bridges have attracted great interests in coastal countries for sea-crossing transportation
infrastructures, such as Norway, Japan, Singapore, etc. Wave and traffic loads are two critical actions to be taken
into consideration in the dynamic analysis of floating bridges, which involves both fluid-structure and vehicle-
bridge interactions. This paper presents a fully coupled computational fluid dynamics (CFD) - finite element
method (FEM) approach to predict dynamic responses of pontoon-type floating bridges when subjected to
combined wave and traffic loads. The sophisticated numerical model is developed using ABAQUS and Star-CCM+
co-simulation software platform, where a two-way data communication scheme is applied between CFD and FEM
solvers. CFD simulation exports the fluid pressure for the determination of structural responses in the FEM solver,
while pontoon motions are fed back to deform the mesh in the CFD solver. Moving vehicles are idealized as mass-
spring-damper mechanical systems and the floating bridge are modeled with beam and plate elements. A simple
traffic model composing of different numbers of trucks is adopted to assess the vehicle-induced dynamic response
of the floating bridge. Research outcome indicates that the traffic load causes downward effects on the dynamic
response of pontoon-type floating bridges. The traffic-induced dynamic impact effect varies with different spans of
the floating bridge. Dynamic responses caused by the simple multi-truck load are relatively small as compared with
wave action effects, and the random traffic load model is suggested to be considered in the engineering practice.
Keywords: Floating Bridge, Moving Load, Vehicle-Bridge Interaction, Dynamic Responses, Co-Simulation.
1 Introduction
Ever-increasing population and economic activities in coastal areas stimulate greater interests and demands for
essential transportation infrastructures to link main cities and islands [1-2]. During the past decade, several world-
famous sea-crossing bridges have been built in China, such as Quanzhou Bay Bridge (2015), Hong Kong-Zhuhai-
Macao Bridge (2018) and Lingdingyang Bridge (2023), etc. In the construction site with deep water and soft seabed,
the floating bridge is considered to be a preferable engineering solution as compared to conventional bottom-founded
bridge structures because it utilizes the natural buoyancy force to balance the vertical loads [3]. Under such
circumstance, the floating solution is able to reduce the construction difficulties and costs significantly by omitting
the need for building huge-size foundations [4-5]. Moreover, floating bridges constructed with modular units can be
relocated, expanded and removed whenever necessary, which can further decrease the costs of maintenance and
reconstruction [6-7]. Ancient floating bridges date back to 480 BC for military purposes, and several modern floating
bridges were built in clam waters in the past century, such as Evergreen Point Bridge, Bergsoysund Bridge, Yumemai
Bridge, etc., while no permanent floating bridges have been constructed in severe sea conditions until now [8].
1
On
ly
fo
rR
ea
din
In-service floating bridges are subjected to various live load actions, including wave load, wind load, traffic load,
etc. Several numerical analysis approaches have been developed to study the dynamic response of discrete pontoon
g
floating bridges. Løken et al. analyzed a scaled model of Bergsoysund bridge under wave actions using the potential
at
flow theory [9], while Petersen and Øiseth proposed a sensitivity-based finite element method (FEM) to analyze the
W
structural response of Bergsoysund bridge [10]. Xu et al. developed a time-domain method to study the dynamic
CF
behavior of a suspension bridge with two floating pylons [11]. Computational fluid dynamics (CFD) methodology
S
provides another time-domain analysis option to simulate flow processes in general and wave actions in particular.
20
The moving load is one of critical live loads in the design of bridge superstructures, and the vehicle-induced dynamic
24
effect has attracted great research interests. The vehicle-bridge interaction system consists of two sub-systems: the
vehicle and the bridge, which can be idealized with mass-spring-damper units and FEM models [12]. It is quite
challenging to properly analyze the discrete pontoon floating bridge when subjected to the combined wave actions
and moving vehicle loads because the whole system involves both fluid-structure and vehicle-bridge interaction
problems. The effects of aforementioned two live loads on dynamic responses of floating bridge superstructures need
further thorough investigations.
This paper presents a co-simulation method using CFD solver Star-CCM+ and FEM solver ABAQUS to predict
the dynamic response of discrete pontoon floating bridges when subjected to combined wave and moving vehicle
loads. A concrete pontoon-type floating bridge case is taken as an example to illustrate the numerical model
development of the vehicle-bridge-wave system. The traffic-induced dynamic amplification effect is further evaluated
through parametric studies, which provides the guidance for a reliable and economical design of floating bridges.
2 Numerical Methodology
A two-way data communication scheme is applied between CFD solver Star-CCM+ and FEM solver ABAQUS to
develop numerical models of discrete pontoon floating bridges. Fig. 1 depicts CFD-FEM co-simulation flow chart for
the analysis work. The external fluid pressure exported from the CFD simulation is used to derive the structural
response in the FEM solver, and structural deformations are fed back into the CFD solver to deform the mesh.
Fig. 2 (a) – (c) show the bridge and vehicle models developed in the ABAQUS software. The bridge superstructure
is modeled with S4R shell elements or B31 generalized beam elements based on different simplification considerations.
The pontoon is modeled with S4R shell elements. The mass and rotational inertia values are assigned to the reference
point attached at the centroid location of the discrete pontoon. Multi-point constraints (MPC) method is used to
connect the bridge superstructure and discrete pontoons. The floating bridge case considered herein owns a uniform
span length of 50 m and an overall length of 300 m with the prestressed concrete twin-box single cross section, whose
boundary conditions are fixed at both ends. Table 1 lists key parameters in the modelling of the beam and pontoon.
The vehicle is modelled as a 15-DOFs mass-spring-damper mechanical system, representing 55 t-weight five-axle
vehicle specified in the Chinese standard "General Specification for Design of Highway Bridges and Culverts"
(JTGD60-2015). The vehicle components are modelled as rigid bodies, suspensions and wheels via spring-dashpot
units, whose parameters have been defined and calibrated by Deng et al. [12]. The five-axle vehicle is modeled with
RB3D2 line elements in the ABAQUS software. The wheel reference points are set to be "hard" contact with the
bridge deck so that the separation is allowed and the jumping effect can be considered.
2
On
ly
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 1. CFD-FEM co-simulation flow chart.
Table 1. Parameters of beams and pontoons.
Components Parameter Unit Value

(a) Vehicle-bridge interaction (refined v.s. simplified)
Mass, m t/m 36.23
Area, S m2 14.68
Beam Moment of Inertia, Ixx m4 72.3
(b) Refined bridge model in ABAQUS Moment of Inertia, Iyy m4 425
Moment of Inertia, Izz m4 23.1
Major axis, a m 50
(c) Simplified bridge model in ABAQUS
Minor axis, b m 20
Depth, h m 9
Draft, d m 5.75
(d) Pontoon model in Star-CCM+ Pontoon Mass, M t 2350
Centroid of Gravity, COG m 4
Roll inertia, Itr t·m 5.02×105
2
Pitch inertia, Itp t·m2 1.105×105
(e) Mesh of fluid domain in Star-CCM+ Yaw inertia, Ity t·m2 5.48×105
Fig. 2. Moving vehicle-floating bridge system model.
The modeling process presented in this paper utilizes the fifth-order Stokes wave model in the Star-CCM+ software
with considering the computational accuracy and efficiency. Appropriate geometric dimensions of the numerical fluid
domain and mesh sizes close to the free surface are determined by Equations (1) - (4) as suggested in the reference
[13]. Fig. 2 (d) – (e) show the pontoon model in Star-CCM+ and the mesh arrangement of the fluid domain. Geometric
sizes of the fluid domain are 600 m (length) × 300 m (width) × 200 m (height) and the basic mesh size is 12 m. For
the range of 1 m in height close to the free surface, horizontal and vertical mesh sizes are 2.64 m and 0.1 m, respectively.
L  2  l (1)
D  2  d (2)
40   m  100 (3)
h n  20 (4)
3
On
ly
fo
rR
ea
din
where L and D are the length and width of the fluid domain.  is the wavelength, l and d are the length and width
of the pontoon, m and n are mesh sizes in the horizontal and vertical directions, h is the wave height.
g
3 Results and Discussion at
W
3.1 Comparison of Different Modeling Techniques
CF
S
The use of shell elements in bridge superstructures increases the computation cost, especially for large-scale bridge
20
models, which is not practical for the numerical parametric study. Therefore, the simplified model (Fig. 2 (c))
24
developed with generalized beam section elements is proposed and compared with the refined analysis model (Fig. 2
(b)). Note that equivalent beam section properties representing the original bridge section configuration are determined
by carefully comparing static responses of both models, such as flexural rigidity EIy and EIz, torsional rigidity GJ, etc.
Analysis results indicate that mode shapes from the first mode to tenth mode are the same for both refined and
simplified analysis models and the corresponding natural frequency values are very close to each other. Fig. 3 depicts
the vertical and lateral displacement time-history responses of interior pontoons in floating bridges under only wave
actions, which is obtained by both refined and simplified analysis models. It can be seen that the displacement
magnitude and time-history curve trends are almost identical for two modeling techniques. Therefore, the simplified
analysis approach with user-defined beam section is reliable and it will be used to perform the numerical parametric
study on discrete pontoon floating bridges when subjected to combined wave actions and moving vehicle loads.
(a) Vertical displacement (b) Lateral displacement
Fig. 3. Time-history responses of interior pontoons using two different modeling approaches.
3.2 Vehicle-induced dynamic amplification effect
In order to explore the dynamic effect of moving vehicles on the response of discrete pontoon floating bridges, two-
truck and six-truck loaded cases are analyzed without considering the road surface roughness. Note that the six trucks
are arranged in three rows with the spacing of 7.2 m between two adjacent vehicles in the longitudinal direction. The
vehicles are symmetrically arranged across the transverse direction and run along the path at a speed of 10 m/s. Fig. 4
and Fig. 5 show vertical displacement time history curves at each span of the discrete pontoon floating bridge when
subjected to two and six moving vehicles. The regular wave action with period of 8.5 s is also considered in the
analysis. The magnitude of static displacements is labelled in the right vertical axis. It is seen that the vehicle load
generally causes downward action effects as compared with the curves without vehicle loads. The downward effect in
the six-truck load case is more obvious due to a heavier load action, but the curve trend does not change significantly.
As compared with the wave action, dynamic responses caused by two or six trucks are quite small. It is essential to
consider the random traffic load in the future analysis to represent the real live load situation.
2
On
ly
fo
rR
ea
din
g
at
W
CF
(a) Two-truck load case (b) First span
S (c) Second span
20
24
(d) Third span (e) Fourth span (f) Fifth span
Fig. 4. Vertical displacement responses at each span of the floating bridge subjected to two moving vehicles and the wave action.
(a) Six-truck load case (b) First span (c) Second span
(d) Third span (e) Fourth span (f) Fifth span
Fig. 5. Vertical displacement responses at each span of the floating bridge subjected to six moving vehicles and the wave action.
4 Conclusion
A CFD-FEM co-simulation method is developed to analyze dynamic responses of discrete pontoon floating bridges
subjected to combined wave and moving vehicle loads. By comparing with the refined shell element model, the
simplified analysis model using beam elements is capable of assessing the dynamic response of floating bridges.
Analysis results indicate that the traffic load causes downward effects on the dynamic response of pontoon-type
3
On
ly
fo
rR
ea
din
floating bridges. Dynamic responses caused by the simple multi-truck load are relatively small as compared with wave
action effects, and the random traffic load model is suggested to be considered in the future analysis to represent the
real live load situation.
g
at
Acknowledgement
W
CF
S
This work is supported by Fundamental Research Funds for the Central Universities (Grant No. 30922010907),
20
National Natural Science Foundation of China (Grant No. 52478180, 51808292), State Key Laboratory of Ocean
Engineering (Shanghai Jiao Tong University) (Grant No. GKZD010089) and Natural Science Foundation of Jiangsu
Province (Grant No. BK20180487).
24
References
1. Luo, C., Jiang, D. Hydroelastic responses of very large floating structures in damage conditions. In: Proceedings of the Third
World Conference on Floating Solutions. Tokyo, Japan. (2023).
2. Jiang, D., Tan, K.H., Wang, C.M., Ong, K.C.G., Bra, H., Jin J., Kim, M. “Analysis and design of floating prestressed concrete
structures in shallow waters,” Marine Structures, 59, 301-320. (2018).
3. Wan, L., Dai, J., Jiang, D., Ang, K.K. “Parametric study and dynamic response analysis of three single curved discrete pontoon
floating bridges,” Ships and Offshore Structures, 17(8): 1788-1801. (2022).
4. Jiang, D., Tan, K.H., Dai, J., Ong, K.C.G., Heng, S. “Structural performance evaluation of innovative prestressed concrete
floating fuel storage tanks,” Structural Concrete, 20 (1), 15-31. (2019).
5. Jiang, D., Tan, K.H., Ong, K.C.G., Heng, S. “Behavior of prestressed concrete floating self-stabilizing fuel storage tanks," In:
Proceedings of 4th Congrès International de Géotechnique Ouvrages Structures Conference, Ho Chi Minh City. (2017).
6. Jiang, D., Tan, K.H., Dai, J., Ang, K.K., Nguyen, H. “Behavior of concrete modular multi-purpose floating structures,” Ocean
Engineering, 229: 108971. (2021).
7. Jiang, D., Tan, K.H., Wang, C.M., Dai, J. “Research and development in connector systems for very large floating structures,”
Ocean Engineering, 232: 109150. (2021).
8. Wan, L., Jiang, D., Dai, J. “Numerical modelling and dynamic response analysis of curved floating bridges with a small rise-
span ratio,” Journal of Marine Science and Engineering, 8, 467. (2020).
9. Løken, A., Oftedal, R., Aarsnes, J. "Aspects of hydrodynamic loading and responses in design of floating bridges." In:
Presented at Second Symposium on Strait Crossings, Trondheim, Norway. (1990).
10. Petersen, Ø. W., Øiseth, O. "Sensitivity-based finite element model updating of a pontoon bridge." Engineering Structures,
150, 573-584. (2017).
11. Xu, Y., Øiseth, O., Moan, T. "Time domain modelling of frequency dependent wind and wave forces on a three-span
suspension bridge with two floating pylons using state space models." Presented at ASME 36th international conference on
ocean, offshore and arctic engineering. Trondheim, Norway. (2017).
12. Deng, L., Duan, L., He, W., Ji, W. "Study on vehicle model for vehicle-bridge coupling vibration of highway bridges in
China." China Journal of Highway and Transport, 31(7), 92-100. (2018).
13. Sun, Z., Liu, G., Zou, L., Zheng H., Djidjeli, K. Investigation of nonlinear ship hydroelasticity by CFD-FEM coupling method.
Journal of Marine Science and Engineering, 9(5): 511. (2021).
4
On
ly
fo
rR
Development and Verification of a Simulation Tool for Floating Offshore
ea
Wind Turbines Using MBDyn
din
g
Tomoaki Utsunomiya1[0000-0003-2081-7547], Ryoya Hisamatsu1[0000-0002-8491-789X], Takayuki Morifuji2, Kodai
at
Uematsu2, Riku Wakasa2, Hisashi Okubo3, Hayami Sato3 and Takuya Kamijo3
W
1 Department of Marine Systems Engineering, Kyushu University, Fukuoka, Japan
CF
2 Department of Naval Architecture and Ocean Engineering, Kyushu University, Fukuoka, Japan
3 Engineering Department-Offshore Wind, Eurus Energy Holdings Corporation, Tokyo, Japan
S
utsunomiya@nams.kyushu-u.ac.jp
20
24
Abstract. For design of Floating Offshore Wind Turbines (FOWTs), time domain simulation of the FOWT
model for many Design Load Cases (DLCs) is essential. For this purpose, several commercial codes for aero-
hydro-elastic-mooring dynamics coupled analysis are under development. In this paper, we attempt to develop
such a code using opensource multibody dynamics solver MBDyn. Particular attention is made to use an elastic
beam element (beam3) in MBDyn, instead of using linear springs to connect rigid bodies as before. Then,
verification of the simulation code and models using MBDyn has been made by comparing with OrcaFlex.
The Volturn semi-submersible model is the model used here for comparisons. This tool would be valuable for
checking and verifying their FOWT models to be used with commercial codes.
Keywords: Floating Offshore Wind Turbine, FOWT, Simulation, MBDyn.
1 Introduction
In order to make floating offshore wind power generation practical and the main power source, there has been a
noticeable trend in recent years toward larger wind turbines to reduce power generation costs. In 2020, IEA Wind
TCP Task 37 released a 15 MW Reference Wind Turbine [1], and the community is expected to develop technol-
ogy for large wind turbines.
The design of floating offshore wind power generation facilities considers complex natural environmental con-
ditions such as wind, waves, and currents, as well as structures such as floats, moorings, nacelles, and blades under
these conditions. In the load analysis, coupled analysis is required after modeling fluid forces and structural ele-
ments under many assumptions in order to handle a large number of design load cases. For this reason, the relia-
bility of modeling and analysis tools is extremely important. In response to this, Utsunomiya et al. [2] considered
that numerical analysis using multibody dynamics (MBD) would be useful to consider the finite amplitude motion
of floats and structures, and developed a design tool using a general-purpose analysis software MSC Adams. The
validity of this tool has been demonstrated for a 2 MW spar-type offshore wind turbine in actual sea areas [2, 3].
On the other hand, one of the practical challenges in dealing with a large number of design load cases is the
limitation imposed by the number of licenses for existing commercial codes for simultaneous parallel analysis.
For this reason, the authors' research group has developed a design tool for offshore wind turbines using the open
source general-purpose MBD analysis software MBDyn [4]. It has been confirmed that this tool is capable of
performing analysis equivalent to MSC Adams for spar-type offshore wind turbines [5].
For large wind turbines above 15 MW scale, the applicable water depth becomes deeper and installation of
wind turbine onto the spar-type floaters becomes more difficult, so it is considered that semi-submersible floaters
are more practicable. Therefore, the purpose of this study is to establish a design tool for semi-submersible off-
shore wind turbines using MBDyn. In this paper, we report on the modeling of the aforementioned IEA 15 MW
offshore wind turbine and the effectiveness of the analysis by comparing with the commercial code OrcaFlex [6].
2 Target Model
The analysis target of this study is the IEA 15 MW Offshore Reference Wind Turbine [1] and its floating platform
VolturnUS-S [7], which are open source and available for the purpose of promoting technological development
in the community. Fig. 1 shows an outline of the design.
On
ly
2
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 1. General arrangement of the VolturnUS-S.
Table 1. Main dimensions of the semisubmersible platform.
Member Item Value

Center Column Height 35.0 m
Outer dimeter 10.0 m
Inner diameter 9.889 m
Lower Hull Length 40.5 m
Outer diameter 10.869 m
Water ballast 93.0 t/m
Outer Column Height 35.0 m
Outer diameter 12.5 m
Fixed ballast 847 t
VolturnUS-S is a semi-submersible floating body consisting of four cylindrical columns and three rectangular
cross-section lower hulls. When the effects of diffraction/radiation due to short-period wave conditions cannot be
ignored, it is necessary to evaluate the hydrodynamic forces by treating the floating body as a rigid body using
potential theory. On the other hand, in wave conditions with long periods and large floating body motion, which
are the range of application of the Morison’s formula, analysis using a three-dimensional elastic frame model that
can consider nonlinearities such as finite amplitude motion and drag, and can directly obtain the cross-sectional
forces of the members, is more faithful and efficient. In this paper, we first construct a case that can be considered
as a three-dimensional frame model using Morison’s formula for wave loading.
The lower hull of the float is a horizontal member with a rectangular cross section, and the viscous effects of
vortex shedding from the bilge are particularly significant. In addition, careful discussion of the effects on the
angle of attack is thought to be necessary. In this paper, the lower hull was first changed from a rectangular
member to a cylindrical member with equivalent mass and cross-sectional area, and an analysis was performed.
The specifications of each member of the semi-submersible float are shown in Table 1. The density of the struc-
tural material of each member is 8.5 t/m3, and the Young's modulus is 210 GPa. In addition, the hydrodynamic
On
ly
3
fo
rR
coefficients are assumed to be the drag coefficient Cd of 0.6 and the added mass coefficient Ca of 1.0 for each
member. In addition, the drag coefficient Cdv = 2.0 for the bottom of the column member, and the added mass
ea
coefficient Cav = 1.0 based on a hemisphere.
din
The mooring system is a multi-point catenary mooring system with mooring chains attached to each outer
column. The water depth is 200 m, and the fairlead is located 6 m from the bottom of the column. The mooring
g
chains are R3 studless chains with a nominal diameter of 185 mm and a total length of 850 m each.
at
W
3 Analysis Method
3.1 About MBDyn

CF
S
In this study, an analysis model was constructed using MBDyn Ver. 1.7.3. General-purpose MBD analysis soft-
20
ware such as MBDyn is a software that mechanically derives equations of motion for any coupled system by
24
inputting information on interactions between rigid bodies (kinematic constraints, elastic forces, etc.) and external
forces, and calculates the motion and internal forces by solving them numerically.
MBDyn is an open source software, but has a high degree of completeness and capability of extension. In our
previous research, it has been confirmed that analysis of spar-type floating bodies is practically possible by adding
buoyancy, fluid forces, and mooring reaction forces as extension modules [5]. In addition, MBDyn has an aeroe-
lastic module implemented because it was developed for rotorcraft in the field of aeronautics, and it has been
suggested that practical analysis is also possible for wind turbines. In addition, it has been confirmed that the
aerodynamic calculation module AeroDyn, which is released by the National Renewable Energy Laboratory
(NREL), can also be introduced as an extension module.
In our previous study, the elastic bodies of the floating body, tower, and blades were modeled as multiple rigid
bodies and springs. However, it has been reported that the analytical results of this model differed from those of
analytical models whose validity has been confirmed [5]. Therefore, in this study, we have tried to use the three-
node beam (beam3) element implemented in MBDyn to analyze the elastic bodies.
Aerodynamic Element. An aerodynamic load calculation element implemented in MBDyn. Aerodynamic loads
are evaluated using blade element theory. On the other hand, when dealing with a rotor, the effect of induced
velocity must be considered using momentum theory. In this module, the induced velocity is calculated by con-
verging the following equation at each time step.
𝑇 2𝜌𝐴𝑢 𝑢 𝑣 𝑤 0 (1)
Here, u is the induced velocity, v is the opposing component of the relative velocity of the rotor plane, and w is
the parallel component. T is the rotor thrust evaluated by the blade element theory considering the induced veloc-
ity, ρ is the air density, and A is the area of the rotor. T in equation (1) refers to the aerodynamic component of the
axial force on the rotor shaft. In other words, it should be noted that equation (1) calculates the induced velocity
under the assumption that the inflow velocity to the rotor is uniform and steady.
User Defined Element. An extension element that users can add to MBDyn. In this study, we introduce a fluid
force element (Spardyn Element) and a mooring reaction force element (Catenary Element) for slender members.
The effect of hydrostatic pressure is applied to the submerged nodes as a nodal load based on the hydrostatic
pressure distribution, not as a vertical buoyancy force. The effect of dynamic hydrodynamic pressure is evaluated
using the Morison’s formula, which considers the relative velocity between the node and the water particles. The
mooring reaction force is evaluated quasi-statically by catenary theory that does not consider elongation and dy-
namic effect, based on the positional relationship between any nodes and a spatially fixed anchor point.
3.2 Floating Body Model
Fig. 2 shows an overview of the modeling of the floating body. The floating body was constructed as a 3D elastic
frame model using beam3 element. The number of elements for each member is five (e.g. 11 nodes) for the lower
hull and four (e.g. 9 nodes) for the column. However, as shown in Fig. 2, the connection part of each column
member with the lower hull is modeled as a rigid body, and the relative displacement of the connection point is
restrained. The fluid force is directly applied to each node by the Spardyn Element. For each end face of the
member, the end face of the lower hull is actually the internal face connected to the column, but a load due to
hydrostatic pressure is applied to ensure consistency with the hydrostatic pressure distribution on the column side.
On
ly
4
fo
rR
On the other hand, the effect of dynamic water pressure is not applied as a fluid force here. This point may need
to be discussed in future when considering actual phenomena. The mooring reaction force is set so that the fair-
ea
lead's position in the global coordinate system can be referenced at each time step by constraining the node rep-
din
resenting the fairlead to the nearby node.
g
at
W
CF
S
20
24
Fig. 2. Model configuration of the platform.
3.3 Wind Turbine Model
The tower is modeled using beam3 element. However, the aerodynamic load acting on the tower is not considered
here. The wind turbine to be analyzed has a direct drive mechanism. Here, the generator is not modeled in detail,
and only the rigid body and constraints are modeled. The drive-train system is represented by the nacelle as a rigid
body, and the rotor shaft is represented by constraining the axial position relative to the nacelle. The blades are
modeled with 16 elements using beam3 element and aerodynamic element. There are four integration points for
the aerodynamic load for each element, and the aerodynamic coefficients and chord length at the integration points
are evaluated by interpolation. The inertia matrix, aerodynamic coefficients, stiffness matrix, Rayleigh damping
coefficients, etc. of the beam required for modeling were input to MBDyn by interpolating and coordinate con-
version from the publicly available OpenFAST input file [8].
3.4 Control System

In this paper, we first focus on confirming the validity of MBDyn for each structural element and load evaluation
module, and only introduce a simple control system. In the wind turbine analyzed, in the reference wind speed
range of 6.98-10.59 m/s, control is performed by a torque controller with the optimal value of Tip Speed Ratio
(TSR) as the target value when the blade pitch angle is 0 degree. In the constructed model, a simple PI controller
for this range was implemented using only the MBDyn input file.
3.5 Comparison Model Using OrcaFlex

OrcaFlex is a commercial software for dynamic analysis of marine structures, and has been especially trusted by
the offshore petroleum industry. In recent years, wind turbine functions have been developed, and analysis of
floating wind turbines has become practical. In this study, we verify the effectiveness of MBDyn by comparing
the analysis results for an equivalent design object between MBDyn and OrcaFlex. An example modeling of the
IEA 15 MW wind turbine, which is the subject of our research, has been published by OrcaFlex. However, since
the floating body is a rigid body and the fluid forces are calculated using potential theory, we modeled it as an
elastic frame structure by combining line elements, as in the case of MBDyn.
4 Analysis Results
4.1 Regular Waves

Regular waves with the wave height of H = 4 m and the period T = 10 s are applied. The results are shown in Fig.
3 for platform motions, sectional forces at the tower base, and the mooring reaction forces. Here, the wind turbine
is modeled only as a point mass. As we can see in Fig. 3, good agreement of the results between MBDyn and
OrcaFlex can be observed.
On
ly
5
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 3. Numerical results for VolturnUS-S model under regular waves of H = 4 m and T = 10 s.
4.2 Regular Waves and Constant Wind
Regular waves with the wave height of H = 1.4 m and the period T = 10 s together with a constant wind speed of
7 m/s are applied. The results are shown in Fig. 4 for platform motions, sectional forces at the tower base, the
rotor speed, the generator torque and the generator power. Here, the wind turbine is modeled as a whole model
using the IEA 15 MW wind turbine specifications. As we can see in Fig. 4, basically a good agreement of the
results between MBDyn and OrcaFlex can be observed. However, some discrepancy is also seen, probably be-
cause of the incomplete aerodynamic modelling of the wind turbine in MBDyn when using aerodynamic element.
This point will be further considered in future research.
On
ly
6
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 4. Numerical results for VolturnUS-S model under regular waves of H = 1.4 m and T = 10 s with a constant wind speed
of 7 m/s.
References
1. Gaertner, E., et al.: Definition of the IEA 15-megawatt offshore reference wind turbine. National Renewable Energy
Laboratory. NREL/TP-5000-75698 (2020).
2. Utsunomiya, T., Sato, I., Kobayashi, O., Shiraishi, T., Harada, T.: Numerical modeling and analysis of a hybrid-spar
floating wind turbine. Journal of Offshore Mechanics and Arctic Engineering, ASME 141(3), 031903 (2019).
3. Tanaka, K., Sato, I., Utsunomiya, T., Kakuya, H.: Validation of dynamic response of a 2-MW hybrid-spar floating wind
turbine during typhoon using full-scale field data. Ocean Engineering 218, 108262 (2020).
4. MBDyn Homepage, https://www.mbdyn.org/, last accessed 2024/06/30.
5. Utsunomiya, T., Okubo, H., Yamada, H.: Development of a simulation tool for floating offshore wind turbines using
MBDyn. In: Proceedings of the Third World Conference on Floating Solutions, pp. 737-750. Springer, Singapore (2024).
6. Orcica: OrcaFlex, https://www.orcina.com/orcaflex/, last accessed 2024/06/30.
7. Allen, C., Viselli, A., Dagher, H., Goupee, A., Gaertner, E., Abbas, N., Hall, M., Barter, G.: Definition of the UMaine
VolturnUS-S reference platform developed for the IEA wind 15-megawatt offshore reference wind turbine. National Re-
newable Energy Laboratory. NREL/TP-5000-76773 (2000).
8. OpenFAST Homepage, https://openfast.readthedocs.io/en/main/index.html, last accessed 2024/06/30.
On
ly
fo
rR
Wave Attenuation Effect of Various Breakwater for Tropical Climate
ea
din
Zhi Yung Tay1[0000-0003-3194-1965] and Nyan Lin Htoo1[0009-0008-2986-7691]
g
1 Engineering Cluster, Singapore Institute of Technology, 138683, Singapore
at
zhiyung.tay@singaporetech.edu.sg
W
CF
Abstract. This study investigates the vital role of floating breakwaters in mitigating wave energy in interme-
S
diate water environments, crucial for shoreline protection and ecosystem preservation. Various floating break-
water types are analyzed for their effectiveness in attenuating the wave force. Through numerical simulations
20
of computational fluid dynamics in ANSYS Fluent, different breakwater cross sections are evaluated in time-
series domain for wave approaching at varying wave periods. The comparative analysis encompasses a range
24
of breakwater designs, including box-type, stepped-box type, pontoon-type, box-with-plates type, dual-box
type and T-shaped breakwater. The transmission of waves when passing through these breakwaters are studied
to assess breakwater efficiency in mitigating the wave forces. The incident and transmitted wave elevations in
each breakwater type observed at a selected time among the upstream and downstream locations are tracked
and compared. The most efficient breakwater type for wave attenuation is then chosen to demonstrate its im-
pact over time. The study aims to inform decision-makers and coastal engineers on selecting optimal floating
breakwater solutions suitable for tropical climate. This research enhances our understanding of wave attenua-
tion, aiding coastal protection and sustainable development while also exploring its impacts on ecosystem
dynamics.
Keywords: floating breakwater, wave attenuation, intermediate water, computational fluid dynamics, tropi-
cal climate
1 Problem Statement
A floating breakwater (FB) is a type of structure used to protect coastal areas, harbors, and marinas from the
impact of waves. Unlike traditional fixed breakwaters, FBs are buoyant and anchored to the seabed, allowing them
to absorb and dissipate wave energy while floating on the water's surface. This design is particularly advantageous
in locations where deep water or soft seabeds make the construction of traditional breakwaters impractical. FBs
are used to enhance coastal protection, improve ecological resilience, and support sustainable development prac-
tices by mitigating wave energy and reducing shoreline erosion.
Numerous research on FB types have been assessed both numerically and experimentally. Lokesha et al.
(2015) conducted experimental research on the stepped-box breakwater design. Huang et al. (2014) and He et al.
(2023) explored the box-with-plates breakwater type. Wang et al. (2024) focused on the dual breakwater design.
The studies by Sun et al. (2023) included the state-of-the-art reviews of wave attenuation capabilities of different
breakwater designs including pontoon and box types. Building upon the relevant prior separate research studies
on FBs, this paper investigates six types of FB section types altogether, such as box-type, stepped-box type, pon-
toon-type, box-with-plates type, dual-box type, and T-shaped FB, to analyze and compare their wave attenuation
effects in the mild seastate of the tropical climate. All FBs share the same two-dimensional footprint boundary,
with width and height combining draft and freeboard. The characteristic width of the FB along the wave propa-
gation direction is 5 meters, the draft is 3 meters, and the freeboard is 2 meters. Detailed geometries of the FBs
are shown in Table 1 and wave properties considered are given in Table 2.
The water depth 𝐷 is assumed to be 10 meters, with incident wave periods 𝑇 of 4s, 5s and 6s. These wave
periods are typical of the relatively benign waters of the Singapore sea which also involves primarily higher fre-
quency waves shorter than 4 seconds. Since the FB is fully effective against high-frequency waves, 𝑇 4s is
chosen as the lower bound and 𝑇 5s and 6s are considered to study the varying FB' effects on the wave attenu-
ation. The incident wave height 𝐻 was set at 2 meters, i.e., 𝐻 2𝐴, where 𝐴 is the wave amplitude. The water
depth category and wavelengths were determined based on the frequency dispersion relation (Phillips, 1977).
According to the wave dispersion equation given in Eqs. (1) to (3) and the corresponding depth-to-wavelength
ratio 𝐷/𝜆 to categorize the wave characteristic, the 10-meter water depth falls into the intermediate water depth
category. The corresponding wavelengths for wave periods of 4 seconds, 5 seconds, and 6 seconds are 24.68m,
36.59m, and 48.41m, respectively.
On
2
ly
fo
rR tanh (for intermediate water depth) (1)
ea 𝜆
din
(for deep water) (2)
g
𝜆
at
𝑇 𝑔ℎ (for shallow water) (3)
W
Higher frequency waves in relatively low water depth and with high wave amplitudes can induce wave break-
CF
ing, especially when analyzed using linear Airy wave theory. To ensure wave breaking and stability limit, man-
datory checks are performed, including relative wave height, wave steepness, Ursell number, and wave regime
S
checks. The 3rd order Stokes wave theory of regular periodic waves is employed, upon consistently satisfying the
20
established checks across wave periods of 4s, 5s and 6s.
A typical numerical fluid domain incorporating a FB is established using the Reynolds-Averaged Navier-
24
Stokes (RANS) equations. The simulations are conducted with the computational fluid dynamics (CFD) software,
ANSYS® Fluent (2023 R2). A two-dimensional numerical wave tank is considered with a clearance of 150 meters
set on both upstream and downstream of the FB. This setup also meets the flow regime requirements and allows
for sufficient wave propagation through the domain while being computationally feasible for multiple runs. The
study analyzes six FB types given in Table 1 for three wave periods.
Table 1. Various floating breakwater sections
Box-type Pontoon type Stepped box-type
Dual box-type Box with plates-type T-shaped-type
Table 2. Wave properties
𝑇 (s) - wave periods 4 5 6

𝐷 (m) - water depth 10 10 10
𝜆 (m) 24.68 36.59 48.41
𝐷/𝜆 0.41 0.27 0.21
Water depth category* Intermediate
*Shallow water depth: 0.05, Intermediate water depth: 0.05 0.5, Deepwater depth: 0.5
The fluid domain with the FB modelled is meshed with refinements to ensure the convergence of the viscous
turbulent model. A refined mesh size of 0.15 meters is used around the free surface water level and wall regions.
The polyhedral meshing is applied, achieving a minimum orthogonal quality standard of 0.2. The Volume of Fluid
(VOF) model is employed for the multiphase simulation of immiscible water and air phases, with an emphasis on
tracking the free surface of the VOF model. The numerical wave tank is set up with wave boundary conditions
using the aforementioned wave parameters. A numerical beach is configured with an effective length of two
wavelengths from the outlet of the numerical wave tank. This setup aids in absorbing wave energy, simulating the
On
3
ly
fo
rR
effect of natural beaches or wave-absorbing structures, thereby preventing artificial wave reflections at the outlet
of the fluid domain. The SST k-omega viscous turbulent model is utilized, and a flat free surface solution of
numerical wave tank is initiated.
ea
din
The required time step size is calculated to be 0.01 seconds. This is based on the longest wavelength of 48.41
meters over 6s (fastest velocity of 8.07 meters per second) and the smallest mesh size of 0.15 meters, giving a
g
required time of 0.0185 seconds. The chosen time step size of 0.01 seconds is smaller than the required time,
at
ensuring sufficient node coverage during wave propagation. The simulation runs for 18,000-time steps with a
0.01-second step size, covering 180 seconds of real time, with solutions exported every second.
W
The FB is assumed to be stationary by assuming that the all three degree of freedoms (i.e., heave, pitch and
CF
surge) is constrained from moving by the taut mooring or pile mooring systems. These mooring methods are
particularly suitable for FBs in intermediate water depth due to their minimal motion which optimizes wave trans-
S
mission reduction. Although taut and pile mooring systems generate greater mooring and foundation forces com-
20
pared to conventional catenary mooring systems, the tropical sea conditions, such as those around Singapore, are
relatively benign. This makes the design of these mooring systems more feasible in this context. For this study,
24
assuming the FB is stationary allows for a clearer and more efficient investigation of various FB shapes and their
performance in wave attenuation, based on the specified environmental parameters.
2.1 Incident and transmitted waves

The CFD simulation is conducted for a duration of 180 seconds for each FB type. All cases are initiated with the
same incident wave height of 2 meters (unit amplitude), allowing for a direct comparison of the transmitted wave
elevations across different FBs to determine the superior attenuation performance. Figures 1, 2 and 3 highlight
the wave attenuations for each FB type over the fluid domain with the incident wave periods of 4s, 5s and 6s,
respectively. These figures illustrate the incident and transmitted waves at the time of 180 seconds after the com-
mencement of wave propagation in the numerical wave tank.
The wave attenuation for 𝑇 4s is similar for all considered FBs due to the shorter wavelengths. This indi-
cates that FBs will remain effective for wave periods shorter than 4s. The transmitted wave coefficients are around
0.1 in all cases, up to 0.15. The stepped and dual box FBs have the best performance with an almost flat transmitted
free surface showing no wave peaks. The box, pontoon and box with plates FBs shows similarly consistent and
low transmitted wave peaks of around 0.1. The T-shaped FB produces a higher transmitted wave peak of 0.15,
indicating less efficiency compared to other types. When 𝑇 increases to 5s, Fig. 2 shows that the wave attenuation
is still effective but with more varying performance observed in different FBs. The stepped box type once again
outperforms the others, having the smallest transmitted wave peak at 0.21. Following in order of efficiency are
the pontoon, typical box, and box with plates, each showing an average transmitted wave elevation of 0.30. The
dual box and T-shaped breakwaters generate the highest transmitted wave elevations, approximately 0.40, sug-
gesting they are least effective in wave attenuation compared to other FBs with same footprint boundaries.
At 𝑇 6s, there is less wave attenuation compared to 𝑇 4s and 5s, due to the FB being less effective when
encountering longer incident wavelength. The stepped breakwater shows the lowest transmitted wave peaks, av-
eraging 0.38m, making it consistently the best performer. This is followed by the pontoon and box with plates
types, with a difference of around 0.1m (0.53m - 0.43m) in transmitted wave elevation compared to the stepped
breakwater. The box breakwater is observed to be moderately effective, preceding the dual box and T-shaped
designs, which are the least effective. The T-shaped type produces the highest transmitted wave amplitude at
0.66m, indicating the poorest performance for wave attenuation. In all cases, the last wavelength near the outlet
of the numerical wave tank is dampened due to the numerical beach effect, which acts as a wave absorber.
2.2 Wave attenuation of the best performing FB

The evolution of wave propagation and attenuation of the best performed stepped box FB is further studied for
various time intervals (22s, 45s, 90s, 135s, 180s) in Fig. 4. The incident wave period selected in this case is 5s.
This also serves to validate the CFD time series simulation of 180s, demonstrating that computations sufficiently
reach steady state by 180s. At 𝑡 45s, the response of the transmitted waves remains in a transient state, pro-
gressing to 135s where it begins to stabilize and continues to display persistent peaks in transmitted waves, thereby
confirming achievement of the steady-state solution.
On
4
ly
fo
3 Conclusion
rR
ea
Based on the analysis outcome, it can be summarized that the stepped type breakwater has the best performance
din
across all the wave periods considered. Pontoon type performs well, ranking below the stepped breakwater. While
not as effective as the stepped or pontoon types, the box with plates design demonstrates similar effectiveness to
g
the standard box, despite having only vertical plate covering half of the draft underwater, rather than a full box-
at
shaped structure. To enhance the assumptions and limitations of the current study, future research should include
more detailed design parameters including the motion response of FB, reflecting the design of slack mooring in
W
addition to stationary, three dimensional CFD approach, hydro-elasticity and validating the findings through ex-
periments.
CF
S
20
24
Note: floating breakwater geometry is not to scale

Fig. 1. Wave elevation along horizontal 𝑥-axis of numerical wave tank for different FBs at 𝑡 180s. 𝑇 4s.

On
5
ly
fo
rR
ea
din
g
at
W
CF
S
20
24


Fig. 4. Wave attenuation of the stepped box breakwater type in wave period 5s at different time intervals: 22s, 45s, 90s, 135s,
and 180s
Acknowledgement
The authors are grateful to the National Research Foundation of Singapore, PUB (Singapore’s National Water Agency) and
the Coastal Protection and Flood Resilience Institute (CFI) Singapore for supporting this work done under the Coastal Protec-
tion and Flood Management Research Programme (CFRP). The conclusions put forward in this publication reflect the views
of the authors alone and not necessarily CFI Singapore or any of the supporting entities.
On
6
ly
fo
References
rR
1.
ea
Ansys® Fluent, Release 2023 R2, User’s Guide, ANSYS, Inc (p. 5552). (n.d.). http://ansyshelp.ansys.com
2.
din
He, F., Li, J., Pan, J., & Yuan, Z. (2023). An experimental study of a rectangular floating breakwater
with vertical plates as wave-dissipating components. Applied Ocean Research, 133(January).
https://doi.org/10.1016/j.apor.2023.103497
g
3.
at
Huang, Z., He, F., & Zhang, W. (2014). A floating box-type breakwater with slotted barriers. Journal of
Hydraulic Research, 52(5), 720–727. https://doi.org/10.1080/00221686.2014.888690
4.
W
Lokesha, Kerpen, N. B., Sannasiraj, S. A., Sundar, V., & Schlurmann, T. (2015). Experimental
CF
investigations on wave transmission at submerged breakwater with smooth and stepped slopes. Procedia
Engineering, 116(1), 713–719. https://doi.org/10.1016/j.proeng.2015.08.356
5.
S
Phillips, O. M. (1977). The dynamics of the upper ocean. In Cambridge University Press.
20
https://doi.org/7319931
6. Sun, B., Zhang, H., Li, C., & Li, Z. (2023). Research Progress and Prospect of Wave Attenuation
24
Performance and Integration of Wave-Energy Converter of Floating Breakwater. Journal of
Environmental Engineering, 149(4), 1–17. https://doi.org/10.1061/joeedu.eeeng-7118
7. Wang, G., Xie, S., Yuan, H., Wang, R., Zhang, T., Liu, X., & Tu, J. (2024). Numerical investigation of
hydrodynamic characteristics of a dual floating breakwater. Ocean Engineering, 294(September 2023).
https://doi.org/10.1016/j.oceaneng.2024.116728
On
ly
fo
rR
Hydrodynamic Analysis of Novel Offshore Floating Concrete Modules and
Mooring System
ea
din
g
Peicen Wang1,2 [0000-0002-4400-1676], Xiaolin Zhao2,3, and Jinping Ou1
1
at
School of Civil and Environmental Engineering, Harbin Institute of Technology, Shenzhen, Shenzhen 518055, China
W
2 Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong
CF
Kong, China
3 Research Institute of Land and Space, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China
S
oujinping163@163.com
20
24
Abstract. This work investigates the hydrodynamic performance of various configurations of concrete floating
modules and mooring systems, which are suited for a shallow water marine environment with a depth of 60
meters. A comparative analysis of the hydrodynamic shapes of floating modules was conducted, including
modules combining flat plate and semi-submersible design, semi-submersible modules, and column-type mod-
ules. This work also compared the catenary chain’s performance, including catenary chains and catenary chains
complemented with segmented weight clumps. The paper underlines the functional characteristics of floating
modules, the performance of distinct hydrodynamic shapes, and both the economic indicators and restoring
force of the mooring systems. Findings suggest that flat plates with circular truncated cone modules aptly fulfill
the hydrodynamic performance requirements and can provide superior load-bearing efficiency. Furthermore,
using catenary chains with segmented weight clumps can improve the motion response performance of the
platforms, under the same mooring chain length and safety requirements, thereby bringing greater economic
benefits. This work can potentially aid in the investigation of the hydrodynamic performance and design of
mooring systems for modular floating structures.
Keywords: Floating Platform Module, Hydrodynamic Performance, Mooring System.
1 Introduction
The accelerated urbanization, rising sea levels due to climate change, and the scarcity of land and resources have
driven global efforts to seek sustainable floating solutions [1]. Concurrently, advancements in marine renewable
energy development and marine engineering have highlighted the potential of Very Large Floating Structures
(VLFS). Extending hundreds of meters or even kilometers, VLFS offer significant advantages such as positional
flexibility, minimal environmental impact, and resilience against rising sea levels. These attributes make VLFS a
viable solution for coastal cities facing land shortages and flooding threats, serving various purposes including
residential areas, industrial zones, airports, and renewable energy installations [2].
However, due to their large structure, VLFS are highly flexible and significantly affected by wave-induced
forces [3]. The hydroelastic effects from the interaction between the structure and fluid are critical. Additionally,
the fabrication and transportation of a single, large VLFS present challenges, leading to the proposal of modular
VLFS. Modular VLFS alleviates large internal bending moments and forces, offering redundancy and scalability.
This study aims to design and research the hydrodynamic shape of individual modules to achieve optimal hydro-
dynamic performance for a multi-module VLFS system.
This paper compares three different platform hydrodynamic shape modules in shallow water: a flat plate with
semi-submersible design platform, a flat plate, a semi-submersible platform, and a column-type platform. It also
proposes a new mooring system that significantly improves dynamic performance without increasing mooring
line length by adding segmented clumps. Using the ANSYS-AQWA program, the hydrodynamic characteristics
of modular platforms under different wave periods were studied, and the hydrodynamic response of the new
mooring system was analyzed across various sea conditions.
2 Hydrodynamic Analysis of Floating Platforms
2.1 Numerical Computation Theory

ANSYS AQWA is a component of the ANSYS engineering simulation software suite, specializing in the hydro-
dynamic analysis of marine and offshore structures. In ANSYS AQWA, the hydrodynamics around marine
On
ly
2
fo
rR
structures are modeled using potential flow theory, which involves solving potential flow equations under specific
boundary conditions. Potential flow theory assumes the fluid is incompressible and irrotational, and the fluid flow
ea
field can be described using a velocity potential function. This velocity potential function must satisfy the Laplace
din
equation:
g
0 (1)
at
Additionally, it is assumed that the motion of the floating body in small amplitude waves is minimal. According
W
to linear wave theory, the velocity potential function must meet the seabed boundary condition, the free surface
CF
condition, the submerged body surface condition, and the radiation condition:
S
0, 𝑎𝑡 𝑧 ℎ (2)
𝑔 0, 𝑎𝑡 𝑧 0
20 (3)
∑ ν 𝑓 𝑥, 𝑦, 𝑧 24(4)
lim 𝜙 0 (5)
→∞
2.2 Design and Parameters of Floating Platforms

This paper designs three types of floating platforms, suitable for a water depth of 60 meters. The side, top, and
3D views of each platform are illustrated in Fig. 1 and Fig. 2. Parameters such as weight, displacement, and
rotational inertia are provided in Table 1. Scheme 1 is a semi-submersible platform with five columns and a flat
bottom pontoon. The central column aims to reduce mid-span deflection and the bending moment of the upper
deck. Scheme 2, based on Scheme 1, removes the central column and changes the flat bottom pontoon to a ring-
shaped pontoon. Scheme 3 is inspired by some column-shaped and hourglass-shaped FPSOs, with the cylinder
contracted at the waterline to reduce the waterline area. For comparison, the displacement of these schemes is set
at approximately 18,000 tons.
50
50 50 10 30 10
5 2
5 2
5 2
24
27
27
15
15
15
2
5
5 40 5
2 10 9 8 9 10 2 2 12 3 16 3 12 2
50
50 50
7
8
9
18
50
50
50
16
18
9
8
7
7 18 18 7 8 9 16 9 8 50
50 50
(a) (b) (c)

Fig. 1. Side and top views of floating platforms, (a) Scheme 1, (b) Scheme 2, (c) Scheme 3.
Fig. 2. 3D models of floating platforms, (a) Scheme 1, (b) Scheme 2, (c) Scheme 3.
On
ly
3
fo
rR Table 1. Parameters of floating platforms.
Scheme number
ea Scheme 1 Scheme 2 Scheme 3
din
Waterline area (m2) 360 449 707
Draft (m) 20 20 17
Displacement (t) 17900
g 17949 18447
Self-weight (t) 9580
at 9431 9843
W
Maximum load weight (t) 8320 8518 8604
CF
Moment of inertia Ixx (kg·m2) 3.20×109 3.28×109 2.40×109
Moment of inertia Iyy (kg·m2) 3.20×109 3.28×109 2.40×109
Moment of inertia Izz (kg·m2) 4.68×109 4.83×109
S 3.16×109
The three designed floating platform schemes are mainly used as supporting platforms for floating factory facili- 20
ties. Therefore, two five-story steel modular assembly buildings are placed on top of the floating platform. The
total weight of these two buildings is approximately 3248.3 tons. In practical calculations, the two buildings are 24
simplified as rectangular prisms. The schematic diagram of the superstructure and the floating platform is shown
in Fig. 3.
Fig. 3. Schematic diagram of the superstructure and the floating platform, (a) left view, (b) front view, (c) isometric view.
The Response Amplitude Operator (RAO) of the platform motion can be calculated using the AQWA-LINE mod-
ule to obtain the RAO for each degree of freedom of the platform. Fig. 4 and Fig. 5 respectively show the pitch
RAO and heave RAO. Due to the strong symmetry of the platform, the roll RAO is omitted in this paper. For the
pitch RAO, different wave directions significantly impact the pitch RAO, with the greatest impact on Scheme 1.
Since the undamped natural frequencies of the pitch for all three schemes are less than 0.2 rad/s, and the frequency
range shown in Fig. 4 is 0.25-2 rad/s, each scheme only has a small peak in the figure. The pitch RAO peaks for
all three schemes are concentrated around 0.65 rad/s, with Scheme 1 having the largest pitch RAO at 0.62. For
the heave RAO, the values for different wave directions are almost identical. Since the undamped natural frequen-
cies of the heave for Schemes 2 and 3 are greater than 0.2 rad/s, a larger peak is observed in the figure, with
Scheme 3 having the highest peak, reaching a heave RAO of 2.1.
0.7 0.7 0.7

0° 0° 0°
0.6 15° 0.6 15° 0.6 15°
30° 30° 30°
45° 45° 45°
0.5 60°
0.5 60°
0.5 60°
75° 75° 75°
0.4 90° 0.4 90° 0.4 90°
0.3 0.3 0.3
0.2 0.2 0.2
0.1 0.1 0.1
0 0 0
0.5 1 1.5 2 0.5 1 1.5 2 0.5 1 1.5 2
Frequency (rad/s) Frequency (rad/s) Frequency (rad/s)
(a) (b) (c)
Fig. 4. Pitch RAO, (a) Scheme 1, (b) Scheme 2, (c) Scheme 3.

On
ly
4
fo
2
rR
0°
2
0°
2
0°
ea
15° 15° 15°
30° 30° 30°
45° 45° 45°
1.5 1.5 1.5
din
60° 60° 60°
75° 75° 75°
90° 90° 90°
1 1 1
g
at
0.5 0.5 0.5
0 0
W 0
CF
0.5 1 1.5 2 0.5 1 1.5 2 0.5 1 1.5 2
Frequency (rad/s) Frequency (rad/s) Frequency (rad/s)
(a) (b) (c)
S
20
Fig. 5. Heave RAO, (a) Scheme 1, (b) Scheme 2, (c) Scheme 3.
3 24
Design and Time-Response Calculation of a Novel Mooring System
3.1 Location Site and Environmental Conditions

The floating platform, which serves as a support structure for floating factory facilities is assumed to be located
at 114° 50' E and 22° 00' N in the South China Sea, where the water depth is approximately 60 meters [4]. The
basic data for this location is sourced from the waveclimate.com database provided by BMT ARGOSS. This
database utilizes measurements from radar altimeters, radar scatter meters, and imaging radar, capturing average
wave parameters within a 50 km radius centered at the specified coordinates. The data has been validated and
calibrated with buoy testing data. Samples were collected every three hours from 1992 to 2016, totaling 73,056
samples. The wave parameters were statistically analyzed, and the 100-year return period sea state conditions for
the time-domain calculations of the floating platform are shown in Table 2 [5].
Table 2. Wind and wave conditions [5].
Significant wave Spectral peak pe-

Return period Wind speed (m/s) Wave spectrum
height (m) riod (s)
100 years 34.22 10.65 14.13 JONSWAP
3.2 Design of a Novel Mooring System

The offshore platform itself does not have thrusters and relies primarily on the mooring system to maintain its
position. Based on the mechanical properties of the mooring lines, the mooring system can be classified into two
main categories: catenary mooring and taut mooring. Catenary mooring is often used in shallower water depths,
and the mooring force is mainly provided by the self-weight of the chain. So, achieving sufficient mooring force
typically requires longer mooring lines, resulting in a larger mooring radius. This can significantly increase ma-
terial usage and costs, as well as enlarge the footprint, making it challenging to arrange the mooring system be-
tween platform modules. To address these drawbacks, this paper proposes a new mooring system arrangement
where multiple weight clumps are placed on the mooring lines. In the initial state, the weight clumps rest on the
seabed. Under the influence of waves or tides, the weight clumps are lifted off the seabed, providing additional
restoring force to the mooring lines.
The mooring lines are studless chains, and the entire mooring system consists of eight mooring lines. The
mooring lines are arranged in pairs, with an angular separation of 15 degrees. Each mooring line is equipped with
four groups of a total of eight concrete weight clumps. The submerged weights of each group are 10, 20, 30, and
40 tons, respectively. These eight weight clumps are evenly distributed along the mooring line, with a spacing of
approximately 33.3 meters. The platform and mooring line arrangement are shown in Fig. 6. The parameters of
the mooring lines and the coordinates of the fairlead and anchor points are shown in Table 3 and Table 4.
Table 3. Parameters of mooring lines.
Maximum ex- Transverse Longitudinal Total length

Mass/unit Added mass
Stiffness (N) pected tension drag coeffi- drag coeffi- (m)
length (kg/m) coefficient (-)
(N) cient (-) cient (-)
219.4 9.415×108 7.693×106 0.8 2.0 1.15 300.0
On
ly
5
fo
rR
Table 4. Coordinate information of fairlead points and anchor points.
Mooring lines Fairlead point (m) Anchor point (m)

ea Mooring lines Fairlead point (m) Anchor point (m)
din
No. 1 (25, 25, 7) (250.2, 191.98, -60) No. 5 (-25, -25, 7) (-250.2, -191.98, -60)
No. 2 (25, 25, 7) (191.98, 250.2, -60) No. 6 (-25, -25, 7) (-191.98, -250.2, -60)
No. 3 (-25, 25, 7) (-191.98, 250.2, -60)
g No. 7 (25, -25, 7) (191.98, -250.2, -60)
No. 4 (-25, 25, 7) (-250.2, 191.98, -60)
at
No. 8 (25, -25, 7) (250.2, -191.98, -60)
W
CF
S
20
Fig. 6. Schematic diagram of mooring lines and clumps, (a) mooring system, (b) single mooring line and clumps.
24
3.3 Time Response Calculation
This section simulates the wind and wave parameters outlined in Table 2 by invoking the AQWA-DRIFT module.
For each design scheme, 10 different random seeds were used to simulate irregular waves, with a total simulation
duration of 12,000 seconds. The unstable values from the first 1,200 seconds were discarded in every final result.
Due to space limitations, this paper presents the motion response and No. 4 mooring line response (maximum
tension of all mooring lines) for Scheme 2 with and without clumps under one random seed, as shown in Fig. 7.
The surge response and mooring response values for all three schemes are displayed in Table 5 and Table 6. Pitch (deg)
Heave (m)
Surge (m)
Mooring line laid length (m)

Mooring line tension (N)
Anchor uplift force (N)
Fig. 7. Motion and mooring response of Scheme 2.
From Fig. 7, and the data in Table 5 and Table 6, it is evident that the introduction of clumps significantly reduces
the maximum and mean values of the surge response, with the maximum value decreasing by over 20% and the
mean value by over 40%. At the same time, the introduction of clumps does not significantly affect the heave or
pitch response. Additionally, adding clumps significantly increases the laid length of the mooring lines and re-
duces the anchor uplifting force. However, it should be noted that the introduction of clumps increases both the
maximum and mean values of mooring line tension. Despite this, the increased tension remains within safety
On
ly
6
fo
rR
factor requirements. Adding concrete weight clumps to each mooring line results in a relatively small increase in
overall material costs, providing good economic value. By comparing the three schemes in Table 5 and Table 6,
ea
it is evident that the mooring system performance of Schemes 1 and 2 is quite similar and significantly better than
din
that of Scheme 3. However, under the same mooring conditions, the surge response of Scheme 2 is reduced by
more than 30% in both maximum and mean values compared to Scheme 1.
g
at
Table 5. Surge and mooring response (without clumps).
W
Scheme Scheme 1 Scheme 2 Scheme 3
CF
Maximum surge (m) 22.01 15.40 25.33
Mean surge (m) 12.12 6.55 13.68
Maximum mooring line tension (N) 2.07×106 1.66×106
S 8.67×106
Mean mooring line tension (N) 5.93×105 5.81×105
20
6.70×105
24
Mooring safety factor (-) 3.71 4.63 0.89
Maximum anchor uplifting force (N) 1.31×105 3.98×104 1.34×106
Minimum laid length (m) 7.13 13.82 3.24
Table 6. Surge and mooring response (with clumps).
Scheme Scheme 1 Scheme 2 Scheme 3

Maximum surge (m) 16.71 10.38 19.22
Mean surge (m) 6.84 2.59 6.05
Maximum mooring line tension (N) 3.51×106 3.59×106 5.68×106
Mean mooring line tension (N) 1.89×106 2.35×106 2.46×106
Mooring safety factor (-) 2.19 2.14 1.35
Maximum anchor uplifting force (N) 7.30×103 6.78×103 1.81×105
Minimum laid length (m) 69.08 70.13 10.95
4 Conclusion
This study conducted a comprehensive analysis of various floating platform configurations and mooring systems
to identify optimal designs for shallow water marine environments with a depth of 60 meters. Among the three
configurations analyzed, the semi-submersible combined with a ring-shaped pontoon demonstrated better hydro-
dynamic performance, showing reduced motion response, making it suitable for offshore applications.
The proposed mooring system, incorporating segmented weight clumps, significantly enhanced the dynamic
performance of the floating platforms without increasing the mooring line length. This system effectively reduced
the surge response. Although the introduction of weight clumps increased mooring line tension, the levels re-
mained within safe limits, ensuring the integrity of the mooring system. The innovative mooring system provides
a cost-effective solution for offshore floating structures.
In conclusion, the findings of this study highlight the potential of using a novel floating platform combined
with a segmented weight clump mooring system to achieve optimal hydrodynamic performance for offshore float-
ing structures. These insights contribute to the design and development of more efficient and cost-effective marine
platforms, supporting the growing need for sustainable solutions in marine engineering.
References
1. World Economic Forum, https://www.weforum.org/agenda/2021/10/floating-buildings-climate-change-rising-sea-lev-
els/, last accessed 2024/06/25.
2. Lamas-Pardo, M., Iglesias, G., Carral, L.: A review of very large floating structures (VLFS) for coastal and offshore uses.
Ocean Engineering, 109, 677–690 (2015).
3. Wang, C. M., Watanabe, E., Utsunomiya, T.: Very large floating structures. Taylor & Francis, Abingdon (2008).
4. Zheng, S., Li, C., Wang, X., Ou, J.: Equivalent static wave-induced loads based on long term responses of a 10-MW
floating semi-submersible wind turbine system. Available at SSRN 4520907.
5. Zheng, S., Li, C., Wang, P., Zhou, S., Xiao, Y.: Wind tunnel and wave flume testing on directionality dynamic responses
of a 10 MW Y-shaped semi-submersible wind turbine. Journal of Renewable and Sustainable Energy, 15(1) (2023).
On
ly
fo
rR
Experimental Study on Hydrodynamic Response of a Modular Floating
ea
Structure System with Chain-type Expansion
din
g
Yanwei Li1, Nianxin Ren1,2, Yaqiong Liu1 and Jinping Ou3
1
at
School of Marine Science and Engineering, Hainan University, Haikou, 570228, China
W
2 State Key Laboratory of Marine Resource Utilization in South China Sea, Hainan University, Haikou, 570228, China
3 School of Civil and Environmental Engineering, Harbin Institute of Technology, Shenzhen, 518055, China
rennianxin@hainanu.edu.cn
CF
S
20
Abstract. Modular floating structures provide a feasible way to solve the land scarcity in and around coastal
24
cities. The present work investigates the hydrodynamic responses of a proposed modular floating structure
system through the scale model test. The proposed system consists of several inner TLP hexagonal floating
structure modules and outermost floating artificial reef modules, forming a chain-type configuration. The
outermost floating artificial reef modules can play the role as a floating breakwater, meanwhile providing
habitat for marine organism. The scale model test was carried out in the wave flume of Hainan University.
Effects of the chain-type modular expansion on the system's dynamic response have been extensively studied
through both regular and irregular wave tests. Experimental results indicate that chain-type modular expansion
notably reduce surge response and enhances the system's survival capability under extreme sea condition. The
primary experimental results in this article can offer valuable guidance for the practical application of modular
floating structures in engineering.
Keywords: modular floating structure, model test, hydrodynamic response, chain-type expansion, floating
artificial reef.
1 Introduction
Global warming is leading to continued shoreline erosion in and around coastal cities. However, the majority of
the world's population is distributed in coastal areas, which urgently requires expanding available space into the
ocean [1]. Modular floating structure has been proved to be a kind of marine structure for effective exploitation
of ocean space, and its characteristic of modular construction provides convenience for building long-life marine
floating platforms [2].
Modular floating structure is a complex multi-body system, and the hydrodynamic coupling effect among mod-
ules significantly influences its behavior. Several numerical and physical model experiments have been developed
to study the hydrodynamics of modular floating structures. Bispo et al. [3] formulated an analytical model for
modular floating structures based on linearized wave theory and small structure response. They obtained the ana-
lytical solution through the velocity decomposition method, validating its high consistency with the results gen-
erated by the AQWA code. Ding et al. [4] conducted an experiment analyzing the responses of an 8-module
modular floating structure under complex wave conditions. The results indicated that an uneven seabed exerts a
significant influence on motion responses and connector loads.
The conventional anchor chain mooring method used for modular floating structures occupies a considerable
sea area, hindering the expansion of the modules [5]. The TLP-type floating wind turbine offers valuable guidance
for positioning modular floating structures. The large vertical stiffness of the tension leg markedly decreases the
structure's responses in heave, pitch, and roll [6]. So far, several numerical and experimental studies have explored
the conceptual design and dynamic characteristics of modular floating structures. Nonetheless, there remains a
scarcity of research reports focusing on TLP-type modular floating structures and its modular expansion, particu-
larly concerning model experiments.
In present work, effects of the chain-type modular expansion on the system's hydrodynamic responses have
been extensively studied by a series of model tests under regular and irregular waves. The floating artificial reefs,
located as the outermost modules in the system, aim to serving as a floating breakwater and enhancing the sur-
rounding marine environment [7].
On
ly
2
fo
2 Experimental setup
rR
ea
The proposed modular floating structure system comprises a chain-type arrangement of hexagonal floating struc-
din
tures interconnected by hinged connectors, with the floating artificial reef modules integrated on the outermost
side. Each hexagonal module utilizes four symmetrically moored tension legs, with a displacement designed to
g
be greater than its mass to ensure adequate pretension. Hexagonal modules are color-coded in red and yellow to
at
aid in locating the waterline. The scale model test was conducted in the wave flume (50 m * 2.5 m * 2m) of Hainan
University, and the scale ratio is set to be 1:80 based on Froude scaling principle. The general layout of the scale
W
model test is shown in Fig. 1, where each module is labeled as Mi (i=1~7). Detailed parameters of the proposed
CF
system are listed in Table 1. The representative photos of the proposed system with different hexagonal modules
are shown in Fig. 2.
S
Wave maker
20 Wave absorbers
24
Incident waves WG 1 WG 2
Hinge connector TLP model Reef model
Z
X
M1 M2 M3 M4 M5 M6 M7
1m Tension leg
(a) Side view
TL1
WG 1 Y WG 2
0.5m
2.5m
X
0.1875m 0.433m 0.04m
(b) Top view

Fig. 1. Schematic diagram of the general layout of the scale model test.
Table 1. Detailed design parameters of the proposed modular floating structure system.
Parameters Full scale Scaled (λ = 1/80) Units

Hexagonal TLP module
Side length 20 0.25 m
Mass; Displacement 6,600,000; 10,650,000 12.89; 20.80 Kg
Water depth; Draft 80; 10 1; 0.125 m
Z_CoG -5 -0.0625 m
Ⅰxx=Ⅰyy, Ⅰzz 9.95×108; 1.25×109 0.304; 0.381 Kg·m2
Adjacent distance 3.2 0.04 m
Floating artificial reef module
Length; Breadth; Draft 20; 15; 10 0.25; 0.188; 0.125 m
Mass; Displacement 800,500; 800,500 1.563; 1.563 Kg
Z_CoG -2 -0.025 m
Ⅰxx, Ⅰyy, Ⅰzz 1.2×108; 9.8×107; 1.0×108 0.037; 0.030; 0.031 Kg·m2
cut-opening ratio 20% 20%
Tension leg dimension
Length; Outer diameter; Thickness 70; 1.2; 0.04 Equivalent steel cable m
Axial stiffness 437,309,697.4 68,329.64 N/m
In the experiment, the material of the model is mainly organic glass. The smooth shaft and vertical support
were used to simulate the hinge connector. Connected adjacent modules can solely exhibit relative pitch motion
around the smooth shaft. The tension legs were simplified as equivalent steel cables, moored at the four corners
of each hexagonal floating structure. Some waterproof tension-compression sensors were installed at the top of
the steel cable to measure the tension force. A stereovision measurement system with two cameras and some
targets was used to measure the 6-DOF motions of multiple modules. In addition, wave gauges (WG) were placed
on both sides of the model system to measure the wave information, as shown in Fig. 1.
On
ly
3
fo
rR
ea
din
g
at
W
CF
Fig. 2. Photo of scaled model for the proposed systems with different hexagonal modules.
3 Experimental results and discussion

S
20
24
Effects of chain-type modular expansion on the dynamic response of systems has been extensively studied through
both regular and irregular wave tests. Due to the significant vertical stiffness of the tension leg platform, its most
prominent response occurs in surge under unidirectional incident waves [8,9]. Meanwhile, there is minimal rela-
tive surge motion between adjacent modules because of the hinge connection. Consequently, solely the surge
results of the up-wave hexagonal module M2 and its maximum tension leg load of the TL1 are presented. All
results are presented in full scale to provide a more comprehensive understanding.
The dynamic response results of systems with different numbers of modules under some representative regular
wave sea conditions (wave height of 3m, wave period of 7s, 10s, 17s) are shown in Fig. 3. Fig. 3a illustrates a
decreasing trend in surge amplitude as the number of modules increases, suggesting that augmenting the modules
can enhance the system's stability. At a wave period of 10 s, the surge amplitude of the 3-module system exhibits
relatively significance. Comparatively, the surge amplitudes of the 5-module, 7-module, and 9-module systems
have decreased by 43.7%, 48.6%, and 62.6%, respectively, in contrast to the 3-module system. Hence, to enhance
the comfort and stability of the system under operational sea conditions, it is advisable to avoid having an exces-
sively small number of modules within the system. Fig. 3b shows that at a wave period of 17 s, each system's
tension leg experience maximum force with no significant variation among them. This suggests that the increase
in the number of modules has a relatively minor impact on the tension leg force. Hence, it can be concluded that
the chain-type modular expansion enhances the system's motion response characteristics without notably ampli-
fying the tension leg load.
3-module 3-module
5-module 5-module
3.0 20
7-module 7-module
9-module 9-module
18
2.5
16
14
Maximum Ft (MN)
2.0
12
Surge (m)
1.5 10
1.0
6
4
0.5
2
0.0 0
T=7s T=10s T=17s T=7s T=10s T=17s
(a) Surge of the M2 (b) Maximum tension leg force of the TL1
Fig. 3. Comparison of the main dynamic responses of the systems with different numbers of modules
To assess the system's safety, the extreme responses of the system with varying numbers of modules under
extreme sea condition (JONSWAP, λ=3.3, Hs=4 m, Tp=10 s) were further investigated. Table 2 lists the extreme
responses characteristic information.
According to Table 2, compared with the 3-module system, the maximum surge amplitude of the 5-module, 7-
module and 9-module systems decreased by about 59.4%, 72.8% and 79.9% respectively, indicating that the in-
crease in the number of modules can improve the system's survivability under extreme sea condition. Furthermore,
in comparison to other systems, the 3-module system displays the highest maximum value and standard deviation
of tension leg load, signifying an increased risk of tension leg failure. To ensure the system's safety, it's advisable
not to have too few modules within the system.
On
ly
4
fo
rR
Table 2. Main extreme response characteristic information of the systems with different numbers of modules un-
der extreme sea condition
ea Surge of the M2 (m) Ft of the TL1 (MN)
9-module din
Max
Mean
0.631
——
13.098
9.754
STD
g 0.135 1.347
at
Max 0.854 14.113
7-module Mean —— 9.730
STD
W 0.155 1.642
CF
Max 1.276 13.578
5-module Mean —— 9.745
S
STD 0.257 1.357
Max 3.140 14.822
3-module Mean ——
20 9.797
24
STD 0.659 1.769
4 Conclusion
This work reports an experimental study on the hydrodynamic response of a modular floating structure system
with chain-type expansion. Effects of the chain-type modular expansion on the system's dynamic response have
been extensively studied through both regular and irregular wave tests. The experimental results show that the
system's surge response demonstrates a downward trend with the chain modular expansion, without notably im-
pacting the tension leg loads. This indicates favorable scalability of the system. In addition, to ensure system
safety and comfort, it is advisable to avoid applying the modular system with too few modules.
Acknowledgements
This research was supported by the National Natural Science Foundation of China (Grant No. 52161041), the
Natural Science Foundation of Hainan Province (Grant NO. 520RC552). The financial supports are greatly
acknowledged.
References
[1] Wang, C., Tay, Z.: Very Large Floating Structures: Applications, Research and Development. Procedia Engi-
neering, 14, 62-72 (2011).
[2] Suzuki, H.: Overview of Megafloat: Concept, design criteria, analysis, and design. Marine structures, 18(2),
111-132 (2005).
[3] Bispo, I., Mohapatra, S., Soares, C.: Numerical analysis of a moored very large floating structure composed
by a set of hinged plates. Ocean Engineering, 253, 110785 (2022).
[4] Ding, J., Geng, Y., Xu, S., Yang, W., Xie, Z.: Experimental study on responses of an 8-module VLFS consid-
ering different encounter wave conditions. Marine Structures, 78, 102959 (2021).
[5] Ja'e, I., Osman, A., Yenduri, A., Nizamani, Z., Nakayama, A.: Optimisation of mooring line parameters for
offshore floating structures: A review paper. Ocean Engineering, 247, 110644 (2022).
[6] Tian, C., Liu, M., Xiao, L., Wei, H.: Investigation for the effect of the column step on the flow-induced motion
of a TLP. Marine Structures, 90, 103400 (2023).
[7] Wang, G., Wan, R., Wang, X., Zhao, F., Lan, X., Cheng, H., Tang, W., Guan, Q.: Study on the influence of
cut-opening ratio, cut-opening shape, and cut-opening number on the flow field of a cubic artificial reef. Ocean
Engineering, 162, 341-352 (2018).
[8] Li, Y., Ren, N., Li, X., Ou, J.: Hydrodynamic Analysis of a novel modular floating structure system integrated
with floating artificial reefs and wave energy converters. Journal of Marine Science and Engineering, 10(8), 1091
(2022).
[9] Ren, N., Ma, Z., Shan, B., Ning, N., Ou, J.: Experimental and numerical study of dynamic responses of a new
combined TLP type floating wind turbine and a wave energy converter under operational conditions. Renewable
Energy, 151, 966-974 (2020).
On
ly
fo
rR
Numerical investigation of hydrodynamic responses of straight floating
ea
bridge with pontoons
din
g
Jihun Song1, Chungkuk Jin2 and Seungjun Kim1
1
at
Department of Architectural, Civil and Environmental Engineering, Korea University, Seoul, Korea
W
2 Department of Ocean Engineering and Marine Sciences, Florida Institute of Technology, Melbourne, USA
CF
rocksmell@korea.ac.kr
S
20
Abstract. Floating bridges offer alternative design solutions for sea-crossing where conventional bridges with
piers and abutments are not feasible. However, their complex hydrodynamic interactions necessitate detailed
24
modeling and analysis to ensure structural safety, durability, and serviceability. Utilizing OrcaFlex software,
this study introduces an advanced computational framework to examine the structural behavior of these bridges
under various ocean environmental loading conditions. Based on previous work that highlights the importance
of considering inhomogeneous waves, we successfully created a realistic model of a straight pontoon bridge
and integrated inhomogeneous wave conditions into the OrcaFlex model. The study mainly focused on the
examination of the effects of different mooring configurations on the structural behaviors under extreme con-
ditions. We found that optimized mooring configurations can substantially improve the bridge's resilience to
hydrodynamic forces. This research is anticipated to contribute valuable insights into the basic and detailed
design of floating pontoon bridges, thereby improving safety and efficiency.
Keywords: Floating bridge, Hydrodynamics, Global performance, Mooring.
1 Introduction
Floating bridges, which rely on natural buoyancy to support the bridge deck, offer a viable alternative to conven-
tional bridge types in locations with deep water or extremely soft seabeds [1-3]. Globally, the advantages of float-
ing bridges have spurred the construction of such infrastructure, including notable examples like the Hood Canal
Bridge, the Bergsøysund Bridge, the Nordhordland Bridge, and the Yumemai Bridge. The Norwegian Public Road
Authority (NPRA) launched the E39 coastal highway project, aiming to reduce travel time by replacing ferry
connections with floating infrastructures. For the Bjørnafjord, which has a width of up to 5 km, various types of
floating bridge concepts have been designed and examined, supported by numerous research activities. Based on
field measurements, the inhomogeneity within wave parameters in the Bjørnafjord was assessed [4-6], and the
impact of inhomogeneous waves on the hydrodynamic responses of a floating curved bridge was explored [7-8].
Some researchers used the Simulating Waves Nearshore (SWAN) model to simulate wave conditions and calibrate
them with field measurement data, followed by increasing the number of data points to perform extreme value
analysis for deriving design wave parameters [9-10]. The same research group also investigated inhomogeneous
load effects on a straight and side-anchored floating bridge, which is the primary literature referred on this study
[11]. Viuff et al. [12-14] studied numerical analysis method for a long side-anchored floating curved bridge with
a stay-cable tower, investigating the uncertainty with a model test. They further studied the uncertainty in various
numerical models of floating bridges to improve experimental methodologies and numerical analyses [15-16].
Despite extensive research, current commercial software does not provide functionalities to implement inhomo-
geneous wave conditions, which are crucial for accurately assessing the structural responses and limit states under
realistic sea states. Previous studies have emphasized the significance of considering inhomogeneous wave char-
acteristics in structural response evaluations and limit state assessments, as they can significantly impact the be-
havior of floating bridges. Therefore, it is essential to integrate inhomogeneous wave conditions into commercial
solvers to improve design effectiveness. Furthermore, the effects of directional inhomogeneity of waves on the
structural responses under various mooring configurations are not well understood. Investigating these effects is
important for optimizing the design and improving the resilience of floating bridges, thus leading to better design
practices and enhanced safety and performance.
In this study, a referred straight floating pontoon bridge was reproduced using OrcaFlex to evaluate wave-
induced hydro-elastic responses of structural components. In addition, an advanced framework to integrate inho-
mogeneous wave conditions into the commercial solver was proposed. Following the approach, various inhomo-
geneous wave cases were simulated for the straight floating bridge model, focused on the directional
On
ly
2
fo
rR
inhomogeneity. Finally, parametric studies considering the characteristics of mooring configurations were con-
ducted to evaluate and improve the resilience capacity against hydrodynamic forces.
ea
2 Theoretical background
din
g
2.1 Motion of equation
at
W
Pontoons are the major components that interact with waves, causing complex wave-induced behaviors of struc-
CF
tural components above the water. In general, the hydrodynamic coefficients of pontoons are evaluated in the
frequency domain analysis based on the potential theory. Then, the pre-evaluated wave load characteristics are
S
mapped on the lumped mass. The governing equations for pontoons of floating bridge in the time domain can be
20
written by Cummin’s equation as
24
𝑀 𝑀 ∞ 𝑢 𝑡 𝑅 𝑡 𝜏 𝑢 𝜏 𝑑𝜏 𝐾 𝐾 𝑢 𝑡 𝐹 𝑡 (1)
where 𝑀 is the inertia matrix, 𝑀 ∞ represents the added mass at infinite frequency matrix, 𝑅 𝑡 𝜏 is the re-
tardation function to involve the energy dissipation corresponding to the velocity by convolutional integral for the
time lag, 𝐾 is the hydrostatic stiffness matrix, 𝐾 is the coupled stiffness matrix of the floating bridge, 𝑢 is the
displacement vector, 𝑢 is the velocity, 𝑢 is the acceleration, and 𝐹 𝑡 is the external force vector including en-
vironmental loads.
2.2 Inhomogeneous wave

In offshore engineering, the ocean wave is described as a function of the power spectrum, directional spreading,
wave heading and individual phase angles of wave components. The Joint North Sea Wave Observation Project
(JONSWAP) spectrum was utilized for generating irregular waves in this study and is expressed as
𝑆 1 0.287ln 𝛾 𝐻 𝑒 𝛾 (2)
where 𝜔 is the peak angular frequency, 𝐻 is the significant wave height, 𝛾 is the extra peak enhancement factor,
𝜎 is the spectrum width parameter which equals to 0.07 for 𝜔 𝜔 and 0.09 for 𝜔 𝜔 .
The directional spreading refers to the distribution of wave energy across different directions, while maintain-
ing the total energy. The specific form with the normalizing constant used by OrcaFlex is given by
𝐷 𝑐𝑜𝑠 𝜃 𝜃 (3)
√
where 𝜃 is the principal wave direction, n is the spreading coefficient to adjust energy concentration towards the
principal wave heading.
For the wave forces acting on the floating body, the first order wave force and the second order difference
frequency forces can be written as
𝐹 ∑ ∑ 𝑅𝑒 |𝑯 𝜔 , 𝜃 | 2𝑆 𝜔 𝐷 𝜃 ∆𝜔∆𝜃 e (4)
𝐹 ∑ ∑ ∑ 𝑅𝑒 |𝑯 𝜔 , 𝜔 , 𝜃 | 2𝑆 𝜔 𝐷 𝜃 ∆𝜔∆𝜃 e (5)
where 𝑯 𝜔 , 𝜃 is the linear transfer functions from the diffraction and radiation analysis, 𝑯 𝜔 , 𝜔 , 𝜃 is
the quadratic transfer functions, 𝑆 𝜔 is the unidirectional wave spectrum, 𝐷 𝜃 is the directional spreading,
𝜀 is the phase angle of wave components, 𝑘 is the wave number calculated by the dispersion relation, 𝜑 is
the phase angle of load transfer functions. If more than one of wave parameters differ by pontoon locations, the
wave condition has inhomogeneity and regarded as the inhomogeneous wave field.
On
ly
3
fo
3 Target floating bridge model
rR
3.1 System particulars
ea
din
The foundational model described by Multiconsult [17] incorporates the stay-cable tower, the seabed terrain char-
g
acteristics of the target site, and additional intricate elements in the design of the floating bridge at Bjørnafjord.
at
This study adopts a straight and side-anchored floating pontoon bridge model, mirroring the numerical framework
proposed by Dai et al. [11] which provides a simplified yet effective representation of the original design's behav-
W
ior. As shown in Fig. 1, the stay-cable tower has been reduced to a line element with equivalent sectional stiffness,
CF
with pontoons, girders, and columns uniformly positioned at the same elevation. Spanning a total length of 4,600
meters, the bridge features pontoons spaced 125 meters apart. The 14.5-meter-tall columns are firmly connected
S
to both the upper surfaces of the pontoons and the girders. Mooring lines are connected to four designated pon-
20
toons, each located 1,000 meters apart, with each set comprising eight strands clustered together.
z
Global axes
x
Sea level A22
24
A30 A35
South end 300m North end
31.25m 93.75m 4,600m

209.375m
Girder Section H1 S1 F1 S1
z’
z’
y’ x’
Pontoon #1 Pontoon #2
Fig. 1. Idealized straight floating bridge in plan and side view
3.2 Numerical model of straight pontoon bridge in OrcaFlex

The numerical model for the floating pontoon bridge was constructed using OrcaFlex, as demonstrated in Fig. 2.
Within OrcaFlex, the local z-axis of the line element aligns with its length, with the default setting being a shear-
flexible beam element. Each line element's beam orientation was configured as illustrated in Fig. 1. To prevent
overlapping connections, Girder S1 was placed with free-end connections at a height of 18 meters, while Girder
F1 connected to Girder S1. All connections were modeled with dummy spring elements which have infinity stiff-
ness. A fixed boundary condition was imposed at the southern end, whereas only axial elongation and transverse
rotation were permitted at the northern end. Pretension was applied to the mooring lines, with 1,400 kN for the
outer clusters and 1,522 kN for the inner clusters. The adjusted length of the mooring lines influences the total
stiffness and natural periods of the system.
Fig. 2. Numerical model of the straight floating pontoon bridge in OrcaFlex.
4 Inhomogeneous wave simulation
4.1 Methodology
OrcaFlex generates wave forces in the time domain with given parameters according to the Eq. (4)-(5). However,
the built-in feature of the software applies identical wave loads at all locations in the fluid field, indicating that
the software only supports the homogeneous wave conditions. Here, we propose a methodology to generate inho-
mogeneous waves by using the applied load feature in OrcaFlex. Basically, OrcaFlex solves the discretized equa-
tion of motion with implicit method after constructing inertial, damping, and stiffness matrix of fully coupled
On
ly
4
fo
rR
structures. Instead of using the built-in feature of generating waves, users can define external forces with tabular
data based on the wave loads at specific nodes. Then, the numerical solver yields identical motion response vectors
ea
discretized for each time step. Similarly, users can define different force vectors for each node with sets of pre-
din
arranged wave parameters to simulate inhomogeneous wave conditions. OrcaFlex provides the environment for
text-editable input files so that users can add a large amount of prepared data in the appropriate input file format
g
with Python codes. This approach is summarized as switching external force vectors with pre-arranged tabular
at
data, skipping the generation of time-series wave forces with wave parameters. The serial process is illustrated in
Fig. 3.
W
The proposed approach was successfully validated with the simulation using the built-in wave generating fea-
CF
ture for homogeneous wave case. The effects of 1st and 2nd difference frequency wave loads, peak period, wave
height, wave headings, directional spreading were examined and showed good agreement, while individual phase
S
angles of wave components were set to be fully correlated.
Python Code
20
24
(Fully coupled floating bridge model)
No constraints for floating bodies
Wave simulation on fixed floating bodies No wave parameters
(P1~P35) Include added-mass & radiation damping
Extraction of wave forces & moments (1st, 2nd)

from the vessels by OrcFxAPI
Displacement & Bending Moment Evaluation
420 Forces (35 Pontoons, 6 dof.)
Total : 420 @ 7000 (500 sec. 0.1step)
Edit input script for the variable data Validation with the homogeneous wave case
Additional script for matching tabular

wave forces to target pontoons
Fig. 3. Flowchart of the proposed methodology.
4.2 Arbitrary inhomogeneous wave cases

This study aims to introduce an advanced framework for examining the structural behaviors of a floating bridge
under inhomogeneous wave conditions using OrcaFlex, rather than considering the appropriate design wave con-
ditions of Bjørnafjord. Referring to the 100-year period inhomogeneous wave conditions given by Dai et al., we
created several arbitrary inhomogeneous wave cases as shown in Fig. 4. Similar to the conventional representation
of wave conditions, the radial axis indicates the frequency range, the angles indicate wave heading, and the spec-
tral density of the wave is presented as a contour plot. For each wave case, only the distribution of wave headings
varied, while the peak period was set to 5.9 seconds and the significant wave height was 2.4 meters. The phase
angles of the wave components were random but fully correlated, and non-directional spreading was configured.
Fig. 4. Arbitrary inhomogeneous wave cases in the conventional representation.
Figure 5 shows the standard deviation of the wave-induced moment at each node of the girder, indicating the
expected dynamic response during the simulation of three inhomogeneous wave cases (IHWC) introduced in Fig.
4 and two homogeneous wave cases (HWC). Directional inhomogeneity was revealed to have a critical effect on
bending moments in the vertical direction. Based on local wave headings, the minor bending moment in the girders
near the south end increased 2 to 4 times in inhomogeneous wave cases. This is due to the pontoon's hydrodynamic
shape, which increases pitch moments and minor girder bending as the wave heading shifts from head sea to beam
sea. This is supported by the fact that the homogeneous wave cases serve as the upper and lower bounds in the
bending moment diagrams. On the other hand, the directional inhomogeneity showed insignificant variance in the
major bending moments and torsional moments of the girders.
On
ly
5
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 5. Wave-induced moments of the girders under the inhomogeneous and homogeneous wave cases.
4.3 Variation in the mooring configuration
Fig. 6. Wave-induced moments of the girders for different mooring configurations.
The current mooring configuration is categorized as a semi-taut type, stretching outwards from the moored pon-
toon. This system provides additional horizontal and vertical stiffness to the moored pontoons while offering
minimal restoring capacity for pitch motion. In the range of wave periods that the floating bridge is examined, the
dominant vibration mode is rotation with respect to the longitudinal axis. Considering the practical limit states of
joints and structural components, two different mooring configurations were investigated under inhomogeneous
wave case 3 as an example. One mooring type consists of a combination of semi-taut and tension legs, while the
other consists solely of tension legs. As shown in Fig. 6, the modified mooring types showed improved resilience
for the pitch of the moored pontoons, reducing wave-induced torsional moments in the center girders by 45%.
On
ly
6
fo
rR
However, the use of solely tension leg moorings increased the major bending moments in the girders by 20%
overall due to the reduced horizontal mooring stiffness. In contrast, the combination of mooring types improved
ea
resilience to torsion while maintaining the capacity for horizontal responses.
din
g
5 Concluding remarks
at
This study aims to introduce an advanced framework to examine the structural behaviors of floating bridges under
W
inhomogeneous wave conditions using OrcaFlex. A straight floating pontoon bridge model was reproduced by
CF
referring to previous works and validated based on the natural periods to capture the behavioral characteristics of
the target model. The methodology for integrating inhomogeneous wave conditions into OrcaFlex by utilizing the
S
script environment was proposed. Using this approach, arbitrary inhomogeneous wave cases were simulated for
the reproduced model, revealing that directional inhomogeneity has a critical effect on the vertical responses of
20
the girders due to the hydrodynamic characteristics of the pontoon. Finally, different mooring configurations were
24
simulated under inhomogeneous wave conditions, indicating that the model has a greater capacity to improve
resilience to torsion and optimize the design.
Acknowledgements
This work was supported by a Korea Agency for Infrastructure Technology Advancement (KAIA) grant funded
by the Ministry of Land, Infrastructure, and Transport (Grant RS-2023-00250727) through the Korea Floating
Infrastructure Research Center at Seoul National University.
References
1. Jang, M., Lee, Y., Won, D., Kang, YJ., Kim, S.: Static Behaviors of a Long-span Cable-Stayed Bridge with a Floating
Tower under Dead Loads. Journal of Marine Science and Engineering, 8(10), 816 (2020).
2. Jang, M., Lee, Y., Kim, S., Kang, YJ.: Dynamic behavior of a long-span cable-stayed bridge with floating towers after
the sudden failure of tethers and cables under irregular waves. Journal of Marine Science and Technology, 28(6), 593-
601 (2020).
3. Won, D., Lee, K., Kang, YJ., Kim, S.: Short-term Fatigue Damage of Tethers of Long-span Floating Cable Supported
Bridges under Harsh Waves. Journal of Marine Science and Technology, 28(6), 602-609 (2020).
4. Cheng, Z., Svangstu, E., Gao, Z., Moan, T.: Field measurements of inhomogeneous wave conditions. Journal of Water-
way, Port, Coastal, and Ocean Engineering, 145, 05018008 (2019).
5. Cheng, Z., Gao, Z., Moan, T.: Numerical modeling and dynamic analysis of a floating bridge subjected to wind, wave,
and current loads. Journal of Offshore Mechanics and Arctic Engineering, 141, 011601 (2019).
6. Cheng, Z., Svangstu, E., Moan, T., Gao, Z.: Assessment of inhomogeneity in environmental conditions in a Norwegian
fjord for design of floating bridges. Ocean Engineering, 220, 108474 (2021).
7. Cheng, Z., Gao, Z., Moan, T.: Hydrodynamic load modeling and analysis of a floating bridge in homogeneous wave
conditions. Marine Structures, 59, 122–141 (2018).
8. Cheng, Z., Gao, Z., Moan, T.: Wave load effect analysis of a floating bridge in a fjord considering inhomogeneous wave
conditions. Engineering Structures, 163, 197–214 (2018).
9. Dai, J., Stefanakos, C., Leira, B.J., Alsos, H.S.: Effect of modelling inhomogeneous wave conditions on structural re-
sponses of a very long floating bridge. Journal of Marine Science and Engineering, 9(5), 548 (2021).
10. Cui, M., Cheng, Z., Moan, T.: Effects of inhomogeneous wave modeling on extreme responses of a very long floating
bridge. Applied Ocean Research, 134, 103505 (2023).
11. Dai, J., Leira, B.J., Moan, T. and Kvittem, M.I.: Inhomogeneous wave load effects on a long, straight and side-anchored
floating pontoon bridge. Marine Structures, 72, 102763 (2020).
12. Viuff, T., Ravinthrakumar, S., Økland, O. D., Gryta, O. A., Xiang, X.: Model test of a hydroelastic truncated floating
bridge with a stay-cable tower. Applied Ocean Research, 135, 103539 (2023).
13. Viuff, T., Xiang, X., Øiseth, O., Leira, B.J.: Model uncertainty assessment for wave- and current-induced global response
of a curved floating pontoon bridge. Applied Ocean Research, 105, 102368 (2020).
14. Cheng, Z., Gao, Z., Moan, T.: Extreme responses and associated uncertainties for a long end-anchored floating bridge.
Engineering Structures, 219, 110858 (2020).
15. Viuff, T., Xiang, X., Leira, B.J., Øiseth, O.: Software-to-software comparison of end-anchored floating bridge global
analysis. Journal of Bridge Engineering, 25 (5), 04020022 (2020).
16. Rodrigues, J. M., Viuff, T., Økland, O. D.: Model tests of a hydroelastic truncated floating bridge. Applied Ocean Re-
search, 125, 103247 (2022).
17. Multiconsult: SBJ-31-C3-MUL-22-RE-100-0 – analysis and design (base case). Multiconsult AS (2017).
On
ly
fo
rR
Dynamic behavior of an operational floating offshore wind turbine
ea
subjected to ship collision
din
g
Xudong Ye1[0009-0002-6108-9523] and Kaiming Bi1[0000-0002-5702-6119]
1
at
The Hong Kong Polytechnic University, Hung Hom, Kowloon, 999077, Hong Kong, China
W
kaiming.bi@polyu.edu.hk
CF
S
Abstract. This paper investigates the dynamic responses of a NREL 5 MW spar-type floating offshore wind
20
turbine (FOWT) under ship collisions. Physics-based finite element models (FEMs) are established by LS-
DYNA. Aerodynamics of FOWT are introduced using the steady wind field simulation results. While hydro-
24
dynamics of floaters are considered by the MCOL method. The collision forces of ship and vibration frequen-
cies of FOWT match well with the reference values, indicating the rationality of the developed model. Four
wind velocities are adopted to represent different operating status for the FOWT. The collision process can be
divided into primary and secondary collisions. The maximum force of the secondary collision only has 60%
of the primary one. The results show the collision forces with 0 m/s and 7 m/s wind speeds are larger than
other cases. This is because the wind thrust delays the collision process through changing the pitch angle of
the FOWT. Also, it is observed that the indentation of the spar has the same regular pattern since it is related
to the collision force. The secondary collision has almost no effect on the indentation.
Keywords: Floating Offshore Wind Turbine, Ship Collision, Operational Condition, Dynamic Response.
1 Introduction
With the development of the offshore wind power industry and the recovery of the global marine economy, the
frequency of ship collisions with offshore wind turbines is increasing. Recently, on April 25, 2023, the Gode Wind
1 wind farm in Germany was struck by a cargo ship, resulting in a large hole on the starboard bow [1]. On Sep-
tember 19, 2024, at the Hornsea One offshore wind farm in the UK, a service operation vessel (SOV) collided
with an operating wind turbine, causing minor damage to the starboard side of the vessel and the turbine founda-
tion [2]. These incidents indicate the necessity of predicting the ship-FOWT collision process and consequences.
Many scholars have conducted research on ship collisions with fixed wind turbines. However, research on
collisions involving floating wind turbines remains relatively superficial. Ren studied the influence of different
impact velocities and hull rigidity on platform damage, without considering the wind-wave coupling effect on the
turbine [3]. Zhang proposed an aero-hydro coupling method to simulate hydrodynamics during ship-FOWT col-
lisions, but his research focused on the rigidity of the colliding object, with fewer scenarios related to the opera-
tional state of the turbine [4].
Existing literature indicates a lack of analysis on the operational floating wind turbines. Therefore, this paper
develops a FOWT model that considers both aerodynamic and hydrodynamic characteristics, considering opera-
tional states of wind turbines. Chapter 2 introduces the FE model and validates its effectiveness. Chapter 3 presents
the collision scenarios and discusses the simulation results.
2 Validation of FE models
2.1 Modeling of ship and FOWT

This paper adopts a 3000 DWT ship to collide with the NREL 5MW spar-type floating offshore wind turbine.
Fig. 1 shows the FE model established in LS-DYNA. For the ship bow, the thickness of the steel plates varies
from 6 mm to 12 mm. The material of the steel is simulated using the keyword
MAT_PIECEWISE_LINEAR_PLASTICITY, with parameters identical to [5].
Details of the FOWT model can be found in [6]. The impacted area of the spar has a diameter of 6.5 m and a
thickness of 40 mm. The blades are simulated using beam elements and can rotate around the hub. The FOWT
uses the same material model as the ship.
On
ly
2
fo
rR
Hydrodynamic characteristics of the ship and the FOWT are considered using the MCOL method while the
viscous damping and mooring system by custom functions. Rigid body dynamics are investigated to balance the
ea
hydrodynamic forces and other forces acting on the ship [7]:

din
M S  M A  y  G  y  FW  FH  FV  FC (1)
g
where 𝑀 is the mass matrix of ship; 𝑀 is the added mass matrix of ship; 𝑦 is the velocity of the ship; 𝐺 is the
at
gyroscopic matrix; 𝐹 is the wave force; 𝐹 is the hydrostatic restoring force; 𝐹 is the viscous force; 𝐹 is the
W
collision force.
CF
Incorporating aerodynamics, a series of wind forces obtained from the steady wind field simulation in FAST
are applied to each node of the blades. The contact between the ship and the FOWT is managed by the keyword
S
CONTACT_AUTOMATIC_SURFACE_TO_SURFACE, with a friction coefficient of 0.3.
20
(a) Global model (b) Ship bow structure
24
0.6m spacing
for frames
1.56m spacing
for keels
(c) Impact site details

FOWT
0.05m mesh size for ship
Ship
0.2m mesh size for spar
Fig. 1. Ship and FOWT models.
2.2 Mesh convergence analysis

According to [8], the ratio of plate thickness to mesh size should be between 1:5 and 1:10. The most suitable mesh
size is 40-80 mm. Fig. 2 compares the indentation of the bow when it collides with a rigid wall under different
mesh sizes. It can be seen that the results are similar for mesh sizes of 25 mm and 50 mm. Balancing computational
accuracy and efficiency, 50 mm is finally selected as the mesh size for the bow. Similarly, the mesh size for the
impacted part of the FOWT is determined to be 200 mm.
Fig. 2. Displacement of bow indentation under different mesh sizes.
2.3 Verification of FOWT’s maneuverability
To validate the effectiveness of the aerodynamic and hydrodynamic simulation methods, the results of LS-DYNA
and FAST models are compared. Table 1 lists the natural frequencies of the FOWT. Good consistency shows the
effectiveness of the MCOL method. Fig. 3 shows the platform surge displacements under different wind velocities,
On
ly
3
fo
rR
and the results obtained by LS-DYNA are very close to those of FAST. This indicates that the developed model
can effectively simulate aerodynamic and hydrodynamic characteristics of FOWT.
ea
din
Table 1. Natural frequencies of FOWT.
g
Motion mode LS-DYNA result (Hz) FAST result (Hz)
at
Surge (Sway) 0.0083 0.0076
Heave 0.032 0.032
Roll (Pitch) 0.036
W 0.032
Yaw 0.12
CF0.12
S
20
24
Fig. 3. Comparison of platform surge motion between LS-DYNA and FAST.
3 Discussion of ship-FOWT collision results
3.1 Collision scenarios

Four collision scenarios are considered with respect to different wind velocities: 0 m/s (no wind), 7 m/s (low
speed), 11.4 m/s (rated speed), and 23 m/s (high speed). Under the 11.4 m/s condition, the blades experience the
maximum wind thrust because of the optimal pitch angle. The moving velocity of the ship is set to 5 m/s, and the
wind turbine reaches a stable state before the collision.
3.2 Collision process
Fig. 4 shows the collision forces between the ship and the FOWT under different wind velocities. It can be seen
that multiple collisions occurred between the ship and the FOWT. The primary collision lasted about 2 seconds,
with the collision force being the greatest under no wind condition and the smallest at the rated speed (11.4 m/s).
This is because there is a large pitch angle of the FOWT, resulting in reducing the severity of the collision process.
This initial pitch angle is caused by the maximum wind thrust. The secondary collision occurred about 5 seconds
after the first one and lasted longer. Its peak force is 60% of the highest value.
Fig. 4. Comparison of collision force under different scenarios.
3.3 Damage assessment

Fig. 5 shows the indentation of the spar during the collision. It can be observed that the spar undergoes significant
plastic deformation within the initial 0.2 seconds, followed by some rebound as the collision force decreases. Due
to the collision forces of primary collision under 0 m/s and 7 m/s conditions are higher, the indentations are 0.05
On
ly
4
fo
rR
m greater than those of other conditions. Besides, the secondary collision will only cause a small fluctuation on
the indentation.
ea
din
g
at
W
CF
S
20
24
Fig. 5. Comparison of spar’s indentation under different scenarios.
4 Conclusion
This paper developed physical-based finite element models to clarify ship-FOWT collision process and the influ-
ence of the operating status. Aerodynamics and hydrodynamics are considered by MCOL method and custom
functions. The developed models are validated by comparing the results between LS-DYNA and FAST. Collision
scenarios with four different wind velocities are simulated. Primary and secondary collisions are observed while
the latter one only has longer duration but lower collision force compared to the primary one. The results show
that cases with lower wind velocities have higher collision forces and spar’s indentations. Also, it should be no-
ticed that the secondary collision has fewer influences on the indentation.
References
1. offshoreWIND.biz, https://www.offshorewind.biz/2023/05/30/cargo-ship-hit-gode-wind-turbine-went-back-into-ser-
vice-in-24-hours-vessel-said-to-have-been-kilometres-off-course, last accessed 2023/05/30.
2. offshoreWIND.biz, https://www.offshorewind.biz/2024/09/23/service-ship-hits-wind-turbine-at-hornsea-one-offshore-
wind-farm, last accessed 2024/09/23.
3. Ren, Y. L., Meng, Q. S., Chen, C., Hua, X. G., Zhang, Z. L., Chen, Z. Q.: Dynamic behavior and damage analysis of a
spar-type floating offshore wind turbine under ship collision. Engineering Structures, 272, 114815 (2022).
4. Zhang, Y. C., Hu, Z. Q.: An aero-hydro coupled method for investigating ship collision against a floating offshore wind
turbine. Marine Structures, 83, 103177 (2022).
5. Fan, W., Zhang, Z. W., Huang, X., Sun, W. B.: A simplified method to efficiently design steel fenders subjected to vessel
head-on collisions. Marine Structures, 74, 102840 (2020).
6. Jonkman, J., Butterfield, S., Musial, W., Scott, G.: Definition of a 5-MW reference wind turbine for offshore system
development. National Renewable Energy Lab (NREL), Golden, CO (United States) (2009).
7. Le Sourne, H., Donner, R., Besnier, F., Ferry, M.: External dynamics of ship-submarine collision. Preprints of 2nd Inter-
national Conference on Collision and Grounding of Ships, pp. 137-144. Copenhagen (2001).
8. Alsos, H. S., Amdahl, J., Hopperstad, O. S.: On the resistance to penetration of stiffened plates, Part II: Numerical analysis.
International Journal of Impact Engineering, 36, 875-887 (2009).
On
ly
fo
rR
Effect of equivalent design wave selection on the stress prediction of FOWT
ea
Semi-submersible hull
din
g
Lei Yang1 and Binbin Li1
1Shenzhen
at
International Graduate School, Tsinghua University, Shenzhen, China
W
libinbin@sz.tsinghua.edu.cn
CF
S
Abstract. Equivalent design wave (EDW) and spectral analysis are two methods to predict the stress in the
20
Semi-submersible hull of floating offshore wind turbine (FOWT). Compared with spectral analysis, EDW save
the cost of finite element calculation, while its accuracy depends on the selection of global response and the
24
method used to select EDWs. To investigate the effect of equivalent design wave selection on the stress pre-
diction, EDW analysis with different selection of governing loads and spectral analysis are conducted for
FOWT Semi-submersible (Semi-sub) hull. In this study, a typical 5 MW Semi-sub FOWT is selected and the
finite element (FE) model of hull is constructed. Hydro-structure interaction analysis of the hull is conducted,
with the hydrodynamic pressure recalculated at structural points, and different pressure components separately
transferred from the hydrodynamic to structural model. The frequency, amplitude, wave heading of EDWs are
selected according to the long-term spectral analysis results of the global response. The results indicate that in
practical EDW analysis, in addition to selecting internal load and inertial load as governing loads according to
the specifications, the pressure, velocity, and hydrodynamic coefficients should also be considered to enhance
the accuracy of EDW predictions. Most of the EDWs are concentrated around a specific frequency and head-
ing, which is related to structural type and geometric dimension. These findings will provide a reference for
the selection of design waves and the design of geometric dimension.
Keywords: Equivalent design wave; Hydro-structural interaction analysis; Long-term spectral analysis;
Global structural response.
1 Introduction
Wind energy has been recognized as a key climate change mitigation technology, which has become increas-
ingly cost-effective and available [1]. In the offshore market, 10.8 GW of new offshore wind was commissioned
in 2023, bringing total global offshore wind capacity to 75.2 GW [1]. As the most prominent offshore wind
technology, FOWTs (floating offshore wind turbine) are considered to have significant incremental potential,
with an installed capacity of only 66.4 MW in 2022, projected to reach 16.5 GW by 2030 [2]. However, high
costs are the main challenge limiting the development of FOWTs, with their costs reaching up to five times
those of fixed offshore wind turbines [3]. In the cost breakdown of FOWT, the substructure and hull are pre-
dicted to account for 37.5% of capital expenditures, making it the highest proportion of all costs [4]. Therefore,
cost-effective design of the hull is meaningful for cost reduction in FOWT.
A cost-effective design involves performing finite element (FE) analysis on the hull. However, most research
on FOWTs has primarily concentrated on their motion response, leaving the structural response relatively unex-
plored. Although recent studies have begun to address the structural response characteristics of FOWT hulls [5,
6], these efforts only focused on the impacts of corrosion, structural elasticity, and material properties on the
structural response. Lack of understanding about the sensitive wave parameter results in conservative structural
design of the floating foundation.
There are two design approaches to conduct structural design for floating foundation, which are spectral anal-
ysis and equivalent design wave (EDW). The spectral analysis approach requires performing spectral analysis on
each element individually, which is usually quite time-consuming. The EDW approach is widely used in the in-
dustry, it significantly reduces the computational cost, and allows to derive simplified formula for hydrodynamic
loading [7].
However, the accuracy of the EDW approach is determined by selection of governing loads. To the best of the
authors' knowledge, the structural response and governing loads of FOWT Semi-sub hull is unclear, which is
difficult to obtain accurate results using EDW approach. So far, the governing loads recommended by standards
for EDW analysis mostly pertain to quadrilateral-shaped Semi-submersible platforms. For triangular-shaped plat-
forms, the governing loads are not well understood. It does inspire the investigation of governing loads for FOWT
Semi-sub hull, which will provide a reference for EDW selection. This study explores the effect of governing
loads selection on the EDW results. To validate the accuracy of EDW results, full spectral analysis is also
On
ly
2
fo
rR
conducted as for comparison purpose. The theory and methodology of linear quasi-static hydro-structural inter-
action analysis, short-term and long-term spectral analysis, and EDW analysis are introduced in Section 2. In
ea
Section 3, OC4-DeepCwind FOWT is selected as a case study to conduct linear quasi-static hydro-structure in-
din
teraction analysis. Extreme von Mises stress of each element is predicted using von Mises stress response ampli-
tude operator (RAO) and the wave scatter diagrams of typical coastal area of Guangdong Province using spectral
g
analysis and EDW analysis. The comparison of extreme von Mises stress between full spectral analysis and EDW
at
analysis are presented in Section 4. The governing loads for Semi-sub hull and the effect of different governing
loads selection on the EDW results are also discussed. Finally, conclusions are drawn in Section 5.
W
2 Theory and methodology
CF
S
20
2.1 Linear quasi-static hydro-structure interaction analysis
24
Concerning hydro-structure issues, fully consistent non-linear hydro-structure calculations are not practically pos-
sible within the reasonable combination of CPU time and accuracy [8]. Therefore, different levels of approxima-
tion methodology are proposed to simplify the hydro-structure issue. This study adopts the linear quasi-static
hydro-structure interaction analysis, which is the most general and efficient method to predict the extreme von
Mises stress. Linear seakeeping analysis and boundary value problems are the basis of hydro-structure interaction
analysis, which are explained in many studies [9, 10]. In order to transfer the hydrodynamic pressure in implicitly
balanced model, the potential at each point in the structural model can be recalculated according to the hydrody-
namic point.
𝜑 𝑥 ∬ 𝜎 𝑥 𝐺 𝑥 ; 𝑥 𝑑𝑆 (1)
Where 𝑥 denotes the structural point and 𝑥 denotes the hydrodynamic point. 𝐺 and 𝜎 are Green's function
and unknown source strength, respectively. In this way, the hydrodynamic pressure at the structural model are
recalculated and different pressure components are separately transferred from hydrodynamic to structural model
[8]. The proposed transferring method can obtain implicit balanced model compared with the interpolation of
hydrodynamic pressure. To improve the efficiency of computation, Gaussian Quadrature integration is used to
integrate the pressure over the structural mesh to obtain the hydrodynamic coefficients. The motion response in
the frequency domain can be obtained by
𝜔 𝑀 𝐴 𝑖𝜔 𝐵 𝐶 𝜉 𝐹 (2)
Where 𝐹 , 𝐵 , 𝐶 are the incident and diffracted forces, added mass coefficients, and radiation damp-
ing coefficients, respectively. These hydrodynamic coefficients can be found by integrating the pressure over the
structural mesh. The inertial force of each element and the hydrodynamic pressure of wet surface can be obtained
according to motion response and acceleration. The real and imaginary components of the nodal force correspond-
ing to each frequency and wave heading are implemented to the FE model for quasi-static analysis. Therefore, the
von Mises stress RAO are obtained to conduct spectral analysis and EDW analysis.
2.2 Short- and long-term spectral analysis

A long-term spectral analysis is performed using wave scatter diagrams and von Mises stress RAOs to predict
extreme stress. This prediction is then utilized for strength verification and to investigate the distribution of ex-
treme stress. The long-term extreme stress distribution can be obtained by cumulating the results from the short-
term analysis. In the short-term analysis, the variable of stress is considered to follow a Rayleigh distribution,
which is stationary. The parameter of Rayleigh distribution has a relationship with the spectral moment of the
structural response, which can be found by
𝑚 𝜔 𝑅𝐴𝑂 𝜔, 𝛽 𝑆 𝜔 𝑑𝜔 (3)
Where 𝛽 , 𝑅𝐴𝑂 𝜔, 𝛽 , 𝑆 𝜔 are the selected wave direction, RAO of selected wave frequency and wave
heading, given wave spectrum, respectively. The short-term extreme stress can be predicted by Rayleigh distribu-
tion with a specific probability.
In the long-term spectral analysis, the scatter diagram provides the probability of occurrence for each state
𝑝𝑟𝑜𝑏 𝑆𝑆 . The cycles of each sea state can be calculated by the total cycles and the probability. The number of
cycles exceeding 𝑋 in each sea state can be written as follows:
On
ly
3
fo
𝑛 𝑋
rR ∑ 𝑛 1 𝑃 𝑋 (4)
ea
where 𝑁 and 𝑃 𝑋 is the total number of sea states and the probability density function in specific sea state,
din
respectively. Battjes wave counting method is used to obtain the long-term cumulative distribution of von Mises
stress. The long-term extreme value of structural response can be determined by a specified probability level.
g
According to the reference [11], the probability in calculating the long-term extreme value is taken as a probability
at
of 50-year return period.
For each sea state in wave scatter diagrams, all wave directions are summed to have the same probability.
W
Therefore, wave directions can be divided into several segments at regular intervals. The distribution of long-term
CF
extreme structural response in each wave direction can also be calculated. Combined with the specific probability
level, the contribution of each wave direction to the long-term extreme value can be obtained. The maximum
S
contribution of wave direction provides a reference to select the wave direction of EDW.
2.3 Equivalent design wave analysis 20

24
EDW analysis is conducted according to results of long-term spectral analysis, as mentioned in Section 2.2. The
EDW is chosen as the wave direction that makes the maximum contribution to the long-term extreme value [12].
Under the wave direction that makes the maximum contribution, the frequency contributing the most to the struc-
tural response RAO is selected to be the wave frequency of EDW. The amplitude and phase of EDW can be
calculated by
𝐴 | , |
(5)
𝜙 𝑅𝐴𝑂 𝜔, 𝛽 (6)
Where 𝜔 and 𝛽 are the selected wave frequency and direction of EDW, respectively. EDW condition can generate
the response stress equivalent to the long-term predictions values of stresses. EDW is chosen based on the long-
term spectral analysis of the governing loads, which is suggested by several specifications, such as ABS and DNV
specification.
3 Numerical modeling
Among various types of Semi-sub hulls, the OC4-DeepCwind is the most extensively studied and representative
[13]. Therefore, the classic OC4-DeepCwind Semi-sub hull of FOWT is selected as a case study. The Semi-sub
hull is modeled using a combination of beam and shell elements: shell elements for the column shell and bulkhead,
and beam elements for the stiffeners and pontoons. Three nodes are constrained, with two nodes restricted in Y
and Z translations, and one node restricted in X and Z translations. The rotor nacelle assembly and tower are
simplified into a single concentrated mass point, defined using a 6x6 mass matrix. The water ballast is modeled
as a fully filled liquid tank, without accounting for the sloshing effect. The structural model of the FOWT Semi-
sub hull is presented in Fig. 1(a).
4 Governing loads of FOWT Semi-sub hull
4.1 Comparison of the extreme von Mises stress using spectral analysis and EDW analysis
In order to investigate the effect of EDW selection on the predicted wave-induced von Mises stress of FOWT
Semi-sub hull, the comparison of extreme von Mises stress using different governing load selection are conducted.
To validate the EDW results, the long-term predicted value based on spectral analysis is also calculated as a
reference. In light of the governing loads of quadrilateral-shaped Semi-sub hull, this study uses similar internal
loads such as split force, shear force, torsion moment, and inertia loads induced by topside acceleration. Fig.1 (b)
illustrates the internal loads and inertial loads of the FOWT Semi-sub hull. It is worth noting that FOWT has three
internal force sections, which are equivalent. For simplicity, only one of these internal forces of section is shown
in Fig.1 (b). In the EDW analysis, the internal forces of all three sections will be considered. In addition to these
internal loads and inertial loads, the global velocity response, hydrodynamic pressure near still water level, and
hydrodynamic coefficients are also selected.
Fig.2 presents the comparison of maximum von Mises stress using different EDW selections, which is validated
by full spectral analysis. The scatter plot of acceleration of topside EDW condition most closely aligns with 𝑦
On
ly
4
fo
rR
𝑥 line, which indicates that the topside acceleration RAO best reflects the variation trend of the RAO of stresses
in most elements. Therefore, the inertial force induced by topside acceleration has the greatest impact on the von
ea
Mises stress, and should be the governing load for FOWT Semi-sub hull.
din
g
at
W
CF
S
20
24
(a) Structural model of FOWT Semi-sub hull. (b) Illustration of the internal loads and inertial loads
Fig. 1. Structural model of FOWT Semi-sub hull.
Fig. 2. The comparison of maximum stress between EDW condition and full spectral analysis.
Table 1. Mean error, RMS error, and the number of elements with an error less than 5% for various EDW
conditions.
Number of ele-
Number of ele-
Items Mean error RMS error ments with an error
ments (𝑦 𝑥)
less than 5%
Split force -22.72% 0.0788 21646 6242
Shear force -21.64% 0.0801 28958 10863
Torsion -23.97% 0.0886 16975 4923
Bending moment -23.90% 0.0841 15828 2185
Column moment -20.51% 0.0738 30238 20371
Topside acceleration 3.08% 0.2198 21719 48175
Pressure 88.22% 3.2810 4027 95793
Velocity -19.61% 0.0958 7241 20209
Hydrodynamic coefficients -15.71% 0.0402 24476 13834
On
ly
5
fo
All governing loads
rR15.60% 0.2395 66220 75497
ea
To quantitatively analyze the impact of different governing loads on the EDW analysis results, Table 1 lists the
mean error, root mean square (RMS) error, and the number of elements with an error less than 5% for various
din
EDW conditions. According to Table 1, the mean error of the topside acceleration is smallest, indicating that
selecting these EDW conditions yields a better alignment with the spectral analysis results. In addition, the mean
g
error of characteristic loads such as the internal loads and global motion are consistently less than zero, which
at
means that these EDW condition underestimate the extreme von Mises stress. The RMS errors of the internal
loads such as split force, shear force, torsion moment, bending moment, and column moment are the smallest,
W
which also indicates that selecting these EDWs enables better prediction of extreme von Mises stress. Pressure
CF
EDW condition is worth noting that the predicted errors are relatively large, but most of the predicted results are
higher than those from spectral analysis. Therefore, choosing these EDW conditions leans towards conservative
S
considerations. The global velocity and hydrodynamic coefficients EDW condition show relatively good perfor-
20
mance, with errors similar to the global structural response. Therefore, it is recommended to use these EDW
conditions in the designing process.
24
In general, none of the governing load can accurately predict extreme von Mises stress for all elements using
EDW analysis. In practical EDW analysis, achieving better predictive results often involves combining different
governing loads. This entails selecting the maximum values from various EDW conditions to predict extreme von
Mises stress.
4.2 Equivalent design wave analysis results

Table 2 presents the EDW analysis results of different governing load. The results show that the EDW amplitudes
mostly range between 4 m and 4.5 m, with the largest EDW amplitude being 4.827 m, obtained from the hydro-
dynamic coefficients EDW condition. However, the maximum EDW amplitude does not imply the maximum von
Mises stress, and the maximum von Mises stress occurs under the topside acceleration EDW condition. Most of
the EDW frequencies are around 0.80 rad/s. According to deep-water wave theory, this frequency corresponds to
a wavelength approximately twice the height of the triangular configuration. Therefore, the sensitive frequency
of von Mises stress is often related to the main dimensions of the structure. Reasonable main dimension design
can be used to avoid the frequencies where wave energy is concentrated. Each governing load determines a dif-
ferent EDW direction, but most of them are near the three principal directions (0°, 120°, 240°), corresponding to
the directions along the three pontoons.
Table 2. EDW analysis results of different governing load.
EDW direc- EDW fre- EDW amplitude Maximum von

Governing loads
tion(°) quency(rad/s) (m) Mises stress (Mpa)
0 0.800 4.277 213.20
Split force 330 0.800 4.274 259.35
30 0.800 4.275 308.33
230 0.825 4.295 248.95
Shear force 110 0.80 4.306 278.76
250 0.80 4.305 275.15
130 0.825 4.283 264.69
Torsion 110 0.800 4.115 266.43
250 0.800 4.105 262.35
0 0.800 4.084 203.58
Bending moment 150 0.775 4.302 298.46
210 0.775 4.296 280.37
Column moment 110 0.800 4.591 297.24
Topside acceleration 270 0.725 4.734 313.96
Pressure 80 0.375 3.492 156.18
Velocity 270 0.675 4.816 281.69
Hydrodynamic coefficients 80 0.700 4.827 298.88
Considering EDW analysis for all governing loads, the results are presented in Table 1 and Figure 2 (b). It can
be observed that a better predictive performance is achieved by considering all governing loads.
On
ly
6
fo
rR
ea
din
g
at
W
CF
S
20
(a) Illustration of sensitive wave headings 24
(b) Comparison of maximum stress using spectral analysis
and EDW analysis considering all governing loads.
Fig. 3. Illustration of sensitive wave headings and the comparison of maximum stress using spectral analysis and EDW anal-
ysis considering all governing loads.
5 Conclusion
This study conducts spectral analysis and equivalent EDW analysis to predict the extreme von Mises stress of
FOWT Semi-sub hull. The extreme von Mises stress obtained from the spectral analysis are compared with that
from EDW analysis, which selected by different governing loads. The local wave-induced stress are obtained by
adopting the linear quasi-static hydro-structure interaction analysis. A novel load transferring method is adopted,
which involves in the recalculation of hydrodynamic pressure at the structural model and separately transferring
of different pressure components from hydrodynamic to structural model. The frequency, amplitude, wave head-
ing of EDWs are selected according to the long-term spectral analysis results of the global response. The main
conclusion are drawn as follow:
 Internal load and inertial load are the governing load for the FOWT Semi-sub hull, including the split force,
shear force, torsion moment, and topside acceleration and so forth.
 In addition to internal load and inertial load, the pressure, velocity, and hydrodynamic coefficients should also
be selected as governing loads in the practical EDW analysis. Velocity and hydrodynamic coefficients EDW
condition achieve good performance in predicting the maximum extreme stress. Pressure EDW condition often
overestimates extreme stress values, which leans towards conservative considerations.
 Most of the EDW frequencies are concentrated around a specific frequency, corresponding to a wavelength
approximately twice the height of the structural triangular configuration. EDW headings are almost near the
three principal directions, corresponding to the directions along the three pontoons.
However, this study only considers regular EDW selection methods during the EDW selection process. More
advanced EDW selection methods, such as irregular EDW selection, remain to be considered.
Reference
1. Jonathan, C., Ben, B., Girish, T., Mohamed, J.A.R.: Global Wind Report 2024. Global Wind Energy Council: Brussels,
Belgium (2023).
2. Rebecca, W., Alvaro, M.: Global offshore wind report 2023. Global Wind Energy Council: Brussels, Belgium 19, pp.10-
12 (2023).
3. Remi, E.: ENERGY TRANSITION OUTLOOK 2023. Det Norske Veritas: Høvik, Norway (2023).
4. Stehly, T., Duffy, P.: 2021 cost of wind energy review, National Renewable Energy Lab.(NREL), Golden, CO United
States (2022).
5. Gao, S., Zhang, L., Shi, W., Wang, W., Wang, B., Li, X.: Dynamic Analysis of a 10 MW Floating Offshore Wind Turbine
Considering the Tower and Platform Flexibility. Journal of Ocean University of China, 23(2), 358-370 (2024).
6. Wang, S., Moan, T., Gao, Z.: Methodology for global structural load effect analysis of the semi-submersible hull of
floating wind turbines under still water, wind, and wave loads. Marine Structures, 91. (2023).
7. De Hauteclocque, G., Derbanne, Q., El-Gharbaoui, A.: Comparison of different equivalent design waves with spectral
analysis. International Conference on Offshore Mechanics and Arctic Engineering (2012).
On
ly
7
fo
rR
8. Malenica, Š., Derbanne, Q., Sireta, F. X., Bigot, F., Tiphine, E., De Hauteclocque, G., Chen, X. B.: HOMER–Integrated
hydro-structure interactions tool for naval and offshore applications. International Conference on Computers Application
in Shipbuilding (2013).
ea
din
9. Chen, X. B.: Hydrodynamics in offshore and naval applications-Part I. Keynote lecture of 6th Intl. Conf. HydroDynamics,
Perth Australia (2004).
g
10. Lee, C. H., Newman, J. N.: Computation Of Wave Eﬀects Using The Panel Method. WIT Transactions on State-of-the-
art in Science and Engineering 18 (2005).
at
11. Igba, J., Alemzadeh, K., Durugbo, C., Henningsen, K.: Through-life engineering services: a wind turbine perspective. Pro-
W
cedia CIRP 22: 213-218 (2014).
12. Liu, Y., Wu, J., Zhu, W., Li, P.: Research on Equivalent Design Waves of Container Vessel Based on Stress Spectrum
CF
Analysis. The 28th International Ocean and Polar Engineering Conference (2018).
13. Gueydon, S.: Effects of Variations on the Experimental Set-up on the Motion Response of a Floating wind Semisubmers-
S
ible (OC4 Type). International Conference on Offshore Mechanics and Arctic Engineering. Vol. 59353. American Society
20
of Mechanical Engineers (2019).
24
On
ly
fo
rR
Studies on Integrating a Carbon Capture System with Hydrogen Fuel
ea
Production in FLNG Power Generation
din
g
We Lin Chan1[0009-0008-0906-5130], Dr Arun Dev2[0000-0002-8955-2519] and Dr Ivan CK Tam3[0000-0002-9111-9749]
1,2,3 Newcastle
at
University in Singapore, Newcastle University,
W
Newcastle upon Tyne NE1 7RU, UK.
CF
W.L.Chan3@newcastle.ac.uk
S
20
Abstract. The burning of hydrocarbon fuels releases greenhouse gases (GHGs) that are linked to climate
change and pose a health risk to people. To offset the marine industry's transition to renewable energy fuel,
24
reduced emissions are going to be required. [1]
LNG fossil gas was seen as the most environmentally friendly option due to its lower carbon emissions com-
pared to other fuels. The increasing cost and demand for LNG have facilitated the exploration of inland gas
deposits. [2]
However, most gas reservoirs are situated in offshore areas, which encourages the growing number of FLNG,
which stands for Floating Liquefied Natural Gas, widely recognised and utilized in both nearshore and off-
shore,
The initial phase of the clean energy trial initiative for FLNG involves meeting the power generation needs
aboard through the utilization of clean energy sources. Adopting a clean energy mindset is crucial for maxim-
izing the sustainable use of hydrocarbon fuel.
The main reason for choosing hydrogens is preferred over other energy sources is because it has a better en-
ergy-to-mass ratio (H/C ratio). The preferred method for hydrogen production is the methane reforming pro-
cess, which is highly efficient in generating large quantities of hydrogen.
The reforming method is determined by the choice of either steam or oxygen as the reactant for natural gas.
This choice leads to three possible processes: steam methane reforming (SMR), partial oxidation (POX), or
autothermal reforming (ATR). [3]
Moreover, restricted space availability which design SMR or ATR will be implemented to enhance the small-
scale plant with CCS as an onboard FLNG.
In conjunction with CCS (Carbon Capturing System), research on the integration of extra Clean Conversion
modules to produce green hydrogen fuel is presented in this study. The goal is to help provide clean fuel for
FLNG's power generation, which has been engineered for offshore conditions and emits no carbon.
Keywords: Carbon Capture System (CCS); Hydrogen; Reforming; FLNG
1 Introduction
1.1 FLNG Power Generation with Green Fuel

To satisfy the FLNG process energy demand, the FLNG completely depends on hydrocarbon fuels. This leads to
growing carbon emissions. To cope with the trend of increasing carbon emissions, it is even more important to
use fossil hydrocarbon fuels sustainably. Research idea of using NG/LNG fuel which has a higher hydrogen/car-
bon ratio than the other fossil fuels to produce hydrogen onboard FLNG. Using hydrogen as a fuel onboard FLNG
can improve power generation efficiency and lower carbon emissions. This is positive for the environment because
of growing concern of carbon emissions. Hydrogen fuel is likely to be favored in the future. Additionally, hydro-
gen is a very environmentally friendly energy carrier. There are numerous ways to produce hydrogen, and it has
potential applications across all end-use sectors. [4]
Hydrogen as Marine Fuel. Hydrogen (H2) is well-known for its adaptability as a clean energy source and for its
potential to completely transform as fuel energy for electricity production. Now, fossil fuels account for 95% of
H2 production due to its affordability and technological maturity, with the SMR process is a fossil-based H2 pro-
duction technique that is additionally cost-effective. As a result, it accounts for 75% of global H2 production. The
lower heating value per mass of this substance is significantly higher than that of methane, gasoline, and coal,
with values that are 2.4, 2.8, and 4 times greater, respectively. The combustion of hydrogen does not result in the
production of CO2 or, in the case of low temperature fuel cells, nitrous oxide, sulphur dioxide, or carbon monoxide.
[5] [6]
On
ly
2
fo
2 Methodology
rR
2.1 Reforming Method
ea
din
The most significant approach for producing hydrogen at a high level is through the process of methane re-
g
forming. The reforming process takes place when natural gas undergoes a reaction with either steam or oxygen.
at
This chemical process results in the production of hydrogen gas. The selection of reaction types dictates the ap-
propriate process model to apply. There are several choices available to enhance research. The methods used for
W
the above reason are Steam Methane Reforming (SMR), Partial Oxidation (POX), and Autothermal reforming
CF
(ATR). The distinction between the SMR and ATR processes lies in the method by which heat is introduced into
the system boundary.
S
The SMR reforming process is a reaction that produces heat. The system relies on external combustion heat for
20
its operation (exothermic reaction). The process of ART reforming is a reaction that absorbs heat. The system
boundary necessitates the presence of internal combustion heat (endothermic reaction). The research analysis will
24
provide a concise explanation of the advantageous methodologies for hydrogen production with different carbon
capture systems through the utilization of natural gas. Moreover, the main challenges for FLNG are the constraints
of space and economic factors. To effectively minimize carbon emissions, it is imperative to choose either the
SMR or ATR design for the small-scale plant on board FLNG. The research scope was defined as the Clean
Conversion module. It is employed to produce Hydrogen for the purpose of generating power in FLNG. [7] [8]
2.2 Steam Methane Reforming Method
Steam methane reforming (SMR) method of hydrogen production from fossil fuel is one of the most exciting
processes in the marine industry. SMR is used in around 40% of the world's hydrogen production It includes
various oxygen transfer materials or oxidants. This reforming technique is a catalytic process with a catalyst-
packed reactor. Methane is reacted with water vapor at temperatures of 700 to 850°C. This is the primary process
of steam reforming. [3]
The reaction product equation of Reform 1 and Meth 1 is syngas, which is a mixture of hydrogen and carbon
monoxide. Syngas contains approximately 12% carbon monoxide. This carbon monoxide can be converted to
more hydrogen by the Water-Gas Shift (WGS) as shown in equation Shift 1 as equation (3). In this step, carbon
monoxide reacts with water vapour over a shift catalyst to produce. Once more, WGS improves hydrogen yields.
In summary, the SMR process is endothermic in thermodynamic terms. It requires or absorbs thermal energy
from its surroundings, usually in the form of heat. The energy input is the limiting factor to maintain the required
reaction temperature in the SMR process. [8]
Reform 1 CH4 + H2O = CO + 3H2 (Conversion) (1)

Meth 1 CH4 + CO2 = 2CO + 2H2 (Conversion) (2)
Shift 1 CO + H2O = CO2 + H2 (Equilibrium) (3)
All cases contain High Temperature Shift (HTS) and Low Temperature Shift (LTS) processes. Using both
processes increased the hydrogen yield. This ensures a higher hydrogen concentration in the gas. More carbon
monoxide reacted with steam, creating hydrogen. Carbon monoxide is very harmful to the atmosphere. Carbon
monoxide reacts with oxygen. This process prevents methane gas from being emitted directly into the atmosphere.
It also reduces the risk of carbon emissions. However, both HTS and LTS ensure a higher hydrogen concentration
in the gas. This may lead to a smaller hydrogen purification unit. It may also help establish a carbon capturing
system. After partial combustion of methane, the gas enters the water-gas shift step. In this stage, carbon monoxide
reacts with steam to create carbon dioxide and hydrogen. The last step will be hydrogen purification.
2.3 Autothermal reforming (ATR) process

The autothermal reforming process is a combination of the Steam Methane Reforming process (SMR). In ther-
modynamic terms, exothermic is a process or reaction that releases energy from the system to its surroundings,
usually in the form of heat. The total reaction is exothermic. Inside the ATR, natural gas reacts with steam to
provide syngas, as in the SMR process, additionally, parts of the fuel react with oxygen.
There are the pre-reformer and main reformer, Steam supplies the process before the gas enters the reformer
stage, there will be production of syngas occurs in the reformer, where hydrocarbons react with steam and oxygen
and create syngas with help from catalysts. [8]
On
ly
3
fo
rR
The chemical equilibriums in the reformer stage are the following reaction equations in the ATR process.
Partial 1
eaCH4 + ½ O2 = CO + 2H2 (Equilibrium) (4)
Reform 1
din
CH4 + H2O = CO + 3H2 (Equilibrium) (5)
g
Shift 1 CO + H2O = CO2 + H2 (Equilibrium) (6)
at
W
3 Theory and Equations
3.1 Steam to Carbon (S/C) Ratio CF

S
20
Steam to carbon S/C ratio is usually used to analyse the steam reforming in both SMR and ATR processes to
optimise the system. The S/C ratio is based on the ratio between the mole of steam and the carbon at the inlet of
24
the system to evaluate the feeding amount of steam to the reformer. If the S/C ratio is lower, will produce the
coke slug formation, the feeding of the steam is important to avoid the thermal cracking of hydrocarbon. If the
S/C ratio increases, the temperature of the reform outlet will be lower, indirectly, reformer consumption energy
will be lower. The design steam to carbon is typically 2.5 with a range between 1.0 and 5.0. To prevent the carbon
deposit and coke deposit formation, the ratio is to be kept equal or above 2.5, or higher. [5] [9]
3.2 Other Thermodynamic variables
Product Yields. The chemical equilibrium analysis of the methane reforming reaction is analysed to predict the
product yields during the reaction. The chemical equilibrium of reactions can be calculated by the CEA Program
which calculates the product concentrations from any set of reactants based on the minimization of Gibbs free
energy and the equation as below, [10] [11]
,,
𝑖, 𝑦𝑖𝑒𝑙𝑑 (7)
,
Where i is the product and j is the feed gas, n is the total moles of species
The product yields of carbon monoxide, carbon dioxide and hydrogen as per the equation
,
CO yield (8)
,
4 Simulation and Model Analysis
Aspen HYSYS simulation software is utilized to run simulations of the hydrogen manufacturing and CCS process.
These include the Small Modular Reactor (SMR) and the Advanced Technology Reactor (ATR). When perform-
ing process modeling, create the LNG systems as steady-state models. Use the "Peng-Robinson" fluid package.
This will facilitate the evaluation and improvement of the underlying design concepts of LNG systems. This study
focused mostly on optimizing and modeling the Steam Methane Reforming (SMR) and Autothermal Reforming
Process (ATR) with the CCS system. The specified flow rate of natural gas is 214.2 kgmole/h for ratio 30% H2
fuel usage for 50MW gas turbine. The pressure decrease in the heat exchanger is expected to be 10 Kpa in the
cooler and 35 Kpa in the reactor. The SMR and ATR process is set up with an air-to-fuel ratio of 2.5. The tem-
perature of the reforming product depends on the amount of oxygen injected. The simulated processes will prior-
itize the production of hydrogen entirely. For example, in specific models, the boundary will align with the dis-
charge of the LT_WGS reactor. There will be a specific level of carbon dioxide (CO2) emissions, and the primary
emphasis of this study on the Carbon Capture System. [7] [12] [11] [13]
Model Development.
The primary determining factors in the reforming process are the steam-to-carbon ratio (S/C ratio) and the oxygen-
to-carbon ratio. In the SMR and ATR process, an excessive intake of oxygen will lead to a decrease in hydrogen
production and an increase in temperature. To regulate the temperature ranges of the feed gas stream, a pre-
reformer, four reactors, a heater, Amine and a chiller will be utilized. The research report will primarily focus on
two models: ATR, and SMR Amine process.
On
ly
4
fo
rR
During the simulation, the various stages of heating and cooling are carried out to achieve specific temperature
ranges. These stages include preheating the feed natural gas from a temperature more than 400°C to a temperature
ea
greater than 700°C, the water gas shift process from a temperature greater than 900°C to a temperature greater
din
than 300°C, and the conversion process at about a temperature greater than 250°C. These processes result in the
production of the desired product. [6]
g
4.1 SMR Process with MEA Amine CCS
at
W
This study applies the control methodology to examine two significant processes and determine the optimal con-
CF
trolled variables for maximizing hydrogen production and minimizing carbon emissions. The techniques under
analysis involve the transformation of natural gas into hydrogen through the SMR process, followed by the capture
S
of CO2 after undergoing treatment in both HT WGS (High-Temperature Water Gas Shift) and LT WGS (Low-
20
Temperature Water Gas Shift) reactors. The capture of CO2 is an essential component in the generation of green
hydrogen from hydrocarbons. Monoethanolamine (MEA) Amine absorption/stripping processes have been used
24
to eliminate CO2. Fig 3 illustrates a carbon dioxide (CO2) capture system stationed downstream of the conversion
of fossil fuels to hydrogen.
Fig.1. SMR Process with Amine CCS Flow Diagram
The MEA CCS methods consist of two columns. In the initial column, referred to as the absorber, carbon
dioxide (CO2) is taken in by a solution of low-concentration amine solvent through rapid chemical reactions. The
end-product solution, referred to as rich amine, is next sent to the second column, often known as the stripper,
where the CO2 component is eliminated. Stripping necessitates a substantial quantity of energy, which is supplied
through the transfer of heat from a recuperative heat exchanger via the cross flow of rich amine and lean amine
(see Fig.1). This energy conservation represents approximately 15-30% of the whole electricity produced by the
facility. The power challenge is essential for efficiently operating this process on board the FLNG to maximize
energy savings in a cost-effective manner. Therefore, the primary goal of this thesis is to create a simple control
system that can efficiently handle on-board operations and operate the process as close to the optimal level as
possible. The study concentrated on the dynamic modeling of the stripper and examined two separate configura-
tions to enhance energy efficiency on board FLNG with a significant electrical requirement. [14]
The authors developed a dynamic model specifically for the absorber and analyzed the effects of a load change
in power generation, from 100% to 50%. An integrated system for hydrogen production and carbon capture and
storage (CCS) that incorporates a stripper and recycle process. The paper has examined how changing the per-
centage of CO2 removal and the concentration of amine (MEA) affect the efficiency of the process and the effec-
tiveness of CO2 capture. Moreover, the consequences of reducing the capacity of hydrogen production have been
investigated.
4.2 ATR Process with DPEG CCS

In comparison to SMR, the ATR is a system that reduce size and complexity. There is no external heat supply
will reduce the fuel costs for the reformer and result in a decrease in CO2 emissions. Another benefit is that the
ATR requires less water than an SMR, increased methane conversion. oxygen contributes to reduce the soot
On
ly
5
fo
rR
formation. Nonetheless, the temperature is constrained by material and construction expenses. It is also presumed
that there is no heat loss in the system, and the compressor's adiabatic efficiency is 75%. ATR reactors are modeled
ea
as Gibbs reactors, which minimize Gibbs free energy. This model designed with as a component splitter is fur-
din
thermore utilized for modeling water separation and H2 production Refinement to demonstrate the potential for
varied separation procedures. CCS process used DEPG and used separators to get separation to be easy step out
g
to get CO2 off (see Fig.2).
at
W
CF
S
20
24
Fig.2. ATR Process with Amine CCS Flow Diagram
5 Result Discussion and Conclusion
5.1 Simulation Result Analysis

Molar Change of Components vs Temperature. The effect of temperature on reforming leads to enhancements in
the process due to changes in the mole fraction of the components. As a result, based on the S/C ratio of 2.5,
temperature change at the process stream point and mole change of mixing components, are plotted as per graph,
(see in Fig.3), Molar changes in terms of temperature variation are presented for all process models. The optimal
operating temperature is to be evaluated to maximise based on the generating of hydrogen.
Fig.3 Molar change of S/C=2.5 vs temperature for SMR_CCS and ATR_CCS process
The results are dependent on the thermodynamic analysis to compare the SMR and ATR Amine processes, the
molar composition change for hydrogen across the outlet stream points are monitored with graph (see in Fig.3).
Both systems are well performing except to consider for floating FPSO space limit that ATR is more considerable
than SMR. Furthermore, simplifications and assumptions made throughout the simulation process in the way of
four reforming processes will be compared under optimum conditions for each process.
On
ly
6
fo
rR
This should all be incorporated with the overall evaluation based on the carbon capture system analysis. The
main process design to see the H2 and CO2 molar flow rate variable directly as carbon capture dashboard which
ea
shown the sour gas and fuel gas mole percentage composition, lean and rich amine loading, inlet and outlet CO2
din
loading, CO2 capture percentage, Net GHG emission and carbon fee as below Table 1. The most important output
of both simulation is the CO2 capture and carbon fee which affects the overall system performance across all
g
processes. To get an efficient design, have to evaluate the temperature which influences the mole fraction and
at
exergy of the reformer, and also separation process in absorber and stripper. Related to the flow is minimise to
meet the hydrogen fuel requirement for the power generation turbine only, in this approach throughout the system
W
will determine to get a smaller and efficient design for FLNG to meet the concern of economy and unit sizing, to
CF
suit the FLNG's space constraint. The review of the various analyses of hydrogen production from all simulation
models is required to identify the performance characteristics of FLNG applications which include energy and
S
exergy efficiency, space constraints, and carbon emissions. The result conclusion is based on the model of the
20
process both in same the environment with all conditions and assumptions. Further extending the investigation to
incorporate more variables can bring one closer to identifying an optimization.
24
Table 1. Simulated carbon capture comparison dashboard for SMR SE Process
Sr Description CCS_SMR(MEA) CCS_ATR (DEPG)

1 Sour Gas CO2 composition (mole %) 15.83 11.21
2 Fuel Gas CO2 composition (mole%) 1.270e-003 3.161
3 Lean CO2 Loading 2.722e-002 3.234e-002
4 Rich CO2 Loading 0.2573 5.479e-002
5 Inlet CO2 (Kg/hr) 7997.1472 8441.4143
6 Outlet CO2 (Kg/hr) 7996.7551 7216.6676
7 CO2 Capture (%) 100 85.49
8 Heat Utility Emissions (Kg/hr) 1208
9 Power Utility Emissions (Kg/hr) 2.934e+004 424.5
10 Inlet Emissions (Kg/hr) 7.216e+004 7.216e+004
11 Outlet Emissions (Kg/hr) 1.308e+004 8777
12 Net GHG Emission (Kg/hr) 1.004e+006 7.233e+005
13 Carbon Fee ($) 5.119e+007 8.680e+008
*According to US SCC regulatory uses an SCC of around $51 per ton
Regarding CO2 capture, SMR with MEA Amine are the best alternatives. SMR Amine is the most efficient
process for producing hydrogen based on hydrogen yields, followed by ATR with DEPG amine. As per the sim-
ulation result, and the comparison graph (see Fig.3), SMR is slightly efficient, however ATR is one feasible so-
lution for FLNG applications in terms of further research to reduce the equipment. The ATR process is the com-
bination of SMR, which requires to control of both oxygen and steam. The oxygen-to-fuel ratio drives carbon
emission, hence, oxygen to fuel ratio used in the simulation will affect to reduce the carbon emission. The reformer
temperature is substantially greater in the ATR reactor than in the SMR reformer because of the absence of steam
and soot or coke formation which can be controlled by the scrubber after the reformer outlet. [14]
5.2 Conclusion
The key objective of this study is to explore the green energy potential of hydrogen fuel as a substitute for con-
ventional fuels through a reforming process, addressing the energy requirements of floating platforms. The com-
prehensive examination of the variance in molar change, efficiency, and carbon capture ratio will determine which
system is preferable. The ATR system is better in terms of carbon fee savings and equipment size reduction;
however, the optimal choice is SMR MEA Amine, as determined by the molar change analysis conducted for
hydrogen production at each stage. To obtain acceptable study findings, it is essential to understand all assump-
tions and simplifications applied in the interpretation of the data. It is recommended to widen the system bound-
aries to provide a more detailed assessment.
On
ly
7
fo
rR
Future system optimization studies may involve reconfiguring the reformer with an enhanced stream subse-
quent to the reactor, so increasing the volumetric flow rate within the reforming reactor; nevertheless, the dimen-
ea
sions of the reformer must be evaluated concerning the space and cost constraints of the FLNG. The expense of
din
hydrogen production will be affected by numerous energy-related variables. This indicates that, following assess-
ment, the SMR and ATR methods for producing hydrogen are more efficient for FLNG.
g
at
W
References
CF
S
[1] IMO, "Greenhouse Gas Emissions. Retrieved from IMO:," Apr 2018. [Online]. Available:
20
http://www.imo.org/en/OurWork/Environment/PollutionPrevention/AirPollution/Pages/GHG-
Emissions.aspx.
[2] GIIGNL, "RETAIL LNG HANDBOOK," 2015.
24[Online]. Available:
https://giignl.org/sites/default/files/PUBLIC_AREA/Publications/giignl_retail_lng_handbook_0.pdf.
[3] A.-M. B.-M. T. M. G. T. Alicia Boyano, "Steam Methane Reforming System for Hydrogen
Production," 10 Feb 2012. [Online]. Available: https://dergipark.org.tr/en/download/article-file/65813.
[4] A. Angus Campbell, Interviewee, LNG Fuel is not a solution for reducing GHGS (Marine & Industrial
_ Magazine Report). [Interview]. Nov/Dec 2016.
[5] R. Gupta, Hydrogen Fuel: Production, Transport, and Storage, Boca Raton: CRC Press, 2008.
[6] H. (. Europe, "Hydrogen Prouction. Retrieved from Hydrogen Europe:," [Online]. Available:
https://hydrogeneurope.eu/index.php/hydrogen-production-0.
[7] A. T. HYSYS®2004.2; Operations Guide, @aspentech, 2004.
[8] L. P. Antonio Vita, "Steam Reforming, Partial Oxidation, and Autothermal Reforming of Ethanol for
Hydrogen Production in Conventional Reactors.," Science Direct, Hydrogen Production:, Aug 31 2018.
[Online]. Available: https://www.sciencedirect.com/topics/chem.
[9] M. Gillespie, "Hyper for hydrogen: our world first for carbon-free fuel. Retrieved from CSIROscope:,"
Nov 2018. [Online]. Available: https://blog.csiro.au/hyper-for-hydrogen-our-world-first-carbon-free-
fuel/.
[10] N. Glenn Research Center, "Chemical Equalibrium with Applications," [Online]. Available:
https://www1.grc.nasa.gov/research-and-engineering/ceaweb/.
[11] A. B. Tam ICK, "Thermal Systems - An Overview. Energies," p. pg. 175., 2020, 14(1),.
[12] V. H. H. A. Silberova B, "Production of hydrogen by short contact time partial oxidation and oxidative
steam reforming of propane.," Journal of Catalysis, pp. 99(1-2): 69-76., 2005.
[13] I. C. T. A. D. WE Lin Chan, "Clean Energy Hypothesis for Future Marine and Offshore," Research
Gate, no. 378684136, 2023.
[14] C. S. a. V. S. K. Liu, Hydrogen and Syngas Production and Purification, USA: Wiley, 2010.
On
ly
fo
rR
Effect of end boundary conditions on the structural response of the
ea
submerged floating tunnel
din
g
Kiwon Jeong 1, Hyunwoo Shin1, Jinwoo Kim1, Seungjun Kim1*
1
at
School of Civil, Environmental and Architectural Engineering, Korea University, Seoul 02841, South Korea
W
rocksmell@korea.ac.kr
CF
S
Abstract. This study focuses on the practical application of submerged floating tunnels (SFTs), emphasizing
20
the need to effectively resist environmental loads while maintaining serviceability and safety. A comprehen-
sive analysis involving various boundary conditions was conducted to understand the behavior of SFTs. The
24
structural responses of SFTs were analyzed through hydrodynamics-based numerical methods, specifically
using ABAQUS-AQUA. The study investigated the dynamic motions and moment distributions under differ-
ent mooring types and boundary conditions. It was found that double-inclined mooring (Type3) displayed the
most stable displacement and the least moment in the Free condition, providing valuable insights for the fun-
damental design of SFTs.
Keywords: Submerged floating tunnel, Tether, Boundary conditions, Hydrodynamics, ABAQUS
1 Introduction
The submerged floating tunnel (SFT) is an innovative underwater transportation infrastructure, suitable for use in
deep waters. The SFT, constructed on land and then installed underwater, simplifies construction and reduces
costs and time. Structural responses are mainly induced by wave loads, affecting both serviceability and stability.
Therefore, appropriate mooring lines and cross-sectional designs are required to sufficiently suppress wave loads
and ensure stability. Studies have been actively conducted to rigorously calculate the structural response of SFTs
to environmental loads and to develop simple yet reasonable methods to estimate this response.
Studies by Jeong et al. (2022), Won et al. (2022), and Østlid et al. (2010) have primarily focused on the impact
of external loads on the tunnel, examining various mooring methods and tunnel cross-sections [1-3]. However,
most research has been limited to a single combination of boundary conditions and mooring types. The dynamic
behavior of SFTs varies complexly depending on combination of boundary conditions at both ends and mooring
methods. More comprehensive research that considers various combination of boundary conditions and mooring
methods is required for more rational design.
2 Analysis method to obtain structural responses of SFTs

2.1 Mathematical formulation
A coordinate system and variables are established as shown in Fig. 1, and the tunnel is divided into a segments
according to the spacing of the mooring lines. Using Hamilton's principle, it's possible to obtain the equations of
motion without considering force balance in the free-body diagram, through repetitive calculations.
Fig. 1. Coordinate system and variables
According to Hamilton's principle, the following Eqn. (1) is established.
𝛿𝑇 𝛿𝑉 𝛿𝑊 𝑑𝑡 0 1
On
ly
2
fo
rR
Where, 𝑇, 𝑉 and 𝑊 is the kinetic energy, potential energy, and energy due to non-conservative forces. There-
fore, by solving the above variational equation, the following Eqns. (2-4) of motion can be obtained.
ea
𝐸𝐼 din 𝐹 (2)
g
𝐸𝐼 𝐹
at 𝑤 𝐵 (3)
𝐺𝐽 W
CF
(4)
S
Here, 𝑚 and 𝐼 mean the self-weight and moment of inertia of the nth tunnel. Additionally, 𝐸 is the modulus of
elasticity, 𝐺 is the shear modulus, 𝐼 and 𝐼 are the second moments of area about the z and y axes, respectively, 𝐽
20
is the polar moment of inertia of the section, 𝑚 is the mass per unit length, 𝐼 is the moment of inertia per unit
24
length, 𝐿 is the length of the nth tunnel, 𝑤 is the self-weight per unit length, and 𝐵 is the buoyancy force per unit
length, 𝐹 , 𝐹 is the horizontal and vertical force components.
The objective of this study is to determine the structural response of tunnels under irregular wave conditions in
actual sea areas. Irregular waves are simulated by defining the energy spectrum using the Joint North Sea Wave
Project (JONSWAP) wave spectrum. In ocean engineering, potential theory is typically used to calculate wave
forces on floating bodies. These potential calculations generally require numerical solutions based on integral
equations. However, it has been shown in numerous studies [4-8] that for cylindrical objects like SFTs, the Mori-
son equation can yield results almost like potential theory in many cases. Therefore, in this study, 𝐹 , 𝐹 are cal-
culated by using Morison equation.
2.2 Hydrodynamics-based numerical analysis method

The equations of motion for the SFTs given by Eqns. (2-4) are nonlinear differential equations, thus necessitating
numerical methods for their solution. Commercial finite element analysis programs like ABAQUS or Orcaflex
have been utilized by many researchers. Particularly, ABAQUS has been used by various research groups glob-
ally, as evidenced in studies like Min et al. 2022 and Xiang et al. 2018 [6,9], and is recognized as a reliable analysis
program. Previous research by Min et al. 2022, which utilized ABAQUS-AQUA to obtain the structural response
of SFTs and compared it with the experimental results from Oh et al. 2013 [6, 10], demonstrated considerable
reliability.
The tunnel is defined as a beam element and modeled as B31 elements, while mooring lines are set as truss
elements, using T3D2 elements. The analysis of the SFTs is conducted in two steps: the initial buoyancy-induced
static bending deformation in Step 1, and Step 2 performs dynamic analysis, obtaining the dynamic response to
waves over 3800 seconds. This dynamic analysis employs the direct time integration method.
3 Structural responses under various mooring and boundary conditions
3.1 Environmental conditions and model specifications

In this study, the Messina Strait in Italy was selected as the target area. To analyze the structural responses of the
SFTs under extreme environmental conditions, 100-year return period irregular waves, representative of extreme
conditions in this area, was considered [11]. Fig.2. illustrates the time series of wave heights impacting the SFTs
over 3800 seconds. The overall environmental conditions and specifications of the tunnel and mooring line are
shown in Table. 1 below. Also, Fig.3 shows the side view of numerical model and front view of the SFT.
Fig. 2. Wave elevation

On
ly
3
fo
rR
ea
din
g
at
W
CF
S
Fig. 3. Numerical analysis model & environment condition
20
Table 1. Tunnel specifications and environmental conditions.
Parameters
1. Environmental condition
Water depth (m)
Values
80.0
24
Significant wave height (m) 13.5
peak wave period (s) 11.5
Draft (m) 40.0
2. For tunnel
Diameter (m) 23.0
Thickness (m) 1
Elastic modulus (GPa) 25.0
Buoyancy–weight Ratio 1.3
Drag/added mass coefficient 1.0/1.0
3. Tether (Mooring)
Diameter (m) 0.95
Thickness (m) 0.04
Elastic modulus (GPa) 210.0
Outside tether inclination angle (degree) 45.0
Tether spacing (m) 40.0
The three mooring types and four boundary conditions are considered to investigate the structural behavior
under different mooring types and boundary conditions. The selected mooring type is shown in Fig. 4. The bound-
ary conditions include free ends on both sides, fixed ends on both sides, hinged ends on both sides, and a combi-
nation of hinged and fixed ends. Transportation infrastructure, such as bridges, typically relies on fixed and hinged
supports as boundary conditions. Fixed supports offer high stability and resistance, though they also induce sig-
nificant bending moments. Hinged supports, on the other hand, provide structural flexibility and efficiently dis-
tribute loads but can result in substantial bending deformations. In this study, we also considered free boundary
conditions to account for the specific characteristics of submerged floating tunnels (SFTs), which are heavily
influenced by mooring lines. This approach effectively reduces the bending moments on the tunnel but may in-
crease the load exerted on the mooring lines.
Fig. 4. Various mooring types

On
ly
4
fo
3.2
rR
Structural responses of the SFTs based on various boundary conditions
ea
din
g
at
W
CF
S
20
(a) Type 1 Sway 24
(b) Type 2 Sway
(c) Type 3 Sway (d) Type 1 Heave
(e) Type 2 Heave (f) Type 3 Heave
Fig. 5. Displacement distribution at peak wave elevation. (460.5s)
In Fig. 5, the sway and heave of the SFT at peak wave height were examined. Under free boundary conditions,
the sway and heave appeared constant for all types, while under hinge and fixed boundary conditions, they exhib-
ited curved forms. This indicates that rigid body motion dominates the total displacement under free conditions,
whereas bending deformation dominates under the other conditions. Additionally, Fig. 5(d) shows the setdown
effect of heave due to large bending sway in the middle position.
On
ly
5
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 6. Comparison of maximum displacement.
In Fig.6, Types 1 and 2, the maximum sway and heave were smaller under conditions where the tunnel end’s
rotation was restricted, such as fixed or hinge-fix. Type 3 exhibited the smallest absolute displacement magnitude
compared to the other types, and the displacements under free boundary conditions were like those under other
boundary conditions. This is attributed to the tether arrangement suppressing the roll of the tunnel.
Fig. 7. Comparison of maximum bending moment
The maximum moment of the SFT tunnel was examined through Fig. 7. For all types, the free boundary con-
dition results in a significantly smaller bending moment compared to the other constrained boundary conditions
due to the dominance of rigid body motion. Considering both Figs. 6 and 7, Type 3 with free boundary condition
appears to be the most reasonable combination because it best suppresses displacement and generates minimal
moments.
On
ly
6
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 8. Maximum tether stress comparison
In addition to the displacement and moment of the tunnel, it is also important to examine the stress occurring
in the mooring lines. Fig. 8 compares the maximum stress for each type and boundary condition against the moor-
ing line design strength specified by the API for extreme condition. Consequently, the tether stress for Type 3
under the free boundary condition, which minimizes displacement and moment, meets the API strength criteria.
4 Conclusion
In this study, the structural behavior of a SFT was investigated under various combinations of boundary conditions
and mooring types to achieve a rational design. The focus was on comparing the displacement, bending moment
of the tunnel and the maximum stress of the tethers, which are directly related to serviceability and safety. The
key results are as follows:
1. For Types 1 and 2, the smallest displacement was observed under boundary conditions with rotational
constraints, whereas Type 3 exhibited the smallest displacement regardless of the boundary condition.
This is due to the tether arrangement effectively restraining roll motion.
2. In terms of bending moment, the free condition resulted in the smallest values for all types. This is be-
cause the free condition induces rigid body motion.
3. From the perspective of displacement and moment, the Type 3 free condition is the most rational
choice. Additionally, under this condition, the stress in the mooring lines was very low, satisfying the
API strength criteria.
References
1. Jeong K, Min S, Jang M, Won D, Kim S.: Feasibility study of submerged floating tunnels with vertical and inclined
combined tethers. Ocean Eng, 2022;265:112587.
2. Won D, Seo J, Kim S.: Dynamic response of submerged floating tunnels with dual sections under irregular waves. Ocean
Eng, 2021;241:110025.
3. Østlid, H.: When is SFT competitive? Procedia engineering, first international symposium on Archimedes Bridge (ISAB-
2010), Vol. 4. Amsterdam: Elsevier; 2010, p. 3–11.
4. Cifuentes S, Kim S, Kim MH, Park WS.: Numerical simulation of the coupled dynamic response of a submerged floating
tunnel with mooring lines in regular waves, Ocean Syst Eng, 2015;5(2):109–23.
5. Seo S, Mun H, Lee J, Kim J.: Simplified analysis for estimation of the behavior of a submerged floating tunnel in waves
and experimental verification. Mar Struct, 2015;44:142–58.
6. Min S, Jeong K, Noh Y, Won D, Kim S.: Damage detection for tethers of submerged floating tunnels based on convolu-
tional neural networks. Ocean Eng, 2022;250:111048.
7. Min, S., Jeong, K., Lee, Y., Kim, S., 2023.: Estimation of unmeasured structural responses of submerged floating tunnels
using pattern model trained via long short-term memory. Ocean Eng, 277, 114284.
8. Jeong, K.; Kim, S.: Structural response of submerged floating tunnels with free-end boundary condition based on an
analytical approach. Appl. Ocean. Res, 2024, 143, 103861.
9. Y. Xiang, Z. Chen, Y. Yang, H. Lin, S., Zhu.: Dynamic response analysis for submerged floating tunnel with anchorcables
subjected to sudden cable breakage, Marine Structures, vol. 59, pp. 179–191, 2018.
10. Oh, S.H., Park, W.S., Jang, S.C., Kim, D.H.: 2013. Investigation on the behavioral and hydrodynamic characteristics of
submerged floating tunnel based on regular wave experiments. KSCE J. Civ. Eng. 33 (5), 1887–1895 (in Korean).
11. Faggiano B, Panduro J, Mendoza Rosas MY, Mazzolani FM.: The conceptual design of a roadway SFT in Baja California,
Mexico. Procedia Eng, 2016;166:3–12.
On
ly
fo
rR
Use of Floating Structures as A Coastal Protection Solution – Challenges
ea
and Opportunities
din
Cheng Chen1 and Xudong Qian1
g
1 Department of Civil and Environmental Engineering, National University of Singapore
at
qianxudong@nus.edu.sg
W
CF
Abstract. According to the latest report by Intergovernmental Panel on Climate Change (IPCC), the global
S
mean sea level will rise by about 0.28 m to 0.98 m by 2100. Under high emission scenarios, IPCC estimates
that about 1.9 million square kilometers of land area would be submerged by 2100. In view of these threats
20
caused by the climate change and sea level rise, this paper presents an overview of using floating structures as
24
a potential solution to protect the coastal region, covering both the challenges to be addressed and opportunities
to tap on the multifunctionality and flexibility of floating structures. The challenges in using the floating struc-
tures as a coastal protection measure include the survivability and durability under the harsh environments,
wave attenuation performance under extreme conditions, integration with renewable energy harvesting, as well
as long term performance monitoring and optimization of the life-cycle costs. The opportunities and ad-
vantages of the floating structures entail the adaptability to sea level rise, flexibility in location, multipurpose
functionality and modularity and scalability. The paper aims to summarize the critical research gaps in using
the floating structures as a coastal protection measure, for the development of a sustainable and resilient coastal
protection strategy.
Keywords: Climate change; floating structures; sea level rise.
1 Introduction
Climate change and sea level rise have posed significant threats to the population in the coastal region. According
to a recent study in Nature Communications [1], up to 630 million people will be living on land below projected
annual flood levels by the end of this century. Coastal protection measures face challenges due to their limited
adaptability against climate change uncertainties. With the escalation of global warming, conventional coastal
protection structures like seawalls or dikes might become insufficient as the sea level continues to increase beyond
initial estimations. Innovative engineering solutions are therefore necessary to provide more resilient and adaptive
coastal protection.
2 Floating Structures as Multifunctional Solutions
The quest for innovative coastal protection strategies in coastal regions has highlighted the potential of floating
structures. With rising sea levels posing a threat to coastal areas, floating structures offer a resilient alternative for
residential, leisure and commercial activities. A Dutch architecture firm has used cross-laminated timber to con-
struct an off-grid office building that floats on water [2], as shown in Fig. 1(a). The building is constructed on a
pontoon of 15 concrete barges. The concrete barges not only keep the building afloat, but also contain pipe systems
that serve as heat exchangers, using the temperature of the harbor water to cool the interior in summer and warm
it in winter. The solar panels on the roof provide the building with its own source of electricity, minimizing the
carbon footprint of the floating office. Several companies have developed floating real estimate products on water
(villas, hotels, etc.) for commercial resale in the United Arab Emirates [3-4], including the world’s largest floating
villa shown in Fig. 1(b).
Among all the floating alternatives developed worldwide for coastal defense, those that have found best ac-
ceptance in recent years are the more environmentally friendly designs. The use of submerged structures has been
a common trend in the floating solutions, to provide habitats for marine life and minimize the visual impact to
marine environment. Dubai will be constructing a submerged floating lido as one of the luxury underwater vessel
resorts [5]. Norway is planning on building the world’s first “submerged floating tunnel” to replace the ferries
between the cities of Kristiansand in the south and Trondheim in the north [6]. Figure 1(c) illustrates the sub-
merged roadway design, which is essentially a pair of concrete tubes submerged about 30 meters below the water
surface. The floating tunnel helps to cut the original travel time almost in half while reducing the environmental
On
ly
2
fo
rR
impact on the area. China, South Korea and Italy are also researching on similar projects to replace conventional
ea
bridges and tunnels [7-9]. The global race on such innovation engineering solutions is intensifying due to the
anticipated climate change.
din
The wave transmitted to a floating structure is reduced because of the reflected or dissipated incident energy.
g
Therefore, the energy converters are integrated with floating structures to harvest energy from wave, tidal and
at
current. Recent designs of wave energy converter focus on optimizing the layout and construction to enhance the
energy conversion efficiency and structural robustness against harsh marine environment [10-12]. In addition to
W
water-related energy, floating structures also provide platforms to harvest other renewable energy sources such as
CF
wind and solar energy, taking advantage of the vast open water surfaces where wind speeds are higher and more
consistent than on land. China has recently launched a floating platform (Fig. 1d) that supports the world’s biggest
S
wind turbines [13]. The platform is assembled at the Huangchuan Wharf in Guangzhou and installed off the coast
20
of Qingzhou, Yangjiang in Guangdong Province. The twin-rotor floating platform has an innovative V-shaped
tower, to which two turbine rotors are affixed. This provides additional yawing power and enhance the production
24
efficiency while reducing the ultimate load. The twin-rotor design facilitates the structural resistance against the
extreme typhoon conditions. While for the conventional energy resources, designs such as the floating nuclear
power plants [14] support the oil and gas exploration activities and offer energy to islands and regions that lack
access to traditional power grids.
Fig. 1. (a) Floating office in Netherland [2]; (b) floating villa in UAE [3]; (c) submerged floating tunnel (to be constructed)
[6]; (d) floating wind turbine in China [13]; (e) floating city in Maldives (under construction) [15]; and (f) sea-scraper (con-
ceptual design) [17].
The concept of floating cities offers future potentials to coastal protection and environmental sustainability. Float-
ing cities are comprehensive urban designs that integrate resourcing, living, leisure, and other essential services.
They provide a sustainable, resilient, and innovative solution to the challenges of urbanization and climate change.
Maldives and Korea have proposed floating city projects to adapt the rising sea levels. Since the Maldives islands
are predicted to be uninhabitable by 2100 due to rising sea levels, the government of the Maldives has launched
the Maldives Floating City project, as indicated in Fig. 1(e). Designed in a pattern similar to the brain coral, the
Maldives Floating City consists of 5,000 floating units including houses, restaurants, shops and schools, with
canals running in between. It is able to house up to 20,000 residents within a 200-hectare lagoon in the Indian
Ocean. The project is currently under construction and set to be completed by 2027 [15]. Korea unveils the pro-
totype for a resilient and sustainable floating community in Busan, which envisions a community of 12,000 people
with the capacity to expand to accommodate more than 100,000 people. The prototype is made of interconnected
platforms, while each modular platform is designed for a specific use, like living space, research facilities or
lodging [16]. Researchers also envision the possibility of underwater sea-scrapers [17]. The sea-scrapers, as illus-
trated in Fig. 1(f), maximize the use of vertical space below the ocean surface, allowing for high-density devel-
opment while minimizing the interference with surface-based activities. The submerged sea-scrapers facilitates
the biodiversity of marine ecosystems while offering unique and immersive living experiences with panoramic
underwater views and proximity to marine life.
On
ly
3
fo
3 The Way Forward – Challenges
rR
ea
The effectiveness of the floating structures as a coastal protection solution requires continued research efforts to
address the following challenges.
din
g
Further research works are necessary to optimize the wave attenuation effectiveness considering the different
at
wave conditions, for example, long waves under storm conditions. The coastal protection against sea level rise
W
requires innovative and adaptive designs of floating structures to be not only self-resilient to the sea level rise, but
also protect the adjacent infrastructure from sea level rise. Possible solutions include integration with the existing
land-based infrastructure near the ports or coastal development.
CF
S
The coastal protection planning and strategy is a long-term mission. The structural integrity and durability of
20
floating structures and their materials to sustain long-term environmental actions and extreme weather conditions
24
become important in ensuring their functionality over the expected service life. Coupled with the durable structural
performance, an efficient maintenance and monitoring scheme, e.g., based on data-driven and sensor-interfaced
digital twins, will be necessary to optimize the lifecycle costs.
The sustainability of floating structures requires integration of renewable energy harvesting as a part of its inherent
design. The potential sources of renewable energies include solar, wind, ocean wave and tidal energies.
The integration of artificial neural networks and machine learning tools has gained increasing interests for evalu-
ating future coastal schemes and enhancing innovative solutions. Artificial intelligence (AI) can play a significant
role in both the design and operation of floating structures, for example, sensor-interfaced AI will be able to detect
anomalies in the structural behavior and forecast when maintenance is needed, which is a part of the predictive
maintenance framework. AI will also be able to interface with the dynamic positioning systems to optimize the
location of the floating structure to maximize the wave attenuation effectiveness. When the floating structure is
integrated with ocean energy converters, AI will be able to optimize the energy production.
4 Conclusions
The development of floating structures has observed tremendous growth over the past decades, with an exciting
journey towards floating cities ahead. The future of floating structures require solutions to tackle sustainability,
carbon neutrality, modularity and scalability, together with a smooth integration with artificial intelligence to
optimize its functionality and operations.
References
1. Kulp SA, Strauss BH. New elevation data triple estimates of global vulnerability to sea-level rise and coastal flooding.
Nature communications, 2019, 10(1): 1-12.
2. Amy Frearson. Powerhouse Company builds floating office in Rotterdam's Rijnhaven. Dezeen. 24 Oct 2022.
https://www.dezeen.com/2022/10/24/floating-office-rotterdam-powerhouse-company
3. Meyer Floating Solutions. Real Estate on Water. https://meyerfloatingsolutions.com
4. The floating seahorse. Luxury Underwater Villas in Dubai. https://thoe.com/the-floating-seahorse
5. Reema Arora. Dubai’s First Underwater Resort: A Floating Venice In The Sea That Will Blow Your Mind. 19 Apr 2023.
https://traveltriangle.com/blog/dubai-first-underwater-resort
6. Norway Underwater Floating Tunnels: Future of Transport. World Famous Things. 29 Mar 2024. https://www.worldfa-
mousthings.com/norway-underwater-floating-tunnels
7. Li H, Cheng X, Pan H. Numerical Simulation of a Submerged Floating Tunnel: Validation and Analysis. Applied Sci-
ences, 2024, 14(9): 3589.
8. Jeong K, Kim S. Structural response of submerged floating tunnels with free-end boundary condition based on an analyt-
ical approach. Applied Ocean Research, 2024, 143: 103861.
9. Foti F, Martinelli L, Morleo E, et al. Dynamic response of submerged floating tunnels: an enhanced semi-analytical ap-
proach. Ocean Engineering, 2024, 294: 116648.
10. Abbaspour, Madjid, and Amir Farshforoush. Optimal arrangements for semi-submersible cylindrical wave energy con-
verters: A study in layout optimization and power extraction efficiency. Ocean Engineering, 2024, 296: 116833.
11. Li Demin, et al. Parameter optimization of a floating two-buoy wave energy converter. Ocean Engineering, 2024, 296:
117043.
On
ly
4
fo
rR
12. Shadmani Alireza, et al. An optimization approach for geometry design of multi-axis wave energy converter. Energy,
2024: 131714.
ea
13. Ameya Paleja. World’s largest twin-rotor floating wind platform launched by Chinese firm. Interesting Engineering. 23
din
Apr 2024. https://interestingengineering.com/energy/oceanx-twin-rotor-floating-platform
14. Pratnashree Basu. Implications of China’s floating nuclear power plants in the South China Sea. Orfonline. 13 May 2024.
g
https://www.orfonline.org/expert-speak/implications-of-china-s-floating-nuclear-power-plants-in-the-south-china-sea
15. Satellite Image Shows Construction of World's First Floating City. Newsweek. 03 May 2024.
at
https://www.newsweek.com/maldives-floating-city-satellite-photos-construction-1897098
W
16. Are Floating Cities the Solution to Rising Seas. The New York Times. 10 May 2024. https://www.ny-
times.com/2024/05/10/style/floating-cities-busan-south-korea.html
CF
17. Nandita. All about Oceanscrapers. Novatr. 1 Apr 2023. https://www.novatr.com/blog/all-about-oceanscrapers Author,
F.: Article title. Journal 2(5), 99–110 (2016).
S
20
24
On
ly
fo
rR
Connector stiffness influence on comfort in large-scale floating
ea
developments
din
g
Rui P. F. Gomes1[0000-0001-5397-8692] and Bart Roeffen1
1
at
Blue21, Paardenmarkt 1A, 2611PA Delft, The Netherlands
W
rui@blue21.nl
CF
S
Abstract. Coastal cities face imminent challenges of land scarcity due to over-population and rising sea levels.
20
Floating structures offer a solution to both of these challenges by creating new living space that adapts to water
level variations. While small-scale floating urban developments have been prevalent in sheltered areas, large-
24
scale implementations in challenging sea conditions hold the key to addressing coastal city issues effectively.
The design of such structures faces technical challenges related to internal loads and human comfort. Typically,
large structures are subjected to high internal loads, while small structures are exposed to larger motions, re-
sulting in a lower level of comfort. The application of compliant mechanical connectors between modules of
large floating structures is an alternative for attaining an adequate balance between internal loads and comfort
level. This study addresses the analysis of a multi-platform floating structure for urban applications subject to
wave action. In particular, it focusses on how the platform comfort level is influenced by the connection bend-
ing stiffness. The interaction between waves and floating structures is computed using a radiation-diffraction
boundary-element method. Each platform is represented as a six degree-of-freedom rigid body, which incor-
porates two buildings. The connectors are modelled as generic six degree-of-freedom springs located between
neighboring platforms. Results demonstrate the dependency of comfort levels and bending moments on the
connector bending stiffness and incident wave conditions. For an adequate combination of wave period and
bending stiffness, the compliant connector provides a good balance between comfort and loads, when com-
pared with flexible and rigid connectors.
Keywords: Multi-module floating urban structures, Compliant connection system, Wave-body interaction.
1 Introduction
Coastal cities worldwide are dealing with severe challenges due to overpopulation and rising sea levels. As urban
populations tend to increase, cities are facing significant problems in housing, infrastructure, and urban planning.
Simultaneously, global warming-induced sea-level rise endangers low-lying areas, intensifying land scarcity and
increasing risks to current developments [1]. These critical issues emphasize the need for innovative and sustain-
able solutions to ensure the resilience and growth of coastal urban environments. Floating structures are seen as
promising solution for these problems, as they offer a flexible and adaptive approach to land scarcity by creating
new living spaces on water, capable of adjusting to sea level variations [2]. Even though small-scale floating
structures have found success in sheltered waters, the upscaling to large-scale developments in more challenging
sea conditions remains relatively unknown and technically complex.
The engineering design of large-scale floating structures is fundamental in addressing these challenges. Engi-
neering considerations must account for internal loads and human comfort, two critical factors that significantly
impact the feasibility and sustainability of these structures [3,4]. Large floating platforms are typically subjected
to high internal loads, affecting their structural integrity and durability [5]. In contrast, smaller floating modules
are more prone to motion induced by waves, which can reduce the occupant comfort of these spaces. Therefore,
achieving a balance between structural integrity and comfort is essential for the success of large-scale floating
urban developments.
Advanced modelling techniques play a crucial role in the engineering design of floating structures. Through
detailed simulations and analyses, engineers can predict the behaviour of these structures under various conditions,
allowing for optimized design solutions [6]. In this study, a multi-body radiation-diffraction boundary-element
method is applied the analyse the interaction between the floating structure and waves. The study focuses on the
importance of compliant mechanical connectors between the modules of large floating structures [7]. These con-
nectors can mitigate internal loads while maintaining an acceptable level of comfort by accommodating relative
motions between modules. The findings of this study emphasize the dependency of comfort levels and structural
loads on the stiffness of the connectors. Results demonstrate that compliant connectors can effectively balance
On
ly
2
fo
rR
the demands of structural integrity and comfort. This balance is fundamental for the practical implementation of
large-scale floating structures in coastal cities.
ea
2 Methodology
din
g
at
2.1 Problem definition
W
This study analyses a large scale urban floating development consisting of 9 platforms interconnected with a
generic six degree-of-freedom spring. The platforms have no gaps between them. Each platform has two four-
CF
story buildings. The problem is solved using the urban design and simulation software HydroMEC+. A three-
dimensional view of the floating structure and buildings and a summary of the characteristics of the floating
structure are presented Fig. 1.
S
20
24
Fig. 1. Three-dimensional view of the 9-platform floating structure and 18 5-floor buildings and summary of the characteris-
tics of the floating urban development.
2.2 Numerical modelling

The fluid-structure interaction problem is analyzed in this study by applying three-dimensional potential flow
theory. Therefore, the formulation presented here considers irrotational flow and neglects viscous effects. Linear
wave theory and linear hydrodynamic interactions are considered by assuming small wave amplitude and small
body motions, respectively. The floating structure consists of an arrangement of several rectangular pontoon plat-
forms, which are linked to neighboring platforms via compliant connectors, which allow a predominantly hinge-
type motion, where the bending stiffness can be adjusted.
The rectangular platforms (length 𝑙, width 𝑤 and height ℎ) are modeled as rigid bodies, each with 6 degrees of
freedom (3 translations – surge, sway and heave - and 3 rotations – roll, pitch and yaw) defined at the center of
gravity of the corresponding body. Waves travel along the positive 𝑥-axis direction for a zero wave direction 𝜃.
The full system is modelled as having 6𝑁 degrees of freedom, where 𝑁 represents the total number of platforms.
The frequency-dependent wave radiation and diffraction hydrodynamic coefficients were computed using a three-
dimensional boundary-element method (BEM) software [8].
All forces relevant to describe the system dynamics are assumed linear. Under this assumption, the system
becomes time invariant and we may write the displacement 𝑥 , velocity 𝑥 and acceleration 𝑥 of each degree of
freedom as
𝐢
𝑥 ,𝑥 ,𝑥 𝑡 Re 𝑋 , 𝐢𝜔𝑋 , 𝜔 𝑋 𝑒 (1)
where 𝑋 is a complex displacement amplitude associated to the 𝑖-th degree of freedom and 𝐢 is the imaginary
unit. Note that dynamic forces can also be described the same way. All displacements of the system can be repre-
sented the vector of complex displacement amplitudes
𝐱 𝑋 ,𝑋 ,𝑋 ,𝑋 ,𝑋 ,𝑋 ,𝑋 ,…,𝑋 ,𝑋 (2)
On
ly
3
fo
rR
where 𝑋 , 𝑋 , 𝑋 , 𝑋 , 𝑋 , 𝑋 represents the 6 degree-of-freedom displacement complex amplitude vector of
body 1 (surge, sway, heave, roll, pitch and yaw), 𝑋 , 𝑋 , 𝑋 , 𝑋 , 𝑋 , 𝑋 is the corresponding vector for body 2,
and so forth for the other bodies.
ea
din
The equations of motion of the 6𝑁-degree-of-freedom system is given by
𝜔 𝐌 𝐀 𝐢𝜔𝐁 𝐂 𝐂 𝐂
g
𝐱 𝐴 𝐟 (3)
at
where 𝐟 is a column vector representing the force coefficient due to wave diffraction, 𝐴 is the incident wave
amplitude, 𝐌 ∈ ℝ is the inertia matrix, 𝐀 ∈ ℝ
W
is the radiation added inertia matrix, 𝐁 ∈ ℝ is
radiation damping matrix, 𝐂 ∈ ℝ
matrix, and 𝐂 ∈ ℝ CF
is the hydrostatic stiffness matrix, 𝐂 ∈ ℝ is the connector stiffness
represents the mooring system stiffness matrix. In this case, mooring forces are applied
directly in the center of gravity, and therefore 𝐂 becomes a diagonal matrix.
S
20
24
Fig. 2. Schematic representation of the model used to simulate multi-platform floating structure with 9 platforms in the 3-by-
3 grid arrangement, and mesh used to discretize the floating structure wetted surface.
2.3 Connector stiffness
The compliant connectors presented in Fig. 2 are modelled as generic six degree-of-freedom springs located be-
tween neighboring platforms. The relative surge, sway and heave displacements between neighboring platforms
at the connector point are constrained by translational springs with stiffnesses 𝐾 , 𝐾 and 𝐾 , respectively. Rela-
tive roll, pitch and yaw are restricted by rotational springs with stiffnesses 𝐾 , 𝐾 and 𝐾 , respectively. The
relationship between the displacement vector of multi-platform system degrees of freedom and the restoring forces
and moments induced at the center of gravity of each platform is given by matrix 𝐂 . This matrix is obtained by
applying Hamilton’s principle [9] to describe the forces induced on the connectors by the motion of the bodies
and by neglecting terms higher than first order [10]. The forces-moment vector acting on a connector is
𝐟 𝐊 𝐱 𝑪𝑨 𝐱 𝑪𝑩 (4)
where 𝐊 ∈ ℝ is a diagonal matrix representing the connector stiffness (𝐊 diag 𝐾 , 𝐾 , 𝐾 , 𝐾 , 𝐾 , 𝐾 ),

𝐱 𝑪𝑨 and 𝐱 𝑪𝑩 are the position vectors of the connector point C due to the motion of bodies A and B, respectively.
2.4 Comfort evaluation

The comfort evaluation is based on the computation of acceleration at critical points in the platform accessible to
occupants. For a given point 𝑷, the three coordinates of its displacement are given by the vector
𝐱 𝑻𝑷 𝐱𝑻 𝐱𝑹 𝐫𝑷 (5)
where 𝐱 𝑻 is a vector representing the three translational degrees of freedom of the corresponding platform, 𝐱 𝑹 is
a vector representing the three rotational degrees of freedom (at the center of gravity) and 𝐫 𝑷 is a vector describing
On
ly
4
fo
rR
the position of point 𝑷 relative to the platform center of gravity. Note that 𝐱 𝑻𝑷 is linear due to the small rotation
assumption. Considering the formulation presented in Eq. (1), the acceleration vector at point 𝑷 is
ea
𝐚𝑷 𝝎𝟐 𝐱 𝑻𝑷
din
(6)
The analysis of comfort requires the evaluation of the system response under irregular wave conditions, and there-
g
fore a stochastic approach is applied. This approach uses the linear response amplitude operators for the compu-
at
tation of statistic representative values of the floating structure dynamics. Let us consider a random sea state
W
represented by the superposition of elementary regular wave components, each with its own amplitude and fre-
quency. This assumption implies that the instantaneous free-surface elevation at a given position can be described
CF
as a stationary, ergodic and Gaussian process [11, 12]. To simulate irregular waves, a JONSWAP spectrum 𝑆 is
considered, whose energy density variation over frequency 𝜔 is dependent on the significant wave height 𝐻 , peak
period 𝑇 and shape factor 𝛾 [12].
S
The energy density spectrum of a the acceleration at point 𝑷 is
20
𝑆𝐚𝑷 𝜔 𝑆 𝜔
𝐚𝑷
24
(7)
Since the system is linear and the water elevation is described by a Gaussian probability density function, then
any linear response 𝑆 can be described by a Gaussian probability density function. The root-mean-square value
of the acceleration at point 𝑷 is
𝐚𝑷𝑹𝑴𝑺 𝑆𝐚𝑷 𝜔 𝑑𝜔 (8)
where the vector 𝐚𝑷𝑹𝑴𝑺 represents acceleration values along the 𝑥-, 𝑦- and 𝑧-axis direction.
3 Results
3.1 Frequency-domain analysis
The frequency-domain approach described in Section 2 was applied to solve the 9-platform dynamic problem
presented in Fig. 2. The computation of the hydrodynamic coefficients considers a regular mesh of the wetted
surface with 3912 panels. Hydrodynamic results are obtained for a total of 81 equally-spaced wave frequencies,
between 0.1 and 2.0 rad/s. The floating structure is moored via 4 spring elements at the center of gravity of bodies
1, 3, 7 and 9, to limit the horizontal motions of the platform. The stiffness coefficients of the spring elements are
𝐾 𝐾 5 10 N/m and 𝐾 10 N m/rad. In order to simulate connector behaving as a hinge with ad-
justable bending stiffness, the spring stiffness of the other degrees of freedom is set with relatively high values.
The stiffness coefficients of the longitudinal connectors are 𝐾 𝐾 𝐾 10 N/m, 𝐾 𝐾
10 N m/rad and the 𝐾 is variable. The stiffness coefficients of the transversal connectors are 𝐾 𝐾 𝐾
10 N/m, 𝐾 𝐾 10 N m/rad and 𝐾 is variable. All results presented here consider a wave direction
equal to zero (aligned with the 𝑥 -axis).
Heave and pitch response amplitude operators (RAOs) are shown in Fig. 3, for a range of wave periods between
4 and 16 s. Results are presented for the three middle longitudinal bodies for three different bending stiffness
values. For 𝐾 10 N m/rad, the connectors have a negligible stiffness and the system behaves similarly to a
hinge (flexible connector). For 𝐾 10 N m/rad, the connectors behave rigidly and restricts completely the
displacement of the relative pitch displacement between platforms (rigid connector). The case with 𝐾
10 N m/rad represents an intermediate stiffness situation (semi-flexible connector – hinge with significant bend-
ing stiffness). Body 2, which is heading the waves, shows generally higher heave and pitch displacements. As
expected, for most periods, the flexible-connector configuration shows the largest pitch and the rigid-connector
configuration displays the smallest pitch.
Fig. 4 shows the irregular-wave normalized bending moment root-mean-square-value at the connectors be-
tween bodies 2 and 5 and between bodies 5 and 8, for sea states with peak period (𝑇 ) equal to 6, 8 and 10 s. The
irregular-wave sea states are described by a JONSWAP spectrum. The bending moment root-mean-square-value
is computed through the stochastic approach presented in Eqs (7-8), for the acceleration root-mean-square value
calculation. For 𝐾 10 N m/rad, no significant bending moment is measured at the connections. For 𝐾
10 N m/rad, the bending moment is independent from 𝐾 , which indicates that the systems behaves rigidly.
Note that for 𝐾 10 N m/rad, the response is weakly dependent on the stiffness 𝐾 . For 10 𝐾 10 ,
On
ly
5
fo
rR
the bending moment varies significantly, suggesting that this intermediate stiffness range has a significant impact
on the system dynamics. The peak at 𝐾 10 N m/rad indicates that the system is near resonance conditions.
ea
Note that the connector between bodies 2 and 5 displays higher bending moment than the connector between
din
bodies 5 and 8, for all wave periods presented.
g
at
W
CF
S
20
24
Fig. 3. Response amplitude operators (RAOs) of heave and pitch as a function of the incident wave period, for bodies 2, 5 and
8 and for different bending stiffness values.
Fig. 4. Normalized bending moment root-mean-square value as a function of the bending stiffness, for the connectors between
bodies 2 and 5 and between bodies 5 and 8, for 3 irregular-wave sea states. Bending moment is normalized with respect to the
incident wave free-surface elevation root-mean-square value.
3.2 Comfort under irregular-wave conditions

The analysis of comfort is based on the evaluation of acceleration root-mean-square (RMS) values of locations in
platform and buildings accessible to occupants. For this analysis, we assume that the acceleration felt by occupants
is induced by irregular-wave action described by an energy density JONSWAP spectrum with a shape factor equal
to 3.3 and significant wave height 𝐻 1.5 m. For this analysis, peak period 𝑇 6, 8, 10 s.
Fig. 5 shows the acceleration RMS value as a function of the connector bending stiffness for the least comfort-
able location in the floating structure. The cases with flexible hinge-type connectors (𝐾 10 N m/rad) shows
maximum acceleration values independent from the bending stiffness value. The same is observed for rigid con-
nectors (𝐾 10 N m/rad). For the intermediate stiffness range (10 𝐾 10 ), the acceleration values
present significant variation, with the largest vertical acceleration RMS value occurring for 𝑇 6 s and 𝐾
10 N m/rad, and the largest horizontal acceleration RMS value occurring for 𝑇 6 s and 𝐾 10 N m/rad.
For this particular configuration, within a range of intermediate stiffness values, it is possible to achieve a better
comfort level (see Fig. 5) than the one with the flexible connectors and a bending moment (see Fig. 4, for connector
2-5) smaller than the one with rigid connectors (𝐾 10 N m/rad). This is the case for 5 10 𝐾
10 N m/rad, for all peak periods. Resonance conditions for the bending moment (at 𝐾 10 N m/rad) and
for the accelerations (at 𝐾 10 N m/rad) are observed for specific intermediate stiffness values.
On
ly
6
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 5. Acceleration root-mean-square value as a function of the bending stiffness, for the point in the floating structure subject
to the largest acceleration.
4 Conclusions
A numerical model of a 300 m long and 150 m wide 9-platform floating structure with the characteristics of a
large-scale urban development was simulated under wave action. Platforms were assembled in a 3-by-3 grid ar-
rangement and neighboring platforms were connected via hinges. At the hinges, a bending stiffness was applied,
and the effect of its magnitude on the dynamic response of the whole 9-body system was evaluated. Three bending
stiffness regions were identified: a low stiffness region where connectors behave as hinges, a high stiffness region
where connector are rigid, and an intermediate region where the system dynamics is highly sensitive to the bending
stiffness value. The analysis of bending moments and comfort showed that for a given wave condition, the selec-
tion of an adequate bending stiffness provides a smaller bending moment than the rigid situation and simultane-
ously the maximum acceleration is smaller than the one of the flexible case. However, an inadequate selection of
the bending stiffness can result in undesirable high values for either the bending moment and acceleration. This
result shows that compliant connectors with an adequate selection of the bending stiffness can provide a good
compromise between structural integrity and comfort but the choice of an inadequate bending stiffness value can
be detrimental for the design.
References
1. Best, K., He, Q., Reilly, A.C., Niemeier, D.A., Anderson, M., Logan, T.: Demographics and risk of isolation due to sea
level rise in the United States. Nature Communications 14(1) (2023).
2. IPCC: Climate Change 2022: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Sixth
Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge (2022).
3. Wang, G., Goldfeld, Y., Drimer, N.: Expanding coastal cities - Proof of feasibility for modular floating structures (MFS).
Journal of Cleaner Production 222 (2019).
4. Wang, G., Rosenfeld, Y., Drimer, N., Goldfeld, Y.: Occupant comfort analysis for rigid floating structures – methodology
and design assessment for offshore dwelling module. Ships and Offshore Structures 16 (2021).
5. Yoon, J.-S., Cho, S.-P., Jiwinangun, R.G., Lee, P.-S.: Hydroelastic analysis of floating plates with multiple hinge connec-
tions in regular waves, Marine Structures 36 (2014).
6. Huang, H., Chen, X.-J. Liu, J.-Y., Ji, S.: A method to predict hydroelastic responses of VLFS under waves and moving
loads. Ocean Engineering 247 (2022).
7. Jiang, D., Tan, K.H., Wang, C.M., Dai, J.: Research and development in connector systems for Very Large Floating
Structures. Ocean Engineering 232 (2021).
8. Ancellin, M., Dias, F.: Capytaine: a Python-based linear potential flow solver. Journal of Open Source Software 4(36),
1341 (2019).
9. Shabana, A.: Dynamics of Multibody Systems. 3rd edn. Cambridge University Press, Cambridge, UK (2020).
10. Sirigu, S.A., Boccanfuso, V., Carapellese, F.: Multi-body Analysis of Modular Floating Islands: Optimisation of Con-
nector Stiffness. In: Ikoma, T., Tabeta, S., Lim, S.H., Wang, C.M. (eds.) Proceedings of the Third World Conference on
Floating Solutions. WCFS 2023. Lecture Notes in Civil Engineering, vol 465. Springer, Singapore (2024).
11. Sarpkaya, T., Isaacson, M.: Mechanics of Wave Forces on Offshore Structures, Van Nostrand Reinhold, New York, USA
(1981).
12. Goda, Y.: Random Seas and Design of Maritime Structures, 3rd edn. World Scientific, Singapore (2010).
On
ly
fo
rR
ea
din
g
at
Research and development of thermoelectric modules
and their application in floating cities
W
CF
Jingming Cai1,2, Yujin Yuan2 and Zhiyang Pei2
S
1
20
Faculty of Civil and Environmental Engineering, TU Darmstadt, Franziska-Braun-Straße,
Darmstadt, Germany
24
2 Key Laboratory of Concrete and Prestressed Concrete Structures of Ministry of Education,
Southeast University, Nan-jing, China

jingmingcai@seu.edu.cn
Abstract. This paper investigates the development and application of thermoe-

lectric (TE) modules, particularly focusing on their integration into floating city
architectures to address unique energy demands and sustain-ability challenges.
TE modules, which operate on the Seebeck effect, convert thermal gradients di-
rectly into electricity. Despite existing challenges, such as the cost and perfor-
mance limitations of current TE materials, recent advancements have improved
their properties, making them viable for practical applications. The focus of this
study is on testing the performance of TE modules, including their mechanical
properties, Seebeck coefficient, and electrical and thermal conductivity. These
are all very important for improving the output power and energy conversion
efficiency. The paper also investigates the innovative use of TE modules in float-
ing cities to enhance energy independence through waste heat recovery and direct
heat-to-electricity conversion, thereby supporting sustainable and autonomous
urban ecosystems. The potential of TE modules for building materials and ther-
mal management in floating cities is also dis-cussed, providing a path forward
for resilient and energy-efficient urban development.
Keywords: Thermoelectric (TE) Modules, Floating Cities, Energy Sustainabil-

ity.
1 Introduction
In recent decades, the imperative to conserve energy and reduce emissions has driven
the global agenda, leading to innovations in building materials and energy conversion
technologies [1]. Floating cities, characterized by their need for self-sufficiency and
resilience, necessitate advanced energy solutions that can operate independently of tra-
ditional power infrastructure [2-4].
As shown in Fig.1, adopting a variety of new energy sources can significantly en-
hance the self-sufficiency of floating cities, achieving zero external electricity pur-
chases. Thermoelectric (TE) modules, which convert thermal energy into electrical en-
ergy through the Seebeck effect, present a promising solution for floating cities [5-6].
These modules can harness thermal gradients to generate electricity, reducing
On
ly
fo
rR
ea
2
din
g
at
dependency on external power sources. In floating cities, it is especially important to
have energy systems that are compact, efficient, and environmentally benign because
to limited space and environmental concerns.
W
CF
Despite the potential advantages, current commercial TE modules face challenges
related to cost, durability, and performance [7-8]. Cement-based TE materials in build-
S
ing applications offer the significant advantages of cost-effectiveness and enhanced
20
structural integrity, making them appealing for integration into energy-harvesting
buildings and reducing reliance on expensive commercial TE modules [9]. However,
24
they face challenges such as a lack of systematic research on their building structural
strength and low TE performance with ZT values, which can severely impair their en-
ergy-harvesting capabilities. Recent advancements in TE materials, particularly the de-
velopment of geopolymer-based TE composites, present a viable alternative [10]. These
materials exhibit improved TE properties, making them suitable for application in float-
ing cities.
In summary, the deployment of advanced TE modules in floating cities is a critical
step towards achieving energy self-sufficiency and environmental sustainability. This
paper explores the mechanical strength and ZT performance of TE modules, with a
particular focus on their application in buildings. It highlights the potential of TE mod-
ules to meet unique energy demands and discusses their integration into building mate-
rials and thermal management systems, emphasizing their suitability in floating urban
environments.
Fig. 1. Energy harvesting for floating cities.

On
ly
fo
rR
ea
din 3
g
2 TE Module Performance Test Method
at
2.1 Mechanical Properties
W
CF
The compressive strength measurements were conducted in accordance with EN 1015-
S
11 [11]. A universal testing machine performed standard prism specimen testing. For
20
the bending tensile strength measurements, three prismatic specimens were subjected
to a three-point loading assessment, and the compressive strength testing was imple-
mented on six prismatic specimens obtained from the bending test (cf Fig. 2).
24
Fig. 2. Mechanical properties test A) Flexural test. B) Compressive test.
2.2 Seebeck Coefficient
Cylindrical molds measuring 50 mm in diameter and 100 mm in height were utilized to

prepare the polymer samples for measuring the Seebeck coefficient. To measure the
voltage differential (ΔV), a digital multimeter was used. Both ends of the module were
surrounded by a copper sheet, and silver paint was applied to enhance the transmission
of electricity between them (cf Fig. 3). The copper sheet interface was utilized to con-
nect the digital multimeter. A digital thermometer equipped with a thermocouple was
utilized to measure the temperature variation (ΔT) of the module. Using a resistance
heater, a temperature difference was created between the two surfaces of the sample,
which produced a corresponding voltage difference that was recorded simultaneously.
On
ly
fo
rR
ea
4
din
g
at
W
CF
S
20
24
Fig. 3. Seebeck coefficient test set-up.
2.3 Electronic Conductivity

To evaluate the electrical conductivity characteristics of the module, the impedance was
measured using an impedance analyzer IM 3570, employing the double-probe method.
The resistance obtained was subsequently substituted into equation (1) to calculate the
electrical conductivity.
𝜌 (1)
where 𝜌 represents the volume resistivity, 𝑅 denotes the resistance, S is the transverse
cross-sectional area of the sample relative to the direction of resistance measurement,
and L indicates the distance between the electrodes.
Fig. 4. Electrical conductivity measurement.
2.4 Thermal conductivity

The thermal conductivity measurements in this paper were conducted using the LFA
467 laser flash apparatus from Netzsch, Germany, employing the transient method. The
vertical configuration of the measurement system places the xenon lamp light source at
the bottom, the sample at the center, and the signal detector at the top. The light source
is controlled by software, allowing for adjustment of the pulse width from 20 µs to 1200
On
ly
fo
rR
ea
din 5
g
at
µs. The xenon lamp uniformly heats the lower surface of the sample, while an infrared
detector measures the temperature rise on the upper surface as a function of time. The
W
test system, treated as an adiabatic system, enables the calculation of the thermal diffu-
sivity coefficient, 𝑎 (mm2 /s), as shown in equation (2).
CF
𝑎 0.1388
S (2)
20
/
In the formula, l represents the thickness of the sample, and 𝑡 / denotes the time re-
24
quired for the temperature to increase by 50%. In addition to the thermal diffusivity
coefficient, the specific heat capacity 𝑐 of the sample can also be determined using a
reference specimen. The thermal conductivity (𝜆) at a given temperature can be calcu-
lated using equation (3).
𝜆 𝑇 𝑎 𝑇 𝜌 𝑇 𝑐 𝑇 (3)
where 𝜌 denotes the density of the material, while 𝑇 represents the temperature
achieved by the sample.
Fig. 5. A schematic of the thermal conductivity testing principle.
To comprehensively assess the TE effect of this module, a dimensionless figure of merit

(ZT) is utilized. This value, crucial for determining conversion efficiency, comprises
the Seebeck coefficient (S), electrical conductivity (σ), thermal conductivity (κ), and
the absolute measurement temperature (T). It is defined as follows.
𝑍𝑇 (4)
The average value of the ZT for this module at ambient temperature has been calcu-
lated.
2.5 TE Module
Materials structured into modules shape TE modules to convert thermal energy into
electrical energy. By connecting p-type and n-type TE legs with metal electrodes, these
On
ly
fo
rR
ea
6
din
g
at
modules improve thermal flow in a single direction perpendicular to the substrate. Typ-
ically, an arrangement combining electrical series and thermal parallel connections in-
W
terconnects several -shaped modules to augment output power.
CF
The ZT primarily determines the theoretical maximum efficiency of TE modules.
However, during the design and fabrication processes, factors at the level, such as the
S
topology, interface elements, and fabrication techniques, significantly influence the ac-
20
tual performance indicators. These indicators include energy conversion efficiency,
maximum power output, and the stability and reliability of the modules.
24
Thermoelectric power generation (TEG) is based on the Seebeck effect. When a
temperature difference is applied across the ends of a TE module—with the hot end and
cold end set at temperatures 𝑇 and 𝑇 , resulting in a temperature difference 𝛥𝑇
𝑇 𝑇 , TE voltage is generated. This voltage can drive load resistance. The two most
critical performance parameters are the power generation efficiency and the maximum
output power [12].
√
𝜂 (5)
√ /
The higher the ZT value of the material, the greater the conversion efficiency of the TE
modules. Currently, the average ZT value of most TE materials systems discovered at
room temperature remains below 1.0, which is a critical issue limiting the enhancement
of power generation efficiency in TEG. The Seebeck voltage generated in the temper-
ature difference circuit between the hot and cold ends determines the maximum output
power. To achieve this maximum power, the external resistance 𝑅 must be matched
with the internal resistance 𝑅 of the module. At this point, the module can achieve its
maximum output power, as shown in equation (6) [13].
𝑃 (6)
In this study, n-type and p-type architectural TE materials were initially prepared. Sub-
sequently, these materials were used to construct 36 pairs of modules on a substrate.
Mechanical strength and TE performance were considered as the basis for TE optimi-
zation, aimed at enhancing the power generation and efficiency of the modules.
3 The Potential Application of TE Modules in Floating Cities
3.1 Energy demands and challenges of floating cities

Floating cities face unique challenges, such as limited space and the need for sustaina-
ble energy. In this context, TE modules offer unique applications tailored to the envi-
ronmental and spatial challenges inherent in such habitats. These modules, leveraging
the direct conversion of heat into electricity, can be crucial in enhancing energy sus-
tainability, optimizing thermal management, and supporting the complex infrastructure
of floating cities.
Floating cities face unique challenges, such as limited space and the need for sus-
tainable energy. TE mod-ules are crucial in these settings, as they can convert the
On
ly
fo
rR
ea
din 7
g
at
marine temperature differentials into a reliable energy source. This ability enhances the
sustainability and energy independence of floating cities, making them more viable so-
lutions to urban and environmental challenges.
W
CF
In the context of floating city buildings, TE modules offer unique applications tai-
lored to the environmen-tal and spatial challenges inherent in such habitats. These mod-
S
ules, leveraging the direct conversion of heat into electricity, can be crucial in enhanc-
20
ing energy sustainability, optimizing thermal management, and supporting the complex
infrastructure of floating cities.
3.2 Applications of TE Modules

24
Integration with Building Materials
TE modules can be embedded into structural elements of buildings, such as roofing
panels or facades, as illustrated in Fig.1. This integration allows the building itself to
function as an energy harvester, utilizing the thermal energy from sunlight. Such appli-
cations not only provide power but also reduce heat gain from solar exposure, contrib-
uting to more stable internal temperatures.
TE Modules in Marine Structures
Floating city buildings can utilize TE modules to harness the thermal gradient between
the warmer surface water and the cooler deep-sea water. This temperature differential
can be a continuous source of energy, ideal for TE modules to generate electricity effi-
ciently. This renewable energy source supports the floating city's power needs, reducing
dependency on traditional energy supplies and minimizing environmental impact.
Efficient Thermal Management
In floating cities, managing indoor climate efficiently is paramount due to exposure to
marine elements and limited space. TE modules can be integrated into the building en-
velopes or windows of floating city buildings to regulate temperature more dynami-
cally. These modules can absorb excess heat from the interior and release it into the
cooler marine environment or vice versa, maintaining comfort without extensive energy
consumption.
Autonomous Energy for Safety and Monitoring Systems
TE modules can power critical safety and monitoring equipment continuously. In float-
ing cities, where reliability is key due to isolation from mainland infrastructure, having
autonomous power sources for emergency lighting, alarms, and structural health mon-
itoring sensors ensures enhanced safety and operational stability.
Reducing Energy Footprint
By converting waste heat from residential, commercial, or industrial activities within
the floating city into usable electricity, TE modules help reduce the overall energy foot-
print. This conversion contributes to the sustainability goals of floating cities by mini-
mizing waste and optimizing resource use.
Overall, the application of thermoelectric modules in floating city buildings repre-
sents a forward-thinking approach to building design and energy management. These
modules not only enhance energy efficiency and comfort but also align with the broader
sustainability and resilience goals of floating urban developments.
On
ly
fo
rR
ea
8
din
g
References
at
W
1. Hassan, Q., Viktor, P., Al-Musawi, T. J., Ali, B. M., Algburi, S., Alzoubi, H. M., ... &
CF
Jaszczur, M.: The renewable energy role in the global energy transformations. Renewable
Energy Focus, 48, 100545 (2024).
S
2. Nguyen, T. T. T.: Promoting sustainability and resilience in vietnam’s floating community,
20
Universität CottbusSenftenberg, (2021).
3. Takahashi, A., Lin, F.-Y. & Yoshida, I. in World Conference on Floating Solutions. 101-
120 (Springer).
24
4. Kirimtat, A., Krejcar, O., Kertesz, A., & Tasgetiren, M. F.: Future trends and current state
of smart city concepts: A survey. IEEE access, 8, 86448-86467 (2020).
5. Nandihalli, N.: Thermoelectric films and periodic structures and spin Seebeck effect sys-
tems: Facets of performance optimization. Materials Today Energy, 25, 100965 (2022).
6. Jouhara, H., Żabnieńska-Góra, A., Khordehgah, N., Doraghi, Q., Ahmad, L., Norman, L., ...
& Dai, S.: Thermoelectric generator (TEG) technologies and applications. International
Journal of Thermofluids, 9, 100063 (2021).
7. Zhou, X., Carmeliet, J., & Derome, D.: Assessment of risk of freeze-thaw damage in inter-
nally insulated masonry in a changing climate. Building and Environment, 175, 106773
(2020).
8. Bu, Z., Zhang, X., Hu, Y., Chen, Z., Lin, S., Li, W., ... & Pei, Y.: A record thermoelectric
efficiency in tellurium-free modules for low-grade waste heat recovery. Nature communi-
cations, 13(1), 237 (2022).
9. Liu, X., Jani, R., Orisakwe, E., Johnston, C., Chudzinski, P., Qu, M., ... & Yazawa, K.: State
of the art in composition, fabrication, characterization, and modeling methods of cement-
based thermoelectric materials for low-temperature applications. Renewable and Sustaina-
ble Energy Reviews, 137, 110361 (2021).
10. Cai, J., Yuan, Y., Xi, X., Ukrainczyk, N., Pan, L., Wang, Y., & Pan, J.: Unveiling the re-
markable potential of geopolymer-based materials by harnessing manganese dioxide incor-
poration. Small, 20(5), 2305360 (2024).
11. EN, T. Mortar Testing Method, Part 11. Measurement of Compressive and Flexural Tensile
Strength of Mortar (2000).
12. Yang, J., & Caillat, T.: Thermoelectric materials for space and automotive power generation.
MRS bulletin, 31(3), 224-229 (2006).
13. Chen, W.-H., Lin, Y.-X., Wang, X.-D. & Lin, Y.-L.: A comprehensive analysis of the per-
formance of thermoelectric generators with constant and variable properties. Applied En-
ergy, 241, 11-24 (2019).
On
ly
fo
rR
A proposal of floating offshore base for wind farm deployed in EEZ
ea
din
Mako Yamamoto1, Saika Iwamatsu1, Kazuhiro Iijima1, Masaaki Shibata1 and Hiroshi Tatsuhara2
g
1 Department of Naval Architecture and Ocean Engineering, Osaka University, Osaka, Japan
at
2 Toda Corporation, Tokyo, Japan
yamamoto_mako@naoe.eng.osaka-u.ac.jp
W
CF
S
Abstract. This study explores the feasibility of offshore floating bases for wind farms deployed in Japan's
Exclusive Economic Zone (EEZ), aiming to achieve the national ambitious goal of installing 30 GW to 45
20
GW of offshore wind power by 2040. Construction of energy islands for wind farms in European waters is
24
seriously discussed. Given the deep waters exceeding 200 meters in Japan's EEZ, floating solutions are
deemed more suitable than the artificial islands, contrasting with European practices. The research investi-
gates three scenarios at varying distances from the coast (50 km, 100 km, and 300 km, respectively), consid-
ering environmental factors such as sea conditions, water depths, and wind characteristics. Referencing to the
foregoing research of the energy islands, functional requirements for each floating offshore base are dis-
cussed and outlined, guiding the layout, area, and dimensions of the facilities. A semi-submersible structure
supported by multiple vertical columns is finally selected from the viewpoint of its response characteristics
in high waves and suitability for open ocean installation. The performance and structural safety under the
design wave conditions are confirmed by numerical hydroelasticity simulations based on the finite element
(FE) model.
Keywords: floating offshore wind turbine, wind farm, floating base, hydroelasticity
1 Introduction
Aiming for installation of "30 to 45 GW of offshore wind power by 2040yr" [1] as declared in Japan's Offshore
Wind Industry Vision, discussions are ongoing regarding the utilization of Japan’s exclusive economic zone
(EEZ) for offshore wind power. Operation and maintenance (O&M) bases, construction bases, and other related
industry bases in the sea areas is deemed necessary. In Europe, the concept of energy islands [2] has been
planned and designed. Construction of one of such islands has even started. Considering the water depth exceed-
ing 200 meters in Japan’s EEZ, floating solutions must be conceived. However, such research has not yet initi-
ated in Japan. This study aims to explore the feasibility of floating bases to be installed in the EEZ.
2 Offshore Bases Conceptualized Overseas
Fig.1. Energy island of Denmark and Doggerbank [2].
As the reference for offshore bases, artificial islands are planned to be constructed in the North Sea of Denmark
and Dogger Bank. In Fig.1., the left side shows the artificial island in Denmark, and the right side shows the ar-
tificial island in Dogger Bank. Facilities to be developed on the islands include base ports, heliports, accommo-
dation facilities, energy storage systems, Power-to-X systems, and high-voltage direct current (HVDC) trans-
mission systems. The Danish artificial island is scheduled for completion in 2030yr, and the Dogger Bank artifi-
cial island is set for 2035yr.
On
ly
2
fo
3
rR
Offshore Bases Needed in Japan's EEZ
ea
When offshore bases suitable for Japan’s EEZ is considered, floating solutions equipped with similar functional-
din
ities as European energy islands are discussed herein. Three installation scenarios at varying distances (50 km,
100km and 300km) away from the coast of Japan’s main island are brought up here. The offshore base located
g
at 50 km from the coast are called Base1, that at 100 km from the coast are Base2, and that at 300 km from the
at
coast are Base3. Based on the offshore wind condition map by New Energy and Industrial Technology Develop-
W
ment Organization, Japan (NEDO) [3], eastern Japan on the Pacific side has strong annual average wind speeds,
leading to the belief that many offshore wind farms will be installed in this area. The offshore bases will be set
CF
up in the sea areas on the Pacific side of eastern Japan. The design significant wave heights are determined by
performing statistical analysis using the wave scatter diagram in the "Wave and Wind Database of Coastal Ja-
S
pan" [4] by the National Maritime Research Institute. The long-term distribution used for the long-term estima-
20
tion is given by fitting the data to the Fisher-Tippet Type I distribution. Assuming a design life of 50 years, the
24
design sea states for checking the structural safety is determined as given in Table 1. The operational limit wave
height is separately determined by referencing to an exceedance probability of 10% and the significant wave
height for checking the operational limit is set 3.5 m.
Table 1. Conditions of base site

Coastal Distance Water Depth Design Significant Wave Height
Base1 50km 500m 11.3m
Base2 100km 2000m 12.7m
Base3 300km 5000m 14.1m
Ref (Doggerbank) 200km 13-30m 11.7m
Ref (Denmark) 100km 25-28m 11-13m
The functions needed for Base1, Base2, and Base3 are summarized in Table 2. Base1, located at 50 km from
the coast, can be reached by boats within a day, so its primary function is to collect electricity from the wind
farms, transmit it to land via subsea cables. Therefore, it has only the minimum facilities, such as a control room
to manage the wind farms and a port for ships carrying personnel and goods. Base2, located at 100 km from the
coast, mainly has transmission, assembly, and repair base functions. A significant feature is that assembly and
commissioning can be carried out at the offshore base instead of making frequent trips by the vessels to build
and construct the wind farm 100 km away. Base3, located at 300 km from the coast, in addition to the functions
of Base2, has the feature of converting electricity into hydrogen or ammonia (P2X), with additional functions
like solar panels and data centers for profitability. It not only has functions to operate wind farms but also to
supplement new values as a floating structure for renewable energy like solar power generation, data centers,
fishery bases and logistics bases. In this study, the solar power generation and data centers are considered.
Table 2. Functions needed for Base1, 2 and 3

Base1 Base2 Base3
Transmission Facility HVAC HVDC -

Transmission and
Substation - ✔ ✔
Transformation
Operation, Labo ✔ ✔ ✔
Accommodation 10 40 45
O&M Generator - ✔ ✔
Warehouse ✔ ✔ ✔
Assembly, Commission-
Assembly - ✔ ✔
ing(P&P)
On
ly
3
fo
Port
rR ✔(small) ✔(large) ✔(large)
Access Heliport
ea ✔ ✔ ✔
Runway
din - - ✔
Storage of electricity
g
✔ ✔ -
Factory of Hydrogen, Am-
at ✔
W
P2X - -
monia
Tank of Hydrogen, Ammo-
nia
-
CF - ✔
S
20
For transmission facilities, wind power generation uses AC to freely raise and lower voltage, but AC has sig-
24
nificant transmission losses. Therefore, when the distance from the coast exceeds 50 km, it is more advanta-
geous to convert to DC before transmitting to land.[5] Hence, Base1 uses AC transmission while Base2 uses DC
transmission. Base3 does not have transmission facilities because all electricity is converted to hydrogen or am-
monia.
For accommodation facilities, considering the number of personnel working within a day [6], Base1, which is
within a day trip distance, is assumed to have 10 people for O&M and transmission facilities. For Base2, 20
people for O&M, transmission facilities, and substations, and 20 people for assembly and repair are assumed.
For Base3, 20 people for O&M, P2X, and substations, and 20 people for assembly and repair are assumed. For
the data center, it is assumed that one manager oversees 5,000 servers, with 5 people managing 20,000 servers at
the base.
For assembly and commissioning, typically, wind turbines are pre-assembled at an onshore base port and
then towed to the installation site. However, for Base2 and Base3, which are located more than 100 km from the
coast, not only assembly but also commissioning, including functional safety confirmation, connection of wind
turbines to the grid, and the start of operation, can be carried out. The gantry crane used for assembly must move
front, back, left, and right, and ensure that the object being lifted is firmly fixed when installing the nacelle and
blades on the tower. Additionally, during assembly, when the rotor is hoisted in the "rabbit" state with the two
blades up, the height is at its lowest, so a gantry crane with a height of at maximum 100 meters is needed. Cur-
rently, detailed discussion on such cranes is not made, however, the development of new construction devices
will be need in the future. The manufacturing of towers, nacelles, and blades are assumed to be done at the ports
along the coast. These can also be towed from other floating factories. Such floating factory unit can be de-
tached from the offshore base, towed to different locations once the wind farm at the site is completed.
On
ly
4
fo
rR
ea
din
g
at
W
CF
S
20
Fig.3. Illustrative images of Base1 (top left), Base2 (top right) and Base3 (bottom).
24
Based on the above functions, the dimensions of the bases are set as follows: Base1 is 100m × 100m, Base2
is 600m × 500m, and Base3 is 1200m × 600m. Illustrations of Base1, Base2, and Base3 are given in Fig. 3.
4 Structural Design and Simulation
Technologies for design and construction of Very Large Floating Structures (VLFSs) were developed in late
1990 to early 2000 years in Japan. VLFS can be classified to pontoon and semi-submersible types. Pontoon
types are relatively susceptible to wave loads and are suitable for calm port sea areas, while semi-submersible
types are less affected by waves and can be installed in open seas with higher wave heights. This study adopted
the semi-submersible structure for the floating offshore base installed in the EEZ.
As the main parameters determining the response of VLFS, there are the deck stiffness per unit width 𝐸𝐼 and
the restoring force coefficient per unit area 𝑘𝑐 . From these two parameters, the characteristic length can be de-
fined as shown in equation (1).[7]
4 𝐸𝐼
𝜆𝑠 = 2𝜋 √ (1)
𝑘𝑐
When the characteristic length matches the wavelength of the incident wave, the maximum quasi-static re-
sponse moment occurs, and this maximum longitudinal bending moment per unit wave amplitude is approxi-
mately given by equation (2).
1
𝑀𝑚𝑎𝑥 = √𝐸𝐼𝑘𝑐 (2)
2
Using equations (1) and (2), the structural arrangement may be examined as follows. Assuming a column
spacing of 60 meters and a column diameter of 25 meters, the restoring force coefficient per unit area is calcu-
lated to be 1340 N/m2. Assuming the deck structure with the height 4m and with the equivalent plate thickness
25mm, the characteristic period at which the wave length is equal to the characteristic length is found to be 17.4
s, which is sufficiently longer than the average wave period. Additionally, using the sectional modulus of the
deck, the bending stress at the deck is evaluated to be 262 MPa under the regular waves with a wave height of
14 meters and at the period of 17.4 s, which is judged to be below the yield stress of steel, 355 MPa. Consider-
ing the mass per unit area is around 3000kg/m2, the draft was set at 22.0 meters, and the air gap for the severest
wave is assumed to be 1.5 meters.
For the local strength, a beam model with both ends clamped is considered. When the deck structure with a
height of 4 meters is subjected to a distributed load of 𝑝 =19,612 N/m² per unit area, calculated from the mass,
the maximum moment occurs at the ends and is given for a beam with length 𝑙 by Equation (3).
𝑝𝑙 2
𝑀= (3)
12
On
ly
5
fo
rR
When 𝑙 = 60 meters, the moment is calculated to be 5.9 × 106 Nm/m. Then, the local bending stress is found
ea
59 MPa, which is well below the yield stress even though some stiffeners may have to be added locally.
This time, the structural feasibility of Base2 is further examined by using numerical simulations. For the eval-
din
uation, a numerical simulation tool based on frequency-domain hydroelasticity theory is used. It couples finite
element for structural domain and potential theory for hydrodynamic domain, respectively and the theoretical
g
details are given in [8]. As shown in Fig. 4, a fluid mesh model and an FE model consisting of beam elements
at
are created, and response analysis in waves is conducted.
W
CF
S
20
24
Fig.4. Illustration of fluid model (left) and structural beam model (right).
Fig. 5 shows the stress response function near the center of the Base2. The characteristic frequency [7] is
found at 0.36 rad/s, corresponding to the first peak. The frequency observed at 0.69 rad/s is the natural fre-
quency of the two-noded vertical bending mode of the structural deformation, and the frequency observed at
0.91 rad/s is the natural frequency of the three-noded vertical bending mode. The resonance peak may further be
reduced by inclusion of drag force terms, which however is not accounted for in the present simulations.
3.50E+01
3.00E+01
stress_amp/ζ (MPa/m)
2.50E+01
2.00E+01
1.50E+01
1.00E+01
5.00E+00
0.00E+00
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
ω[rad/s]
Fig.5. RAO of stress at central part of the Base2.
Using the stress response function and the spectrum of irregular waves (ISSC spectrum), a short-term predic-
tion was made under sea conditions equivalent to the maximum wave height in 50 years. In the actual sea area,
it is assumed that approximately 1,000 waves persist under the same sea conditions which lasts two-three hours.
Therefore, the 1/1000 maximum expected value can be considered the maximum value experienced under those
sea conditions. The 1/1000 maximum response expected value is calculated to be 293.9 MPa, which is below
the yield stress of steel (355 MPa), confirming the structural feasibility.
5 Conclusions
This study has aimed to establish a foundation for future discussions on offshore bases for wind farms deployed
in EEZ through the following investigations:
・The energy islands proposed in EU are examined and referenced for the facilities on the offshore bases.
On
ly
6
fo
rR
・The geographical and location conditions such as the distance from the coast (50km, 100km, 300km, respec-
ea
tively), water depth, and wind conditions for three different offshore bases are assumed. The necessary facilities
for each offshore base based on overseas proposals are examined to determine the layout, area, and dimensions
din
of the bases, and to design the offshore bases. The primary functions include: collecting and transmitting elec-
tricity from the wind farm to land at Base1; transmission, O&M base, assembly and repair at Base2; and con-
g
verting electricity to hydrogen or ammonia (P2X), solar panels, and a data center at Base3.
at
・A large semi-submersible type, is selected for the floating structure of the base, by considering its response
W
characteristics. The structural arrangements such as plate thickness, deck height, and column dimensions are
determined.
CF
・Its feasibility from the structural engineers’ viewpoint is confirmed based on the response functions of heave,
pitch, and stress using a numerical simulation tool based on hydroelasticity theory.
S
Many future works need to be done. They may include: an improvement of the numerical model by giving
20
appropriate viscous damping and reflecting more accurate geometry of the bases with cut-outs for port facilities;
24
an evaluation of capacity utilization factor of, for example, the assembly or O&M function of the base; a devel-
opment of efficient construction procedure of the base itself. Studying on creating calm sea area in the port area
for higher workability is a necessary and interesting issue. Finally, the CAPEX and OPEX of the base should be
clarified.
References
1. Ministry of Economy, Trade and Industry.: Green growth strategy through achieving carbon neutrality in 2050. (2021).
2. COWI.: Cost benefit analyse og klimaaftryk af energiøer i Nordsøen og Østersøen, JANUAR (2021).
3. NeoWins, https://appwdc1.infoc.nedo.go.jp/Nedo_Webgis/index.html, last accessed 2024/07/16.
4. Statistical Database od Winds and Waves around Japan, https://cloud.nmri.go.jp/apps/wwjapan/service/serve/wsTDY-
iMbcx6XQhcnof6UXrnegvAICDEJV2TM0Y8lY4bBnzpGdqKdS4gvxJnYoMryE/, last accessed 2024/07/16.
5. https://www.meti.go.jp/shingikai/energy_environment/chokyori_kaitei/pdf/001_05_00.pdf, last accessed 2024/07/16.
6. https://www.jstage.jst.go.jp/article/jwea/35/3/35_85/_pdf/-char/ja, last accessed 2024/07/16.
7. Suzuki, H., Yoshida, K.: A Consideration on the Dynamic Behavior and the Structural Design of Large Scale Floating
Structure. Journal of the Society of Naval Architects of Japan, 178, 473-483 (1995).
8. Iijima, K., Yoshida, K., Suzuki, H.: Hydrodynamic and Hydroelastic Analyses of Very Large Floating Structures in
Wave. Proceedings of OMAE1997, Vol.4, 139-145 (1997).
On
ly
fo
rR
Unlocking the Global Potential of Urban Development Offshore
ea in the Mediterranean Region

Floating Solutions
din
Gil Wang1 [0000-0002-7086-7967], Lieke Bikker2 [0009-0002-6281-4643], and Tomer Fishman2 [0000-0003-4405-2382]
g
at
1 CAMERI - Coastal and Marine Engineering Research Institute, Technion I.I.T, Haifa, Israel
2 Institute of Environmental Sciences (CML), Department of Industrial Ecology, Leiden University, Leiden, Netherlands
W
wang@technion.ac.il
CF
S
Abstract. Recent research on floating solutions represents a significant advancement in promoting responsible
20
human activities in the nearby marine environment and offshore regions. However, little is currently known
about the actual scale of the potential for using floating solutions worldwide. The upscaling course of floating
24
solutions is powered by the development needs of society and is restricted by natural environmental conditions
and physical constraints. The main motivational drivers for floating solutions include land scarcity, population
demographics, sustainability goals, recreational activities, and disaster mitigation. On the other hand, con-
straints primarily stem from issues such as wave energy, extreme weather events, zoning regulations, and
bathymetric considerations. Thus, when considering both motivational drivers and constraints as layers of data,
a model can be developed in a geographic information system (GIS) to help answer where floating solutions
could be applied on a global scale. This study explores some of the methodological frameworks necessary for
creating a global GIS model, focusing on the technical potential of floating solutions in the context of offshore
dwelling.
Keywords: GIS, Floating Structures, Offshore Urban Development.
1 Introduction
1.1 Background
Extending urban development into the adjacent marine environment has great potential for a more balanced rela-
tionship between man and planet, in the near and distant future [1]. The existing pressures in coastal urban regions,
coupled with rising populations, threaten this equilibrium, leading to challenges that impact both people and nature
[2]. Development needs often consume large amounts of natural resources and landscapes to accommodate new
roads, infrastructures (e.g., energy, water, and communication), housing, industry, and commerce. This results in
a massive loss of biodiversity and quality of life [3]. The tension between available space resources and develop-
ment needs is further exacerbated in coastal areas suffering from land scarcity, flooding risks, and earthquakes
[4]. Using a parametric GIS analysis model, our goal is to evaluate the worldwide feasibility of floating solutions
as an alternative space for urban development [5, 6]. The current research focuses only on the Mediterranean
Region showcasing the methodology and approach for global application to be used with a worldwide model.
1.2 Motivations and technological constraints

The GIS analysis is intended to address two primary questions. The first question addresses where floating solu-
tions are a viable option for offshore expansion, based on a data set that examines motivational aspects. The
second question addresses whether these locations can accommodate the associated technological constraints, or
in other words if the threshold requirements with respect to the performance criteria of the floating structure can
be met. The GIS model examines both questions to map the location with the greatest potential for the develop-
ment of floating solutions. The model is a parametric investigation tool that may examine various design inputs
reflecting the unique characteristics of any floating solution design inputted by the user.
The motivational aspects reflect the driving factors for offshore urban development which include sea level
rise, land scarcity issues, limited potential growth, flooding threats, and the risk of earthquakes near major coastal
communities. The constraints are dictated by the local environmental conditions such as wave climate and ba-
thymetry along spatial zoning considerations such as shipping routes, and nature reserves to name only a few.
Other constraints relate to the legal status of the potential location from municipal and regional perspectives,
including borders and jurisdiction areas. The chosen floating solution layout (design and arrangement) is crucial
for this analysis, as each design has distinct mechanical and structural properties for securing both ultimate and
On
ly
2
fo
rR
service limit states, thereby setting its performance criteria. In addition, the main particulars are also key features
for safety (hydrostatics) and comfort (hydromechanics), along with the compliance to more shallow waters where
draft is a limiting factor.
ea
1.3 Demonstrator din
g
The chosen demonstrator for this current study is the Modular Floating Structures (MFS) concept encompassing
at
all equal modules, type Module9000 [7–10]. This layout has two principal configurations: ‘MFS’ and ‘MFS +
W
Floating Seawall’. The MFS is suitable for calm wave climate, up to a significant wave height of ca. 2 m [8], and
CF
the MFS + Floating Seawall is suitable for moderated wave energy with significant wave height up to 4 m, but it
includes a survival condition of 1 to 100 years storm with a significant wave height of 8.5 m [9]. The latter is
S
equipped with a protective Floating Seawall configuration made of floating breakwaters [11]. Fig. 1. Shows the
demonstrator layout, the chosen geographical location, and the Floating Seawall concept tested in the Ship Hy-
dromechanics Laboratory in TU Delft.
20
(A) (B)
24
(C)
Fig. 1. (A) Typical MFS city block, made from ca. 50 modules all type Module9000 [9], (B) Mediterranean Region, one of
the nine marine regions explored using the GIS Analysis , (C) Physical testing of the Seawall concept [11] used for the
alternative of MFS+ breakwaters.
On
ly
3
fo
2 Descriptive methodology
rR
2.1 Design of the GIS model
ea
din
This GIS model uses data layers of identified relevant factors as an input for establishing the technical potential.
g
It combines them and generates maps showing on a predefined grid size where, if, and what configuration of MFS
at
model can be used. The three options are color coded, and the grid size is equivalent to the schematic spatial
arrangement given in Fig. 1. Using this data the technical potential calculations can be performed to estimate the
W
maximal achievable area where MFS could potentially be implemented. The ArcGIS Project and Python code
CF
used for data preprocessing are available upon request.
2.2 Model development, verification and validation

S
20
The GIS model is developed by combining the motivational factors for urban development offshore with the
24
limiting factors related to the performance criteria of the MFS design. The environmental conditions limiting the
technical potential of MFS include bathymetry, wave statistics, and wind climate. The zoning constraints include
marine protected areas and shipping routes (military areas, and existing infrastructure are currently outside the
scope of the model). The motivational factors for the potential of MFS include urban growth rates, topography
(e.g., slopes in proximity to coastal cities), land protected areas, and disaster mitigation such as exposure to sea-
level-rise, flooding, and earthquakes. Data validation refers to the process of examining the quality, accuracy, and
uncertainty of the used data sources before importing and processing the data. In the current study, data was
collected from academic publications to ensure the quality. Ideally, raw data sets are used, and minimal data
processing is applied within the GIS model. Accuracy can be affected by the level of resolution at which spatial
data is made available. When data is provided with a coarser resolution, the divergence between the dataset value
and the true value may be more pronounced, resulting in decreased accuracy – as is the case for the raw data
extreme wave and wind values. Regarding data uncertainty, the use or inclusion of future projected data is im-
portant to consider. When dealing with future projections of urban growth and SLR, there are inherent uncertain-
ties, as these projections are based on theoretical models. Data verification refers to the process of checking
whether the data used accurately represents the actual data, mainly by cross-referencing it against published liter-
ature.
2.3 Data preprocessing

After data validation and verification, the next step is to import the spatial data obtained in ArcGIS Pro (version
3.1.2) as individual layers. The datasets are obtained in different formats, resolution and geographical coordinate
systems. The different formats include shapefiles, raster data, and tables. Following the definitions of Esri [12]
shapefiles are simple formats storing the location, shape, and attributes of a geographic feature. These geographic
features can be represented as points, lines, or polygons. A raster is a matrix of equally sized cells, arranged in
rows and columns, where each cell contains a value representing specific information in the data set. Spatial
resolution is associated with the raster data and refers to the size of cells in a raster. Lastly, the geographical
coordinate system of the data defines where the data are located using coordinates. Both resolution and the coor-
dinate system are important to consider, as these can determine how accurately and consistently spatial data can
be measured, displayed, and analyzed.
When using multiple data layers in a GIS model or analysis, coordinate systems and resolutions should ideally
be the same. As such, before analyzing these layers together, preprocessing of the individual layers was needed
for coordinate alignment, cell resolution, and geographical coverage extent. The chosen geographical coordinate
system used as the basis throughout the GIS analysis in this thesis was WGS84, with degrees as the unit of meas-
urement. This choice is rooted in the fact that this is a widely recognized coordinate system, thus the coordinate
system used for most of the obtained datasets, and to maintain consistency throughout the GIS analysis.
The resolution of raster layers used in the GIS model refers to the size of individual grid cells within a raster
layer. Cell resolution is carefully selected with respect to the original raster data, to accommodate the global scale
extent, while maintaining a desired level of detail for the coastal shoreline in a representative cell of 0.0125 ° by
0.0125 °. This cell dimension choice is based on the schematic spatial arrangement of the MFS (Fig. 1). The
resolution of 0.0125 ° by 0.0125 ° varies roughly between 1.3 by 1.3 km at 0 degrees latitude - at the equatorial -
to 1.3 by 0.5 km at 67 ° latitude north/south – the arctic circle. The GIS model includes the geographical scope
which includes territorial, internal, and archipelagic waters. A shapefile representing these waters was collected
online and used as the spatial extent of the data layers.
On
ly
4
fo
3 Results
rR
3.1
ea
Floating urban development in the Mediterranean Region
din
The calculated suitable area in the GIS model is based on the motivational input and constraint layers outlined in
g
Section 1.2 above and it accounts only for the geographical area close to the relevant coastal cities (cities with
at
≧300,000 inhabitants), at a fixed radius of 35 km from the center of the city. The results of the global potential
analysis in the Mediterranean Region show that the useable area for floating urban solutions is more than 18,000
W
square km, an area the size of Slovenia or approximately half the size of Switzerland or Denmark. However, the
CF
key difference when comparing these areas is that the entire calculated area, in the model results, is within the
metropolitan rim, adjacent to the main cities. Thus, this is the net area relevant for future urban development,
S
excluding unusable surfaces such as mountain ridges, national parks, deserts, or even industrial terrain. Fig. 2
20
shows the investigation region (including relevant coastal cities), and the results of the calculated potential.
24
(A)
(B)
Fig. 2. (A) Main results showing the coastal cities considered in the GIS analysis, and (B) the global potential
map for the Mediterranean Region.
3.2 Taranto, Italy (40°28’0’’N, 17°14’0’’E)
The vast potential for floating solution is further revealed when zooming in on a random or an explicit coastal
city. Looking at Taranto, Italy, we can see that it has the potential of adding an additional 646 square km using
On
ly
5
fo
rR
MFS technology – three times its current size! The results of the GIS analysis are presented in Fig. 3 A. Taranto
is one of 12 Italian coastal cities analyzed in the GIS model, and it has the second highest potential suitable area
ea
for floating development after Pescara (with 766 square km). Taranto was chosen randomly to evaluate the model
din
outcome, and to explore the results using the methodology and approach outlined above. To validate whether
Taranto is a suitable location from a technical perspective, the wave climate (which is a dominating factor for
g
MFS feasibility) was cross-referenced with published research examining the wave energy at a 20 m water depth
at
contour of the Mediterranean basin [13]. This research uses the Climate Data Store of the Copernicus Climate
Change, while the GIS model uses the US Geological Survey (USGS) in partnership with Esri [12]. Good agree-
W
ment was found between the two models showing a moderate wave climate with medium to low energy. This is
CF
a crucial point to ensure the performance criteria of the MFS with respect to occupant comfort in waves. Fig. 3 C
provides the average MFS acceleration, in heave motion, used to determine occupant comfort criteria limits in
S
waves [14, 15]. Closer examination of this specific location shows additional motivational factors beyond the
20
scope of the GIS model. Taranto is located near the largest integrated cycle steel complex in Europe and has been
declared an area at high risk of environmental crisis by the Italian Council of Ministers [16]. Thus, using floating
24
urban development in the future could help improve some of the local health and environmental conditions. For
example, by creating additional urban green spaces and expanding the city towards the bay area, the exposure to
long-term PM10 (particulate matter measuring ≤ 10 µm) can be reduced as shown in Fig. 3 B.
(A) (B)
(C)
Fig. 3. (A) Taranto, Italy, 646.407 suitable square km for MFS + Seawall from an investigated area of 1,437.36
square km. (B) Long-term residential exposure to PM10 overlaid on a Google Map. Red-to-blue color coding
indicates high-to-low PM10 concentrations (the dots represent residences) [16]. (C) Average MFS module
acceleration in heave (with Floating Seawall) in significant wave heights of up to 4 m – corresponding to the
local wave climate, presented on the main criteria limits for occupant comfort [8, 9].
On
ly
6
fo
4 Conclusions
rR
ea
Using our developed global GIS model in the Mediterranean region immediately reveals the vast potential of
din
floating solutions for sustainable future growth. The total calculated potential area for urban MFS is 18,000 square
km, which can accommodate 100 million people, at a fix density of 6,000 inhabitants per square km, and has the
g
technical carrying capacity of over 300 million people [9]. A real game-changing figure in terms of reducing the
at
tension on the available space resources and development needs in coastal areas. By performing a random spot
check in Taranto, Italy, the model was validated against cross-referenced data to evaluate the main constraint for
W
its performance, the local wave climate. In addition, site-specific insights provided an additional motivational
CF
driver for MFS, related to air quality and human health. Thus, using this GIS model enables researchers and
planners to use it both for global strategies on a regional scale and tactical strategies at a specific site location. In
S
both cases, the GIS model developed can be used to improve the design of floating solutions to conform to global
20
or local requirements.
Hopefully, this model will provide a foundation for collaborations between researchers, industry, governments,
24
and institutions looking to include the adjacent marine environment as a viable alternative for sustainable future
urban growth.
References
1. Glavovic, B.C., Dawson, R., Chow, W., Garschagen, M., Haasnoot, M., Singh, C., Thomas, A.: Cross-Chapter Paper 2:
Cities and Settlements by the Se. In: Climate Change 2022: Impacts, Adaptation and Vulnerability. Contribution of
Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. pp. 2163–2194.
Cambridge University Press, Cambridge, UK and New York, USA (2022).
2. Cohen, M.: A Systematic Review of Urban Sustainability Assessment Literature. Sustainability. 9, 2048 (2017).
https://doi.org/10.3390/su9112048.
3. Simkin, R.D., Seto, K.C., McDonald, R.I., Jetz, W.: Biodiversity impacts and conservation implications of urban land
expansion projected to 2050. Proc. Natl. Acad. Sci. 119, e2117297119 (2022). https://doi.org/10.1073/pnas.2117297119.
4. de Graaf-van Dinther, R., Ovink, H.: The Five Pillars of Climate Resilience. In: de Graaf-van Dinther, R. (ed.) Climate
Resilient Urban Areas: Governance, design and development in coastal delta cities. pp. 1–19. Springer International
Publishing, Cham (2021). https://doi.org/10.1007/978-3-030-57537-3_1.
5. Wang, C.M., Wang, B.T.: Floating Solutions for Challenges Facing Humanity. In: Reddy, J.N., Wang, C.M., Luong,
V.H., and Le, A.T. (eds.) ICSCEA 2019. pp. 3–29. Springer, Singapore (2020). https://doi.org/10.1007/978-981-15-
5144-4_1.
6. Wang, C.M., Watanabe, E., Utsunomiya, T. eds: Very Large Floating Structures. CRC Press, London ; New York (2007).
7. Wang, G., Goldfeld, Y., Drimer, N.: Expanding coastal cities – Proof of feasibility for modular floating structures (MFS).
J. Clean. Prod. (2019). https://doi.org/10.1016/j.jclepro.2019.03.007.
8. Wang, G., Rosenfeld, Y., Drimer, N., Goldfeld, Y.: Occupant comfort analysis for rigid floating structures – methodology
and design assessment for offshore dwelling module. Ships Offshore Struct. 1–16 (2020).
https://doi.org/10.1080/17445302.2020.1718267.
9. Wang, G., Drimer, N., Goldfeld, Y.: Modular floating structures (MFS) for offshore dwelling a hydrodynamic analysis
in the frequency domain. Ocean Eng. 216, 107996 (2020). https://doi.org/10.1016/j.oceaneng.2020.107996.
10. Wang, S.: Simplified analytical solutions to the yaw dynamics of modular floating structures. Ocean Eng. 276, 114206
(2023). https://doi.org/10.1016/j.oceaneng.2023.114206.
11. Wang, G., Bar, D., Schreier, S.: The potential of end-of-life ships as a floating seawall and the methodical use of gap
resonance for wave attenuation. Ocean Eng. 298, 117246 (2024). https://doi.org/10.1016/j.oceaneng.2024.117246.
12. Esri: What is GIS?, https://www.esri.com/en-us/what-is-gis/overview, (2023).
13. Simonetti, I., Cappietti, L.: Mediterranean coastal wave-climate long-term trend in climate change scenarios and effects
on the optimal sizing of OWC wave energy converters. Coast. Eng. 179, 104247 (2023).
https://doi.org/10.1016/j.coastaleng.2022.104247.
14. ISO - 6897: 6897:1984, Guidelines for the evaluation of the response of occupants of fixed structures, especially buildings
and off-shore structures, to low-frequency horizontal motion (0,063 to 1 Hz). (1984).
15. ISO-10137: 10137:2007, Bases for design of structures — Serviceability of buildings and walkways against vibration.
(2007).
16. Martín-Olmedo, P., Ranzi, A., Spadaro, J.V., Portier, C., Pearce, N.: Health impact assessment of steel plant activities in
Taranto, Italy. , Copenhagen: WHO Regional Office for Europe (2023).
On
ly
fo
rR
Case study analysis: a game-changer in the identification of requirements
ea
for floating architecture
din
g
Livia Calcagni1[0000-0002-8020-6054]
1
at
Department of Planning, Design and Technology of Architecture, Sapienza University of Rome, RM, 001965, Italy
W
CF
Abstract. The paper provides extensive case study research involving existing projects distributed across dif-
S
ferent scales, countries, water types, times, and functions. Identifying and cataloguing available and exemplary
20
existing projects lays the foundations for creating an open-source archive of projects and solutions that allows
the development of standard requirements and guidelines for floating building design. A preliminary screening
24
based on specified criteria led to selecting 25 case studies, mainly designed for permanent inhabitation or
public use, that are analysed following a requirement-structured rubric. The requirement-structured rubric and
case study analysis protocol are evaluated by experts in different disciplinary domains and refined accordingly.
The case studies illustrate the different technical and technological advancements in floating architecture, the
challenges and opportunities associated with floating buildings and their aggregation in districts, and most
importantly, their compliance with each requirement. A multi-criteria evaluation matrix returns best practices
among the case studies according to their compliance with the requirement rubric and the priority distribution
among requirements according to their fulfilment among the case studies. The multi-criteria matrix proved to
be a helpful tool for comparing the different requirements and projects to make informed decisions in a struc-
tured and systematic manner.
Keywords: Case study, Requirements, Floating building design, Multi criteria.
1 Introduction
In the most virtuous European and international contexts, water-based development is gaining increasing at-
tention and is becoming a component of city plans for sustainable development and climate adaptation [1,2].
However, a significant gap in the field concerns the taxonomy and regulation of floating architecture. Since float-
ing architecture is not considered a full-fledged building typology, it is not expected to comply with norms that
regulate land-based buildings. There are no comprehensive guidelines for floating architecture, and the limited
existing design frameworks primarily emphasize safety and comfort, driven by the demanding conditions of ma-
rine environments. Identifying and cataloguing available and exemplary existing projects lays the foundations for
creating an open-source archive of projects and solutions that allows the development of standard requirements
and guidelines for floating building design. Case study research provides information grounded in practice that
has not necessarily already been taken into consideration within the theoretical and regulatory sectors [3]. The
case studies analysis provides accurate information regarding highly significant requirements and the most used
and widespread structural and technological components, materials, and construction processes, allowing the de-
velopment of general considerations on state of the art within the practice field.
The review presented in this paper covers multiple cases and a comparative analysis to draw a set of cross-case
conclusions. It is both descriptive and exploratory in purpose, as it aims at generating hypotheses for later inves-
tigation. In fact, on the one hand, it collects an extensive database of projects and presents them in a structured
analytical format; on the other, it is intended for theory-building. More specifically, it serves to further develop
and integrate an existing Performance-based Design Support Framework for floating building design, previously
developed by the author [4], with inputs from the practice field. A multi-criteria matrix (MCM) allows the iden-
tification of best practices among case studies and of a priority order amongst requirements.
1.1 Methods
Workflow. The case study assessment protocol is articulated into six steps and uses a research-from-design ap-
proach to further integrate and update the framework for floating building design (PDSF).
Step 1 Definition of the objectives for the selection of case studies and sampling methods. Case studies were
selected through selective sampling, as it represents the most appropriate method if only a limited number of
primary data sources can contribute to the study, which is the case for floating buildings. The selection of the
On
ly
2
fo
rR
projects included a wide-ranging exploratory phase in which over 100 floating buildings built worldwide were
considered. A first restriction was achieved by excluding cases with highly conditioned attributions and cases that
ea
did not adhere to the following conditions: built (still existing); realised over the last 20 years; located in inland
din
and intermediate sheltered waters; compliant with minimum occupancy and indoor comfort conditions; data avail-
ability. A second boundary condition involved their compliance with at least four of the following objectives:
g
ecosystem preservation, resource circularity, energy efficiency, wellbeing conditions, and economic feasibility.
at
The 25 selected case studies represent a database snapshot considering the distribution between residential and
non-residential projects. Ten housing projects - ranging from single housing units to multi-unit or entire districts
W
– were selected to focus the attention on permanent inhabitation. The remaining 15 projects include 5 examples
CF
of temporary accommodations like touristic facilities and 10 non-residential projects hosting leisure, food produc-
tion, infrastructure, tertiary, cultural-educational, and commercial functions. The goal of examining various func-
S
tions, beyond residential buildings, is to create a framework for living and its essential services.
20
Step 2 Development of the requirement-structured rubric for the case study assessment. The projects are presented
within the scope of the research with a particular focus on their (primary) function, scale, morphology, compliance
24
to selection criteria, material and construction components of the critical technical units, and eventually, require-
ment compliance. An analysis sheet and a requirement-structured assessment rubric was developed to provide a
simplified and structured reading method to facilitate the comparison between the different design solutions. The
assessment form combines quantitative and qualitative analysis and a specific section is assigned to the project's
compliance with each objective, schematically highlighted through the requirement-structured assessment rubric.
The assessment rubric is based on the PDSF.
Step 3 Evaluation and validation of the case study assessment rubric. The case study assessment form was sub-
jected to validation by a diverse group of scientific experts and practitioners involved in the fields of floating
architecture, hydraulic and mechanic marine engineering, ocean engineering and ship technology, ecology and
natural science, urban ecology, and environmental architecture. This collaborative validation process ensures that
the assessment rubric is robust and comprehensive, reflecting the collective expertise of the research community.
Step 4 Refinement of the case study assessment rubric. Once the reviewing process ended, changes and variations
were made to the case study assessment rubric according to the feedback received through the evaluation forms.
The final requirement rubric can be found in Calcagni et al. [4].
Step 5 Case study assessment. The selected projects were analysed based on the refined case study assessment
rubric. General information is acquired mainly through desk research integrated with field research (observational
analysis and direct interviews with the main actors involved in the design and construction process). The data is
used to provide general information about the project (identification data, site features, functional typology, di-
mensional aspects, and construction components) and to assess the project's compliance with the objectives de-
fined in Step 1. Grey literature is also used for the evaluation of the project's compliance with each requirement.
The requirement compliance is further assessed through observational research and direct interviews.
Step 6 Construction of the multi-criteria evaluation matrix. A MCM is built, placing the case studies on the x-axis
and the classes of requirements on the y-axis. The requirements are included in each case study sheet within the
section "requirement compliance." The construction of the matrix returns two results simultaneously. On one
hand, it assesses and prioritises different projects on the basis of various criteria (performance requirements) help-
ing decision-makers to objectively evaluate each project's benefits and drawbacks, and to identify best practices
among the case studies according to their compliance with the requirements. On the other hand, by defining rele-
vant criteria (presence/absence within selected projects), the matrix identifies a priority order amongst the classes
of requirements according to their observance within the practice field, providing a percentage of fulfilment for
each requirement and thus highlighting the importance of each requirement compared to the others.
2 Case study analysis
2.1 Case studies

The residential case studies include: Ijburg, Amsterdam; Schoonship, Amsterdam; Tatami house, Amsterdam;
Harnaschpolder, Delft; Harlem Shuffle, Haarlem; Urban Riggers, Copenhagen; Seattle floating home; Floating
homes in Nassauhaven; Pvrok, Střelecký Island; Maldive floating house. The non-residential case studies include
10 buildings hosting public or private functions including infrastructure, commercial, cultural-education, food
production, leisure, tertiary and greenery: GCA Floating Office, Rotterdam; Floating Pavillion, Rotterdam; Float-
ing Farm, Rotterdam; Mansa Floating Hub - MFS IV, Mindelo Bay; Jellyfish barge, Florence; Meripaviljonki Sea
Restaurant, Helsinki; Theater L’Île Ô, Lyon; Ferry Terminal, Terengganu; Joyous floating sauna, Oslo; Brock-
holes Visitor Centre, Lancashire. The other five include tourism related functions with a particular focus on
On
ly
3
fo
rR
hospitality in the sense of temporary living facilities: Marina Azzurra Resort, Lignano Sabbiadoro; DD16, Mos-
cow; Ar-che, Brandenburg; Anthénea, Dora; Botel 2.0, Porto Ceresio. The analysis of such different examples
ea
highlights the diverse approach and common challenges encountered in floating building design, especially con-
din
cerning the legal status and regulatory standards. The debate about the juridical and financial status of floating
houses and their classification as either boats or houses is ongoing in several countries. In the Netherlands, floating
g
houses are legally recognized as homes if they are moored to the ground (waterbed) in a specific location [5]. This
at
has led to banks offering mortgages and insurance boosting market demand and trust among potential house buy-
ers. Moreover, floating houses require an ordinary building permit and should fulfil ordinary building standards.
W
Other countries, such as Finland, Norway, France, and Italy, lack specific codes for floating buildings, making
CF
permit acquisition complex and time-consuming, often resulting in structures being classified as houseboats, boats
or barges. A common challenge regards the difficulties encountered in buying a plot of water or obtaining con-
S
struction permits to build on water, that significantly extends project timelines.The construction of the Meripavil-
20
jonki Sea Restaurant, for instance, took up to ten years, mostly due to bureaucratic delays in obtaining permits.
Public floating buildings, like restaurants and museums, face stricter regulations than private ones, leading to
24
their registration as buildings that comply with land-based laws, unlike residential floating units typically classi-
fied as houseboats. For instance, Botel 2.0 is registered as a navigation unit in northern Italy to secure legal per-
mits, despite meeting land-based building requirements, while Anthenea is built with boat certification but does
not comply with land-based rules.
The analysis revealed that only a few projects, like Botel 2.0 and the Floating Office in Rotterdam, include
onboard instructions for safe building use, detailing occupancy limits, emergency procedures, and resource con-
sumption. Climbing and holding devices are not provided if other measures like parapets along the perimeter,
prevent falling into the water. Case studies show that tourist leisure accommodations or housing units prioritize
water access for recreational purposes rather than safety reasons. Also climate and microclimate conditions
strongly affect which requirements are observed. For instance, thermal variation resistance of pipelines is only
met in areas prone to water freezing which require frost resistance, such as IJburg, Oslo sauna, or DD16.
Ultimately, the case studies have shown the great number of actors involved, directly or indirectly, in floating
development, categorized into three groups: public actors, private actors, and society. Public actors, such as mu-
nicipal committees and environmental agencies, operate at both the process and governance levels. Private actors
include architects and developers directly involved in projects, alongside investors and port authorities indirectly.
Society actors primarily encompass users and residents, as well as mortgage providers, housing corporations, and
environmental activist groups at an indirect level.
2.2 Multi criteria analysis: best practices

Unlike conventional MCMs, this matrix can be read in both directions and all requirements are equally rated
with a common weight of 1. According to the evaluation of the case studies (alternatives) based on the require-
ments (criteria) (Fig.1), all case studies address more than 60% of the requirements.
Fig. 1. A The graph shows the degree of compliance of the case studies (alternatives) with the design requirements (criteria).
On
ly
4
fo
2.3
rR
Multi criteria analysis: priority order among requirements
ea
The bee-swarm plot diagram in figure 2 displays the distribution of items (classes of requirements) over a
continuous dimension (score). The only requirements met by all case studies (100%) are the following: Mechan-
din
ical resistance to static actions, Mechanical resistance to dynamic actions, Structural continuity with sub-struc-
g
ture, Structural fire integrity, Indoor temperature level and control, Indoor humidity level and control, Ventilation
at
control, Artificial illumination level and control, Occupancy rate, Ease of use and maneuver, Hygroscopicity,
Freeboard, Watertight integrity, Mooring arrangements, Anchoring provisions and arrangements, Maintainabil-
W
ity-repairability, Pipeline watertight integrity. These requirements mainly address safety, wellbeing, and buoy-
CF
ancy. Structural and fire safety are followed by all designers due to their mandatory nature across fields like
architecture and offshore engineering. Wellbeing most conventional requirements like thermal, acoustic, and vis-
S
ual comfort, are met because they align with on-land building regulations and, therefore, followed by designers
when building on water. Motion comfort is widely fulfilled (98%), while biophilia, active design, or behavioural
20
or community engagement are generally not considered a priority in design and account respectively for 86%,
24
42%, and 38%. About 90% of projects address plant system maintainability, repairability, and watertight integrity.
Other essential requirements, accounting for 96%, are Overtopping reduction (clearance above water) and
Non-slip resistance within the safety class, Mobility (towing arrangements) within usability class, Atmospheric
agents resistance (management class), Plant system integration (integrability class), Avoid interference with pro-
tected areas (Environmental regeneration class), and Adaptability of pipe-lines to water fluctuations (plant system
class). Requirements like Overtopping reduction (that prevents the threat of waves or tides washing over the deck)
and Non-slip resistance (of the walking surfaces in contact with water) are observed by many projects because
they are part of the safety class of demand despite being specifically related to the water environment. Require-
ment Mobility (towing arrangements) is undoubtedly crucial for the building's maintenance, repair, and relocation.
Therefore, it is not surprising that most case studies are equipped with towing arrangements.
Requirements like Atmospheric agents resistance (of exposed materials and components) and Adaptability of
pipelines to water fluctuations are extremely important for floating buildings as the first ensures their long-term
durability, and the latter their daily survival and adaptation to the changing site conditions. Adaptable pipelines
are essential for ensuring the constant operation of the building's plant systems under all weather and climate
conditions. Most case studies reasonably observe requirement Plant system integration, as it guarantees the overall
efficiency of the building. Protected areas are regulated by specific regulations, which overrule building codes.
Therefore, any construction must comply with the protected areas' regulations. Since the case studies are located
in countries where regulations that protect these areas are in force, this requirement could be removed as it is
implicit. However, since the framework could be applied anywhere, keeping the requirement as a reminder in the
PDSF may be necessary.
Under keel clearance is revealed to be highly observed by designers (92% of case studies). This requirement
is crucial as it is connected to several aspects, including safety, environmental impact, and mobility. It prevents
the building from grounding, minimizes the impact of the building on the seabed and marine life, and allows it to
manoeuvre the building easily and safely in shallow waters.
The requirements of the Environment regeneration class are accomplished by 68% on average. Only slightly
more than one-third of the case studies address requirement Foster biodiversity. Between fifty and sixty percent
of the case studies meet the requirements Avoid impingement, entrainment, entanglement, and impairment of bi-
ostructure and aquatic vegetation, Avoid unnecessary reduction/obstruction and facilitate incoming sunlight in
water, and Reduce underwater noise sources (hydroacoustic energy). More attention is paid to Reduce light pol-
lution and avoid underwater illumination at night, which is met by one-eighth of the case studies.
Rational use of resources, as depicted in the graph in the figure, is the least observed class of demand. Circular
use of materials or Water collection, treatment and reuse are met by only half of the case studies, while Waste-
water treatment optimization by slightly more than one-third. It is interesting to highlight how 64% of the projects
use renewable energy resources, while only 30% use marine renewable energy. The only requirements observed
by most case studies (86%) are Use of bioclimatic passive solutions and Safe waste storage and disposal.
Only 26% of the case studies have an on-board user manual. This requirement comes from the NTA 8111 rules
in force in the Netherlands. Therefore, most projects realized after 2008 in the Netherlands should fulfil it. How-
ever, the NTA 8111 requires to provide the user of the floating structure with a manual containing the load limi-
tations, with no reference to how to operate the floating building safely and efficiently, how to maintain the float-
ing building properly, and how to respond to emergencies.
Among the other requirements that are worth noting is requirement Collision risk reduction arrangements, that
is only met by 24% of case studies. This requirement comes from maritime regulations and refers to moving
objects. The case studies that meet this requirement are equipped with fender devices mainly to allow easy boat
docking rather than to protect the floating foundation bodies and decks from eventual collision with other
On
ly
5
fo
rR
structures or vessels. This is mainly due to two reasons. On one hand, most floating buildings are located in areas
with low levels of boat traffic and thus lower risk of collision. On the other, the buildings are kept apart by mooring
ea
systems that limit the horizontal movements of the structures. Aside from this, several projects do not comply
din
with any specific water-related regulation besides conventional architectural codes that reasonably do not include
anti-collision arrangements.
g
Requirements Emergency and signal lighting and Sink risk prevention indicators are addressed in only 8% of
at
the projects. According to building codes, the first is mandatory only for public buildings. This is the most likely
explanation for it not being met by most case studies. The case studies that are public and meant for public use
W
are only three: the Theater L’Île Ô, the Ferry Terminal, and the Brockholes Visitor Centre. As no information is
CF
available for the visitor centre regarding this requirement, it is only met with certainty by two case studies (8%).
Emergency and signal lighting and Sink risk prevention indicators can be expensive to install and maintain. This
S
can make them a less attractive option for some developers, especially those working on limited-budget projects
20
and residential or private buildings that are not obliged to comply with these requirements according to on-land
building regulations. Sink prevention indicators come from the shipping field. They could be extremely useful to
24
ensure the safety of occupants and the structure by providing early warning and, thus, sufficient time to evacuate
and take other precautionary measures. However, they are not considered by designers, as most projects comply
with on-land-based building regulations that do not include this requirement as the buildings do not encounter this
risk.
The least observed requirement is Pest and dangerous animal prevention, likely due to its conflict (mutual
exclusion) with several other requirements like Biophilia, Foster biodiversity, or Toxic emission control of mate-
rials. Designers and developers may prioritize sustainability and ecological design principles, leading to less em-
phasis on traditional pest and animal control methods in favor of alternatives that minimize environmental impact.
Additionally, many case study sites are in sheltered waters near urban areas with low wildlife, resulting in a lower
need for protection against dangerous animals. However, this doesn’t diminish the importance of including this
requirement in a design framework, as certain locations may attract insects and pests.
Fig. 2. Bee-swarm plot diagram displaying the fulfilment of the classes of requirements.
2.4 Theoretical integrations from practice

Despite being mentioned by very few regulations (UNI 8289; CLC SOR/2010-120), Furniture integrability (usa-
bility class) is often observed in projects (more than two-thirds of the case studies). Functional/spatial flexibility,
which has no regulatory system supporting it but only scientific literature, has been proven extremely important
as it is fulfilled by 80% of the case studies. Case study n°25 (Botel 2.0) has an automated domotic system that
warns users when they exceed consumption limits. This system could be considered a digital version of the
onboard user manual for the information related to the efficient energy operation of the building. This suggests
how Onboard user manual could work with real-time monitoring systems requirement and integrated real-time
On
ly
6
fo
rR
information and alarms. The requirement CO2 absorption design solutions has been further integrated with ex-
amples from case studies that use photocatalytic materials and natural materials with embodied carbon properties.
ea
The case studies n° 2 (Schoonship), n°10 (Maldive floating city prototype), n° 11 (Floating Office Rotterdam), n°
din
12 (Floating Pavilion), n° 15 (Jellyfish barge), and n° 25 (Botel 2.0) have highlighted how, being in a water-based
habitat, Water collection, treatment, and reuse should also embrace systems that extract and treat water from the
g
surrounding environment. Desalination processes can remove high salt concentrations from seawater, while treat-
at
ments like filtration, aeration, and disinfection can eliminate bacteria, algae, and chemicals from lake and river
water to provide water for domestic use. Despite being mentioned only by certification systems and not by regu-
W
lations, Limit water consumption is observed in 18% of the case studies. Several projects include water-saving
CF
taps equipped with technologies that limit water consumption, recirculating showers, and similar devices. This
performance requirement, while not explicitly addressed in codes and guidelines, is of utmost importance within
S
international and Europeans goals and objectives.
20
24
3 Conclusions
The database does not claim to provide a complete collection of all existing floating buildings. However, it sets
up an initial repertoire of projects that can be further and constantly updated by other designers using the same
methodology. Considering how floating buildings are relatively rare compared to traditional buildings, this col-
lection provides a decent number of projects to draw significant conclusions. Overall, the case study analysis has
confirmed the framework's requirements, leading to its validation within practice. The MCM proves to be an
effective tool for comparing projects based on performance requirements. Overall, the case study research high-
lights 6 major issues.
1. Legal and regulatory challenges: floating buildings face significant regulatory and legal barriers, par-
ticularly regarding their classification as houses or boats. These challenges affect mortgage availability,
insurance, and the time required to obtain construction permits.
2. Sustainability and compliance: non-residential and tourist accommodations generally have higher com-
pliance with safety, accessibility, and sustainability requirements compared to residential buildings.
Shared systems in larger developments also contribute to higher compliance rates.
3. Material use and technical innovations: projects employ a range of materials and innovations showing
the potential for diverse and sustainable construction solutions in floating environments.
4. Priority on safety and durability: structural integrity, safety, and well-being, are the top priorities
across the case studies, demonstrating how floating buildings are equivalent to land-based building in
terms of compliance with these essential requirements.
5. Environmental preservation: while many projects meet energy efficiency and resource circularity
goals, biodiversity protection and underwater noise reduction are less commonly fulfilled, indicating
areas for improvement in integrating environmental considerations.
6. Real-time monitoring: some projects highlight the potential of real-time monitoring systems to enhance
energy management and improve floating building operation efficiency.
These findings emphasize the need for further development of international standards and guidelines and the
importance of drawing on practical, built examples to inform and develop these guidelines.
References
1. Ernst, L., Dinther, R. E. D. G., Peek, G. J., & Loorbach, D. A. Sustainable urban transformation and sustainability transi-
tions ; conceptual framework and case study. Journal of Cleaner Production, 112, 2988–2999 (2016).
2. Dal Bo Zanon, B., Roeffen, B., Czapiewska, K. M., de Graaf-Van Dinther, R. E., & Mooij, P. R. Potential of floating
production for delta and coastal cities. Journal of Cleaner Production, 151, 10–20 (2017).
3. Groat, L. N., & Wang, D. Architectural research methods. John Wiley & Sons (2013).
4. Calcagni, L., Hensel, D.S., Hensel, M.U., Battisti, A. A Performance-Based Design Framework for Floating Architecture.
Trade-Offs and Correlations Between Requirements for Multiple Criteria Decision-Making Optimization. In: Ikoma, T.,
Tabeta, S., Lim, S.H., Wang, C.M. (eds) Proceedings of the Third World Conference on Floating Solutions. WCFS 2023.
Lecture Notes in Civil Engineering, vol 465. Springer, Singapore (2024).
5. Vermande, H. (2009). Drijvende woningen en de bouwregelgeving, Handreiking voor ontwikkelaars, bouwers en
gemeentelijke plantoetsers. Floating Houses and Construction Legislation, Guidelines for Developers, Contractors and
Municipalities).
On
ly
fo
rR
ea
din
g
at
Floating Cities and Waste Management: A Case Study of a Waste
W
CF
Collection Vessel
S
20
Artur Karczewski1[0000-0002-8797-4355], Maciej Reichel1[0000-0002-9876-9584], Hanna Pruszko10000-0003-1902-2659], Ewa
Marcinkiewicz2[0009-0002-5152-9150]
1
24
Department of Mechanical Engineering and Ship Technology, Gdansk University of Technology, Gdansk, Poland
2 Seatech Engineering LTD, Gdansk, Poland
artkarcz@pg.edu.pl
Abstract. Floating cities, conceived as a remedy for overcrowding and the challenges posed by rising sea
levels, present distinct hurdles, particularly in waste management. The case study of a Waste Collection Vessel
(WCV) seeks to examine these associated challenges. It delves into the assumptions, limitations, and risks
involved in addressing garbage management in large floating habitats. The analysis encompasses both formal
situations and technical solutions. The culmination of this study is the proposal for a "floating garbage truck"
- a zero-emission autonomous catamaran designed to be an intelligent vessel equipped with advanced technol-
ogies. It will be highly flexible and ideally suited for applications in floating cities, capable of tasks such as
removing surface garbage, collecting garbage containers from habitats, addressing oil spills, and cutting weeds.
The design of WCV ship was developed because of the ZeroWastePorts project.
Keywords: floating structure; waste management; WCV; autonomous technologies
1 Introduction
In the face of global challenges such as climate change, rising sea levels, and environmental pollution, the con-
cept of floating cities is gaining significance. Floating cities are visionary solutions to global problems like over-
population, rising sea levels, and land scarcity [1]. These innovative, floating habitats will offer adaptable, sus-
tainable living spaces that open new possibilities for habitation, development, and environmental protection [2],
[3].
Despite many advantages, like any major human settlement, floating cities will have to address the issue of
waste management also. That includes waste segregation and their collection, transportation, processing, and
recycling [4]. Waste will accumulate in the inner parts of floating cities, as well as in ports, marinas, and associ-
ated water infrastructure. Additionally, there is a heightened risk of oil spills due to operational accidents during
loading, unloading, and bunkering, and heavy ship traffic complicates cleaning efforts in these areas.
These factors indicate the need for an ecological vessel dedicated to cleaning the aquatic environment, both in
the waters surrounding the city's edges and in ports. This vessel's tasks would include regular cleaning of float-
ing debris and pollutants, cutting surface-level weeds, and emergency actions such as neutralizing or eliminating
oil spills caused by leaks.
This article presents the conceptualization process of such a vessel, the Waste Collecting Vessel (WCV). It is an
autonomous catamaran, 14 meters in length, with zero emissions. This intelligent vessel will be equipped with
advanced technologies, highly flexible, and perfectly suited for sheltered terminal and port applications. It will
be capable of rapid response and water cleaning from pollutants, as well as recovering any oil contaminants.
On
ly
2
fo
2 Methodology
rR
ea
A standard method for designing a small craft was adopted, supplemented at the preliminary stage by analyses of
din
the following unusual elements of the design process [5], [6], [7]:
 Modularity: The ship should have the ability to be reconfigured depending on the type of task.
g
 Modes of exploitation: The ship performs contrasting functions, several ships should be designed in
one, like a multitasker.
at
 Degree of autonomy: The ship should exhibit the highest possible degree of autonomy.
W
 Task grouping: The ship's tasks should be grouped according to its modularity.
CF
S
2.1 Modularity
20
Depending on the type of task to be performed, the operating ship will consist of a different number of modules:
24
 A single middle module (SET 1).
 Bow and middle modules (SET 2).
 Middle and aft modules (SET 3).
 A complete set of three modules (SET 4).
The bow module has been designed in three variants – 1a (floating weir oil skimmer), 1b (smooth drum type oil
skimmer), 1c (weed harvest option), therefore SET 2 will be available in the variants SET 2a, SET 2b and SET
2c. Only one of them can be used at a time. The middle and stern modules are designed in only one variant – for
seaweed collection.
2.2 Modes of exploitation

Ship should be designed for the following modes of exploitation.
 Waste collection from water (TYPE 1): Vessel designed to collect waste and garbage from water
surface (to a depth of approx. 20 cm) while driving autonomously. For this operation ship will need
SET 1 and occasionally SET 3 (if there is a lot of waste on a water surface).
 Oil spill recovery from water surface (TYPE 2): Vessel designed to collect oil spills from water sur-
face. Depending on area depth and size of oil spill ship is to use one of two skimmers, variants a or b.
For oil spill recovery there is need to use SET 2 (skimmer module).
 Seaweed harvesting (TYPE 3): Vessel designed to cut and/or collect seaweed in port area. For this
operation there is need for SET 2c (harvest module) and occasionally SET 4.
 Collecting garbage from other ships (TYPE 4): Vessel designed to collect garbage and waste from
other ships. For this operation there is need for SET 3. Ship will park side-to-side to the next vessel
so it can throw waste into cargo boxes on the deck of the ship.
2.3 Degrees of autonomy

The vessel, is designing, should operate in one of four degrees of autonomy [8]:
 Degree One (M): Ship with automated processes and decision support: Seafarers are on board to op-
erate and control shipboard systems and functions. Some operations may be automated and at times
be unsupervised but with seafarers on board ready to take control.
 Degree Two (R): Remotely controlled ship with seafarers on board: The ship is controlled and oper-
ated from another location. Seafarers are available on board to take control and to operate the ship-
board systems and functions.
 Degree Three (RU): Remotely controlled ship without seafarers on board: The ship is controlled and
operated from another location. There are no seafarers on board. Some vessels functions will be re-
motely controlled, and other become automated.
 Degree Four (A): Fully autonomous ship: The operating system of the ship can make decisions and
determine actions by itself.
The extent of support from on-board personnel and the level of automation will depend on the task to be per-
formed. The vessel will have a lower level of autonomy in certain operations, remaining under human control.
On
ly
3
fo
2.4 Task grouping
rR
ea
To optimize the number and types of modules for the designed ship, tasks were grouped accordingly. The number
of modules used will depend on the planned amount of garbage to be collected and the intended degree of auton-
din
omy for the ship. Considering the possible types of operations, module configurations, and degrees of autonomy,
g
the ship will be able to use for 26 tasks listed below. Each task will be appropriately programmed, and depending
on the selected task, the corresponding ship reconfiguration procedure will be activated.
 Waste collection from water:
at
W
o TASKS 1 – 4
CF
If the ship will be used to collect incidental waste from a specific area daily, it can be ex-
pected that only the middle module (SET 1) will be sufficient for such a task. This middle
S
module will be a self-contained unit equipped with all the devices needed for independent
navigation and waste collection facilities with a small tank for the collected waste. Depend-
20
ing on the needs the middle module (or the middle and aft modules) could be operate in any
24
of the 4 levels of autonomy. In the event of a failure a single module as a fully autonomous
ship is, due to its smallest size, less dangerous for other users than a unit in other configura-
tions.
o TASKS 5 – 8
If it is necessary to collect more garbage than the permissible volume of the tank of the mid-
dle module, the ship will work as the middle module + stern module configuration (SET 3).
The set consisting of the middle and stern modules will be able to operate in all 4 degrees of
autonomy, like TASK 1 -4.
 Oil spill recovery:
o TASKS 9 – 14
Oil spill removing will be possible using the middle module connected to the bow module
(SET 2), on which oil collection devices will be mounted 1a or 1b.
In the case of collecting oil spills, the ship will be able to operate in 3 degrees of autonomy:
 human-operated ship with automated processes and decision support,
 remotely controlled ship with seafarers on board,
 remotely controlled ship without seafarers on board.
For safety reasons, the ship will not be able to operate as an unmanned autonomous vessel for these tasks.
 Weeds cutting and collecting:
o TASKS 15 – 18
For weeds cutting task will be used the middle module and the bow module (SET 2) equipped
with dedicated devices or all three modules (SET 4) in case when there are a lot of weeds to
be harvested. In these tasks the vessel can operate as:
 conventionally human-operated ship,
 remotely controlled ship from the deck or from the quay (or operation station).
Due to the elevated risk of damage caused by the bow module, fully automatic work, and remotely controlled
ship without seafarer on board is not recommended in this task.
 Collecting garbage from other ships:
o TASKS 19 – 26
Garbage collecting from other ships will be possible using only middle module (SET 1) or
middle module connected with aft module (SET 3). In this task, ship can operate in all 4
autonomic levels of control.
A summary of all tasks with respect to the degree of autonomy, the number of intended modules and the type of
operation to be done is shown in Table 1.
Table 1. A summary of all assumptions in range waste collection in design process.

MODES OF TASK SET AUTONOMY
EXPLOITATION NUMBER 1 2 3 4 M R RU A
a b c
TYPE 1 1 + +
2 + +
3 + +
4 + +
On
ly
4
fo 5
6 rR +
+
+
+
7
ea + +
din
8 + +
TYPE 2 9 + +
g
10 + +
at
11 + +
12 + +
13 +
W +
CF
14 + +
TYPE 3 15 + +
S
16 + +
20
17 + +
18 + +
TYPE 4 19
20
21
+
+
+
24
+
+
+
22 + +
23 + +
24 + +
25 + +
26 + +
3 Results – Waste Collection Vessel
The result of the design work is zero emission autonomous catamaran, 14 metres in length. The designed ship
consists of three separate modules: the bow module (1), the middle module (2), and the stern module (3) (see
Fig. 1). Only the middle module has its own propulsion. The other two modules, when connected to the middle
module, are either towed or pushed. It is possible to control the ship in various module configurations, but al-
ways using the middle propulsion module. All working equipment on the different modules will be directly ac-
cessible from the deck. Each module is designed with a single flat deck, allowing personnel to move around eas-
ily and have access to all onboard and operational equipment.
The middle module (2) will serve as the propulsion module for the entire ship assembly. Below deck, it will
space of the propeller, battery rooms, and electrical room. This module will also be equipped with waste collec-
tion equipment, a conveyor belt for transporting collected pollutants, and a small onboard tank, making it capa-
ble of collecting surface waste as an independent unit.
The unpowered stern module (3) will be used exclusively for transporting the tank containing surface-collected
waste, gathered algae, or waste collected from other ships. It is equipped with an additional tank, which, com-
bined with the tank in the middle module, doubles its capacity.
The ship's superstructure is located on the middle module. It includes a control station equipped with a seat and
a console panel for direct manual control. The station is covered to ensure operator comfort and will be mechan-
ically and electronically secured against unauthorized access.
On
ly
5
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 1. General view of the ship configuration and modules.
4 Discussion
Integration of modular design principles in shipbuilding processes represents a significant advancement in ma-
rine engineering. This study utilizes the flexibility and adaptability of modular construction to meet diverse op-
erational requirements and environmental challenges. The ability to reconfigure the ship according to task type,
as described in the article, underscores the importance of modularity in achieving versatile and multifunctional
marine units.
The concept of modularity, particularly the ability to switch between different configurations such as SET 1,
SET 2, SET 3, and SET 4, demonstrates an important level of innovation. This approach not only enhances op-
erational efficiency but also promotes sustainable development by reducing the need for multiple specialized
units. The specific design of bow, middle, and stern modules tailored for tasks such as waste collection, oil spill
removal, and algae harvesting provides a practical example of modularity applied in real-world scenarios.
Furthermore, discussing levels of vessel autonomy in accordance with [8], highlights the evolution of maritime
operations towards increased automation. The four autonomy levels provide a structural framework to under-
stand how ships can transition from manual operations to fully autonomous systems. Gradual integration of au-
tonomous capabilities ensures operational safety and reliability while leveraging technological advancements.
Task grouping based on operational requirements and autonomy levels is another key aspect discussed in the
article. By categorizing tasks such as waste collection and oil spill removal, the study effectively matches modu-
lar configurations to intended functions and autonomy levels. The task grouping strategy optimizes ship perfor-
mance and enhances its ability to adapt to diverse marine environments and operational demands.
5 Conclusions
Floating cities are emerging as a visionary solution to global challenges such as climate change, rising sea levels,
and land scarcity. However, like any large human settlement, floating cities must tackle common issues such as
waste management. This includes segregation, collection, transport, processing, and recycling of waste. Waste
will accumulate on solid areas of floating cities, as well as in ports, marinas, and associated water infrastructures.
These premises underscore the need for environmentally friendly water cleaning ships, both in the waters sur-
rounding city edges and in ports. Their task would involve regular, periodic cleaning of floating debris and
On
ly
6
fo
rR
pollutants, surface weed cutting, as well as intervention actions when necessary, such as neutralizing or eliminat-
ing oil spills. These ships must be advanced and innovative to meet future challenges. Therefore, this article pro-
ea
poses an approach to ship design based on modularity, combined with various levels of autonomy, presenting a
din
comprehensive solution to contemporary challenges in ship design. To promote efficiency and sustainable devel-
opment, new standards in ship design and operations must be established. The detailed analysis and practical
g
examples presented in this research serve as valuable reference points for future advancements in this field.
at
W
Acknowledgement
CF
The research was supported by the National Center for Research and Development (NCRD) within the project
S
ZeroWastePorts: Zero Emission Waste Collecting Vessel to Use in Ports and Close-to-Shore Areas. Agreement
no PL-TW/IX/43/ZEROWASTEPORTS/2022. The project visualization was supported by the Academic Com-
20
puter Centre in Gdansk (CI TASK). All support is highly appreciated by the authors. Authors thank the reviewers
24
for their valuable comments.
References
[1] B. Neumann, A. Vafeidis, J. Zimmermann i R. Nicholls, „Future Coastal Population Growth and
Exposure to Sea-Level Rise and Coastal Flooding - A Global Assesment,” PloS ONE, 2015.
[2] Blue21, „Sustainable Floating Cities as a Future Solution to Sea Level Rise,” In Sustainable
Devolopment Goals Report 2019, 2019.
[3] D. Fornari, G. Landolfi, G. Paci i S. Ceccacci, „Floating Architecture: The Challenge of a New Urban
Development on Water,” Sustainability, 2018.
[4] J. Smith, A. Lee i S. Kumar, „Waste Management in Floating Cities: Challenges and Opportunities,”
Journal of Environmental Management, 2022.
[5] P. Gełesz, A. Karczewski, J. Kozak, W. Liwin i Ł. Piatek, „Design Methodology for Small Passenger
Ships On the Example of the Ferryboat Motława 2 Driven by Hybrid Propulsion System,” Polish Maritime
Research, pp. 67-73, 2017.
[6] Choi, Minjoo; Lee, Jaekyeong;, „Modular Ship Design: Rapid Prototyping and Enhancing Efficiency
through Design Modules,” w 15th International Marine Design Conference (IMDC-2024), Amsterdam,
2024.
[7] S. O. Erikstad, „Design for Modularity: Volume 1: Optimisation of Ship Design and Operaion Life
Cycle,” w A Holistic Approach to Ship Design, Springer, 2019, pp. 2-28.
[8] IMO, „MSC.1/Circ.1638 Outcome of the regulatory Scoping Exercise for the Use Maritime
Autonomous Surface Ships (MASS),” IMO, London, 2021.
On
ly
fo
rR
Hydroelastic analysis of a large container ship with forward speed based on
ea
multi-module method
din
g
Yusong Ye1,2, Shixiao Fu1,2,*, Shiyuan Zhang1,2
1
at
State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai, 200240, China
W
2 Collaborative Innovation Centre for Advanced Ship and Deep-Sea Exploration, Shanghai, 200240, China
CF
shixiao.fu@sjtu.edu.cn
S
20
Abstract. In this study, a beam-connected-discrete-modules (BCDM) hydroelastic method is employed to
conduct hydroelastic analysis of large container ships with forward speed. The large container ship is discre-
24
tized into several rigid modules connected by a hull girder. Considering the hydrodynamic interactions between
modules, the multi-body hydrodynamic theory is adopted to obtain the hydrodynamic coefficients of each
module. Subsequently, the hydroelastic equation is established by coupling the hydrodynamics of modules
with the structural stiffness of the hull girder. The BCDM method replaces the boundary value problem solved
in the modal space with extensively studied multi-body hydrodynamics. In particular, the derivation of gener-
alized restoring matrices associated with flexible modes could be avoided, which has been ongoing debate for
more than two decades. The effect of flexible deformation on a large container ship is investigated. To account
for the influence of forward speed, the wave natural frequency is replaced by the encounter frequency. The
results show that the flexible deformation increases sectional forces, especially at the structural vibration fre-
quency.
Keywords: Beam-connected-discrete-module method, Large container ship, Global hydroelasticity, Forward

speed.
1 Introduction
With the enhancement of shipping demand, large container ships play an increasingly important role in ship-
ping. However, due to the large dimensions, the structural vibration frequency of the ship decreases. When ships
advance in waves, the encounter frequency will approach the structural vibration frequency, which excites spring-
ing response and increases fatigue damage significantly.
To ensure the structural safety of the container ship, it is necessary to carry out the hydroelastic analysis of
large ships with forward speed. The modal superposition method is a popular approach to solve the problem. Wu
et al. [1] employed the three-dimensional frequency domain linear hydroelastic method to conduct a series of
numerical calculations on a submerged ellipsoid near the wave surface and a semi-ellipsoidal elastic surface hull
navigating on the wave surface. Tian [2] established a second-order nonlinear hydroelastic analysis method for
sailing ships considering the steady flow, and analyzed the hydroelastic response of the SWATH ship. Duan et al.
[3] investigated hydroelastic response of ship with speed in time domain based on the TEBEM method. Chen et
al. [4] established a three-dimensional time-domain nonlinear hydroelastic analysis method in irregular oblique
waves, and validated the method by comparing the experimental and numerical results of a high-speed trimaran.
Zhang [5] integrated the computational advantages of the two-dimensional to extend three-dimensional time-do-
main hydroelasticity theory, and put forward the determination method of nonlinear design wave loads on ships
with large bow flare. Hu et al. [6] used the second-order nonlinear three-dimensional frequency domain hydroe-
lasticity method to calculate the springing response of a large ore ship. Despite the widely use of the modal su-
perposition method, there are still some limitations, such as the controversial generalized hydrostatic coefficient
[7,8].
In addition to the aforementioned superposition method, in recent years, the beam-connected-discrete-module
(BCDM) [9] method for analyzing the hydroelastic response of very large floating structures (VLFS) and ships
has been developed. This method avoid the problems of the controversial generalized hydrostatic coefficient. Zhu
et al. [10] analyzed the hydroelastic response of the VLFS with moving loads by the BCDM method. Bispo et al.
[11] investigated a moored VLFS composed by a set of hinged plates, and calculated the effects of the hinged
plates on the structural displacement and the mooring cable loads. Wei et al. [12] employed the BCDM method
to calculate the hydroelastic response of the pontoon bridge in inhomogeneous sea environment, and analyzed the
influence of the inhomogeneous waves with spatial varying wave heights/periods. Li et al. [13] extended the
BCDM method to calculate hydroelastic response of a floating wind turbine with a large pontoon, and the effects
On
ly
2
fo
rR
of elastic deformation on bending moment are analyzed. Based on the second-order hydrodynamics of multi-body,
Li et al. [14] considered the effect of the first-order hydroelastic response on the flow field and extended the
ea
BCDM method to the second-order hydrodynamic. Zhang et al. [15] developed a two-step stress analysis method
din
to incorporate the global hydroelasticity and local bending effect based on the BCDM method, which extended
the method to the stress analysis.
g
Previous studies of the BCDM method focused on no-speed hydroelastic analysis. For the hydroelastic analysis
at
with forward speed, Bakti et al. [16] and Zhang et al. [17] used Wigley hulls to investigate hydroelastic responses
of ship with forward speed based on BCDM method. In this paper, the BCDM method is utilized to analyze the
W
hydroelastic response of a large container ship with forward speed, realizing the engineering application of the
CF
BCDM method in ship with forward speed. The large container ship is discretized into several rigid modules,
which are connected by a hull girder. The hydroelastic equation is established by coupling the hydrodynamics of
S
modules with the structural stiffness of the hull girder. In order to consider the effect of speed, the wave natural
20
frequency is replaced by the encounter frequency. Subsequently, the hydroelastic response of the container ship
is investigated.
2 Theoretical background
24
According to the BCDM method, the large container ship is firstly discretized into several rigid modules con-
nected by a hull girder. Then, the multibody hydrodynamic theory is adopted to obtain the wave excitation force,
added mass, and damping coefficient acting on each module. After that, the multibody hydrodynamic force is
coupled with the structural stiffness of the beam to establish the frequency domain hydroelastic equation (see Fig.
1). Due to the warping phenomenon in large container ships, a thin-walled beam element is adopted [18].
Fig. 1. Flow diagram of the hydroelastic analysis of the container ship.
𝜔 𝑀 𝐴 𝜔 𝑖𝜔 𝑐 𝐶 𝜔 𝐾 𝑘 𝑢 𝑓 𝜔 (1)
On
ly
3
fo
rR
where N is the number of thin-walled beam nodes with seven DOF; [k] is the stiffness matrix of the thin-walled
beam; [c] is the structural damping matrix of the thin-walled beam; [M], [A], [C], [K] are the mass, added mass,
ea
damping coefficient and hydrostatic restoring coefficient matrices of the rigid bodies, respectively; {u} is the node
din
displacement vector; and {f} is the wave excitation force vector.
In order to consider the effect of the forward speed, the encounter frequency 𝜔 is used to replace the wave
g
natural frequency ω when calculating [A], [C] and {f}. The relationship between ω and 𝜔 is
at
𝜔 𝜔 1 𝑐𝑜𝑠 𝜃 (2)
W
where U is the ship speed, 𝑔 is the gravity acceleration, and θ is the wave direction. Equation (1) changes to
CF
𝜔 𝑀 𝐴 𝜔 𝑖𝜔 𝑐 𝐶 𝜔 𝐾 𝑘 𝑢 𝑓 𝜔 (3)
Although this approach only considers the encounter frequency and ignores the steady flow caused by ship
S
speed, it is widely used in engineering and can analyze the effect of speed on the flexible deformation [1916].
20
3 Numerical model
24
A 14,000 TEU container ship is considered in this study (see Fig. 2). The main particulars of the full loaded
container ship are shown in Table 1.
Fig. 2. Model of the 14,000TEU container ship.
Table 1. Main particulars of the full loaded container ship.
Designation Value
Length (m) 335.9
Breath (m) 51.0
Depth (m) 29.9
Draft (m) 14.5
Mass (t) 173509.4
Center of gravity in X-axis (m) 165.9
Center of gravity in Z-axis (m) 22.7
Roll inertia (ton ⋅ m2) 33570000
Pitch inertia (ton ⋅ m2) 1160000000
Yaw inertia (ton ⋅ m 2)
1154000000
Products of inertia about xz axis (ton ⋅ m2) 1608000
The 1D thin-walled beam model is established according to the mass and stiffness distribution of the container
ship. In order to prove that the 1D beam model can reflect the vibration characteristics of the container ship, the
structural dry modes based on the one-dimensional beam model are calculated and compared with the 3D FE
model, as shown in Table 2 and Fig. 3. The results demonstrate the efficacy of the established 1D beam model.
Table 2. Natural frequency based on 1D beam and 3D FE model.
Natural frequency (Hz)

Mode number Mode shape Error
3D FE 1D beam
7 1st Tor. & H.B. 3.073 3.123 -1.601%
8 2nd Tor. & H.B. 3.424 3.487 -1.807%
9 1st V.B. 4.096 4.016 1.993%
On
ly
4
fo
10
rR
3rd Tor. & H.B. 6.126 6.057 1.139%
ea
11 2nd V.B. 7.351 7.684 -4.334%
din
g
at
W
CF
S
20
24
Fig. 3. The first 5 flexible mode shapes of the container ship in dry conditions.
4 Hydroelastic results
According to the above methodology, the global hydroelastic response of the container ship at service speed
(22kn) and zero speed is considered. To investigate the flexibility effect on sectional forces, the results are also
given under the assumption of rigid body.
Fig. 4 (a) and (b) display the amplitude of vertical bending moment at mid-ship position and vertical shear
force at the position of 𝑥/𝐿 = 0.25 under θ = 180° wave direction respectively. For rigid ship, the results of the
sectional forces at 22kn speed are basically the same as those at zero speed, which indicates that the effect of
speed is not significant. For flexible ship, when the encounter frequency is close to the structural vibration fre-
quency, springing response is excited, and the sectional forces increase significantly. As shown in Fig. 4 (a), the
response amplitude at the first-order vertical bending resonance frequency is about 1.2 times of the maximum
amplitude of the response at the non-resonance frequency. In addition, for the vertical shear force, as shown in
Fig. 4 (b), the peak occurs at both the first-order and second-order vertical bending resonance frequency. However,
in the non-resonant frequency interval, the sectional forces of the flexible ship are consistent with the rigid ship.
(a) 10 (b) 10
2.0 1.00
Flexibility Effect on Sectional Forces
Vertical Bending Moment (MNꞏm/m)
Flexible: 22 knots 0 knots

Vertical Shear Force (MN/m)
Rigid: 22 knots 0 knots

1.5 1st V.B.
0.75 1st V.B.
1.0 0.50
0.5 0.25 2nd V.B.
0.0 0.00
0.5 1.0 1.5 2.0 2.5 0.5 1.0 1.5 2.0 2.5
(rad/s) (rad/s)
Fig. 4. Effect of structural stiffness on the sectional forces of navigating and freely floating container ships under wave direc-
tion of θ = 180°: (a) vertical bending moment at mid-ship position; (b) vertical shear force at the position of 𝑥/𝐿 = 0.25.
Fig. 5 displays the amplitude of vertical bending moment, horizontal bending moment and torsional moment
at midship and vertical shear force at 𝑥/𝐿 = 0.25 under θ = 135° wave direction respectively. Similar to the case
of θ = 180°, when the encounter frequency is close to the structural vibration frequency, the elastic response is
excited and the sectional forces increase. It is worth noting that for large container ships, due to the low torsional
structural vibration frequency, the torsional modes are excited when sailing in oblique waves, which also leads to
higher sectional forces, as shown in Fig. 5 (c) and (d). Both the first-order and second-order torsion & horizontal
bending modes are excited. The resonance peak is almost equal to the maximum amplitude of the response at the
non-resonance frequency. Therefore, extra attention must be paid to the torsional loads of large container ships.
On
ly
5
fo
rR
(a) 10 (b) 10
3.0 1.0
Flexibility Effect on Sectional Forces

ea
Flexible: 22 knots 0 knots
Vertical Shear Force (MN/m)

Rigid: 22 knots 0 knots 1st V.B.
2.4 0.8
din
1st V.B.
1.8 0.6
g
1.2 0.4
at
2nd V.B.
0.6 0.2
0.0
0.5 1.0 1.5 2.0 2.5
0.0
W 0.5 1.0 1.5 2.0 2.5
CF
(rad/s) (rad/s)
(c) 10 (d) 10
1.0 20
S
Horizontal Bending Moment (MNꞏm/m)
(MNꞏm/m)
1st Tor.&H.B.
20
0.8
15
1st Tor.&H.B.
0.6
24
Torsional Moment,
2nd Tor.&H.B. 10
2nd Tor.&H.B.
0.4
5
0.2
0.0 0
0.5 1.0 1.5 2.0 2.5 0.5 1.0 1.5 2.0 2.5
(rad/s) (rad/s)
Fig. 5. Effect of structural stiffness on the sectional forces of navigating and freely floating container ships under wave direc-
tion of θ = 135°: (a) vertical bending moment at mid-ship position; (b) vertical shear force at the position of 𝑥/𝐿 = 0.25; (c)
horizontal bending moment at mid-ship position; (d) torsional moment at mid-ship position.
In order to investigate the effect of structural stiffness on the sectional forces along the longitudinal direction
of the ship, Fig. 6 illustrates the amplitude of vertical bending moment, horizontal bending moment and torsional
moment along the longitudinal direction of the container ship under θ = 135° wave direction respectively. Two
types of wave natural frequency are considered: 𝜔 = 0.5 rad/s and wave natural frequency which will excite elastic
modes (for vertical bending 𝜔 = 1.4 rad/s and for torsion & horizontal bending 𝜔 = 1.3 rad/s). When 𝜔 = 0.5 rad/s,
the encounter frequency is away from the structural vibration frequency. In Fig. 6 (a), the vertical bending moment
of the flexible ship is basically consistent with that of the rigid ship. Furthermore, in Fig. 6 (b) and (c), the hori-
zontal bending moment and torsional moment of the flexible ship are slightly higher than those of the rigid ship.
Therefore, flexible deformations can be ignored in long wave length. Besides, for both rigid and flexible ships,
the forward speed will increase the horizontal bending moment and torsional moment. However, as shown in Fig.
6 (d)-(f), when the encounter frequency approaches the structural vibration frequency, the elastic modes are ex-
cited, which induces significant increase of sectional loads. Therefore, it is necessary to conduct the hydroelastic
analysis of large container ships with forward speed.
(a) (b) (c)
10 10 10
1.2 0.6 10
Flexibility Effect on Responses along Ship
Torsional Moment (MNꞏm/m)
Flexible: 22 knots 0 knots , = 0.5 rad/s , = 0.5 rad/s

Rigid: 22 knots 0 knots 8
0.8 0.4
, = 0.5 rad/s 6
4
0.4 0.2
2
0.0 0.0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
(d) (e) (f)

10 10 10
2.4 1.00 16
Torsional Moment (MNꞏm/m)
, = 1.4 rad/s , = 1.3 rad/s , = 1.3 rad/s

0.75 12
1.6
0.50 8
0.8
0.25 4
0.0 0.00 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Fig. 6. Effect of structural stiffness on the sectional forces along the longitudinal direction of the container ship under the wave
direction of θ = 135°: (a) vertical bending moment when 𝜔 = 0.5 rad/s; (b) horizontal bending moment when 𝜔 = 0.5 rad/s;
(c) torsional moment when 𝜔 = 0.5 rad/s; (d) vertical bending moment when 𝜔 = 1.4 rad/s; (e) horizontal bending moment
when 𝜔 = 1.3 rad/s; (f) torsional moment when 𝜔 = 1.3 rad/s.
On
ly
6
fo
5 Conclusions
rR
ea
The beam-connected-discrete-module (BCDM) method is applied to the hydroelastic analysis of large con-
din
tainer ships with forward speed. In this method, the large container ship is divided into several rigid modules
connected by a hull girder. The global hydroelastic equation is established by coupling multibody hydrodynamics
g
and beam structural dynamics equation. The BCDM method avoid the problem of the controversial generalized
at
restoring matrices by using extensively studied multi-body hydrodynamics. Besides, to account for the speed
effects, the wave natural frequency is replaced by the encounter frequency.
W
Based on the BCDM method, the hydroelastic response of a 14,000 TEU container ship is investigated. The
CF
results show that when the encounter frequency approaches the structural vibration frequency, the elastic modes
of the large container ship will be excited, resulting in higher sectional forces, especially in the case of oblique
S
waves, which will excite the torsion & horizontal bending vibrations. Therefore, the hydroelastic response must
20
be considered for safety checking in the design process of large container ships.
In the future research, the influence of steady wave caused by forward speed can be considered in the hydro-
24
dynamic calculation to extend BCDM method in large ship hydroelastic analysis with forward speed.
References
1. Wu, Y., Du, S.: Influence of speed and steady flow on the hydroelastic mechanical properties of navigating hull structures.
Ship Building of China, S1, 90-99 (1998).
2. Tian, C.: Study on the theory and applications of nonlinear hydroelasticity of ships with forward speed, Doctor thesis
(2007).
3. Duan, W., Wang, H., Chen, J., Wang, H.: TEBEM for springing responses of a container ship with forward speed and
nonlinear effects in time domain. Engineering Analysis with Boundary Elements, 140, 406-420 (2022).
4. Chen, Z., Gui, H., Dong, P., Yu, C.: Numerical and experimental analysis of hydroelastic responses of a high-speed
trimaran in oblique irregular waves, International Journal of Naval Architecture and Ocean Engineering, 11, 409-421
(2019).
5. Zhang, K.: Research on nonlinear wave loads’ time-domain hydroelastic analysis method and application of ships with
large bow flare, Doctor thesis (2018).
6. Hu, J., Wu, Y., Tian, C., Wang, X., Zhang, F.: Hydroelastic analysis and model tests on the structural responses and
fatigue behaviours of an ultra-large ore carrier in waves. Proceedings of the Institution of Mechanical Engineers, Part M:
Journal of Engineering for the Maritime Environment, 226(2), 135-155 (2012).
7. Malenica, Š., Bigot, F.: On the generalized motions/deformations of the floating bodies. Proceedings of the 35th Interna-
tional Workshop on Water Waves and Floating Bodies. South Korea (2020).
8. Pan, Z., Zhang, S., Fu, S.: Integrated hydrodynamic-structural analysis of flexible floating structures. Ocean Engineering,
310, 118644 (2024).
9. Wei, W., Fu, S., Moan, T., Lu, Z., Deng, S.: A discrete-modules-based frequency domain hydroelasticity method for
floating structures in inhomogeneous sea conditions, Journal of Fluids and Structures, 74, 321-339 (2017).
10. Zhu, L., Shao, F., Xu, Q., Sun, Y., Ma, Q.: The Effects of Moving Load on the Hydroelastic Response of a Very Large
Floating Structure, Mathematical Problems in Engineering, 6062586, 9 (2018).
11. Bispo, I.B.S., Mohapatra, S.C., Guedes Soares, C.: Numerical analysis of a moored very large floating structure composed
by a set of hinged plates, Ocean Engineering, 253, 110785 (2022).
12. Wei, W., Fu, S., Moan, T., Song, C., Deng, S., Lie, H.: A time-domain method for hydroelasticity of a curved floating
bridge in inhomogeneous waves. Journal of Offshore Mechanics and Arctic Engineering, 141(1), 014501 (2019).
13. Li, L.: Full-coupled analysis of offshore floating wind turbine supported by very large floating structure with consideration
of hydroelasticity, Renewable Energy, 189, 790-799 (2022).
14. Li, S., Fu, S., Zhang, S., Moan, T.: Second-order hydroelastic analysis of a flexible floating structure under spatially
inhomogeneous waves. Marine Structures, 86 ,103306 (2022).
15. Zhang, S., Fu, S., Li, S., Moan, T. Xu, Y., Pan, Z.: Frequency-domain hydroelastic stress analysis considering local
bending effect based on a two-step procedure, Marine Structures, 95, 103580 (2024).
16. Bakti, F. P., Jin, C., Kim, M. H.: Practical approach of linear hydro-elasticity effect on vessel with forward speed in the
frequency domain. Journal of Fluids and Structures, 101, 103204 (2021).
17. Zhang, S., Fu, S., Pan, Z., Han, K., Ye, Y.: Springing responses of ships with forward speed based on a multi-module
hydroelastic method. Marine Structures, (Under Review).
18. Senjanović, I., Tomašević, S., Vladimir, N.: An advanced theory of thin-walled girders with application to ship vibrations,
Marine Structures, 22, 387-437 (2009).
19. Jiao, J., Chen, Z., Chen, C., Ren, H.: Time-domain hydroelastic analysis of nonlinear motions and loads on a large bow-
flare ship advancing in high irregular seas, Journal of Marine Science and Technology, 25(2), 426-454 (2020).
On
ly
f or
Re on floating offshore wind turbine platform
Hydrodynamic investigation
integrated with porousashell and OWC wave energy converter

din
g
at
Dezhi Ning1, 2 and Sijia Deng1, 2
1
W
State Key Laboratory of Coastal and Offshore Engineering, Dalian University of Technology, 116024 Dalian, China;
2Dalian
CF
Key Laboratory of Offshore Renewable Energy, Dalian University of Technology, 116024 Dalian, China
S
dzning@dlut.edu.cn
20
24
Abstract. The motion response of a floating offshore wind turbine (FOWT) platform is closely related to
ensuring its safety. It is significant to decrease the motion response in the design of FOWTs. By integrating an
external porous shell and internal Oscillating Water Column (OWC) devises with a floating wind turbine
platform, a new-formed platform is designed to reduce the motion response. A numerical model, based on the
nonlinear higher-order boundary element method (HOBEM), investigating the coupling effect of the platform-
porous shell-OWCs was developed and validated by comparison with the physical experiments. The results of
the motion response were compared with those of the conventional platform without the integrating parts. The
motion response of the two kinds of platforms was simulated in the time domain. The study demonstrates a
floating platform integrated with porous shell and OWC to achieve reduced motion response is feasible and
can serve as a new design approach for the development of deep-sea offshore floating wind turbines.
Keywords: FOWT; Platform; Porous shell; OWC; Motion response.
1 Introduction
Offshore wind turbines are extensively deployed along coastlines and in oceans to harness wind power.[1]
Compared to fixed offshore wind turbines, floating turbines exhibit more complex responses due to their six
degrees of freedom in motion. Their motion is a critical factor in the operation and maintenance of turbines, as
reducing the motion response can effectively lower maintenance costs and increase the energy yield of wind
projects. Numerous studies have focused on mitigating the motion response of turbines. Currently, extensive
research is aimed at reducing the motion response amplitude. Rao et al. [2] conducted a 1:100 scale model
experiment to investigate the heave damping and added mass of a classical spar with a heave plate, comparing
experimental results with numerical observations. Hegde et al. [3] explored the impact of heave plate diameter
and position relative to draft ratio on damping, identifying an optimal configuration. Yang et al. [4]installed two
tuned mass dampers on the spar and nacelle of a floating offshore wind turbine (FOWT) to control vibration
response. The stiffness and damping coefficients of the TMD were optimized, and the damping effect under wind
and wave loading was investigated.
In recent years, many researchers have combined wind turbines with wave energy converters to reduce
turbine motion. An innovative integration of a cylindrical WEC device into a seabed-mounted wind turbine
foundation has been proposed for the combined exploitation of wave and ocean wind energy in coastal areas [5].
Perez-Collazo et al. [6] conducted a feasibility study on the multi-purpose integration of an OWC device with a
jacket-frame substructure. Perez-Collazo et al. [7], Zhou et al. [8], and Cong et al.[9] systematically studied the
hydrodynamic flow around an OWC WEC fitted to a fixed wind turbine monopile. An exterior shell was installed
concentrically around the wind turbine monopile to form the OWC chamber, trapping water and air inside.
On
ly
fo
rR
In addition to studies on reducing the motion response of FOWTs, porous structures are frequently used in the
ea
marine industry to dissipate energy and reduce the kinematic response of structures. Porous structures are
din
commonly employed in fixed and floating breakwaters [10, 11] and are also used to enhance the motion damping
of marine structures [12–16]. Mackay et al. [17] investigated the use of porous materials for reducing loads on
g
fixed and offshore structures and confirmed that porous structures have significant potential for load reduction
at
and motion damping on floating structures. Saha et al. [18] examined the scattering phenomena in water waves
W
from a bottom porous composite cylinder with an annular porous cap, finding that wave forces on an inner
concentric column diminished when the porous effect parameter was low.
CF
S
Based on the review of the above studies, the concept of platform-porous shell-OWC was developed. This
20
study aims to propose a new method to reduce the amplitude of the motion response of the offshore floating wind
turbine spar platform. The remainder of the manuscript is structured as follows: Section 2 presents the description
24
of the hybrid system. Section 3 outlines the mathematical formulation and numerical methodology. The reliability
of numerical model is verified in section 4. Section 5 illustrates the research results. Finally, the research
conclusions are discussed in Section 6.
2 Description of the hybrid system
The hybrid system consisting of a FOWT system, the OWC device and porous shell, which have the rigid
connection with each other. The OWC device lay in the middle of the main column. The porous shell is integrated
with the vertical semisubmersible platform at the incident wave height region. The configuration about the whole
FOWT system are consisted a new form floating platform structural and mooring system properties. Details of
the hybrid system are provided in Figure 1.
(a) (b)
Fig. 1. The hybrid system including floating platform, mooring lines, porous shell, and the mooring lines.
The semisubmersible platform is employed. This semisubmersible platform consist of two cylinders
vertically. The diameter of the top and bottom cylinder are 12m and 24m respectively. The draft of the float is
20m. Three mooring lines are selected for station keeping of FOWT and anchored to sea bed at a water depth of
200 m. The mooring lines are attached at the bottom of the platform, with 120 apart in each lines. The properties
of the mooring system are summarized in Ref [19]
The OWC devices are rigidly integrated into the floating platform, motioning in sync with the platform. The
PTO is modeled as linear damping without considering the complexity of the various control strategies. The
porous shell outer diameter is 24 m, the inner diameter 23.8 m, the wall thickness 0.2 m and the porosity 𝜏 0.3.
The wall height is 14m. The openings in the porous cylinder are rectangular and arranged on a regular square grid.
On
ly
fo
3
rR
Mathematical formulation and numerical methodology
ea
For generality, the constrained motion equations of a hybrid system consisting of a new form platform. ϕ is
din
complex velocity potential satisfying the Laplace equation,
𝜕 𝜑 𝜕 𝜑
g
at
0 (1)
𝜕𝑥 𝜕𝑧
W
The corresponding boundary conditions of all velocity potentials can be described as,
∂𝜑 ∂𝜑
, on 𝑆 and 𝑆 CF (2)
S
∂𝑛 ∂𝑛
where
20
denotes the normal derivative on the solid surface. To analyze the wave motion in a finite domain, a
24
sponge layer is introduced to absorb the reflected waves from the platform. Using Taylor's expansion, the
kinematic and dynamic boundary conditions on the free surface 𝑆 can be expressed as
∂𝜂 ∂𝜑 (3)
𝑓 𝜈 𝜂 𝑚 1&2
∂𝑡 ∂𝑧
⎧ 1𝑃 𝜇
∂𝜑
𝑚 1
∂𝜑 ⎪ 𝜌 ∂𝑛
𝑔𝜂 𝑓 𝜈 𝜑 (4)
∂𝑡 ⎨ ∂𝜑 ∂𝜑
⎪ 𝜇 𝜂 𝑚 2
⎩ ∂𝑛 ∂𝑛
The air pressure 𝑃air can be linked to the square of the flow velocity[20]:
𝑃air 𝑡 𝜇 |𝑢 𝑡 |𝑢 𝑡 , (5)
where u is the air flow velocity, μ2 is the nonlinear pneumatic damping coefficient which characterizes the OWC
turbine damping.
4 Validation
4.1 Wave-porous shell interaction
The novel aspect of the design proposed in this study consists of the addition of the concentric porous structure
around the floating platform to attenuate the platform motion. Therefore, the performance of the model with the
addition of this component should be validated. To achieve the model validation, a numerical model consisting of
a solid cylinder with an external diameter of a = 250 mm and a porous cylinder with an external diameter of b =
375 mm is created. In the porous component, the holes have a circular form of radius r, satisfying a porosity τ =
0.3. The wall thickness of the porous structure is 3 mm and the effect of the wall thickness are ignored in the
numerical calculation as shown in Fig. 2. The model is located in a water depth of 1 meter and the height of the
porous structure extends above the maximum amplitude of the incident waves. By comparing the results obtained
from present numerical calculations with the results of analytical solutions using the boundary element method
[17], the reliability of the shell part in present model is established as presented in Fig. 3.
On
ly
fo
rR
ea
din
g
at
W
(a)
CF (b)
S
Fig. 2. The validation mesh of the porous shell: (a) side view and (b) top view.
20
24
Fig. 3. The comparison of present model and analytical solutions.
4.2 Wave-OWC device interaction
The novel aspect of the design proposed in this study consists of the integration of the OWC devices, so the
performance of the model with the addition of this component should be validated. Fig. 4. Shows the mesh of
the FOWT-OWC model. Fig. 5 presents the comparison of present model and experimental result [21]. The
variations in the heave-motion, Δξ denotes to the displacement difference in heave modes. In general, the
results of the numerical model are close to the experimental results, which proves the validity of present
model.
(a) (b)
Fig. 4. The validation mesh of the OWC device: (a) side view and (b) top view.
On
ly
fo
rR
Fig. 5. The comparison of present model and experimental result.
5 eand
The coupled effect of OWC
adporous shell
ing
A regular wave environment case is adopted to test the platforms motion response. The wave height (H) and the
at
wave period (T) are assumed to be 1m and 7s, respectively. The wave propagates along the x-axis. Because both
W
the structure and the waves are symmetrical along the x-axis, the sway, roll and yaw modes shift too little to be
CF
negligible. Hereafter, the surge, heave and pitch mode motion responses are discussed, as shown in Fig. 6. In the
surge mode, the motion response are basically same, where slight differences are shown in the peaks and troughs.
S
In the heave and pitch modes, the motion responses of hybrid platform are decreased obviously.
20
24
(a)
(b)
(c)
Fig. 4. The motion response comparison of conventional platform and the hybrid platform: (a) surge, (b) heave, and (c) pitch.
6 Conclusions
In this study, a novel hybrid platform with OWC devices and porous shell are proposed. Based on the HOBEM
method, the motion responses of the conventional and the hybrid platform are compared. The result shows that
the hybrid platform motion in heave and pitch modes deceased obviously, which is helpful to lower maintenance
costs and increase the energy yield of wind projects.
On
ly
fo
Acknowledgements
rR
ea
This work is supported the National Natural Science Foundation of China [Grant Nos. U22A20242 and 52271260].
din
References
g
at
W
1. Council, G.W.E.: GWEC global wind report 2024. Global wind energy council. (2024)
CF
2. Rao, S.K., Mahesh J. Nallayarasu, S. Bhattacharyya: Numerical and experimental studies of heave damping and added
mass of spar with heave plates using forced oscillation. Applied Ocean Research. 111, (2021)
3.
S
Pooja, H., Nallayarasu, S.: Investigation of heave damping characteristics of buoy form spar with heave plate near the
20
free surface using CFD validated by experiments. Ships and offshore structures. 18, 1650–1667 (2023)
4.
24
Yang, Y., Bashir, M., Michailides, C., Li, C., Wang, J.: Development and application of an aero-hydro-servo-elastic
coupling framework for analysis of floating offshore wind turbines. Renewable Energy. 161, 606–625 (2020).
5. Pérez-Collazo, C., Greaves, D., Iglesias, G.: A review of combined wave and offshore wind energy. Renewable &
Sustainable Energy Reviews. 42, 141–153 (2015)
6. Perez-Collazo, C., Greaves, D., Iglesias, G.: A Novel Hybrid Wind-Wave Energy Converter for Jacket-Frame
Substructures. Energies. 11, (2018)
7. Perez-Collazo, C., Pemberton, R., Greaves, D., Iglesias, G.: Monopile-mounted wave energy converter for a hybrid wind-
wave system. Energy Conversion & Management. 199, 111971.1-111971.13 (2019)
8. Zhou, Y., Ning, D., Shi, W., Johanning, L., Liang, D.: Hydrodynamic investigation on an OWC wave energy converter
integrated into an OWT monopile. Coastal Engineering. 103731 (2020)
9. Cong, Y., Dezhi Liu: Wave power absorption by an oscillating water column (OWC) device of annular cross-section in a
combined wind-wave energy system. Applied Ocean Research. 107, (2021)
10. Huang, Z., Li, Y., Liu, Y.: Hydraulic performance and wave loadings of perforated/slotted coastal structures: A review.
Ocean Engineering. 38, 1031–1053 (2011)
11. Dai, J., Wang, C.M., Utsunomiya, T., Duan, W.: Review of recent research and developments on floating breakwaters.
Ocean Engineering. 158, 132–151 (2018)
12. Downie, M.J., Graham, J.M.R., Hall, C.: An experimental investigation of motion control devices for truss spars. Marine
Structures. 13, 75–90 (2000)
13. Molin, B.: On the added mass and damping of periodic arrays of fully or partially porous disks. Journal of Fluids and
Structures. 15, 275–290 (2001)
14. Tao, L., Dray, D.: Hydrodynamic performance of solid and porous heave plates. Ocean Engineering. 35, 1006–1014
(2008)
15. An, S., Faltinsen, O.M.: Linear free-surface effects on a horizontally submerged and perforated 2D thin plate in finite
and infinite water depths. Applied Ocean Research. 37, 220–234 (2012)
16. An, S., Faltinsen, O.M.: An experimental and numerical study of heave added mass and damping of horizontally
submerged and perforated rectangular plates. Journal of Fluids & Structures. 39, 87–101 (2013)
17. Mackay, E., Shi, W., Qiao, D., Gabl, R., Davey, T., Ning, D., Johanning, L.: Numerical and experimental modelling of
wave interaction with fixed and floating porous cylinders. Ocean engineering. 242 (2021)
18. Saha, S., Bora, S., Das, S.: Time-dependent water wave scattering by a bottom-mounted porous compound cylinder fitted
with an annular porous lid. Waves in Random and Complex Media. 1–22 (2023)
19. Robertson, A., Jonkman, J., Masciola, M., Song, H., Goupee, A., Coulling, A., Luan, C.: Definition of the
Semisubmersible Floating System for Phase II of OC4. (2014)
20. Zhou, Y., Ning, D., Shi, W., Johanning, L., Liang, D.: Hydrodynamic investigation on an OWC wave energy converter
integrated into an offshore wind turbine monopile. Coastal Engineering. 162, 103731 (2020).
On
ly
fo
rR
21. Zhou, Y., Ning, D., Chen, L., Mayon, R., Zhang, C.: Experimental investigation on an OWC wave energy converter
ea
integrated into a floating offshore wind turbine. Energy Conversion and Management. 276, 116546 (2023).
din
g
at
W
CF
S
20
24
On
ly
fo
rR
ea
din
A frequency-domain modeling approach for hydroelastic responses of
g
pontoon-type floating bridges: an experimental validation
at
W
Minghao Cui1,2, Zhengshun Cheng1,2*, Peng Chen1,2 and Torgeir Moan3,4
1
CF
State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University,
S
Shanghai 200240, China
2 Hainan Research Institute, Shanghai Jiao Tong University, Sanya 572024, China
20
3 Norwegian University of Science and Technology, Trondheim 7491, Norway
24
4 Ningbo University, Ningbo 315211, China
zhengshun.cheng@sjtu.edu.cn
Abstract. Very large floating bridges are distinguished by their capacity to cross wide and deep waters without
the need for fixed foundations. Code-based design checks of floating bridges require a great number of simu-
lations to ensure their serviceability and safety under various load conditions. However, time-domain simula-
tion requires tremendous computational efforts. A promising alternative is to utilize a frequency-domain ap-
proach. In this study, a frequency-domain hydroelastic modeling approach is developed for a general pontoon-
type floating bridge, in which the pontoons are modeled as rigid bodies whereas the superstructure is repre-
sented using beam elements. Hydrodynamic interactions between pontoons and viscous loads are considered.
This study aims to validate the method by utilizing a public-available dataset of a truncated floating bridge
tested in SINTEF Ocean basin. Time-domain simulations based on SIMA are also conducted as comparisons.
It is shown that the proposed approach offers reasonable estimates of modal properties and RAOs. Moreover,
with linearized second-order loads and hydrodynamic interactions considered, it captures low- and wave-fre-
quency resonant peaks with adequate accuracy.
Keywords: Floating bridge, Frequency-domain approach, Experimental validation, Hydroelasticity, Hydro-

dynamic interaction and viscous loads.
1 Introduction
Floating bridges are large-scale infrastructures supported by floating pontoons. Without the need for fixed foun-
dations, these bridges are particularly suitable for spanning wide and deep waters, such as fjords, or areas with
very soft seabeds [1].
Modern floating bridges can extend for several kilometers, which significantly increases the overall flexibility,
leading to complex hydroelastic responses under diverse loading conditions. In code-based designs, numerous
simulations are required to represent their dynamic behaviors and ensure adequate serviceability and safety.
Time-domain simulations are typically employed to capture nonlinear and transient effects. However, due to
the large dimensions and coupling between the environment and the flexible structure, each simulation is ex-
tremely time-consuming. Consequently, a frequency-domain approach is promising, but ensuring accuracy
through appropriate modeling and linearization techniques poses a further challenge.
Therefore, this paper introduces a frequency-domain approach for the hydroelastic responses of floating
bridges. The focus is set on its extensive applications and experimental validation regarding eigenmodes, RAO,
response spectra, and statistics. A publicly available dataset is utilized, based on the model tests of a truncated
straight floating bridge conducted by Rodrigues et al. [2] at the SINTEF Ocean basin.
2 Frequency-domain modeling of the experimental model
The experimental model applies a truncated segment scaled by 1:33 with the full-scale length exceeding 1000
meters. As shown in Fig. 1, the bridge consists of a continuous girder with ballast bars and rigid clamps, boundary
brackets and bar elements, nine pontoons (of two types), and taut moorings for lateral stiffness. The model is fixed
at the right hand with a bar element and ball joints. Details on the model description are referred to [2].
On
ly
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 1. Frequency-domain (FD) modeling of the tested floating bridge (viewed from incoming wave direction). Details on
the modeling approach can be found in our previous study [3].
The FD approach assumes minor deformations of the pontoon-column pairs, such that the motion equations of the
girder and the pontoons can be established based on the beam model and the multi-rigid-body hydrodynamics,
respectively. With multibody interaction, wave viscosity, and second-order wave loads incorporated, the ultimate
hydroelastic equation of motion can be solved for eigenvalues, response statistics and response spectra within
only a few minutes.
In addition, another numerical model is established by the commercial software SIMO-RIFLEX (SIMA) [4].
The widely applied simulation tool applies Cummins’s theory for coupled time-domain analyses, where structural
dynamics and hydrodynamics are considered based on the finite element method and the potential flow theory,
respectively. Each seed of a 3h simulation requires a solution time of over ten hours.
3 Experimental validation
Fig. 2 shows an overview of the tested bridge and relevant instruments [2]. Due to the existence of equipment and
cables, the mass properties of pontoons and clamps at different pontoon-axis vary slightly; however, they are
assumed to have a uniform distribution in the TD and FD methods. The stiffness of the bridge girder is calibrated
by comparing static pull-out results between TD and tested results.
Fig. 2. Experimental model and instrumentations photographed by Rodrigues et al. [2].
In Fig. 3, eigenvalues of the entire system are estimated by the FD approach with frequency-dependent added
mass of the pontoons, serving as a basis for model calibration and dynamic analysis. The modal shapes are gen-
erally close to those shared in [2], indicating a reasonable representation of the mass and stiffness term. Located
in low- and wave-frequency, these modes can be excited by wave loads and increase the responses significantly.
On
ly
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 3. Eigen-periods and model shapes by the frequency-domain approach considering frequency-dependent added mass of
the pontoons.
Response amplitude operators (RAOs) of pontoon motions are shown in Fig. 4. During white noise analysis,
hydrodynamic interactions and viscosity are not considered for both TD and FD models. Besides, the taut moor-
ings are modeled in SIMA as bar elements with Morison’s damping term, while for the FD model a simplified
stiffness matrix is used. The differences in damping terms result in the peak differences for surge and pitch RAO,
while a remarkable peak in heave is caused by the interaction effect.
(a) P1 (b) P5
Fig. 4. RAOs of surge, heave, and pitch motions of pontoon (a) P1 and (b) P5 by the model test (Test), the time-domain sim-
ulation (SIMA), and the proposed frequency-domain approach (FD).
Fig. 5 shows the standard deviations of pontoon motions subjected to irregular wave actions. The viscosity and
multi-body interaction in the experiments are represented in the FD approach. Satisfactory consistency is found
between FD and Test results in surge of all pontoons and in heave and pitch of only central pontoons.
JONSWAP spectrum: Hs = 1.6m, Tp = 5.3s

On
ly
fo
rR
Fig. 5. Standard deviations of surge, heave and pitch motions of pontoon P1, P3 and P5. Hydrodynamic interaction and wave
viscous damping are considered in tests and by the FD approach.
ea
din
The differences are revealed by response spectra in Fig. 6. For surge, the low-frequency resonances are dominating
and well captured by the FD approach. A significant peak in heave motion, corresponding to 6.6s, is caused by
g
hydrodynamic interactions for both FD and Test model, while for P1 the heave PSD is less consistent. The pitch
at
PSD is lower for Test of P1, indicating larger damping effects related to the mooring lines.
W
CF
S
20
24
(a) P1 (b) P5
Fig. 6. Power spectral density (PSD) of surge, heave and pitch motions of pontoon P1 and P5.
4 Conclusion
This study performs an experimental validation of an efficient FD hydroelastic approach for floating bridges based
on a publicly available dataset. The tested floating bridge is represented by SIMO-RIFLEX and by the FD ap-
proach for a comprehensive comparison between the model test, TD and FD approaches.
It is demonstrated that the FD method effectively estimates modal properties and derives closer RAOs against
tested results owing to its ability to account for wave viscosity and multi-body effects. In addition, standard devi-
ations of pontoon motions are compared under irregular wave actions and well consistency is found between FD
and Test results, except for the side pontoon that is affected by the nonlinearities of mooring lines and boundary
conditions. Response spectra further reveal that the FD method captures linear wave responses, low-frequency
resonances as well as wave-frequency resonances induced by the hydrodynamic interaction of pontoons.
Acknowledgment
The research was supported by the Hainan Provincial Joint Project of Sanya Yazhou Bay Science and Technology
City, Grant No 2021JJLH0018, and the National Natural Science Foundation of China, Grant No 52201330 & No
42176210.
References
1. Moan, T., Eidem, M.E.: Floating bridges and submerged tunnels in Norway—the history and future outlook. In: Proceed-
ings of the World Conference on Floating Solutions, pp. 81-111. Springer, Singapore (2020).
2. Rodrigues, J.M., Viuff, T., Økland, O.D.: Model tests of a hydroelastic truncated floating bridge. Applied Ocean Research.
125, 103247 (2022).
3. Cui, M., Cheng, Z., Chen, P., Moan, T.: A frequency-domain hydroelastic approach for modeling and stochastic dynamic
analysis of pontoon-type floating bridges. Engineering Structures. 318, 118740 (2024).
4. SINTEF Ocean. SIMA documentation, version 4.2.0 (2022).
On
ly
f or and optimization on the performance of wave
Re
Numerical investigation
ad converter array
energy
ing
at
W
Mengjie Wang1, Ling Wan1*, Chi Zhang2, Zhiyung Tay3
1Faculty
CF
of Maritime and Transportation, Ningbo University, Ningbo, China
S
2 Technology Centre for Offshore and Marine, TCOMS, Singapore
20
3 Engineering Cluster, Singapore Institute of Technology, Singapore
24
Abstract: To reduce the emission of the greenhouse gas, exploration of renewable energy sources with abundant reserves
could be viable solutions. Wave energy as renewable resource, has a substantial potential in the ocean. Nevertheless, the
advancement of wave energy utilization technologies encounters obstacles such as suboptimal energy conversion efficiencies
and high costs. To overcome these obstacles, dynamic properties and designs of single wave energy converter (WEC) and
WEC array need to be studied. This work investigated floating WEC array composed of point absorbers (PA), focusing on the
coupling effects among WECs in an array configuration. By employing fundamental theories and computational techniques
related to ocean structures in the frequency, the study performs numerical simulations using ANSYS-AQWA software. In
frequency domain, the study focuses on the effects of array types, float spacings, float parameters, and on the energy capture
performance of the WEC array. A comparison was made between the energy capture efficiency of the WEC array and a single
WEC, indicating that the WEC array demonstrated a superior capability to harness wave energy resources effectively.
Keywords: wave energy converter; wave energy converter array; design parameters; energy capture performance
1 Introduction
With global attention on the climate change due to the rising CO2 level, generating electricity from renewable
sources is once again an important area of research [1].Wave energy utilization has garnered significant interest
from governments and industry due to its high energy density, which surpasses that of solar and wind energy
among renewable sources.[2]. Currently, three main wave energy converters (WEC) types are used for harnessing
wave energy: overtopping (OP) devices, oscillating bodies (OB), and oscillating water column (OWC) [3]. Among
the myriad devices, OWC have gained extensive research attention owing to their structural simplicity, robust
survivability, and ease of maintenance [4]. However, most existing OWCs utilize a single float, limiting power
generation efficiency and increasing costs, hindering commercialization [5]. Consequently, arranging multiple
WECs into an array has become a key research focus in wave energy converters [6,7].
To improve the energy capture efficiency of array WECs, many scholars have studied the hydrodynamic
characteristics and optimized the array layout based on their findings. Siddorn et al. [8] studied the hydrodynamic
performance of a four-float cylindrical array using numerical analysis, and provided the expressions of wave
excitation force, added mass, radiational damping and other parameters acting on each float. Babarit [9] examined
how float spacing affects absorbed power, finding that larger spacing reduces interaction impact on energy
absorption, with regular waves having a greater influence than irregular waves.Agamloh et al. [10] studied the
interaction between two cylindrical floats and waves by using a three-dimensional numerical flume simulator, and
found that changes in wave parameters and float spacing would greatly affect the interaction between floats. In
addition, there are many works that combines WEC with wind turbines [11], that plays a role to stabilize the
On
ly
fo
floater motion.
rR
ea
In this study, a heave-type WEC array is proposed with flexible designs tailored to power demand, layout area,
din
and wave conditions. To enhance efficiency and stability, it investigates the effects of array form, float parameters,
wave incidence direction, and float spacing on energy capture characteristics. The findings will provide practical
g
insights for optimizing WEC arrays and understanding their impact on large-scale wave fields.
at
2 Mathematical model
W
CF
In this study, we only consider WEC with heave degree of freedom in frequency-domain, therefore, viscous force,
S
mooring force and other nonlinear forces are ignored. The kinematic equation of the array float in the frequency
domain is established based on the linear potential flow theory.
20
𝐹 𝑗𝜔 𝐹𝑝𝑡𝑜 𝑗𝜔
24
According to Newton's second law, the kinematic equation of the array float in the heave direction is:
𝑀𝑍 𝑗𝜔 1
where, M is the mass matrix of the array float, 𝑍 𝑗𝜔 is the heave motion displacement matrix of the array float,
𝐹 𝑗𝜔 is the PTO force matrix of the array float, and 𝐹 𝑗𝜔 is the hydrodynamic matrix of the array float in
the heave direction, which is expressed as:
𝐹 𝐹 𝐹 𝐹 2
Further analysis, 𝐹 includes wave excitation force matrix 𝐹 in the heave direction of the array float, wave
radiant force matrix 𝐹 in the heave direction and the hydrostatic restoring force matrix 𝐹 of the array float are
obtained. They can respectively be expressed as:
𝐹 𝑀 𝜔 𝑍 𝑗𝜔 𝐵 𝜔 𝑍 𝑗𝜔 3
𝐹 𝐾 𝑍 𝑗𝜔 𝜌𝑔𝐴 𝑍 𝑗𝜔 4
In equation (3) and (4): 𝑀 𝜔 is the added mass matrix of the array float in the heave direction. 𝐵 𝜔 is the
radiational damping coefficient matrix of the array float in the heave direction. 𝐾 is the hydrostatic restoring
force coefficient matrix of the array float; 𝐴 is the array float waterplane area matrix.
In this study, only linear PTO damping force is considered, so 𝐹𝑝𝑡𝑜 can be expressed as:
𝐹𝑝𝑡𝑜 𝐶𝑝𝑡𝑜 𝑍 𝑗𝜔 5
By combining the above equations, the kinematic equation of the array float in the frequency domain can be
obtained as:
𝜔 𝑀 𝑀 𝜔 𝑍 j𝜔 j𝜔 𝐵 𝜔 𝐶𝑝𝑡𝑜 𝑍 j𝜔 𝜌𝑔𝐴 𝑍 j𝜔 𝐹 j𝜔 6
The average output power of the array floats can be expressed as:
1
𝑃 𝜔 |𝑍 j𝜔 | 𝐶𝑝𝑡𝑜 |𝑍 j𝜔 | 7
2
Array interaction q factor is defined as:
Σ 𝑃
q 8
𝑁𝑃
This study assumes that all devices in the array are exactly the same. In the equation, 𝑃 is the output power of
a single device layout.
3 Numerical model
For the hydrodynamic analysis of point absorption wave energy converter, it is necessary to establish a reasonable
hydrodynamic model. In this study, ANSYS-AQWA is used to establish the hydrodynamic model of floating body
and calculate the hydrodynamic parameters of the float, then the calculation results are imported into ORIGIN
and programmed to solve the energy capture power of the array float.
On
ly
fo
rR
In this study, four identical floats were selected to form two array configurations with linear shape and square
ea
shape, respectively, as shown in Figure 1. In the simulation analysis, deep water condition with unit wave height,
din
and incident wave frequency of 0~4rad/s are assumed. To simulate the PTO system, linear damping is set in the
vertical motion direction of the array float.
g
at
W
Body2 Body4
Body1 Body2 Body3 Body4
CF
S Body1 Body3
20
Fig. 1. Linear layout and square layout
24
After establishing the panel model in AQWA, the frequency domain analysis is carried out. The diagonal length
of the panel size is small enough (<<λ/6) [12] that can capture the effects from the smallest wave length analyzed.
4 Hydrodynamic performance of wave energy converter
The WEC design parameters are shown in Table 1, the effects of array layout, WEC spacing, wave incidence
direction and float number on the hydrodynamic performance of the array wave energy converter are studied.
Table 1. Design parameters of wave energy converter array
Parameter value
Float radius R 2.5m
Float draft B 4m
Water depth H 1000m (Deep water condition)
Number of floats N 2-4
Float spacing L 4R~10R
WEC array layout Linear layout, Square layout
Incident wave frequency 𝜔 0~4rad/s
PTO Damping coefficient C 10000N/(m/s)
4.1 Influence of array form on hydrodynamic performance
Numerical simulation parameters: the float is a cylindrical float, the float spacing is 4R, the number of floats N is
4, and the rest parameters remain unchanged in Table 1. Considering the symmetry of array form, the incident
wave direction of linear array is 0°,15°, 30°, 45°, 60°, 75°, 90° when numerical simulation is carried out. When
the square array is numerically simulated, the incident wave direction is 0°, 15 °, 30 ° and 45°. Figure 2 shows the
capture power of the two array layouts at the 90° wave direction. Table 2 and Table 3 show the peak energy
capture power of the two array layouts.
On
ly
fo
rR
ea
din
g
at
W
CF
S
(a) Capture energy power (Linear layout)
20
(b) Capture energy power (Square layout)
24
Fig. 2. Capture energy power of two array layouts at 90° wave direction (heave mode only)
Table 2. Captured power peak of linear array layout under different wave directions
Peak capture power(kW)

Wave incident direction
0° 75.69 75.463 76.77 70.17
15° 77.61 73.73 76.55 68.35
30° 82.74 69.45 76.60 64.92
45° 87.00 66.03 79.51 65.80
60° 83.68 68.68 87.57 74.16
75° 76.23 79.74 95.43 80.44
90° 76.95 92.57 92.58 76.96
Table 3. Captured power peak of square array layout under different wave directions
Peak capture power(kW)

Wave incident direction
0° 81.60 81.61 74.20 74.21
15° 81.55 85.09 79.86 67.40
30° 83.79 89.38 86.88 60.91
45° 85.09 90.50 90.49 58.23
From Table 2, Body1's peak capture power increases between 0° and 45°, while Body2, Body3, and Body4's
decrease. At 60° to 75°, Body1's power drops as the others increases. At 90°, Body1 and Body4 have similar peak
power, as do Body2 and Body3. Table 3 shows that at 0°, the peak power of Body1 and Body2, as well as Body3
and Body4, is nearly identical due to the same wave force distribution. As the wave incidence angle increases,
Body1, Body2, and Body3's power increases, while Body4's decreases.
4.2 Influence of float spacing on hydrodynamic performance
Numerical simulation parameters: the float is a cylindrical float, the incident wave direction is 90°, the number of
floats N is 4, the float spacing ranges from 4R to 10R, and the rest parameters remain unchanged in Table 1. The
energy capture power of the two types of array floats varies with the layout distance as shown in Figure 3.
On
ly
fo
rR
ea
din
g
at
W
CF
S
(a) Linear layout
20
(b)Square layout
24
Fig. 3. The energy capture power of the two types of array floats varies with the layout distance
As can be seen from Fig.3, with the increase of layout spacing, the energy capture power of the array floats
decreases first and then increases, and the energy capture power of any spacing of the linear layout is greater than
the generating power when the floats move alone. The square layout reaches its maximum value when the spacing
is 8R.
4.3 Influence of number of floats on hydrodynamic performance
Numerical simulation parameters: the float is a cylindrical float, the array form is linear layout, the incident wave
direction is 90°, the float spacing is 4R, the number of floats N range from 2 to 5, and the rest parameters remain
unchanged in Table 1. The average output power of two arrays with different number of floats is obtained by
simulation and shown in Figure 4.
(a) q factor curve (N = 2) (b) q factor curve (N = 3)

On
ly
fo
rR
(c) q factor curve (N = 4) (d) q factor curve (N = 5)
ea
Fig. 4. Array interaction q factor curve when the number of floats N = 2~5
din
As shown in Fig. 4, increasing the number of floats amplifies q factor fluctuations around 1. When q > 1, adding
g
more floats improves energy capture efficiency; however, if the incident wave frequency falls within the range
at
where q < 1, it negatively impacts efficiency. Therefore, selecting the appropriate number of floats based on the
W
sea area's wave information is crucial for optimizing the WEC array's energy capture performance.
5 Conclusion and future work CF

S
20
A WEC array is proposed to supply power for moderate ocean operations, such as fish farms and ocean
24
observation. Using linear potential flow theory and ANSYS-AQWA software, the energy capture characteristics
of the WEC array are studied.
Different float layout significantly impacts the hydrodynamic performance of the WEC array. Selecting an
optimal layout based on wave conditions is crucial for improving performance. As float spacing increases, energy
capture power initially decreases, then increases, peaking at 8R. After this point, the peak frequency of energy
capture declines with greater spacing. Additionally, increasing the number of floats enhances hydrodynamic
interactions, as indicated by increased fluctuations in the q factor around 1.
There are various aspects that should be further explored in the future. More parameters, e.g., PTO parameter
effect, dynamic responses in time domain, where various nonlinear forces are considered, as well as control
strategies in time domain should be included to increase the power output. In addition, only numerical results are
shown here, the results should be verified experimentally in the future.
References
1. Drew, B., Plummer, A., Sahinkaya, M.: A review of wave energy converter technology. Proceedings of the Institution of
Mechanical Engineers, Part A: Journal of Power and Energy, 223, 887-902 (2009).
2. Clement, A., Mccullen, P., Falcao, A.: Wave energy in Europe: current status and perspectives. Renewable and
Sustainable Energy Reviews, 6(5), 405-431 (2002).
3. Falcao, A.: Wave energy utilization: a review of the technologies. Renewable Sustainable Energy Rev., 14 (3), 899–918
(2010).
4. Heath, T.: A review of oscillating water columns. Phil. Trans. Math. Phys. Eng. Sci., 370 (1959), 235–245 (2012).
5. Shi, Y. T., Liu, Y. J., Zhao, L. Z.: Overview of array float wave power generation technology. Energy and the
Environment, 2, 12-14 (2021).
6. Shi, H. D., Wang, C. K.: Progress and prospect of marine energy technology in China. Solar Energy, 3, 30-37 (2017).
7. Tunde, A., Hua, L.: Ocean wave energy converters: status and challenges. Energies, 11(5), 1250-1275 (2018).
8. Siddom, P., Eatock Taylor, R.: Diffraction and independent radiation by an array of floating cylinders. Ocean Engineering,
35, 1289-1303 (2008).
9. Babarit, A.: Impact of long separating distances on the energy production of two interacting wave energy converters.
Ocean Engineering, 37(8-9), 718-729 (2010).
10. Agamloh, E., Wallace, A., Von, J.: Application of fluid-structure interaction simulation of an ocean wave energy
extraction device. Renewable Energy, 33(4), 748-757 (2008).
11. Wan, L., Moan, T., Gao, Z., Shi, W.: A review on the technical development of combined wind and wave energy
conversion systems. Energy, 294,130885 (2024).
12. DNV.: Recommended practice dnv-rp-f205: global performance analysis of deepwater floating structures. (2019).
On
ly
fo
rR
Heave vibration control of very large floating structures using vertical
ea
tuned liquid column dampers
din
g
Hao Ding1,2[0000-0001-7058-3561], Kaiming Bi1,2[0000-0002-5702-6119], and Ruisheng Ma3[0000-0003-4872-9485]
1
at
Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hong Kong, China
W
2
Research Institute for Sustainable Urban Development (RISUD), The Hong Kong Polytechnic University, Hong Kong,
CF
China
3 Key Laboratory of Urban Security and Disaster Engineering of Ministry of Education, Beijing University of Technology,
Beijing, China
kaiming.bi@polyu.edu.hk
S
20
24
Abstract. In offshore engineering, tuned liquid column dampers (TLCDs) are widely used to mitigate the
horizontal and pitch motions of floating structures. However, their application in controlling heave motion
remains unexplored due to the working mechanism of TLCDs. This study explores the feasibility of using non-
submerged TLCDs in very large floating structures (VLFSs) to mitigate heave motion. An analytical model of
a VLFS-TLCD system is developed in the time domain. Five wave loading scenarios are considered for a
VLFS measuring 300 m 100 m, including conditions in the Gulf of Mexico (GM) and the South China Sea
(SCS), 10-year return period waves in the North West Shelf of Australia (NWSA) and the SCS, and 100-year
return period waves in the SCS. An optimization design program using a genetic algorithm is developed to
determine the optimal control parameters of TLCDs. Results using conventional submerged fixed heave plates
(FHPs) are also included for comparison. The findings highlight the effectiveness and control characteristics
of non-submerged TLCDs in mitigating the heave motion of VLFSs in the time domain.
Keywords: Heave vibration control, Tuned liquid column damper, Very large floating structure, Wave.
1 Introduction
Very large floating structures (VLFSs), which could provide stable and dependable platforms for diverse func-
tions/activities, are deemed as good solutions to solve the land shortage problem of coastal countries and/or cities.
There are two primary types of VLFSs: the semi-submersible platform (SSP)-type and the pontoon-type [1]. The
unique structure and shape of SSPs make them have advantages on versatility, deepwater operation, relocation,
and mobilization [2], and this study mainly focuses on this type of VLFSs.
To reduce wave-induced heave response of VLFSs, numerous control strategies, such as fixed heave plates
(FHPs) [3], tuned heave plates (THPs) [4], and rotational inertia dampers (RIDs) [5], were proposed over the past
several decades. These devices are installed in the water, and the harsh offshore environment may cause a variety
of issues such as corrosion, marine growth, etc., which significantly reduces the control effectiveness while in-
creases the maintenance cost. It will be ideal if the control system could be installed within the VLFSs. The large
dimension of VLFSs actually makes this request achievable.
Tuned liquid column damper (TLCD) is a cost-effective vibration control device that utilizes liquids as the
additional mass. However, to the best of the authors' knowledge, its application in marine engineering has been
confined to mitigate the lateral or pitch motion of offshore wind turbines and floating platforms [6, 7]. There has
been no reported research on employing TLCDs for heave vibration control in marine engineering. This scarcity
is primarily due to its inherent limitation in directly providing damping effects against vertical vibrations. Very
recently, a vertical tuned liquid column damper (VTLCD) system was proposed by the authors [8]. The VTLCD
control system, consisting of a pair of inclined U-shaped containers, springs, dashpots, as well as supporting bases,
can reduce vibrations of structures in the vertical direction owing to its geometric design. VTLCD is proposed to
control the heave motion of VLFSs in the present study. It is obvious that, seawater could be used as the liquid in
this damper, it is thus cost-effective and environmentally friendly. Moreover, distinct from existing vibration
control designs for offshore platforms, which typically locate control devices below sea level, this device can be
installed within the VLFS as discussed above, which could significantly reduce the maintenance cost. In this
study, the feasibility and effectiveness of VTLCDs on the heave response mitigation of SSP-type VLFSs will be
explored. The control effectiveness of VTLCDs will be examined.
On
ly
2
fo
2 Analytical model
rR
ea
Fig. 1 shows the schematic diagram of the analytical model of an SSP-type VLFS equipped with VTLCDs. As
din
shown, the control system is installed in the bottom part of the VLFS. In the present study, as we focus on the
heave motion, to simplify the analysis, the coupling effects from surge, sway, yaw, pitch, and roll motions are
g
ignored. The control system is assumed being installed at the middle of VLFS, namely, eccentricity is not consid-
at
ered in the analysis. The VLFS-VTLCD system (subjected to a wave-induced force 𝑓 ) contains three DOFs: (i)
the displacement of the SSP-type VLFS in the heave direction, denoted as 𝑧 ; (ii) the container displacement in
W
the direction of slope, denoted as u; and (iii) the liquid column displacement, denoted as y. The mass of the VLFS
CF
includes two parts: the physical mass 𝑚 and the frequency-dependent added mass 𝑚′ 𝜔 . The added mass is
due to the interaction between the VLFS and surrounding seawater, and herein, 𝜔 denotes the excitation frequency
in the unit of rad/s. Similarly, the damping of VLFS also includes two parts: the frequency-independent damping
S
20
𝑐 and the frequency-dependent radiation damping 𝑐 ′ 𝜔 . The frequency-independent damping primarily arises
from internal loss mechanisms, whereas the frequency-dependent radiation damping encompasses the dynamic
interaction between the VLFS and its surrounding environment, involving the conversion of vibrational energy
into wave energy, which is then dissipated. The restoring stiffness of the VLFS provided by the surrounding 24
seawater is denoted as 𝑘 . The reference base represents the support provided by the seawater, rather than the
seabed. In the VTLCD device, the tilt angle of the supporting surface is assumed as 𝜃. The liquid column length
in the inclined pipe is 𝐻 , while the liquid column lengths in the long and short vertical pipes are 𝑉 and 𝑉 , re-
spectively (the lengths are measured when the system is in a static state). The restoring stiffness, damping and
mass of the container are 𝑘 , 𝑐 , and 𝑚 , respectively. Two symmetrically distributed VTLCDs are used with the
aim of generating control forces only in the heave direction on the VLFS when subjected to wave-induced forces.
The equations of motion for the VLFS-VTLCD system can be expressed as [8, 9]:
𝑚 𝑦 𝑐 𝑦 𝑘 𝑦 𝜌𝐴 𝑉 𝑉 𝑠𝑖𝑛 𝜃 𝐻 𝑢 0 (1)
𝑚 𝑚 𝑢 𝑐𝑢 𝑘 𝑢 𝑚 𝑚 𝑧 𝑠𝑖𝑛 𝜃 𝜌𝐴 𝑉 𝑉 𝑠𝑖𝑛 𝜃 𝐻 𝑦 (2)
2𝜌𝐴 𝑦 𝑦𝑦 𝑠𝑖𝑛 𝜃 0
𝑚 2𝑚 2𝑚 𝑧 𝑐 𝑧 𝑘 𝑧 2 𝑚 𝑚 𝑢 𝑠𝑖𝑛 𝜃 4𝜌𝐴 𝑦 𝑦𝑦 𝑓 𝑓 (3)
in which 𝑚 , 𝑐 , and 𝑘 are respectively the liquid mass, damping, and stiffness of one VTLCD device, and they
are:
𝑚 𝜌𝐴 𝑉 𝑉 𝐻 (4)
𝑐 0.5𝜌𝐴𝛿|𝑦| (5)
𝑘 2𝜌𝐴 𝑧 𝑢 𝑠𝑖𝑛 𝜃 𝑔 (6)
where 𝜌 is the density of liquid inside the VTLCD; A is the cross-sectional area of the liquid columns; 𝛿 is the
flow resistance coefficient of the VTLCD; g is the acceleration of gravity; and 𝑓 is the radiation force due to the
change in momentum of the fluid resulting from the VLFS heave motion. Based on Cummins’s work [10], the
term 𝑓 in Equation (3) can be approximated as:
(7)
𝑓 𝑡 𝑚 𝑧 𝑡 𝐾 𝑡 𝑡 𝑧 𝑡 𝑑𝑡
where 𝑚 is the infinite-frequency added mass of the VLFS in the heave direction; and 𝐾 is the impulse response
function of the radiation in the heave direction. The integral term in Equation (7) represents the fluid-memory
model, capturing the energy transfer from the VLFS motion to the radiated waves.
SSP-type VLFS zs
Vertical tuned liquid column dampers
Container
gravitational acceleration
mc
y Liquid
Direction of
V2 V2
V1 zs Center of gravity
V3
u zs
Bracket m0+m'0(ω)
θ Sea wave
Equipped in
the VLFS
cv+czr' (ω) k0
f0
Reference base
Fig. 1. Analytical model of an SSP-type VLFS with VTLCDs.

On
ly
3
fo
rR
Equation (3) assumes that the VTLCDs are installed within the VLFS. In fact, during the processing and man-
ufacturing of the VLFS, it is possible to replace a portion of the structure’s weight with VTLCDs, to maintain the
ea
total mass of the VLFS unchanged. In this situation, Equation (3) can be rewritten as:
din
𝑚 𝑧 𝑐 𝑧 𝑘 𝑧 2 𝑚 𝑚 𝑢 𝑠𝑖𝑛 𝜃 4𝜌𝐴 𝑦 𝑦𝑦 𝑓 𝑓 (8)
3 Prototype of VLFS g
at
W
The SSP-type VLFS adopted in this work was designed by the China Ship Scientific Research Center (CSSRC)
CF
[11, 12], and it is shown in Fig. 2. The structure is composed of three parts: (i) one deck with a dimension of 300
m 100 m 6 m, (ii) ten columns (each has a height of 16 m and a diameter of 18 m), and (iii) five pontoons
S
(each is 30 m 96 m 5 m). The basic information of the VLFS is listed in Table 1. The damping ratio of this
20
VLFS in the heave direction is experimentally identified as approximately 5.5 % in previous studies using a small-
scale model (experimental scale 1:100) [12].
300 24
100
Deck
5 16
18
18
Column
15
30 96
Unit [m] Pontoon
(a) elevation view (b) side view
Fig. 2. Schematic drawings of the SSP-type VLFS designed by the CSSRC. (a) Elevation view; (b) side view.
Table 1. Full-scale principal particulars of the SSP-type VLFS designed by the CSSRC.
Designation Unit Value
Length over all m 300
Breadth m 100
Depth m 27
Draft m 15
Displacement ton 101072
Centre of gravity above base m 19
Radius of roll gyration m 28.47
Radius of pitch gyration m 86.31
Radius of yaw gyration m 89.77
4 Wave loading scenarios
Without loss of generality, to investigate the impacts of wave height and period on the heave motion control of
VLFS using VTLCDs, five wave loading scenarios (denoted as WL1-WL5) are considered. These five wave load-
ing scenarios represent various conditions and are summarized in Table 2, including working conditions in the
Gulf of Mexico (GM) and the South China Sea (SCS), 10-year return period waves in the North West Shelf of
Australia (NWSA) and the SCS, and 100-year return period waves in the SCS [13, 14]. The terms 10-year and
100-year waves indicate the likelihood of occurrence for waves of a specific magnitude, and are commonly used
in engineering and design to account for extreme wave events.
Table 2. Summary of five typical sea wave loading conditions.
Wave load Description Wave height 𝐻 (m) Wave period 𝑇 (s)

WL1 GM: working condition 3.96 9.00
WL2 SCS: working condition 6.00 11.20
WL3 NWSA: 10-year wave 10.00 12.50
WL4 SCS: 10-year wave 11.10 13.60
WL5 SCS: 100-year wave 13.30 15.50
On
ly
4
fo
5 Design parameters
rR
ea
To ensure optimal performance of the VTLCD device, it is necessary to optimize the parameters of the device. In
din
the present study, similar to many previous studies, the standard deviation 𝜎 of the structural response is selected
as the optimization objectives. The optimized values of the VTLCD parameters can be obtained via the genetic
g
algorithm using MATLAB. The following parameters can be optimized in MATLAB: (i) liquid mass ratio 𝜇 , (ii)
at
container (without liquid) mass ratio 𝜇 , (iii) liquid frequency tuning ratio 𝜙 , (iv) container (filled with liquid)
W
frequency tuning ratio 𝜙 , (v) flow resistance coefficient 𝛿, (vi) damping ratio of container 𝜁 , (vii) initial depth
ratio of liquid columns 𝜒, and (viii) angle of inclination 𝜃 (see Ref. [8]). The optimization results for 𝜒 0.3, 𝜃
70° , 𝜇 1.25% and 𝜇 6.25% are listed in Table 3.
CF
S
In Section 6, the results of using conventional FHPs are included for performance comparison with VTLCDs.
The analytical model of a VLFS equipped with a FHP, which is derived in [15], is used in the current analysis.
20
To ensure a fair comparison, we assume that the initial investment for both FHP and VTLCD devices is compa-
24
rable. The VTLCD design we adopt here is based on the following parameters: 𝜇 1.25% and 𝜇 6.25%.
This indicates that the total liquid mass ratio is 12.5% and the total container mass ratio is 2.5%. Assuming the
liquid within the VTLCD is seawater, which is considered cost-free, and the U-shaped container is fabricated from
steel with a unit cost of 𝛾, the total investment for the liquid and the U-shaped container would amount to
0.025𝛾𝑚 (𝑚 being the physical mass of the VLFS). Considering that the fabrication of the VTLCD involves
additional components such as springs, dashpots, and sliding tracks, and assuming the cost of these components
equals that of the U-shaped container, the initial total investment for the VTLCD system will be 0.05𝛾𝑚 . When
the initial investment for the FHP (made of steel) system is equivalent to 0.05𝛾𝑚 , it is straightforward to deduce
that the ratio of the FHP mass to the VLFS mass stands at 0.05. To obtain the specific dimensions of the FHP,
several reasonable assumptions are further made: (i) the shape of the FHP employs a conventional circular design;
(ii) to ensure that the FHP possesses sufficient stiffness, its thickness is set to be 0.075 times its diameter; (iii) the
draft of the VLFS remains unchanged before and after the installation of the FHP, and thus the physical weight of
the FHP is equivalent to the buoyancy force it experiences, which is directly proportional to the volume of sea-
water displaced by the FHP; and (iv) the connecting component between the FHP and the VLFS also displaces a
portion of seawater, the volume of seawater displaced by this connecting part is 10% of the volume displaced by
the FHP. Given these premises, the dimensions of the FHP can be determined (diameter: 42.4 m), and the dynamic
response of the VLFS-FHP system can then be computed utilizing the theoretical model developed in [15].
Table 3. Optimized control parameters of VTLCDs.
Wave load 𝛿[-] 𝜙 [-] 𝜙 [-] 𝜁 [-]

WL1 4.000 1.019 1.031 0.016
WL2 9.983 1.039 1.026 0.028
WL3 4.000 1.051 1.034 0.034
WL4 4.002 1.039 1.030 0.022
WL5 4.000 0.996 1.038 0.024
In Fig. 3, the heave displacement time histories of the VLFS controlled by VTLCDs under WL2-WL5 are plotted.
To scrutinize the control effects more clearly, we have selected a time window of 150 seconds for presentation.
VTLCDs are effective in reducing the heave response of the VLFS. It is important to note that due to the narrow
control bandwidth of VTLCDs, which primarily act around the baseline frequency 𝜔 (it is defined as the angular
frequency that corresponds to the maximum structural response in the spectrum for the uncontrolled cases), they
may not provide continuous vibration mitigation. When the frequency of the wave excitation deviates from this
baseline, the effectiveness of VTLCDs is significantly reduced, as exemplified in the interval from 900 s to 940 s
in Fig. 3 (b). During this interval, the dominant frequency of the VLFS response, at 0.449 rad/s, is 12% lower than
the baseline frequency of 𝜔 0.511 rad/s, leading to a markedly diminished effectiveness of VTLCDs due to
this frequency mismatch.
On
ly
5
fo
rR
ea
din
g
at
W
CF
(a) WL2 (b) WL3
S
20 Uncontrolled
24 With VTLCD
With FHP
(c) WL4 (d) WL5
Fig. 3. Heave motion time histories of the VLFS controlled by VTLCDs. (a)-(d) corresponds to WL2-WL5, respectively.
We define two vibration reduction metrics for the DOF 𝑧 within the time domain:
max |𝑧 𝑡 | (9)
𝑅 1
max |𝑧 𝑡 |
std 𝑧 𝑡 (10)
𝑅 1
std 𝑧 𝑡
where 𝑅 and 𝑅 are the reduction ratios of the peak value and the standard deviation value, respectively; the
superscripts c and u denote the results of controlled and uncontrolled cases, respectively. Based on Equations (9)
and (10), the values of vibration reduction metrics under WL1-WL5 within the time domain are summarized in
Fig. 4. By examining the control effectiveness, it is discernible that given similar initial investment costs, VTLCDs
may offer slightly superior control performance compared to FHPs. Furthermore, VTLCDs, like FHPs, exhibit
versatility in adapting to various sea conditions, making them a competitive option. In addition, it is observed that
both VTLCDs and FHPs exhibit control efficiencies typically fluctuating around 10%, indicating modest effec-
tiveness. This is largely attributed to the inherent high damping ratio of approximately 5.5% in the heave direction
within the considered VLFS, as detailed in Section 3. As a result, adding external vibration control devices may
yield less pronounced improvements due to this substantial intrinsic damping.
7 Conclusions
In this study, we explore the feasibility of VTLCDs for heave motion control in VLFSs. The VTLCD system is
composed of a pair of inclined U-shaped containers (filled with seawater), springs, dashpots, as well as supporting
slopes and is fixed in the VLFS. We assume that the ballast mass can effectively be provided by the mass of
VTLCDs, thereby integrating heave motion control into the system design of VLFSs. An analytical model of a
VLFS (300 m 100 m, designed by the CSSRC) equipped with VTLCDs is developed. Without loss of general-
ity, five wave loading scenarios are considered. VTLCDs successfully reduce the heave response of VLFSs. With
equivalent initial investment expenses, VTLCDs may exhibit slightly enhanced control performance over FHPs.
It is noted that being non-submerged, VTLCDs are not susceptible to fouling by marine life, and they benefit from
more straightforward installation and maintenance processes, which can subsequently lead to reduced overall
project expenses. Overall, the non-submerged VTLCD presents a viable solution for controlling the heave motion
of VLFSs.
On
ly
6
fo
rR
ea
(a) (b)
14.4
15 15
14
VTLCD VTLCD
din
FHP FHP
12.2
11.9
g
12 12
10.8
10.6
10.3
at
Reduction ratio (%)
9.9
9.6
9.1
W
8.8
9 9
8.1
CF
7.5
7.5
7.4
7.2
7.2
7.1
7.0
6.1
5.4
S
6 6
20
24
3 3
0 0
1 2 3 4 5 1 2 3 4 5
Wave load number (-)
Fig. 4. Control efficiency of the FHP or VTLCDs in the time domain under WL1-WL5. (a) and (b) correspond to the reduction
ratios of the maximum absolute value and the standard deviation value in structural displacement time histories, respectively.
References
1. Wang, C., Tay, Z.: Very large floating structures: applications, research and development. Procedia Engineering 14,
62–72 (2011).
2. Chandrasekaran, S.: Offshore semi-submersible platform engineering. CRC Press, Location (2020).
3. Yue, M., Liu, Q., Li, C., Ding, Q., Cheng, S., Zhu, H.: Effects of heave plate on dynamic response of floating wind
turbine Spar platform under the coupling effect of wind and wave. Ocean Engineering 201, 107103 (2020).
4. Liu, K., Ou, J.: A novel tuned heave plate system for heave motion suppression and energy harvesting on semi-
submersible platforms. Science China Technological Sciences 59, 897–912 (2016).
5. Ma, R., Bi, K., Hao, H.: A novel rotational inertia damper for amplifying fluid resistance: Experiment and mechan-
ical model. Mechanical Systems and Signal Processing 149, 107313 (2021).
6. Colwell, S., Basu, B.: Tuned liquid column dampers in offshore wind turbines for structural control. Engineering
Structures 31, 358–368 (2009).
7. Zhou, Y., Qian, L., Bai, W.: Sloshing dynamics of a tuned liquid multi-column damper for semi-submersible floating
offshore wind turbines. Ocean Engineering 269, 113484 (2023).
8. Ding, H., Bi, K., Song, J., Fang, X.: Vertical vibration control of structures with tuned liquid column dampers.
International Journal of Mechanical Sciences 279, 109502 (2024).
9. Ma, R., Bi, K., Hao, H.: Heave motion mitigation of semi-submersible platform using inerter-based vibration isola-
tion system (IVIS). Engineering Structures 219, 110833 (2020).
10. Cummins, W.: The impulse response function and ship motions (1962).
11. Wu, Y., Ding, J., Li, Z., Ni, X., Wu, X., Tian, C.: Hydroelastic responses of VLFS deployed near islands and reefs.
In: International Conference on Offshore Mechanics and Arctic Engineering, American Society of Mechanical En-
gineers, pp. V009T12A14 (2017).
12. Wang, Y., Tang, Z., Wang, L., Wang, X.: Linear and quadratic damping coefficients of a single module of a very
large floating structure over variable bathymetry: physical and numerical free-decay experiments. Journal of Ocean
Engineering and Science 7, 607–618 (2022).
13. Liu, K., Liang, H., Ou, J.: Numerical investigation of a tuned heave plate energy-harvesting system of a semi-sub-
mersible platform. Energies 9, 82 (2016).
14. Travanca, J., Hao, H.: Control of wave-induced vibrations on floating production systems. Ocean Engineering 141,
35–52 (2017).
15. Ma, R., Bi, K., Hao, H.: Mitigation of heave response of semi-submersible platform (SSP) using tuned heave plate
inerter (THPI). Engineering Structures 177, 357–373 (2018).
On
ly
fo
rR
ea
din
The 4th World Conference on Floating Solutions (WCFS 2024)
g
2-4 December 2024, Hong Kong, China
at
W
Sustainable Floating City as a Solution to
Rising Sea Levels: CF
An Integrated Project for Northwest Coast of Brazil S
20
24
Allan Gomes1, Fernando Borguetti1, Guilherme Moraes 1, Sena Takahashi2,
Tomoki Ikoma3 and Kazuo Nishimoto1
1Polytechnic School of the University of Sao Paulo, Sao Paulo, Brazil

2Department of Architecture, Waseda University, Tokyo, Japan
3Department of Oceanic Architecture & Engineering, CST, Nihon University Chiba, Japan
Abstract. This work presents an innovative integrated project for a floating city
designed to address the challenges posed by rising sea levels, particularly in Re-
cife, Brazil. The project encompasses comprehensive naval engineering con-
cepts, with a focus on hydrostatics, hydrodynamics, anchoring, and renewable
energies. It draws from a diverse range of sources for the various topics covered
and involved sourcing suppliers for the city's components. Additionally, a part-
nership with Nihon University was established for in-depth structural and hydro-
dynamic analyses. Using the Oceanix model as a starting point, this research
delves deeper into the essential engineering aspects necessary for developing this
complex system. The primary focus is on integrating solutions, analyzing tech-
nical aspects, and ensuring the sustainability of the floating city. It is noteworthy
that this work is open to future enhancements, aiming to conduct economic fea-
sibility analyses and further explore the individual considerations of each com-
ponent addressed in this document.
Keywords: Floating city, Naval Architecture, Sea Level Rise, Sustainability,

Hydrostatics, Hydrodynamics, Renewable Energies, Integrated Solutions, Re-
cife, Brazil.
1 Introduction
This paper introduces a strategic response to mitigate the adverse effects of rising sea
levels, with a particular focus on the feasibility of implementing a floating city as a
potential housing solution for future displaced populations. The objective is to apply
floating systems engineering principles to examine the technical feasibility of a Very
Large Floating Structure (VLFS) in Recife, one of the Brazilian cities most affected by
this phenomenon. By investigating the feasibility of floating infrastructure, this study
On
ly
fo
rR
ea
2
din
g
at
aims to provide a sustainable and innovative solution that can adapt to the dynamic and
increased impacts of climate change on coastal urban areas.
W
The paper will cover real case studies, Oceanix Busan, deep case analysis, problem
CF
development, project requirements, weight and center analysis, stability assessments,
energy module considerations, anchoring strategies, hydrodynamic analyses, and the
S
final presentation of results. The multifaceted methodology, integrating naval architec-
20
ture principles with global concepts, underscores the comprehensive nature of the
study. This approach is pivotal in effectively addressing to several challenges associ-
ated with rising sea levels in Recife.
24
2 The problem formulation
Sea-level rise has become a major global concern due to global warming, intensified
by anthropogenic effects. On a global scale, tangible consequences have already been
observed:
 A sea level rise of 20 cm since 1900, with the potential to reach 1 meter in
the next 50 years.
 Increased frequency and intensity of coastal flooding, coastal erosion, and
soil salinization.
 The World Bank estimates that coastal cities could incur losses exceeding
1 trillion USD by 2050.
In the Brazilian context, the Ministry of the Environment estimates that approxi-
mately 60% of the Brazilian population could be affected by the effects of sea-level rise
due to living in coastal regions. Additionally, a study by the Federal University of Per-
nambuco (UFPE) highlighted that Recife is one of the most vulnerable cities in the
country due to its low average altitude of only 4 meters above sea level. In a scenario
of a 0.5-meter sea-level rise, it is estimated that the flooded area could be around 25.38
km², potentially affecting 175,000 people based on census density, highlighting the im-
portance of finding a pivotal solution for this worst-case scenario.
3 Weights and Centers of General Arrangement
The weights and centers estimations were made by breaking down the different func-
tional modules designed for the floating city (general arrangement). For each module,
a should-weight analysis was made to build up the estimated weight by its subcompo-
nents. Module “A2 – Food Production and Storage”, for example, followed a rationale
of defining the average daily food intake per person, the crop yield of different crops
and the weight of aeroponic equipment. The results are listed below:
Table 1. Estimated weight of functional modules
Module Weight [ton] KG [m]

A1 - Freshwater Production, Treatment and Storage 5.257 2,95
A2 – Food Production and Storage 87 1,64
On
ly
fo
rR
ea
din 3
g
A3 – Waste Treatment
at 234 0,69
W
A4 – Residential Towers 14.726 14,99
CF
A5 – Floating Structure 5.308 3,56
Total 25.611 9,97
S
4 Structural Arrangement Analysis
20
24
The structural analysis for the studied case should start with a thorough analysis of the
classification societies, the suitability of the existing rules for the context of the floating
city, and their applicability to the design scenario of the primary, secondary, and tertiary
structures. With this analysis, it was possible to determine the estimated number of
structural components, thicknesses, dimensions, and the approximate weight of the
structures.
The support structure for the designed neighborhood includes four main compo-
nents: a lower hexagonal structure, internal support beams, bulkheads, and an upper
deck shaped like an equilateral triangle.
Fig. 1. Main structural components

On
ly
fo
rR
ea
4
din
g
5 Energy Generation
at
W
Aiming to establish a clean and renewable energy matrix, several energies generation
CF
sources for the city location were analyzed: solar, wind, tides, waves, and currents. Es-
S
timations based on local data are crucial to determine the optimal energy matrix for the
floating city. Selecting an energy source for the floating city requires understanding
20
specific environmental conditions. This section analyzes geographical, climatic, and
24
oceanographic factors that affect energy source viability and efficiency.
5.1 Local Environmental Conditions and Energy Matrix Selection

The coastal Northeast region, where the floating city is proposed, has favorable wind
conditions with speeds between 2.5 and 3.0 m/s and peaks of 12.0 to 15.0 m/s. Solar
energy is also viable, with annual solar radiation ranging from 5,000 to 5,500 Wh/m²
per day. However, tidal and tidal stream energies are limited due to low tide variations
and insufficient wave amplitudes.
Considering Recife's total energy consumption and population, necessary daily en-
ergy are calculated. A hybrid energy approach, combining wind and solar generation
with energy storage, is proposed to supply the city's energy requirements.
6 Intact and Damaged Stability
Since the structure is floating system, one crucial safety and comfort requirement for
the floating city is its stability. Therefore, the initial, intact and damaged stability (by
lost buoyance method) were evaluated for the project. The key metrics evaluated were:
Table 3. Estimated weight of functional modules
Module Weight [ton] KG [m]

Draft 5.257 2,95
2º Moment of Area (Waterplane) 16.688.488 m4
Displaced volume ∇ 24.551 m3
Metacentric Height (GM) 670.72 m
Flooding angle 𝜃 3,36 °
Righting lever (GZ) AUC to 𝜃 1,15 m-rad
For the damaged stability, the complete flooding of each compartment (Fig. 2) was
evaluated. Due to the high value of the metacentric height, the maximum heel/trim an-
gle was 0.95° (when compartment 1 was flooded), which is less than the flooding angle
𝜃 3,36°.
On
ly
fo
rR
ea
din 5
g
at
W
CF
S
20
Fig. 2. Compartmentalization 24
7 Hydrodynamics
7.1 Wave Spectrum Consideration

For modeling the wave spectrum observed in the region of Recife (PE), it was necessary
to consider a standardized sea spectrum that determines the relationship between wave
height and period. For this project, the frequency domain spectrum was considered due
to the ease of wave modeling with fewer parameters, using JONSWAP ("Joint North
Sea Wave Observation Project") as it is the most universally used type, commonly em-
ployed for modeling on the Brazilian coast.
This type of spectrum model for waves uses only two parameters: significant wave
height and peak period. Based on historical observations in the Recife region, approxi-
mately 10 km from the coast, these parameters have been estimated at 1-3.5 meters for
wave height and 7.5 seconds for the peak period. The Python implementation of the
JONSWAP spectrum is illustrated in the figure below:
Fig. 3. Angular and Frequency Domain Spectrum
7.2 Transfer Function

With the wave spectrum in the frequency domain obtained, it was established the trans-
fer function (spectral crossing) that will be responsible for determining the response
On
ly
fo
rR
ea
6
din
g
at
curve of the floating unit. This response function can be obtained using the module of
the Response Amplitude Operator (RAO).
W
For the initial analysis of the RAO of the structure, modeling was conducted using
CF
WAMIT software considering rigid body. This resulted in frequency domain curves
considering the heave, pitch, and roll movements of each floating parts. It is important
S
to note that this preliminary analysis considered only the floating neighborhood, repre-
20
senting a single floating unit and the damping effect generated by the anchoring system
was considered negligible, as its primary function was initially directed solely towards
24
preventing the drift movement of the structure. Considering this, the following RAO
curves in the frequency domain were obtained, considering 5 different incidence angles,
as illustrated in the figure below
Fig. 4. RAOs for Heave, Roll and Pitch
7.3 Results Analysis

The analysis of response spectra allows us to conclude that:
 Heave: The response spectrum showed little variation in heave motion due to
the high restoring coefficient of the structure, with a larger variation in vertical
motion equal to 0.71m. However, it was emphasized the importance of creating
an anchoring system that ensures the natural period is outside any risk of reso-
nance between the structure and the wave period of the region.
 Roll and Pitch: As explained in the stability chapter, one of the systemic con-
cerns of the designed neighborhood was to ensure that sea conditions would not
cause significant angulation. In this context, the response spectra showed that,
despite the wave period of the region falling within a range of maxima, the val-
ues are negligible, with maximum variation values reaching 0.013 rad (0.771º)
due to the large proportion of the floating structure.
In summary, hydrodynamic analyses have shown that the structure remains stable
dynamically under the hydrodynamic conditions of the Recife region and 10 km from
the coast. However, since the structure is very large, the hydroelasticity analysis will
be required when several module will be connected each other.
On
ly
fo
rR
ea
din 7
g
8 Mooring System
at
8.1
W
Anchoring System Selection and Mooring Line Sizing
CF
Designing large floating structures involves critical decisions, particularly in selecting
S
the anchoring system. Challenges include accommodating their significant size, envi-
20
ronmental loads, shallow depths, limited mooring cable space, and anchor installation
complexities.
24
For the floating city, the Taut Legs system was chosen due to the shallow depth and
proximity to the coast. This system employs angled ties and tensioned chains to mini-
mize the anchor footprint, which is crucial in areas with high vessel traffic. An alterna-
tive catenary arrangement, where part of the mooring line rests on the seabed, was con-
sidered but deemed less efficient due to its larger footprint. The chosen mooring system
must withstand environmental forces, primarily constant northeastern winds in the se-
lected coastal region, with minimal impact from waves and currents.
A simplified approach involves sizing the mooring lines to account for the forces
acting on the city. The mooring system is pre-tensioned to absorb energy and reduce
movement. The primary consideration in the analysis was the horizontal wind force
combined with the pre-tension established in the moorings. The estimated force was F
= 7.377 kN, leading to the use of 12 moorings attached to the vertices of the hexagon
to withstand this force.
Fig. 6. Final mooring line diagram
8.2 Anchor Selection

Choosing anchors for floating structures involves assessing various loading conditions
and seabed soil properties, requiring thorough geotechnical analyses. For the floating
city prject, driven piles with diameters of 1-3 meters are selected for their strong re-
sistance, capable of supporting over 5,000 tons. These piles, ideal for systems like Taut
Legs, are installed using pile-driving hammers in shallow depths. Anchor fixation is
achieved through shear with the soil, utilizing the piles' hollow design effectively.
On
ly
fo
rR
ea
8
din
g
9 Hydroelasticity Response
at
W
CF
9.1 Analysis Method
S
In this study, eigenvalue analysis was carried out using the finite element structural
analysis software Femap with NX Nastran® , and flow force elastic response analysis
20
was carried out using eigenfunctions and eigenvalues in the manner of Sato et al.6) .
24
The computational conditions for the unit-connected large floating structure used in the
analysis are shown in Table 3 and an overview of the unit-connected floating structure
is shown in Fig. 7.
Table 3. Calculation conditions
High strength concrete

Floating body material
(cheaper than steel)
Floating Body Thickness B [mm] 1000
Young's modulus [N/mm2] 4.0x104
Poisson's ratio 0.2
Mass density[N/mm2] 2.5x10-10
Depth of water [m] 15.0
Angle of incident waves [°] 0.0
Incident wave period [sec.] 4.0 to 20
連結部分
B
1) Ground plan 2) Cross-section view
Fig. 7. Outline of unit-connected floating structure
The computational floating body mesh partition is shown in Figure 8. The eigenfunc-
tions and eigenvalues obtained from the eigenvalue analysis were used to analyse the
hydroelastic response at incident wave periods from 4 s to 20 s.
On
ly
fo
rR
ea
din 9
g
at
W
CF
S
20
Fig. 8. Floating body mesh model for calculation
24
9.2 Vertical displacement calculation points overview
To evaluate the elastic deformation in waves, 16 points were selected from the nodes
of the mesh data and flow force elastic response analysis was carried out using a calcu-
lation programmed for 8-node isoperimetric elements. 6 of the 16 points, A-F, were
selected, including points with high loads and the inside and outside of the connecting
unit. The six evaluation points are shown in Fig. 9.
Fig. 9. Assessment points

-2 -2 -4
[10 ] [10 ] 2.5 [10 ]
1.0 1.0
2.0
0.8 0.8
0.6 1.5 0.6

|/a|
|/a|
|/a|
0.4 1.0 0.4
0.2 0.5 0.2
0 0 0
0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4
/L /L /L
a) Point A b) Point B c) Point C

[10-3] 3.0 [10-3] [10 ]
-2
3.0 1.0
0.8
2.0
2.0 0.6
|/a|
|/a|
|/a|
0.4
1.0 1.0
0.2
0
0 0 0.1 0.2 0.3 0.4
0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4
/L
/L /L
d) Point D e) Point E f) Point F

Fig. 10. Vertical displacement amplitude at each point due to elastic deformation in
waves
On
ly
fo
rR
ea
10
din
g
at
The maximum values of vertical displacement at each of the A~F assessment points
are shown in Fig. 10. The vertical axis of the graph is shown against the dimensionless
W
value of the deformation, while the horizontal axis is shown as the ratio of wavelength
CF
to the length of the floating body. Here, λ: incident wave wavelength, L: total length of
the floating body, a: incident wave amplitude.
S
Similar changes were observed in the wavelength-to-float length ratio at points A
20
and D, E and F respectively. Particularly large deformation was observed at point B,
with a deformation of 0.025 at λ/L = 0.39. This result can be attributed to the fact that
24
point B is the evaluation point located furthest in front of the incident wave. The next
largest deformation was observed at point FH, with a deformation of 0.011 at λ/L=0.39.
The results show that similar changes were observed at points A and D, E and F on the
outer side of the floating structure for each wavelength range, suggesting that elastic
deformation in waves depends on the direction of the incident wave, and that the shal-
low water depth of 15 m in the planned area of this study makes the wavelength rela-
tively small compared to the total length of the floating structure, so the motion The
water depth in the planned study area is shallow (15 m), so the wavelength is relatively
small compared to the total length of the floating structure, which is thought to have
resulted in the micro-scale motion.
10 City Planning Proposal
The unit-connected floating structure proposed in this study was found to have minimal
displacement at key locations due to elastic deformation. Therefore, it is considered
that the discomfort people feel when living at sea is not significant. Based on these
results, an overview of the actual urban design is shown in Figure 11.
a) layout plan b) Overall perspective
c) Medical institutions, research

d) living space
institutions
On
ly
fo
rR
ea
din 11
g
at
W
CF
S
e) educational institution f) amusement facilities 20
Fig. 11. Urban planning proposal
24
11 Conclusion
In summary, the system was proved to be feasible in the naval architecture aspects and
safety aspects, considering the weights and centers, structural analysis, energy genera-
tion system, stability, hydrodynamics- hydroelasticity system, and mooring system.
References
1. Author, F.: Article title. Journal 2(5), 99–110 (2016).
2. Author, F., Author, S.: Title of a proceedings paper. In: Editor, F., Editor, S. (eds.)
CONFERENCE 2016, LNCS, vol. 9999, pp. 1–13. Springer, Heidelberg (2016).
3. Author, F., Author, S., Author, T.: Book title. 2nd edn. Publisher, Location (1999).
4. Author, F.: Contribution title. In: 9th International Proceedings on Proceedings, pp. 1–2.
Publisher, Location (2010).
5. LNCS Homepage, http://www.springer.com/lncs, last accessed 2016/11/21.
6. Fumiya Sato, Tomoki Ikoma, Koichi Masuda and Tomoka Kaneko: Study on the behaviour
characteristics in waves of a large circular floating structure with a moon pool, Conference
proceedings, pp. 385-386, (Marine Architecture 10006), 2011.8.
7. Kouichi Masuda, Tomoki Ikoma, Fumiya Sato, Akio Kuroyanagi, Akio Kobayashi, Juro
Hiroumi, Yukio Nobukawa, Mikazu Nakanishi, Yasuhiro Umemura and Toshikatsu Saito:
Proposal of Floating Plan Planning Considering Response in Waves of a Multifunctional
Large Floating Body NU-FLOAT, 23rd Symposium on Ocean Engineering, Japan Society
of Marine Engineering and Ocean Technology, 2012.
On
ly
fo
rR
Numerical Investigation on Motion Response of Modular Semi-submersible
ea
Floating Platform in Waves
din
g
Lidong Zhang1, Jia Wang1, Xuanlie Zhao1,2[0000-0002-4184-9831], Jing Geng1,2, Zhiqiang Hu3 and C. Guedes Soares2,4
1
at
College of Shipbuilding Engineering, Harbin Engineering University, Harbin, China
W
2 International Joint Laboratory of Naval Architecture and Offshore Technology between Harbin Engineering University and
CF
University of Lisbon, Harbin, P.R. China
3 School of Engineering, Newcastle University, Newcastle, United Kingdom
4 Centre for Marine Technology and Ocean Engineering (CENTEC), Instituto Superior Técnico, Universidade de Lisboa,
Lisbon, Portugal
S
20
xlzhao@hrbeu.edu.cn
24
Abstract. Modular semi-submersible floating platforms have recently gained popularity in offshore renewable
energy applications. In this paper, the hydrodynamic responses of such platforms are investigated numerically.
A constraint matrix is derived that characterizes the motion relation of the interconnected multi-module system.
The frequency-domain motion equation is formulated using the Lagrange multiplier method. Parametrical in-
vestigations are conducted to explore the hydrodynamic performance of the modular semi-submersible floating
platform in waves. The effects of incident wave angle and the spacing between adjacent modules are further
investigated. The numerical results of the hydrodynamic responses of the multi-module system are presented,
and the forces acting on the connectors are analyzed. It is found that the maximum responses in heave and
pitch degree-of-freedom are observed for the module at the lee side. The motion responses of the modules are
relatively larger at the lower frequency range. In general, the vertical forces acting on connectors at the middle
and high-frequency ranges are larger than at the low-frequency range. The findings can provide valuable ref-
erences for designing and constructing of floating modular structures.
Keywords: Modular Floating Platform, Lagrange Multiplier Method, Motion Response, Potential Flow The-
ory.
1 Introduction
Developing of offshore airports, terminals and multi-purpose platforms are crucial to explore marine resources.
However, as the structural size increases, the structural deformation exerts a pivotal influence on the motion re-
sponse of the floating body. Many investigations have been conducted to explore the hydrodynamics of very large
floating structures (VLFS) or multi-body systems. Wu [1] developed a three-dimensional linear hydroelastic the-
ory to analyze the motion and deformation of an arbitrary-shaped floating body in waves. Kiyokazu and Shoichi
[2] developed a numerical analysis code to estimate the elastic motions and the connecting loads of box-shaped
VLFS. Watanabe et al. [3] performed a hydroelastic analysis of a pontoon-type circular VLFS using the modal
expansion method. Bispo et al. [4] presented a numerical model associated with wave interaction with a moored
articulated very large floating structure and analyzed the hydrodynamic performance on the floating structure for
different design parameters.
Du and Ertekin [5] numerically investigated four-module floating structure's motion response and loading in
multi-directional random waves. Lee and Newman [6] proposed a frequency-domain motion equation for a float-
ing system consisting of multiple modules, and the response and loading of the floating system are analyzed. Sun
et al. [7] employed the Lagrange multiplier method to formulate the motion equation of the interconnected floating
bodies. Jin et al. [8] and Han et al. [9] used the open-source tool WEC-Sim to model a two-body hinged raft wave
energy converter and investigated the wave power extraction performance of the device.
This paper focuses on a floating system consisting of four interconnected modules (see Fig. 1). The frequency-
domain motion equation is formulated based on the Lagrange multiplier method. Using a numerical method, par-
ametrical investigations focus on the analysis of the hydrodynamic responses and forces on connectors.
On
ly
2
fo
2 Numerical method
rR
ea
Considering the hydrodynamic coupling of a multi-module system, the 6-degree-of-freedom (6DoF) motion equa-
din
tion of a modular semi-submersible platform without the connection can be formulated as:
  (M  a)  ib  C ε  f
g
2
(1)
at
Here i is the imaginary unit; ω is the wave angular frequency; M, a, b, C,  and f denote mass matrix, added mass
W
matrix, radiation damping matrix, restoring force matrix, 6DoF motion response vector and wave excitation force
CF
vector, respectively. a, b and f are obtained by using the potential-flow solver NEMOH [10].
Motion Equation (1) can be simplified as Kε  f , where K   2 ( M  a)  i b  C , ε  [ε1 , ε 2 ,  , ε n ]T ,
εi   ,  , , 
(1) (2) (6)
S
 , i  1, 2, , n , n represents the number of the modules,  ( j)
corresponds to the com-
20
i i i i
plex response of the i-th module in the j-th DoF. j  1, 2, , 6 corresponds sequentially to the surge, sway,
heave, roll, pitch and yaw DoF, respectively.
24
The sketch of the multi-module system is shown in Fig. 1. The central module (i.e., Module #2) is surrounded
by three modules (i.e., Module #1, #3, #4). Hinge #1 denotes the connector between Module #1 and Module #2.
The role of Hinge #1, #2 and #3 can be found in Fig. 1. Note that the connectors considered in this paper are
column hinges. Based on the Lagrange multiplier method [7], the corresponding motion equation can be derived
as:
K DT   ε   f 
      (2)
D 0   λ  0 
Here D denotes the constraint matrix; the force vector of the hinges is expressed as λ  [ λ 1 , λ 2 ,  , λ m ]T , where
λ i  i(1) , i(2) , , i(5)  , i  1, 2,  , m , m stands for the hinge number; i( j ) corresponds to the complex load
of the i-th hinge in the j-th DoF. For instance, 1( j ) , j  1, 2, , 5 corresponds sequentially to x'-, y'-, and z'-
direction complex forces, as well as the x'- and z'-direction complex torques of the local coordinate system, o'-
x'y'z', acting on the Hinge #1, respectively.
Fig. 1. Top-view of the modular semi-submersible floating platform. β is incident wave angle, l denotes the spacing between
adjacent modules. o-z, o'-z0 and o'-z' axes point upwardly.
Fig. 2 presents the gravitational centres of Module #1, Module #2 and the corresponding coordinate of the
connector Hinge #1. The coordinate system of o'-x0y0z0 is parallel with the global system o-xyz. And the o'-x' axis
in the o'-x'y'z' system is aligned with the connecting line between the two gravitational centres. In this instance,
the angle, α, is formed by the o'-x0 and o'-x' axes.
On
ly
3
fo
rRz z2
y y2
ea z0
din
y0 o2
x
o z' y' (xc2, yc2, zc2)
x2
g
z1 y1 x' Module #2
at
o' 
(x1, y1, z1) x0
W
Hinge #1
o
CF
(x 1, y
c1 c1, zc1) x1
Module #1
S
20
Fig. 2. The coordinates of two adjacent modules (i.e., Module #1 and Module #2) and the corresponding connector (i.e.,
Hinge #1).
It is known that the relative pitch motion in the o'-x'y'z' system is released under action of the connectors. The
constraint matrix D, which characterizes such kind of motion, can be derived as:
24
1 0 0 0 ( zc1  z1 ) ( yc1  y1 ) 1 0 0 0 ( zc 2  z1 ) ( yc 2  y1 ) 
 cos  sin  
  sin   0 1 0 ( zc1  z1 ) 0 ( xc1  x1 ) 0 1 0 ( zc 2  z1 ) 0 ( xc 2  x1 ) 
cos   (3)
  0 0 1 ( yc1  y1 ) ( xc1  x1 ) 0 0 0 1 ( yc 2  y1 ) ( xc 2  x1 ) 0 
D   1  
  0 0 0 1 0 0 0 0 0 1 0 0 
 cos  sin   0 0 0 0 1 0 0 0 0 0 1 0 
 1  
0 0 0 0 0 1 0 0 0 0 0 1 
Here (xc1, yc1, zc1) and (xc2, yc2, zc2) denote the gravitational centre coordinates of Module 1 and Module 2, respec-
tively, and the coordinate (x1, y1, z1) represents the hinge point of the connector.
3 Description of modular semi-submersible floating platform
As shown in Fig. 1, the plane of the modular semi-submersible floating platform contains four identical floating
modules. The sketch of the module is depicted in Fig. 3, and the corresponding parameters are shown in Table 1.
Fig. 3. The identical module unit of a modular semi-submersible floating platform.
Table 1. The geometrical parameters of the floating platform.
Term Dimensions (m)

Diameter of Upper column 1, Upper column 2 and Upper column 3 4.0
Diameter of Base column 1, Base column 2 and Base column 3 8.0
Diameter of Upper brace 1.0
Diameter of Base brace 1.0
Height of Deck 1.0
4.1 Effect of incident wave angle

In this subsection, the effect of incident wave angle on the hydrodynamic responses of the multi-module system
is investigated by considering β = 0°, 30° and 60°. Fig. 4 presents the results of heave and pitch DoF responses
On
ly
4
fo
rR
for an isolated platform. It can be demonstrated that the incident wave angle slightly influences the trend of heave
and pitch motion. As for the heave motion, it is observed that the peak value is located at ω = 0.65 rad/s. A spike
ea
is observed at ω = 0.50 rad/s in the trend of pitch motion against frequency.
din
4.0 0.4
ß = 0? ß = 0?
g
ß = 30? ß = 30?
3.0 0.3
at
ß = 60? ß = 60?
W
2.0 0.2
1.0 0.1 CF
S
20
0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
24
 (rad/s)  (rad/s)
(a) (b)
Fig. 4. Results of heave (a) and pitch (b) responses for isolated module in β = 0°, 30° and 60°.
Fig. 5 demonstrates the results of the heave and pitch responses of each module for the multi-module system
when β = 0° and l = 3 m. Compared to the isolated case, the hydrodynamic interference among different modules
modifies the trend of heave and pitch responses, although to a limited extent. Multiple peaks are found at ω = 0.55
rad/s and 0.65 rad/s in Fig 5 (b). Furthermore, it is found that the maximum motion responses in heave and pitch
DoF are observed at Module #3, which is located on the lee side.
Fig. 6 shows the impact of incident wave angle on the hydrodynamic responses of Module #3. It is observed
that the peak values of the heave and pitch responses decrease with the increase of incident wave angle. It is
noteworthy that the amplitude of heave motion corresponding to β = 60° is slightly larger than that corresponding
to β = 0° and 30° within the frequency range of 0.80 rad/s < ω < 1.50 rad/s. However, the pitch response of
Module #3 decreases as the incident wave angle increases.
4.0 0.24
i =1 i=2 i=1 i=2
i =3 i=4 i=3 i=4
3.0 0.18
2.0 0.12
1.0 0.06
0.0 0.00
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
 (rad/s)  (rad/s)
(a) (b)
Fig. 5. Results of heave (a) and pitch (b) responses for modular semi-submersible floating platform. (β = 0°, l = 3 m)
4.0 0.24
ß = 0? ß = 0?
ß = 30? ß = 30?
3.0 0.18
ß = 60? ß = 60?
2.0 0.12
1.0 0.06
0.0 0.00
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
 (rad/s)  (rad/s)
(a) (b)
Fig. 6. Results of heave (a) and pitch (b) responses of Module #3 in β = 0°, 30° and 60°. (l = 3 m)
On
ly
5
fo
4.2 Effect of module spacing
rR
ea
The effect of module spacing (l = 1 m, 3 m and 5 m) is investigated in this subsection, where β = 0° is considered
here. Fig. 7 shows us the variation of heave and pitch motion for different module spacings. The modifications of
din
the heave and pitch responses are primarily reflected within the frequency range of 0.50 rad/s < ω < 0.80 rad/s.
g
It is noted that the peak values of heave response are 3.47 m, 2.55 m and 4.55 m for l = 1 m, 3 m and 5 m,
respectively. Correspondingly, the peak values of pitch response are 0.08 rad, 0.10 rad and 0.54 rad.
6.0
at0.8
l=1m
W l=1m
CF
4.5 l=3m 0.6 l=3m
l=5m l=5m
3.0 0.4
S
20
24
1.5 0.2
0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
 (rad/s)  (rad/s)
(a) (b)
Fig. 7. Results of heave (a) and pitch (b) responses for Module #1 in l = 1 m, 3 m and 5 m. (β = 0°)
Fig. 8 illustrates the results of vertical forces acting on connectors by considering β = 0° and l = 3 m. Due to
the symmetrical arrangement, the vertical forces acting on Hinge #1 and Hinge #3 are identical. It's worth noting
that the force acting on Hinge #2, which is located at the lee side, is larger than that on Hinge #1/Hinge #3 at the
tested frequency range.
0.4
i=1
i=2
0.3
i=3
0.2
0.1
0.0
0.0 0.5 1.0 1.5 2.0 2.5
 (rad/s)
Fig. 8. Results of vertical forces acting on different connectors. (β = 0°, l = 3 m)
0.60 0.8
l=1 m l=1m
l=3 m l=3m
0.45 0.6
l=5 m l=5m
0.30 0.4
0.15 0.2
0.00 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
 (rad/s)  (rad/s)
(a) (b)
Fig. 9. The vertical forces acting on Hinge #1 (a) and Hinge #2 (b). (l = 1 m, 3 m and 5 m)
On
ly
6
fo
rR
Fig. 9 describes the vertical forces acting on Hinge #1 (i.e., 1(3) ) and Hinge #2 (i.e., 2(3) ) for l = 1 m, 3 m
ea
and 5 m. A spike is found for both 1(3) and 2(3) at ω = 0.55 rad/s for l = 5 m. However, such kind of spikes
din
cannot be found for case of l = 1 m and 3 m. With the exception of the spike observed at the lower frequency
range, it can be found that 1(3) and 2(3) at the middle and high-frequency range (i.e., 1.00 rad/s < ω < 2.50
g
rad/s) are larger than those observed at the lower frequency range (i.e., 0.05 rad/s < ω < 1.00 rad/s). It can be
at
concluded from the calculations that medium and short waves may trigger a greater force acting on the connector.
W
5 Conclusions
CF
S
The hydrodynamic performance of a modular semi-submersible floating platform in waves is investigated numer-
20
ically in the frequency domain based on the Lagrange multiplier method. The effect of incident wave angle and
24
module spacing are discussed in this paper. The conclusions are summarized as follows:
1) the peak motion response amplitude decreases in heave and pitch DoF with the increase of incident wave angle;
2) the maximum motion response is found for the lee side module (i.e., Module #3) when β = 0°;
3) the larger vertical forces acting on connectors are found at the middle and high frequency range.
References
1. Wu, Y.: Hydroelasticity of floating bodies. Thesis, Brunel University, (1984).
2. Kiyokazu, Y., Shoichi, H.: On the hydroelastic response of box-shaped floating structure with shallow draft. Journal of
the Society of Naval Architects of Japan, 183, 227-237 (1998).
3. Watanabe, E., Utsunomiya, T., Wang, C.M., Hang, L.T.T.: Benchmark hydroelastic responses of a circular VLFS under
wave action. Engineering Structures, 28, 423-430 (2005).
4. Bispo I. B. S., Mohapatra S. C., Guedes, S. C.: Numerical analysis of a moored very large floating structure composed by
a set of hinged plates. Ocean Engineering, 253: 110785, (2022).
5. Du, S., Ertekin, R.C.: Dynamic response analysis of a flexibly joined, multi-module very large floating structure. In:
Proceedings of the Institute of Electrical and Electronics Engineers, pp. 1286-1293. Honolulu, Hawaii, USA (1991).
6. Lee, C. H., Newman, J. N.: An assessment of hydroelasticity for very large hinged vessels. Journal of Fluids & Structures,
14(7), 957-970 (2000).
7. Sun, L., Taylor, R.E., Choo, Y.S.: Responses of interconnected floating bodies. IES Journal Part A: Civil and Structural
Engineering, 4(3), 143-156 (2011).
8. Jin, S., Wang, D., Hann, M., Collins, K., Conley, D., Greaves, D.: A designed two-body hinged raft wave energy con-
verter: From experimental study to annual power prediction for the EMEC site using WEC-Sim. Ocean Engineering,
267:113286, (2023).
9. Han, Z., Jin, S., Greaves, D., Hann, M., Shi, H.: Study on the energy capture spectrum of a two-body hinged-raft wave
energy converter. Energy, 132057, (2024).
10. Babarit, A., Delhommeau, G.: Theoretical and numerical aspects of the open source BEM solver NEMOH. 11th European
Wave and Tidal Energy Conference, Nantes, France (2015).
On
ly
fo
rR
Development of a 6-DOF Actuating System for Implementing Real-Time
ea
Hybrid Simulation of Floating Bridge Systems under Wind and Wave
Loads
din
g
at
Giheon You1[0009-0001-0844-1228], Minyeop Kim1[0009-0006-5750-7453] and Yunbyeong Chae1[0000-0002-6460-229X]
1
W
Seoul National University, Gwanak-gu, Seoul 08826, Republic of Korea
CF
ybchae@snu.ac.kr
S
20
Abstract. This study developed both analytical and experimental substructures to perform real-time hybrid
24
simulation (RTHS) for floating bridge systems subjected to wind and wave loads. The upper structure influ-
enced by wind loads is physically tested using a scaled model in a wind tunnel, while the lower structure
subjected to wave loads is modeled analytically. The wind loads are measured with a 6-axis load cell, and the
corresponding target displacements are calculated assuming rigid body motion. These real-time responses are
computed in MATLAB Simulink and applied to the experimental substructure mounted on a 6-degree-of-
freedom (DOF) actuating system. The 6-DOF actuating system is controlled using the Speedgoat real-time
target machine and the adaptive time series (ATS) compensator. This study proposes a method for experimen-
tally analyzing the dynamic response of the pontoon-bridge pylon system using RTHS.
Keywords: Real-Time Hybrid Simulation, Floating Bridge System, 6-DOF Stewart Platform.
1 Introduction
An innovative design for a floating bridge, such as a cable bridge with pylons supported by a TLP (Tension Leg
Platform) system, has been proposed in Norway’s transportation plans to address the challenge of crossing deep
and wide fjords [1]. These floating bridge systems should consider many conditions that are not accounted for in
conventional bridges or existing types of floating bridges. Accurate performance evaluation is crucial for the
design, construction, and operation of these infrastructures. Consequently, new analytical methods need to be
developed to address these challenges. In particular, experimental verification of wind-wave coupled response is
necessary for floating bridge systems, which are influenced by both wind and wave loads.
Generally, in model tests of floating bodies, the upper and lower structures follow different kinematic and
dynamic similarities, making it difficult to accurately simulate both aerodynamic and hydrodynamic forces.
Ransley (2022) conducted real-time hybrid model tests in a wave tank to analyze the effects of wind and wave
loads on the response of a floating offshore wind turbine [2]. The hydrodynamic loads were applied to a model-
scale floating platform, while the aerodynamic effects on the upper structure were analytically modeled and im-
plemented through a fan system.
In this study, real-time hybrid simulation (RTHS) is suggested using wind tunnel, where the structure above
the pontoon, influenced by wind loads, is physically tested in the wind tunnel, while the lower structure, subjected
to wave loads and other hydrodynamic forces, is analytically modeled.
The motion of a floating body is described with 6-degree-of-freedom (DOF) rigid body motions, assuming that
the floating body moves like a rigid structure. For RTHS, a 6-DOF actuating system is required to apply real-time
displacements to the scaled model in the wind tunnel. The target displacements of the analytical structure are
applied to the scaled model through the real-time displacement control, and the force at the base on the scaled
model is measured using a 6-axis load cell and fed into the equations of motion of the entire system, which are
solved numerically to obtain the command displacements at the next time step.
2 Development and Verification of Analytical Substructure
2.1 Modeling of the Analytical Substructures

The submerged structure of the floating body was assumed to have six degrees of freedom in motion as a rigid
body. Each degree of freedom is referenced to a coordinate system with the origin at the center of gravity of the
body, the 𝑥-axis in the direction of the incoming waves, and the 𝑧-axis directed opposite to the direction of gravity.
Translational motions along the 𝑥, 𝑦, and 𝑧 axes are referred to as surge, sway, and heave, respectively, while
On
ly
fo
2
rR
rotational movements about each axis are referred to as roll, pitch, and yaw. In this study, it is assumed that the
ea
motion of the floating body is very small, and the slope of incident waves, represented by the product of wave
din
number (𝑘) and amplitude (𝐴), is also very small, allowing the application of linearized theory. Based on this
assumption, the motion of a floating body with six degrees of freedom can be described by the frequency-domain
g
equation of motions given in equation (1) [3].
∑ 𝜔 𝑀 𝐴 𝜔 𝑖𝜔𝐵 𝜔 𝐶
at
𝜂̅ 𝜔 𝐹 𝜔 𝐹 𝜔 (1)
W
CF
Where, 𝜔 represents the frequency, 𝑀 the inertia, 𝐴 the added mass, 𝐵 the radiation damping, 𝐶 the hy-
drostatic stiffness, 𝐹 the Froude-Krylov force, and 𝐹 the diffraction force. Each of these coefficients can be
S
obtained by solving the potential problem with the ocean environment and the floating body as boundary condi-
20
tions. In this study, the coefficients were obtained through analysis using Ansys AQWA. To apply equation (1),
which is in the frequency domain, to RTHS, it needs to be transformed into the time domain equation. Cummins
24
proposed the Cummins Equation for the time domain motion of floating bodies, as given in equation (2), which
includes the convolution integral, by separating added mass and radiation damping using the values at infinite
frequency, and then applying inverse Fourier transform as follows [4];
𝐌 𝐀 𝐱 𝑡 𝐑 𝑡 𝜏 𝐱 𝜏 𝑑𝜏 𝐂𝐱 𝑡 𝐅 𝑡 (2)
where, 𝑡 represents the time, 𝐌 is the mass(inertia) matrix, 𝐀 is the added mass matrix at infinite frequency, 𝐑
is the retardation function, 𝐂 is the hydrostatic stiffness matrix, 𝐱 is the displacement vector of the body, and 𝐅 is
the external force. Each of these coefficients can be mathematically derived from the frequency domain coeffi-
cients.
The wave excitation force was modeled using the transfer function of amplitude (𝑃 𝜔 ) and phase lag (𝜑 𝜔 )
of sum of the Froude-Krylov force and diffraction force. For example, the wave force for the 𝑗-th degree of free-
dom due to a regular incident wave 𝜁 𝑡 𝐴 cos 𝜔𝑡 𝜀 is given by equation (3) [5].
𝐹 𝑡 𝐹 𝑡 𝑃 𝜔 𝐴 cos 𝜔𝑡 𝜀 𝜑 𝜔 (3)
In the case of irregular waves, the JONSWAP spectrum is used to represent the waves as a sum of components
with various frequencies. The wave loads generated by irregular waves are calculated as the sum of the wave load
component at each frequency.
The wave loads must also consider second-order wave loads, which arise from the superposition of two fre-
quencies. These loads appear at a frequency equal to the sum or difference of the two frequencies. The amplitude
of the second-order wave load is given by the product of the amplitudes of each frequency component and the
Quadratic Transfer Function (QTF). The second-order wave load generated by two waves, 𝜁 𝑡
𝐴 cos 𝜔 𝑡 𝜀 , 𝜁 𝑡 𝐴 cos 𝜔 𝑡 𝜀 , is given by equation (4);
𝐹 𝑡 𝑃 𝜔 , 𝜔 𝐴 𝐴 cos 𝜔 𝜔 𝑡 𝜀 𝜀 𝜑 𝜔 ,𝜔 (4)
where, 𝑃 is amplitude QTF matrix and 𝜑 is phase QTF matrix.
The damping of the floating body is predominantly influenced by the effects of Morison drag. The Morison
drag acting on a slender member is exerted in a direction perpendicular to the longitudinal axis of the member.
For an object moving in water, the Morison drag is expressed in terms of velocity 𝐮 and the velocity of the object
𝐯 as shown in equation (5);
𝑑𝐅 𝐶 𝜌𝑑𝐴 𝐮 𝐯 |𝐮 𝐯| (5)
where, 𝑑𝐅 denotes the drag force acting on a small cross-sectional area 𝑑𝐴 of the slender member, 𝐶 denotes
the drag coefficient, and 𝜌 represents the fluid density. The velocities of the fluid and the body, 𝐮 and 𝐯, include
only the components perpendicular to the length of the member. For all Morison members constituting the floating
body, the forces were calculated using the Morison equation, and these forces were then converted to forces acting
at the center of gravity of the floating body to account for Morison drag.
On
ly
fo
2.2 rR
Verification of the Analytical Substructures
3
ea
The analytical substructure modeling was verified using the Offshore Code Comparison Collaboration Continua-
din
tion (OC4) [4]. The geometry and mass properties of OC4 are as shown in Table 1, where SWL stands for Sea
g
Water Level and CM stands for Center of Mass.
at
Table 1. Geometry and mass properties of OC4
Geometry
WMass properties
Platform mass
CF 1.3473E+7 kg
S
CM location below SWL 13.46 m
20
Roll inertia about CM 6.827E+9 kg-m2
Pitch inertia about CM 6.827E+9 kg-m2
Yaw inertia about CM
24
1.226E+10 kg-m2
Using Ansys AQWA, the diffraction analysis was performed to obtain the frequency dependent coefficients of
the structure, including radiation damping, added mass, and wave excitation. The time domain coefficients were
obtained by transforming the frequency dependent coefficients, which are shown in Fig.1. Based on the obtained
time domain coefficients, the Cummins equation was formulated and solved using the Newmark method.
Fig. 1. Time domain coefficients of OC4
The response amplitude operator (RAO) was determined from the structure’s steady state response to regular
waves and compared with the frequency domain analysis results. Irregular wave responses were verified by com-
paring the response spectrum, which is the product of the wave spectrum and the square of the RAO. The response
spectrum was obtained through Fast Fourier Transform (FFT) and compared with Ansys AQWA analysis results,
averaged over 30 repetitions. Verification results for the heave responses to regular and irregular waves are shown
in Fig. 2.
Fig. 2. Verification results for regular wave response (left) and irregular wave response (right)
Since the damping of the floating body’s motion is predominantly governed by Morison drag, the modeled
Morison element was verified through free decay tests of the floating body. To validate the effectiveness of the
modeling, the same parameters were input into the commercial software OrcaFlex, and the results were compared
On
ly
fo
4
rR
with those from the analytical substructure. The time-displacement results of the heave and pitch free decay are
shown in Fig. 3.
ea
din
g
at
W
CF
S
20
24
Fig. 3. The time-displacement results of the heave and pitch free decay test
As shown in Fig. 3, the result of the free decay test with an initial displacement of 1m in heave and 0.05 rad in
pitch closely matched the OrcaFlex results, indicating that the Morison members have been accurately modeled.
3 Experimental Substructures
3.1 Experimental Setup

The experimental test setup consists of a 6-DOF actuating system, a 6-axis load cell, a target PC, a test model,
and a wind tunnel. Wind loads are generated in the wind tunnel and measured at the base of the model using the
6-axis load cell. Measured loads are input into the target PC and applied as external loads to the analytical struc-
ture. All calculations are performed within the target PC, including the numerical integration and the real-time
displacement control of the 6-DOF actuating system. This process is illustrated in Fig. 4.
Fig. 4. Configuration of the experimental setup
3.2 6-DOF Actuating System

The 6-DOF actuating system, consisting of an upper and base plate connected by six linear motors, adjusts the
upper plate’s displacement through motor length adjustments, allowing the 6-DOF movement. Inverse kinematics
calculates motor lengths from 6-DOF displacements, and forward kinematics computes the 6-DOF displacements
from motor lengths. Inverse kinematics uses the reference configuration coordinates to calculate the upper joint
coordinates with target displacements, determining the motor lengths from the change of distance between the
upper and lower joints. In the reference configuration, the coordinates of the upper and lower joints of the 𝑖-th
motor are 𝐔 and 𝐁 , respectively, and when the 6-DOF displacement of 𝐮 〈𝑢 , 𝑢 , 𝑢 , 𝜃 , 𝜃 , 𝜃 〉 is applied,
the coordinates of the upper joint 𝐔 and the length of the 𝑖-th motor 𝐿 are determined as shown in equation (6);
𝐔 𝐑 𝐔 𝐇 𝐓 𝐇, 𝐿 ‖𝐔 𝐁‖ (6)
On
ly
fo
rR
where, 𝐑 is the 3-dimensional rotation matrix due to the rotation of upper plate, 𝐑
5
𝐑 𝐑 𝐑 , and 𝐓 is the trans-
ea
lational displacement vector 𝐓 〈𝑢 , 𝑢 , 𝑢 〉 .
din
Determining the 6-DOF displacement from the measured motor lengths requires an iterative calculation pro-
g
cess. The gradient descent was utilized to apply forward kinematics. In the 𝑛-th iteration, the sum of squared error
at
function 𝐹 𝐮 is calculated between the estimated length of the 𝑗-th motor 𝑙 for the assumed platform displace-
ment 𝑢 and the measured length 𝐿 . At this point, to proceed with a new assumption in the direction that mini-
W
mize 𝐹 𝐮 , the gradient of 𝐹 𝐮 , ∇𝐹 , was used to calculate the new estimate of 𝐮 . When 𝐹 𝐮 converges
CF
to a sufficiently small value, 𝐮 at that point was determined as the 6-DOF displacements. This process is described
in equation (7);
S
20
𝐹 𝐮 ∑ 𝐿 𝑙 , ∇𝐹 , 𝑢 𝑢 𝑘 𝐹 ∇𝐹 (7)
24
where, 𝑘 is the learning rate, which adjusts the magnitude of the gradient to control the speed and stability of the
iterative calculation.
Minimizing time delay is crucial for high-accuracy RTHS. The adaptive time series (ATS) compensator was
applied as the time delay compensation algorithm [7]. As shown in Fig. 5, the ATS compensator was applied to
each motor.
Fig. 5. ATS application process for the 6-DOF actuating system
Fig. 6 compares the 𝑧-direction displacement over time with and without applying the ATS compensator for
the target displacement when the platform is subjected to a sinusoidal input with an amplitude of 10mm and a
frequency of 1Hz. As a result of applying the ATS compensator, both the time delay and amplitude errors were
significantly reduced compared to the case without the ATS compensator. Before applying the ATS compensator,
the time delay was approximately 152ms with an amplitude error of around 3.8mm, but after applying the ATS
compensator, the time delay reduced to under 2ms and the amplitude error to approximately 0.1mm.
Fig. 6. Comparison of the displacement with the target displacement
4 Conclusions and Further Study
This study developed and verified analytical and experimental substructures for RTHS on a floating structure with
upper and lower structures. The analytical modeling of the lower structure, subjected to wave loads, used the time-
domain Cummins equation. Coefficients were obtained by transforming frequency-domain coefficients from An-
sys AQWA. The Morison drag, calculated using the Morison equation, was considered by defining Morison ele-
ments, which consist of the start and end points, cross-sectional area, and drag coefficients. The experimental test
setup for simulating the 6-DOF response of the floating system includes the target PC for control, wind tunnel, 6-
axis load cell, and 6-DOF actuating system. Inverse kinematics controlled the system, and the ATS compensator
minimized time delay and amplitude errors. Future research will consider damping from the Morison effect and
secondary wave load effects, as well as the model catenary mooring cables for nonlinear mooring effects.
On
ly
fo
6
rR
Validation of the experimental substructure will include physical measurements using devices like laser displace-
ment sensors.
ea
din
g
References
at
1. Samferdselsdepartement, D. K.: Meld. St. 33 Nasjonal Transportplan 2018–2029. departementenes sikkerhets- og ser-
W
viceorganisasjon, Oslo, Norway (2017).
2. Ransley, E. J., et al.: Real-time hybrid testing of a floating offshore wind turbine using a surrogate-based aerodynamic
emulator. AMSE Open Journal of Engineering 2. (2022)
CF
3. Lewis, Edward V.: Principles of naval architecture. Vol. 3, 2nd revision. Society of Naval Architects and Marine Engi-
neers. Jersey City, NJ (1989).
S
20
4. Cummins, W. E.: The impulse response function and ship motions. Schiffstechnik 9, 101–109 (1962).
5. Ansys Inc.: Aqwa theory manual. Canonsburg, PA. (2022).
24
6. Robertson, A., Jonkman, J., Masciola, M., Song, H., Coupee, A., Coulling, A., Luan, C.: Definition of the semisubmers-
ible floating system for phase II of OC4 (No. NREL/TP-5000-60601). National Renewable Energy Lab. (NREL), golden,
CO (United States) (2014).
7. Chae, Y., Kazemibidokhti, K., Ricles, J. M.: Adaptive time series compensator for delay compensation of servo-hydraulic
actuator systems for real-time hybrid simulation. Earthquake Engineering & Structural Dynamics, 42(11), 1697-1715
(2013).
On
ly
fo
rR
Creation of Wave-Free Space by Annular Elastic Floats
ea
din
S. Akasaka, T. Iida*[0000-0002-3566-0223] and K. Irifune
g
Department of Naval Architecture and Ocean Engineering, Osaka University, Osaka, Japan
at
*
iida@naoe.eng.osaka-u.ac.jp
W
CF
Abstract. Various facilities such as offshore wind farms have been constructed at sea in recent years. In gen-
S
eral, offshore facilities cannot avoid damage caused by waves. This may result in accidents, such as vessels
drifting and capsizing as well as collision with marine structures. If wave-free space can be created on the free
20
surface of the sea, we can reduce such wave accidents. Additionally, it leads to further development in offshore
24
utilization such as floating cities. In this study, we aim to create a wave-free space within an annular floating
structure consisting of multi-layered elastic plates. We establish the numerical simulation method to solve the
wave field around this plate based on the eigenfunction matching method. In addition, we derive the inside
wave energy in the annular plate. To achieve wave-free space within the plate, the evolutionary optimization
method, Covariance Matrix Adaptation Evolution Strategy (CMA-ES), is employed for seeking the optimum
plate design. We designate two parameters associated with each layer of the thin plate as design variables and
optimize them to minimize the wave energy inside the circular structure. Optimization is carried out at the
specific wave number. We compare the wave displacement around the circular float structure in two cases: (1)
when only the wave energy inside the structure is minimized and (2) when both the inside wave energy and
the outside scattered wave energy are minimized. Since the plate is the annular ring, the wave-free performance
is omnidirectional.
Keywords: Wave Free Space, Annular Elastic Floats, CMA-ES.
1 Introduction
In recent years, various facilities such as offshore wind farms have been constructed at sea and floating cities are
conceptualized. There has been increasing interest in marine utilization. However, offshore facilities cannot avoid
damage caused by waves. This may result in accidents, such as vessel drifting, capsizing, and collisions with
marine structures. Furthermore, one of the reasons why floating cities haven’t been realized is the effect of the
waves. Then if wave-free space can be created on the free surface of the sea, we can reduce such wave accidents.
Additionally, it leads to further development in offshore utilization such as floating cities.
Newman [1] presented the phenomenon called cloaking, wherein the scattering of incident waves by the floats
doesn’t propagate outside the float array, thereby reducing drift forces using multiple floats. Additionally, Iida et
al.[2][3] achieved cloaking at specific wavenumbers through numerical simulations using a higher-order boundary
element method that considers mutual interference and model experiments. However, the cloaking phenomenon
depends on the angle of incident waves with multiple floats, Thus, Iida et al. [4] proposed a model in which the
cloaking phenomenon is independent of the direction of incident waves by using a multi-layered thin plate struc-
ture, thereby reducing the wave drifting force at a specific wave number.
In this study, we aim to create a wave-free space using a multi-layered structure of thin plates arranged annular
configuration expanding upon previous research. To achieve this, it is necessary to minimize the wave energy
inside the annular elastic floating body. Therefore, in this study, we utilize the CMA-ES optimization algorithm
[5]. We consider the two parameters associated with each layer of thin plates as design variables and aim to
minimize the wave energy inside the floating body.
2 Formulation of the problem
2.1 Floats placement
This research aims to create a wave-free space inside an annular elastic float on the sea. A coordinate system is
set up as shown in Fig.1. The thin plate is made up of 𝑁 concentric layers, and the parameter of the i-th layer
counting from the outside is denoted by a superscript (i). The inner radius from the central axis is 𝑎, and the outer
annular radius of the outermost layer is 𝑏. The thickness 𝑡 of the thin plate is assumed to be sufficiently small
compared to the wavelength 𝜆. The water depth is assumed to be constant at z=−ℎ.
On
ly
fo
rR
ea
2
din
g
at
W
CF
S
20
24
Fig. 1. Coordinate the system of the problem. The plate is assumed elastic and thin. The composite plate consists of 𝑁 hori-
zontally concentric annular rings, and the inner radius from the central axis is 𝑎 and the outermost radius of the plate is 𝑏.
2.2 Governing equation and boundary conditions equations

This study assumes a linear potential theory. The governing equation and boundary condition equations are pre-
sented. These equations are considered on undisturbed free surfaces of the water with the positive 𝑧-axis pointing
upward and 𝑔 is the acceleration of gravity.
∇ 𝜙=0 −ℎ ≤𝑧 ≤ 0 (2. 1)
𝜕𝜙
=0 𝑧 = −ℎ (2. 2)
𝜕𝑧
𝜕 𝜙 𝜕𝜙
−𝑔 =0 𝑧=0 (2.3)
𝜕𝑡 𝜕𝑧
𝜕 𝜙 𝜕𝜙 1 𝜕 𝜕 𝜙
−𝑔 = 𝑚 + 𝐷∇ 𝜙 𝑧=0 (2.4)
𝜕𝑡 𝜕𝑧 𝜌 𝜕𝑧 𝜕𝑡
In these equations (2.1) is the mass conservation, (2.2) is the kinematic boundary on the bottom, (2.3) is a line-
arized free surface condition, and (2.4) is the linearized equation for the thin plate. (2.3) is obtained when 𝑝 = 0
for the pressure distribution in the liner-free surface equation, and (2.4) is obtained when the distributed load in
elastic waves is applied. Where 𝐷 = 𝐸𝑡 ⁄12(1 − 𝜈 ) is the stiffness modulus, 𝐸 is Young’s modulus, 𝜈 is Pois-
son’s ratio, and 𝜌 is the density of the thin plate.
Next, Nondimensionalization [6] is performed as follows, using the length a of the inner annular radius of an
annular in (2.5).
𝑥 𝑔 𝜙 ℎ
𝑥̅ = , 𝑡̅ = 𝑡 , 𝜙= , 𝐻= (2.5)
𝑎 𝑎 𝑎 𝑎𝑔 𝑎
For simplicity, the bar representing the dimensionless value will be omitted hereafter. Thus, the governing
equation and the continuity equation can be rewritten as
𝛻 𝜙=0 −𝐻 ≤𝑧 ≤0 (2.6)
On
ly
fo
rR
ea
din 3
𝜕𝜙
g
at
=0 𝑧 = −𝐻 (2.7)
𝜕𝑧
𝜕𝜙
W
CF
= 𝐾𝜙 𝑧 = 0 (𝑟 ≤ 1, 𝑏 ≤ 𝑟) (2.8)
𝜕𝑧
S
𝜕𝜙
𝛽 ( ) 𝛻 + 1 − 𝐾𝛾 ( ) = 𝐾𝜙 𝑧 = 0 (1 ≤ 𝑟 ≤ 𝑏) (2.9)
20
𝜕𝑧
() ()
24
where 𝛾 ( ) = 𝜌 𝑡 (𝜌 𝑎) and 𝛽 ( ) = 𝐷 ( ) ⁄(𝜌 𝑔𝑎 ) are dimensionless flexural rigidity and mass of 𝑖-th
ring, respectively.
2.3 Dispersion relation and velocity potential
Drive solutions that satisfy the governing equation and boundary condition equations. First, the dispersion relation
between the water waves on the free surface inside and outside the annular floating and the elastic wave in the
floating body is written as
𝐾 𝜅 tanh 𝜅 𝐻
= (2.10)
𝛽 ( ) 𝜅 − 𝐾𝛾 ( ) + 1 −𝜅 tan 𝜅 𝐻
K is the dimensionless frequency, and 𝜅 is the wave number without distinction between water and elastic
()
waves. In the following, 𝑘 is the wavenumber of the water waves and 𝜇 is the wavenumber of the elastic
waves. 𝑘 has an infinite number of positive and real solutions, where 𝑘 indicates progressive waves, and
𝑘 𝑘 ⋯ correspond to local waves. Similar, 𝜇 has an infinite number of positive and real solutions, where
() () () ()
𝜇 indicates progressive waves, 𝜇 , 𝜇 ⋯ indicate local waves, and additionally two complex solutions 𝜇 and
()
𝜇 represented damped waves. The dispersion relation in the case of a water wave is when 𝛽=0 and 𝛾=0.
The velocity potential that satisfies these equations is written as
𝐴 √
𝜙(𝑟, 𝜃, 𝑧, 𝑡) = Re 𝛹 (𝑟, 𝑧)𝑒 𝑒 (2.11)
𝑖√𝐾
And
⎧ ( )
𝑖 𝐽 (𝑘 𝑟)𝑓 (𝑧) + 𝑎 𝐻 (𝑘 𝑟)𝑓 (𝑧) + 𝑎 𝐾 (𝑘 𝑟)𝑓 (𝑧)
⎪
⎪
⎪ 𝑏 ( ) 𝐽 𝜇 ( ) 𝑟 𝐹 ( ) (𝑧)+𝑐 ( ) 𝐻 ( ) 𝜇 ( ) 𝑟 𝐹 ( ) (𝑧)
⎪
𝛹 = () () () () () () (2.12)
⎨+ 𝑏 𝐼 𝜇 𝑟 𝐹 (𝑧) + 𝑐 𝐾 𝜇 𝑟 𝐹 (𝑧)
⎪ , ,
⎪
⎪
⎪ 𝑑 𝐽 (𝑘 𝑟)𝑓 (𝑧) + 𝑑 𝐼 (𝑘 𝑟)𝑓 (𝑧)
⎩
( ) ( )
where 𝐽 (𝑘 𝑟) is a Bessel function of the first kind, 𝐻 (𝑘 𝑟) and 𝐻 (𝜇 𝑟) are Hankel functions of the first
() () ()
kind, and 𝐼 (𝜇 𝑟) and 𝐾 (𝜇 𝑟) are deformed Bessel functions of the first and second kinds. 𝑓 (𝑧), 𝐹 are
written as
On
ly
fo
rR
ea
4
din
⎧ cosh 𝑘 (𝑧 + ℎ) g ⎧ cosh 𝜇 (𝑧 + ℎ)
at
⎪ cosh 𝑘 ℎ 𝑛=0 ⎪ cosh 𝜇 ℎ 𝑛=0
()
W
𝑓 (𝑧) = , 𝐹 (𝑧) = (2.13)
⎨cosh 𝑘 (𝑧 + ℎ) ⎨cosh 𝜇 (𝑧 + ℎ)
𝑛≥1 𝑛 = −1, −2, 𝑛 ≥ 1
CF
⎪ cosh 𝑘 ℎ ⎪ cosh 𝜇 ℎ
⎩ ⎩
S
Velocity potentials are described in three different ways: water waves outside the float, elastic waves on the
() ()
20
float, and water waves inside the float. The unknown coefficients 𝑎 , 𝑏 , 𝑐 and 𝑑 in the velocity potential
24
are solved by the eigenfunction matching method using the boundary conditions equation around the float [4].
2.4 The energy around the floats
The energy around the float to create a wave-free space is derived. First, the wave energy inside the annular float
is derived. The average energy inside the annulus is the sum of the kinetic energy and the potential energy. Con-
sidering the time average and applying the velocity potential of the water waves inside the float, the average
energy inside the annulus is written as
𝐸= |𝑑 |
1 (𝑚 − 1)
⎡ 𝐻 (𝐽 (𝑘 )) + 1 − 𝐽 (𝑘 ) ⎤
⎢ 16 𝑘 ⎥
⎢ 1 (𝑚 + 1) ⎥
× ⎢+ 𝐻 (𝐽 (𝑘 )) + 1 − 𝐽 (𝑘 ) ⎥ (2.14)
16 𝑘
⎢ ⎥
⎢ + 1 (4 − 𝐻) (𝐽 (𝑘 )) + 1 − 𝑚 𝐽 (𝑘 ) ⎥
⎣ 8 𝑘 ⎦
where 𝐻 = (1 + 2𝑘 ℎ⁄sinh 2𝑘 ℎ). By bringing this close to zero, it is possible to create a wave-free space.
Next, consider the energy of the scattered waves [2] generated by the disturbance of the incident waves by the
float. The energy of the scattered wave is derived from the principle of conservation of energy. 𝑊 that work done
by the fluid on the object from the time-averaged rate of change of the total energy in a given fluid domain is
obtained. Applying the velocity potential of the water waves outside the float to it, the energy is written
1
𝑊 =− {Re[𝑖 𝑎 ∗ ]|𝑎 | } (2.15)
𝐶 √𝐾
where 𝐶 = 𝑘 ⁄𝐾 + (𝑘 − 𝐾 ) ℎ. Note that because 𝑎 is the amplitude of scattered waves, the second term in
(2.15) indicates the energy transferred to scattered waves. As a result, the energy of scattered waves 𝑊 is given
as
𝑊 = |𝑎 | (2.16)
By bringing this close to zero, cloaking can be achieved.
3 Evolutionary optimization of the plate
To create a wave-free space, CMA-ES [5] was used for optimization. The optimization was performed in two
cases: the objective function is (1) 𝐹 = 𝐸 , (2) 𝐹 = 𝐸 + 0.05𝑊 and the results are compared. The energy of scat-
ted waves was corrected by a factor of 0.05 in this study due to the larger scale of the scattered wave energy
On
ly
fo
rR
ea
din 5
g
compared to the wave energy inside the annular. In addition, in performing the optimization, the variables are 𝛾
at
and 𝛽, which indicated the mass and stiffness of the float. These two variables are subject to the following as
0.01 ≤ 𝛾 ≤ 0.5 and 0.01 ≤ 𝛾 ≤ 0.5. The values of 𝛾 and 𝛽 for each layer are optimized to minimize the wave
W
energy when the frequency of the incident wave 𝑘 = 1.0, which is dimensionless as the inner annular radius of
CF
the float 𝑎 = 1, outer annular radius 𝑏 = 5, number of layers of the thin plate 𝑁 = 8, and water depth ℎ = 𝜆.

S
20
24
First, the optimized wave energy in case (1) is 𝐸 = 2.98 × 10 , while the optimized wave energy and scattered
wave energy in case (2) are 𝐸 = 1.31 × 10 and 𝑊 = 4.28. Fig.2 shows the wave displacement of the float and
the areas around it. In case (1), a wave-free space is observed inside the float, but the waves around it are disturbed
by the scattered waves. On the other hand, in case (2), waves are slightly generated inside the float, but no scattered
waves are generated and a phenomenon quite close to cloaking can be observed. Fig.3 shows the variation of
variables 𝛽 and 𝛾 in the direction 𝑟. As in Iida et al. [4], the 𝛽 varies from layer to layer, but 𝛾 varies only slightly
from layer to layer in two cases. This suggests that the stiffness coefficient 𝛽 has more control over the optimiza-
tion than the mass coefficient 𝛾. Finally, Fig.4 shows the variation of the wave energy inside the float and the
scattered wave energy versus the wavenumber of the incident wave. The wave energy and scattered wave energy
change as close to zero as possible when the wavenumber of the incident wave 𝑘 = 1.0. However, the addition
of the scattered energy reduces the accuracy of the optimization.
(a) (b)
Fig. 2. Wave displacement around the floating body. (a) show the wave displacement in case (1) when optimized only wave
energy inside float. (b) show in case (2) when optimized wave energy inside float and scattered energy
(a) (b)
Fig. 3. Spatial distributions of mass γ and flexural rigidities β for the optimized composite plate. (a) show in case (1) when
optimized only wave energy inside float. (b) show in case (2) when optimized wave energy inside float and scattered energy
On
ly
fo
rR
ea
6
din
g
at
W
CF
S
20
24
(a) (b)
Fig. 4. Variation of energy versus wave number of the incident wave. (a) show in case (1) when optimized only wave energy
inside float. (b) show in case (2) when optimized wave energy inside float and scattered energy. Left vertical axis: inner en-
ergy, right vertical axis: scattered energy.
5 Conclusion
In this study, we created a wave-free space using the annular elastic float. We derive the wave energy inside the
float and scattered energy and optimize them to approach zero. In the optimization process, the mass coefficient
𝛾 and stiffness coefficient 𝛽 of the float were treated as variables, specifying a certain incident wave number. The
evolutionary optimization technique, specifically CMA-ES, was employed for this study. The optimization results
were compared between scenarios where only the wave energy inside the float was optimized and where both
wave energy inside the float and scattered energy were optimized. In comparison to optimizing only the wave
energy inside the float, optimizing both inside wave energy and scattered energy resulted in lower optimization
precision. In optimization, there are still problems that need to be improved. Additionally, concerning variables 𝛾
and 𝛽, while the values of 𝛽 varied significantly across layers, 𝛾 exhibited minimal changes with little discernible
variation. It is presumed that 𝛽 has a significant influence on the optimization process of this study.
References
1. Newman, J. N. (2014). Cloaking a circular cylinder in water waves. European journal of mechanics-B/fluids, 47, 145-
150.
2. Iida, T., Kashiwagi, M., & He, G. (2014). Numerical confirmation of cloaking phenomenon on an array of floating bodies
and reduction of wave drift force. International Journal of Offshore and Polar Engineering, 24(04), 241-246.
3. Iida, T., Kashiwagi, M., & Miki, M. (2016). Wave pattern in cloaking phenomenon around a body surrounded by multiple
vertical circular cylinders. International Journal of Offshore and Polar Engineering, 26(01), 13-19.
4. Iida, T., Zareei, A., & Alam, M. R. (2023). Water wave cloaking using a floating composite plate. Journal of Fluid Me-
chanics, 954, A4.
5. Hansen, N. (2016). The CMA evolution strategy: A tutorial. arXiv preprint arXiv:1604.00772.
6. Meylan, M. H. (2002). Wave response of an ice floe of arbitrary geometry. Journal of Geophysical Research:
Oceans, 107(C1), 5-1.
On
ly
fo
rR
Comparison between the practical floating PV generation and simulation
ea
models
din
g
Haojie Luo1[0000-0003-1783-7213], Suijie Liu1[0009-0005-6622-6250], Zhenyu Dou1[0009-0009-7126-9828], Qi Wu1[0000-0002-8550-5387],
at
Sunliang Cao1,2,3,*[0000-0001-9589-8914], Hongxing Yang1,3 [0000-0001-5117-5394]
W
1
Renewable Energy Research Group (RERG), Department of Building Environment and Energy Engineering, Faculty of
CF
Construction and Environment, The Hong Kong Polytechnic University, Kowloon, Hong Kong SAR
2 Research Institute for Smart Energy (RISE), The Hong Kong Polytechnic University, Kowloon, Hong Kong SAR
3 Research Institute for Sustainable Urban Development (RISUD), The Hong Kong Polytechnic University, Kowloon, Hong
Kong SAR
S
20
sunliang.cao@polyu.edu.hk
24
Abstract. With the limited availability of land resources in urban areas and the increasing energy demands,
there is a growing interest in exploring alternative renewable energy solutions in coastal regions. This study
focused on the potential of utilising nearshore space by designing a micro-floating PV panel prototype and
evaluating its performance under actual ocean conditions in Hong Kong. The primary objectives were to mon-
itor photovoltaic generation and investigate the characteristics of the generation data while also assessing the
environmental impact on the performance. For deck-on PVs, PV temperature significantly affects power gen-
eration. On clear days, peak PV temperature at noon can reduce generation by 5.8% despite a 39.9% increase
in solar radiation. Ambient temperature and wind speed impact PV temperature, with a 6.8 °C increase reduc-
ing generation by 35% even with a 31.3% increase in radiation. The enclosed floater temperature was heated
heavily due to the strong greenhouse effect under intense solar radiation. The open-circuit voltage showed a
general correlation with PV generation. The simulation-experiment comparison identifies key factors impact-
ing PV performance. Solar radiation and shade effects bring experimental uncertainty. Surroundings influence
PV temperature, with barge radiative heating and floater conductive heating causing differences between ex-
periments and simulations. Barge radiative heating is more significant, resulting in a 16.5% PV generation
difference during the afternoon on a clear sky for deck-on PVs.
Keywords: Floating solar system; Nearshore space; Sustainable energy solution.
1 Introduction
1.1 Background
Floating photovoltaic (PV) generation is gaining attention as an alternative to land-based solar installations,
especially in land-constrained regions. Deploying floating PV systems on water bodies like reservoirs and lakes
offers space efficiency, increased energy generation, and reduced environmental impact. Hong Kong's extensive
coastline and islands provide a unique opportunity for floating PV systems [1]. These systems can overcome land
constraints, utilizing seawater for solar energy generation. Positioned on the sea surface, they maximize sunlight
exposure, achieving higher efficiency than land installations. In Hong Kong's push to reduce its carbon footprint
and embrace cleaner energy sources, integrating floating PV systems into the energy mix is crucial. These systems
play a vital role in meeting renewable energy goals, cutting greenhouse gas emissions, and enhancing energy
security by diversifying electricity sources [2] [3]. Compared to land-based setups, floating PV systems offer
environmental benefits by reducing land conversion and preserving ecosystems. They also help maintain the cit-
yscape while bolstering the local renewable energy mix. However, successful implementation demands meticu-
lous planning, design, and assessment. Recent global floating PV projects highlight the technology's potential,
with the global market expected to exceed 6 GW by 2031. By then, over 15 countries are projected to surpass 500
MW of cumulative installations, with Indonesia, India, and China leading the trend [4]. Real-world projects offer
valuable insights into performance, efficiency, and challenges, while simulation models aid in assessing feasibility
and predicting system performance under various conditions. [5] [6].
1.2 Objective
This study aims to compare practical floating PV generation with simulation models to assess their effectiveness
and identify any discrepancies. By analysing real-world experiments and simulation results, the research aims to
understand floating PV's performance and simulation model accuracy. The study will use data from existing
On
ly
2
fo
rR
floating PV installations, focusing on deck-on and micro floating PV systems in seawater. It will evaluate energy
generation and efficiency under diverse weather conditions to identify key factors influencing system effective-
ea
ness and assess real-world potential. Furthermore, the study will review and analyse simulation models used to
din
predict the performance of floating PV systems. By comparing simulation outcomes with practical data, it aims
to evaluate model accuracy and reliability, identifying any discrepancies for improvement. The research will ex-
g
plore factors influencing performance differences, such as environmental conditions, system design, and energy
at
losses. Understanding these elements will refine models for more accurate future predictions.
2 Outline of the project W

CF
S
2.1 Deck-on photovoltaic system
20
As shown in 错误!未找到引用源。, the deck-on PV system is set on a boat staying at the To Kwa Wan Typhoon
24
Shelter. The system consists of 5 PV panels with 450W capacity each, all these panels are connected to a maxi-
mum power point tracking (MPPT) and a charge controller. A set of battery packs is connected to the PV and an
inverter through the controller to utilize the energy for supporting the external load. A remote data uploader is
installed to collect and monitor the system data including voltage, current and power of PV panels and inverter,
together with the back surface temperature of PV panels. The time-resolution of the online database is two
minutes.
To record the actual maximum power generation of the system. Three LED lights of 0.06 kW are installed to
consume extra energy from the battery packs. The impacts of the back surface temperature of PV panels influ-
enced by ambient temperature and wind speed on energy generation under different weather conditions (clear
sky, scattered cloudy sky and overcast sky) are studied. The comparisons of the energy generation data of PV
panels obtained through testing and simulation under different weather conditions are also conducted.
Fig. 1. The location of deck-on photovoltaic sys- Fig. 2. The two test model of schematic design of
tem. floating PV system.
2.2 Micro floating PV system
The micro floating PV system has two test models, as shown in Fig. 2. The first is a closed-circuit model, where
one or two floating PV panels charge a portable battery (Battery 1), with Battery 2 acting as a load connected via
a DC cable. Battery 1’s input power indicates PV generation, and both charging and discharging powers are mon-
itored every 15 minutes, along with FSOC, floater angle, and various temperatures. The second model is open-
circuit, where each PV panel is connected to a sensor recording voltage every minute. All panels, each with a 40W
capacity, are placed on the sea surface in Tsing Yi harbour, Hong Kong (Fig. 3 and Fig. 4).
错误!未找到引用源。错误!未找到引用源。错误!未找到引用源。错误!未找到引用源。错误!未找到引用
源。
On
ly
3
fo
rR
ea
din
g
at
W
CF
Fig. 3. The location of floating PV panel system. Fig. 4. Site scene photo.
S
Using the recorded data, the relationship between PV panel temperature, floater temperature, seawater tempera-
20
ture, floater angle, and PV generation, as well as open-circuit voltage, is examined. Differences between the test
data and computer simulations under three weather conditions (clear sky, scattered clouds, and overcast) are ana-
lyzed, with results presented in Section 3.
24
2.3 Simulated photovoltaic system
Simulations of the floating PV systems are conducted to compare the simulation model result and real-world PV
generation for uncertainty characterisation. The simulation is performed in TRNSYS 18, a commercial tool to
simulate the behaviour of transient systems. Type 567 is chosen to simulate the floating PV dynamic generation.
Type 99 is adopted to provide the weather information. The weather information is collected from the Hong Kong
Observatory database. Weather information includes the wind velocity, wind direction, solar irradiance, and am-
bient temperature. The seawater temperature is collected from the experiment data by Infrared imaging. The sys-
tem diagram is depicted in Fig. 5. Based on the data of weather and seawater temperature, the simulation model
exports the generation and temperature. The real experiment is affected by the motion and slope of the floating
PV which are not considered in the simulation.
Fig. 3. Schematic of the simulation model.
3 Result and discussion
3.1 Performance of Deck-on Photovoltaic

The analysis is based on the data under three typical weathers. Due to the orientation of the deck and shading
effect by other objects, only the data of afternoon is utilized for analysis. 29th May, 2nd June and 30th May are
considered as typical clear, scattered cloudy and overcast sky respectively, the data of these days are shown in 错
误!未找到引用源。, 错误!未找到引用源。 and Fig. 8 separately. In a clear day, after the temperature of PV
panels reaches the peak around 44.7 °C at 13:40, the power generation would decreased by 5.8% to 1430.3 W
while the global radiation increased by 39.9% to 1087.0 W/ m2 at 13:50. This phenomenon can also be noticed in
a scattered cloudy day, after the PV temperature reaches the peak of that day around 52.5 °C at 13:20, the power
generation of PV panels remains almost the same even when the global radiation increases by 13.0%.
The impact of air temperature and wind speed on the power generation of the deck-on PV system is also inves-
tigated. Compared the period from 13:00 to 14:00, the average global radiation in 错误!未找到引用源。 is
940.7 W/m2 and is 31.3% higher than that in 错误!未找到引用源。. However, the corresponding average
power generation in 错误!未找到引用源。 is 35.0% lower than that in 错误!未找到引用源。. This may
because the air temperature and wind speed differences. The corresponding average air temperature of 错误!
未找到引用源。 is 31.7 °C and the average wind speed is 14.9 m/s. Those of 错误!未找到引用源。 are
On
ly
4
fo
rR
26.4 °C and 22.6 m/s respectively. These result in the 6.8 °C higher average PV temperature of the noon in 错
误!未找到引用源。 than that in 错误!未找到引用源。, from 41.1 °C to 47.9 °C. This 16.5% increase of the
ea
PV temperature caused by air temperature and wind speed constrains the power generation by 35.0% even
din
when the global radiation has a 31.3% increase.
g
at
W
CF
S
20
24
Fig. 4. Deck-on PV results with clear sky Fig. 5. Deck-on PV results with scattered cloudy sky.
Fig. 6. Deck-on PV results with overcast sky.
3.2 Performance of Floating Photovoltaic
The floating PV tests, conducted in Tsing Yi harbour on 8th May and 29th May, under scattered clouds and typical
clear sky respectively, showed that the systems worked as expected in Fig.9 and Fig.10. The PV panels generated
power successfully, continuously charging BAT1 and reducing the FSOC decline due to BAT2's charging. How-
ever, a gap between BAT1's discharging power and BAT2's charging power was noted due to charging loss. PV
generation and open-circuit voltage fluctuated due to environmental changes like cloud cover and wind.
Fig. 9. Data of 8th May test. (a) Close-circuit test data; (b) Open-circuit test data.
Fig. 10. Data of 29th May test. (a) Close-circuit test system data; (b) Close-circuit test temperature and genera-
tion data; (c) Open-circuit test data.
On
ly
5
fo
rR
ea
Several patterns can be observed from the data: PV and floater temperatures positively correlated with BAT1
din
input power, indicating higher PV generation with increased solar intensity. However, seawater temperature re-
mained stable due to its high specific heat capacity. Peak PV generation occurred between 13:00 and 15:00, with
g
a maximum at 14:00 on both test days, reaching 28.3W on a cloudy day and 37.5W on a clear day. The floater
at
temperature was generally higher than the PV temperature on clear days, likely due to the greenhouse effect within
the floater. Open-circuit voltage fluctuated between 14.2 and 15.4 V, with higher voltage corresponding to in-
W
creased PV generation. The second test on 29th May also recorded floater angles, showing that waves influenced
CF
PV generation, a topic for further research.
S
3.3 Comparison between the actual experiment and simulation
20
The experiment and simulation results are investigated in this section. Fig. 7 depicts the results for deck-on PVs.
24
The power and temperature of PVs are presented. On May 29th, the solar radiation is relatively high. From 13:00
to 18:00, the simulation and experiment PV power decrease, due to the solar radiation. However, the experiment
power fluctuates and has two peaks, while the simulation directly drops, due to the shade effect. The experiment
temperature has the similar trend with the experiment power curve. The experiment PV temperature changes
dramatically compared to the simulated PV temperature. The cause of the temperature difference is that the am-
bient temperature on the barge is high, and the radiation from the barge leads to the heating of the PV cells. When
the weather is cloudy, the simulated PV power is higher than the experiment PV power. The cause of this power
difference is the shade from the cloud. The experiment PV temperature is higher before 14:00, and then the ex-
periment and simulation PV temperatures are similar. These results indicate the radiative heating of barge is sig-
nificant when the solar radiation is high.
Fig. 7. Comparison of deck-on PV experiment and simulation results on (a) May 29th, and (b) June 2nd.
Fig. 8 depicts the simulation and experiment results for floating PV. On May 8th, the solar radiation is high.
The simulated and experiment power have similar value after 14:00. The sudden drop of PV power between
13:00 to 14:00 is attributed to the shade. The simulation and experiment temperatures have the difference
reaching 20 ℃. The potential cause is the floater temperature rise due to solar radiation. The floater temperature
reaches over 50 ℃ and hinders the heat dissipation of PV cells. On May 29th, the weather is cloudy. The
experiment PV power is similar to the simulation but fluctuates after 14:00. The cause is the temperature change
of PV floater., which affects the PV cell temperature and the efficiency. The experiment temperature is higher
than the simulation, caused by the floater heating.
Fig. 8. Comparison of floating PV experiment and simulation results at (a) May 8th, and (b) May 29th.
On
ly
6
fo
4 Conclusions
rR
ea
Through a thorough comparison between real-world floating PV generation data and simulation models, this re-
din
search aims to bridge the existing gap between theoretical understanding and practical application, enabling stake-
holders to make informed decisions and maximise the benefits of floating PV technology.
g
at
For the power generation of the deck-on PV system, the impact of the PV temperature is significant. Once the
PV temperature reaches the peak at noon on a clear day, the power generation of the system may even have a
W
5.8% reduction when the global solar radiation has a 39.9% increase. Furthermore, the ambient temperature
CF
and wind speed have a serious effect on the deck-on PV power generation by influencing the PV temperature.
A 5.3 °C increase in the air temperature and a 7.7 m/s decrease in wind speed can raise the PV temperature by
S
6.8 °C, which can bring a 35% reduction in the PV power generation even when the global radiation has a
20
31.3% increase.
24
As for the experimental test of micro floating PV systems, several preliminary patterns can be observed. The
high intensity of solar radiation leads to a larger floating PV generation, while brings higher PV panel and
floater temperatures. The intense solar radiation trigger greenhouse effect in the floater, showing that the se-
lection of floater type plays a key role in reducing the temperature of floating PV back surface and enhancing
the panel performance. The comparison between close-circuit and open-circuit tests reveals a general trend that
high PV generation under close-circuit conditions corresponds with high PV voltage under open-circuit test in
the same working environment.
Comparison of simulation and experiment results reveal the potential key impact factors affecting the PV per-
formance. The first is the uncertain solar radiation in real applications. The shade of clouds and other construc-
tions significantly reduce the PV generation. Secondly, the PV temperature is highly dependent on the sur-
roundings. The radiative heating from barge, and the conductive heating from the floater lead to the PV gener-
ation differences of experiments and simulations. Based on the results, the radiative heating from barge is more
significant, and leading to 16.5% PV generation difference by summing up the PV generation from 13:00 to
18:00 on May 29th for deck-on PVs.
By evaluating and comparing practical generation data with simulation models, this research contributes to the
development of sustainable and efficient energy solutions in locations with similar climates and geography.
The findings can guide future design improvements for integrating floating solar systems in urban coastal areas.
References
1. World Bank Group. Where Sun Meets Water: Floating Solar Market Report,
https://documents1.worldbank.org/curated/en/579941540407455831/pdf/Floating-Solar-Market-Report-
Executive-Summary.pdf, last accessed 2024/07/01.
2. Environment Bureau. Hong Kong Climate Action Plan 2030+,
https://www.hkgbc.org.hk/eng/engagement/file/ClimateActionPlanEng.pdf, last accessed 2024/08/01.
3. Environment Bureau. Hong Kong Climate Change Report 2015. 2015,
https://www.eeb.gov.hk/sites/default/files/pdf/ClimateChangeEng.pdf, last accessed 2024/08/01.
4. Global floating solar to top 6GW threshold by 2031, https://www.woodmac.com/press-releases/global-
floating-solar-to-top-6gw-threshold-by-2031/, last accessed 2023/07/01.
5. Elminshawy NAS, Osama A, El-Damhogi DG, Oterkus E, Mohamed AMI.: Simulation and experimental
performance analysis of partially floating PV system in windy conditions. Solar Energy. 2021;230:1106-21(2021).
6. Zhou S, Cao S, Wang S.: Realisation of a coastal zero-emission office building with the support of hybrid
ocean thermal, floating photovoltaics, and tidal stream generators. Energy Conversion and Management.
2022;253:115135 (2022).
On
ly
fo
rR
Floating urbanism, the bold vision of streetless cities.
ea
din
Tania Feldzer [0009-0003-5947-3427], John-Salvator Liotta [0000-0003-3504-935X], Emmanuel Dufrasnes [0000-0002-7341-8470]
g
1
School of architecture La Cambre Horta, Research Laboratory ‘Habiter’, ULB (Université Libre de Bruxelles),
at
Bruxelles, Belgium
tania.feldzer@ulb.be
W
Abstract:
CF
S
To build upon the discussions initiated by Rutger de Graaf on floating urbanism at the recent WCFS conferences, this document
aims to not only delve into the potential of reimagining street spaces in floating cities but also to shed light on the innovative
20
concept of "floating urbanism," which envisions cities free from traditional streets. It critically examines and reimagines the
24
foundational principles of urban planning, positing the provocative idea that cities might not require streets to serve as their
backbone.Thought an experiment made during the SIP (Pedagogical Week Innovation) in ULB (University Libre de Brux-
elles) in March 2024, the students were invited to work on conceptual exercises to rethink street functions in urban design,
based on the idea that the absence can create new perspectives of mobility. From Tim Ingold's work, students were tasked to
reimagining urban mobility and space in existing cities built without streets, Catalhoyuk in Turkey, Ganvié in Benim and Uros
in Peru. The methodology consists of three steps: first, developing the DNA of each given city model. Then, by using creative
thinking techniques, such as mind mapping and SCAMPER, generate innovative ideas on 5 key mobility concepts: mobility
of memory, mobility in time, mobility for networks and generative infrastructure, and mobility for division. Finally, by over-
layering the 3d models obtained, we have produced a series of prototypes of innovative mobility in cities. The text shows how
floating urbanism offers new potentialities, emphasising ecological management, community spaces, adaptability to climate
change, and integrated mobility networks in time motion.
Keywords: Floating Urbanism; Mobility; Network
Introduction
For the architectural conception of a holiday village in Tunisia in the beginning of the 1960s, Cantal Dupart
(Cantal Dupart is a French urban architect, professor emeritus at the CNAM of Paris, former member of ANRU,
Nice Modernist Prize 2002, Director of the Urbanisme magazine in 1990-1991, Co-initiator of Grand Paris in
2010.), a young architecture student at the time in Les Beaux-Arts de Paris, had proposed to not build roads
between the sleeping rooms and the shared spaces. His strategy was to let the visitor themselves trace the path
that would suit them best. Thus, an area of earth and sand was deliberately left on the ground so that over weeks
and years a vernacular path emerged, traced by spontaneous acts or so-called "desire line." What is remarkable in
this initiative is not so much that the visitor became an actor in the production of public space but that circulation
was not the founding act of the project's layout. It outcome a street, traced without determinism and initially
conceived by voluntary omission. Streets are not just means of communication but also integral elements of the
urban fabric, shaping the interactions and experiences of city dwellers. Cantal Dupart's bold experiment in Tunisia
highlights an organic and participatory approach to urban design, questioning the role of the architect and planner
in the creation of public space. Reconsider the street or its absence is the theme we would like to address in this
article by asking the provocative question: can we conceive urbanism without streets? If the question arises, it is
because such configurations already exist or have existed in the past in cities like Çatalhöyük in Turkey or Ganvie
in Benin. According to our hypothesis, the street is thought of as a circulation before being a space, consumed and
consuming, often dedicated to the technology of the car. The exercise of an urbanism without streets allows us to
rethink spatial organization logic, movements, mobilities, and social interactions. It structures the city, determines
a spatial organization, a social policy, organizes voids and solids, addresses the question of density, and the eco-
logical footprint of public space.To build upon the discussions initiated by Rutger de Graaf on floating urbanism
at recent WCFS conferences, this document aims not only to explore the potential of reimagining street spaces in
floating cities but also to shed light on the innovative concept of "floating urbanism," which envisions cities free
from traditional streets. De Graaf, in his work on urban water management, emphasizes the importance of adapting
existing infrastructures to climate challenges. The proposed research aims to further explore the reflections on
new cities by Blue21 by reexamining key notions of "urban structure" and "property rights."
This text proposes to explore how we can think differently about our movements in models of existing cities built
without streets and to test our hypotheses through an experiment conducted during the pedagogical innovation
week organized by the author at ULB (Université Libre de Bruxelles) in March 2024.
On
ly
2
fo
Methodology
rR
ea
To address the questions posed in this article, we first propose to explore, through a historical and analytical study,
din
how urban issues have been treated in terms of street design and its limits of understanding. Then, we will detail
the experimentation developed during the workshop and the proposed method to rethink the characteristics of the
g
street differently. Finally, we will analyse the results of the experimentation and the insights they may provide on
at
urban design and public spaces.
1 Literature review
W
CF
The street as a structuring space
Eric Alonzo, in his bibliographic work on the history and theory of the way [1], sheds light on their evolution and
S
how streets have shaped our cities. As early as the 15th century, in his treatise De Re Aedificatoria, Leon Battista
20
Alberti assigned a particular status to the street, considering it a work of art to be treated as an object of conven-
ience and beauty[2]. However, Alonzo criticizes the so-called "modern" conception of streets, which he sees as a
24
technological alienation in service of the car. The invention of the car led to the functional separation of flows and
the rectification of routes, creating a spatial and conceptual isolation of main roads, kept at a distance from built-
up areas. This isolation, as David Mangin points out, is rendered impassable by the proliferation of support infra-
structures for trains, cars, buses, and bicycles[3]. Mangin highlights how this configuration leads to sectoral ur-
banism, contributing to social segregation and urban sprawl[4].
The absence of streets: an experiment from the 60s

The work of Aldo Van Eyck and Piet Blom in the 1960s questions the solid/void duality by freeing itself from the
space of the street [5]. Piet Blom viewed each house as a cell within a larger urban organism, promoting social
interactions and community while maintaining the intimacy of personal space. Blom's concept of circulation man-
ifests through the creation of public spaces as places for meetings and exchanges[6]. These spaces are actively
created by the community, opening a reflection on the possibility of rethinking land use and the place of public
space[7].
Fig 1. Pratical planning exercice, study project 1959, Piet Blom, published in Forum, vol.15 no.5.
Conclusion of the Analytical Research

Imagining the street as an impermanent and de-hierarchized space allows us to question the notions of mobility
and circulation. Rethinking the land and real estate status of private and public spaces can open up new perspec-
tives in urban design.
2 Experimentation during the SIP

The methodology is based on the idea of considering the street by its absence and viewing it not as a structuring
element but as an intrinsic and singular entity. Inspired by key concepts from the anthropologist Tim Ingold, we
explore the street as memory, as social interaction, as the development of a new temporality, as a network, and as
a division of property[8]. Each work group was assigned a city and where engage to create new hypothesis of
mobility. The steps included creating the city's DNA, formulating planning hypotheses, and analyzing the com-
bined results.
On
ly
3
fo
2.1
rR
Hypotheses of Mobility by Function
Mobility as a Space of Memory
ea
Lines of movement weave knowledge and memories, constituting living paths. Urban spaces serve as receptacles
din
for collective and individual memory [9]. considers displacement as public spaces impregnated with memorial
elements that transcend their immediate use[10].
Mobility as a Space of Social Interaction

g
at
Social spaces emerge from streets and communication networks [11]. These streets actively constitute social space
W
by connecting people and ideas, fostering encounters and neighborhood life[12].
Temporary Mobility
CF
S
Streets represent pathways through time, where each movement and trace is a manifestation of temporality. Lines
of movement inscribe time into space, creating continuity between past, present, and future[13]. Mobility is both
20
a physical and temporal experience, influencing how spaces are lived and perceived [14].
Mobility as a Network
24
Streets facilitate understanding of connectivity and networks, forming a complex web of relationships and ex-
changes. Connectivity is essential in contemporary urban experiences. Students hypothesized that streets can
evolve based on network necessities[15].
Mobility as a Guide and Spatial Division

Lines delimit and structure space, creating distinct zones[16]. They guide movement and interaction, influencing
power dynamics and identity[17].
2.2 Research Supports and proposals.
Catalhöyük
The Neolithic city of Catalhöyük is characterized by the absence of streets and movement over rooftops. The
study questions the hierarchy of spaces and the relationship between private and public spaces. Mobility could be
perceived as an embodiment of its "life lines," moving vertically rather than horizontally. This perspective sug-
gests a form of connectivity and social network based not on physical configurations but on interpersonal and
spatial relationships.
Towards a De-hierarchized Architecture
Historically, around 7500 B.C., spaces were not hierarchized. Projecting this observation to 2100, we could en-
visage a standardization influenced by our contemporary housing models. These models, powered by generative
artificial intelligence like "Architectures AI," could favor the return to non-nominative and non-hierarchized hab-
itats accessible to all. Mobility in these future habitats would be vertical over prolonged durations, adapted to
changes in family units, while horizontal mobility would be facilitated by universal accessibility to public and
outdoor spaces.
Fig. 2 Drawings of long-term mobility recompositions in Catalhöyük, source: author based on the work of students
Christopher Luamba, Jeremy Mukala, Alina Maria Petrescu, Alix Revaz.
On
ly
4
fo
Ganvié rR
ea
The lake city of Ganvié is built on stilts, with pirogue-based movement. Adaptation to seasonal variations and
din
gender-based community distribution are key aspects. The transformation of current practices offers innovative
opportunities, notably through the use of organic waste as resources for creating new circulation spaces. This
g
approach rethinks mobility by integrating territorial changes, where stilt constructions allow an expansion of con-
tact surface between dwellings, optimizing space usage. A reversal of the traditional relationship between built
at
structures and land is envisaged through topographical manipulations that adapt the territory to the built environ-
W
ment. These topographical interventions aim to harmonize architecture with its natural environment, promoting a
more organic and sustainable integration.
Rethinking Functions
CF
It is crucial to design functions related to mobility. The establishment of floating platforms, whose functions are
S
not predefined, offers a framework for participatory mobility management. These modular and flexible structures
20
allow space to adapt to the changing needs of users, facilitating a more collective and dynamic approach to urban
planning.
24
Fig. 4. Territorial plan of Ganvié, source: author based on the work of ULB students Inès Francois, Alexis Gazel,
Valentina-Claudia Popa, Kristine Sampang.
Uros
The floating city of Uros is composed of micro-islands made of reeds, with variable movement. Mobility is re-
thought without a sense of immobility, promoting social interactions and increased connectivity.
On
ly
5
fo
rR
ea
din
g
at
W
CF
S
20
Fig. 5. Territorial plan of Uros, Scan of the movement of islands over the years. source: author
The Free Plan of Uros

24
Inspired by the burolandschaft theory developed in the 1960s by the brothers Eberhard and Wolfgang Schnelle,
"Spaces were defined by the functions of the people who worked in them and their relations to colleagues."[18].
The project drawn by the students is inspired by the free plan office, proposing a game where the participant is
invited to create their own scenario. The project extends the idea further by integrating the formation of clusters
based on urban functions. In 1974, Yona Friedman proposed a formula to rethink the uses according to their
distances (d) and their frequencies (w), which he brought together in a path matrix to evaluate the effort value (E)
𝑦 𝑛
of each entity. Thus, Ex = ∑𝑦 1 𝑑𝑥𝑦. 𝑤𝑦 [19]. In the proposal, each clusters Ex=0 group around a central axis,
forming a hub of activity, while circulation flows are redefined to bypass these groupings. The parameters of this
configuration are controlled by a script developed with Grasshopper, using vector fields to adjust the size, radius
of clusters, as well as numerical proximities and the number of relationships between clusters. Circulation in these
spaces is no longer determined by fixed axes but by clusters of constantly evolving functions, reflecting a more
flexible and responsive approach to spatial organization. This model offers an adaptive response to changing user
needs, while promoting increased interaction and connectivity between different functional spaces.
Fig. 6 Cluster of islands generated by parametric drawing in Grasshopper, field attractor. Source: author.
On
ly
6
fo
rR
Conclusion ea
din
g
The results raise obvious technical and political feasibility questions while opening up the field of possibilities.
They also question the fundamental idea: should we reproduce the same city model on water as on land? These
at
approaches encourage reinventing the notions of "plots," "islets," real estate, and privacy in urban composition.
W
They also promote more active user participation in configuring their environments, fostering spatial democracy
where each individual can influence the form and function of their living space. These models prefigure a future
CF
where architecture and urbanism not only meet functional requirements but also aspire to create spaces that pro-
mote collective and individuality while respecting ecological imperatives. Can mobilities be design by a new
S
evaluation of function not only dictated by use but by frequency, usage and distance based on an improvement of
Yona Friedman theory of space organisation?
20
References
24
1. Alonzo, É.: L’architecture de la voie: histoire et théories. École d’architecture de la ville & des territoires ; Éditions
Parenthèses, Champs-sur-Marne : Marseille (2018).
2. Alberti, L.B., Rivera, J.: De re aedificatoria. Akal, Madrid (1991).
3. Les flux, l’architecture et la ville: Entretien avec David Mangin, architecte-urbaniste, par Éric Charmes et Benjamin
Pradel. Flux. n° 63-64, 89–99 (2006).
4. Mangin, D.: La rue en toutes franchises: Pouvoirs. n° 116, 131–147 (2006).
5. Di Palma, V., Periton, D., Lathouri, M. eds: Intimate metropolis: urban subjects in the modern city. Routledge, London ;
New York (2009).
6. Gaffikin, F., Mceldowney, M., Sterrett, K.: Creating Shared Public Space in the Contested City: The Role of Urban De-
sign. Journal of Urban Design. 15, 493–513 (2010).
7. Merx, S.: Public Pie Performing public space. Performance Research. 16, 132–137 (2011).
8. Ingold, T.: Lines: a brief history. Routledge, London ; New York (2016).
9. Lynch, K.: The image of the city. M.I.T. Press, Cambridge, Mass. (2008).
10. Aceska, A.: Loose Space: Possibility and Diversity in Urban Life – Edited by Karen A. Franck and Quentin Stevens. Int
J Urban Regional Res. 36, 411–412 (2012).
11. Gehl, J.: Cities for people. Island Press, Washington, DC (2010).
12. Jacobs, J.: The death and life of great American cities. Vintage Books, New York (1992).
13. Cresswell, T.: On the move: mobility in the modern Western world. Routledge, New York, NY (2006).
14. Amin, A., Thrift, N.J.: Cities: reimagining the urban. Polity Press, Cambridge (2012).
15. Hillier, B., Hanson, J.: The social logic of space. Cambridge University Press, Cambridge (2005).
16. Massey, D.B.: For space. SAGE Publications Ltd, Place of publication not identified (2012).
17. Soja, E.W.: Thirdspace: journeys to Los Angeles and other real-and-imagined places. Blackwell, Cambridge, Mass
(1996).
18. How the 21st century office was born in post war europe, https://workplaceinsight.net/how-the-21st-century-office-
was-born-in-post-war-europe/.last accessed 2024/05/25
19. Friedman, Y.: Toward a scientific architecture. MIT Press, Cambridge, Mass (1975).
On
ly
fo
rR
A green concrete material and its new biomimetic sustainable ocean
ea
floating structure
din
g
Yufei Wu 1and Xiaoxu Huang 1
1
at
School of Civil and Transportation Engineering, Shenzhen University, Shenzhen, China
W
yufei.wu@szu.edu.cn
CF
Abstract. The authors and their colleagues have recently developed a new concrete technology – compression-
cast concrete. Compression casting technology condenses concrete, which makes it stronger and more durable.
S
It is a practical technology that not only greatly improves the performance of concrete materials, but also
20
reduces the cost and greenhouse gas emissions of related concrete structures. The compression-cast concrete
technology has been developed for a major research project – a new biomimetic sustainable marine floating
24
structure. The goal of the project is to develop a new type of offshore floating structure characterized by supe-
rior structural performance, low construction and maintenance costs, high sustainability, and good survivability
in harsh marine environments. To achieve these objectives, the team has developed several new technologies,
including compression-cast concrete technology, bionic honeycomb floating unit, flexible floating unit con-
nectors, new mooring systems, new ocean energy harvesting systems, and an effective and practical method
for surviving in harsh sea conditions. These new technologies are described in this article.
Keywords: Compression cast concrete; floating structure; sustainability; survivability.
1 Introduction
Concrete is regarded as the most widely used construction material with significant carbon emissions. It is reported
that 400 kg CO2eq carbon emissions are generated when producing 1 m3 concrete [1]. It is an urgent task to reduce
CO2 emissions for the construction industry. In the past few decades, many techniques have been proposed to
reduce the amount of cement in concrete production, such as partially substituting cement with by-products (e.g,
blast silica fume, fly ash, and industrial wastes), and complete replacement of cement with green cement alterna-
tives (e.g., alkali-activated geopolymers and limestone calcined clay cement) [2-4]. However, the high cost hin-
ders their application in practical engineering. Recently, the authors’ research group has developed a novel com-
pression casting technology (CCT) (see Fig. 1). With this new technology, the strength and durability of concrete
can be improved significantly, thus reducing the cost of concrete [5-8]. Notably, compared to normal casting
(NC), lower carbon emission is generated for CCT, providing an effective approach for the sustainable develop-
ment of the construction industry [1].
Fig. 1. Compression casting process.
To promote the application of CCT in practical engineering, the performances of compression-cast concrete
(CCC) are first investigated comprehensively, including mechanical properties, microstructures, durability, and
carbon reduction benefits. Then, this green concrete material is applied to floating structures with a biomimetic
shape. For this new kind of floating structure, the required information for the design has been investigated in-
cluding bionic honeycomb floating unit, flexible connectors for units, and new mooring systems. To achieve
On
ly
2
fo
rR
energy self-sufficiency for floating structures, new ocean energy harvesting systems are studied. In addition, an
effective and practical method for surviving in harsh sea conditions is proposed.
ea
2
din
Compression-cast concrete (CCC)
g
at
2.1 Mechanical properties
W
As shown in Fig. 2, compared with NC, the compressive strength and elastic moduli of CCC increased by 26%-
83% and 26%-69%, respectively [9]. The remarkable improvements in the mechanical properties of CCC are
mainly attributed to the reduced air and water contents and the denser microstructure of the concrete [5].
CF
S
60 40
Compressive strength of CCC (MPa)
20
Elastic modulus of CCC (GPa)

35
50
40
30
25
24
30
20
20 Test results
Test results 15
Fitted curve Fitted curve
10 10
10 20 30 40 50 60 10 15 20 25 30 35 40
Compressive strength of NC (MPa) Elastic modulus of NC (GPa)
(a) Compressive strength (b) Elastic modulus
Fig. 2. Improvement of mechanical properties of CCC over NC.
30 120
NC
NC
Relative modulus of elasticity (%)
CCC-5
25 CCC5
CCC-15
Carbonation depth (mm)
100 CCC-15
20
80
15
60
10
40
5
20
0 -25 0 25 50 75 100 125 150 175 200 225 250
3 7 14 28
Days Number of freeze-thaw cycles
(a) carbonation depth (b) relative modulus of elasticity
25
22.6
20 19.1
16.3
DRcm /1e-12 m2/s
15
10 9.1
6.7 6.9
5
0
CCC-5#1 CCC-5#2 CCC-5#3 NC#1 NC#2 NC#3
(c) Chloride diffusion coefficient (d) Strength retention v.s. temperature
Fig. 3. Durability of CCC.
On
ly
3
fo
2.2 Durability
rR
Carbonation ea
din
As shown in Fig. 3(a), the carbonation depths are much smaller for CCC specimens compared with NC specimens:
after 28 days of CO2 exposure, the carbonation depth of CCC specimens is 82.6% and 91.7% lower for the casting
g
pressure of 5 and 15 MPa, respectively, compared with NC specimens. The result also shows that the effect of
casting pressure on the carbonation depths is insignificant.
at
W
CF
Freeze-thaw (FT) cycles
Fig. 3(b) presents the retention values of the dynamic elastic modulus of concrete after FT cycles. It can be found
S
that the effect of the casting pressure on the deterioration ratio of the dynamic elastic modulus is obvious. The
20
deterioration ratio decreases as the casting pressure increases. This phenomenon can be explained by the fact that
the denser internal structure and higher strength of the cement matrix under higher casting pressure lead to the
improved retention of dynamic modulus of elasticity and enhanced FT resistance.
24
Chloride penetration
Fig. 3(c) shows that the chloride diffusion coefficients of CCC are smaller than those of NC, indicating that the
CCT can postpone the process of chloride penetration. Due to the reduced porosity and densified microstructure
of CCC, the inhibition of chloride migration is improved correspondingly.
High temperature
Fig. 3(d) shows the retention rate of compressive strength (i.e., the ratio of strength at the current temperature to
the initial strength at room temperature) for different specimens. The strength retention of CCC is larger than NC:
compared to NC specimens at 600 ℃, the strength retention of CCC increased by 26.4% and 47.9% with casting
pressures of 5 MPa and 15 MPa, respectively.
2.3 Cement saving

Fig. 4 presents the experimental results of strength enhancement and cement reduction contributed by CCT. For
cement content larger than 253 kg/m3, the strength enhancement of CCC is significant, as shown in Fig. 4(a). The
improvement of strength is the basis for the reduction of cement in concrete for the same strength requirement.
To achieve the equivalent concrete strength, the cement content in CCC is much lower than that in NC, as shown
in Fig. 4(b). Thus, CCT can remarkably reduce cement content in concrete production without loss of strength
and durability.
55
50
Cement saving (%)
45
40
35
30
25
30 35 40 45 50 55
Compressive strength (MPa)
(a) Compressive strength (b) Cement saving
Fig. 4. Strength enhancement and cement reduction contributed by compression cast.
On
ly
4
fo
3
rR
Marine floating structure with CCC
3.1 Bionic honeycomb floating unit

ea
din
Based on the concept of a bionic honeycomb, hexagonal floating units constructed by CCC are proposed by the
g
author’s research group, and the optimal geometric configurations (small-scale unit and medium-large unit) are
at
obtained for different sea states. As illustrated in Fig. 5, the new type of floating unit exhibits several advantages:
(a) the floating units are constructed by CCC, which is more economical in manufacturing and maintenance
W
compared to traditional steel structures; (b) different with square modules that are confined to linear expansion,
CF
hexagonal floating units allow for dynamic and flexible expansion without gap according to specific requirements
(see Fig. 5-c); (c) the floating unit is designed as a prefabricated structure, significantly reducing construction and
S
installation time as well as associated costs; (d) the function of diving and floating can be realized for the floating
20
unit as needed.
24
(a) small-scale unit (b) medium-large unit (c) floating platform

Fig. 5. Honeycomb floating unit and platform.
3.2 Flexible floating unit connectors

The existing connections between float units mainly include rigid and flexible connectors. Stress concentration
may exist for rigid connectors while most flexible connectors are mechanical mechanisms with inconvenient dis-
assembly and limited buffer capacity. Based on the concept of the intervertebral disc, a new kind of flexible
connector is developed with inflatable (or fluid) rubber strips, as shown in Fig. 6. By selecting the fluid type and
pressure inside the rubber strip, and setting the fluid in and out of the regulator and the automatic jump valve
system, the stiffness, damping and maximum pressure of the connector can be controlled for different working
conditions. The proposed ductile connector significantly reduces impact loads.
(a) Concept of flexible connectors (b) Flexible connector for floating unit
(c) Experimental results

Fig. 6. Design of flexible connectors for floating units.
3.3 New mooring systems

For a conventional taut mooring system, due to the cyclic loading of waves, significant fluctuations in tension
occur in mooring lines, reducing the cable’s service life. To overcome this issue, a constant tension pulley mooring
On
ly
5
fo
rR
system has been proposed for floating structures. As shown in Fig. 7, the proposed mooring system consists of
clump weight, pulley, and wires. The advantages of the pulley mooring system are obvious: 1) the force in the
ea
mooring line is controlled by weight, avoiding line breaking and reducing anchoring costs; 2) the fluctuation of
din
tension force is small, increasing the fatigue life of the mooring system.
g
at
W
CF
S
20
(a) Pulley mooring system for floating unit
24
(b) Pulley mooring system for floating platform
Fig. 7. Design of the pulley mooring system.
3.4 New ocean energy harvesting systems

Based on novel piezoelectric nanomaterials, a new ocean energy harvesting system is proposed [10], as shown in
Fig. 8. The system consists of four energy trapping layers: wind and solar energy capture layer on the upper
surface; wave energy trapping layer in the middle part; wave energy trapping layers at the bottom and tail. The
new floating structure can effectively convert the natural energy of the sea (wave energy, tidal energy, wind en-
ergy, solar energy, etc.) into electrical energy storage, and realize the energy self-sufficiency of the floating struc-
ture.
Fig. 8. Design of energy harvesting system for floating structures.
3.5 Method for surviving in harsh sea conditions

The survivability of floating structures in harsh sea conditions is a critical factor affecting their safety, stability,
and large-scale applicability. A method for surviving in harsh sea conditions is developed for floating structures
combined with a submersible and floating mooring system. As shown in Fig. 9, the surviving system consists of
a taut mooring system and a buoyancy adjustment device (i.e., ballast tank and winch). By increasing or decreasing
ballast water and retracting or releasing mooring lines, the floating and submersible states can be switched auton-
omously to encounter the variation of sea states. The simulation results show that, as the diving depth increases,
the motion responses of the floating structure and the tension in the mooring lines gradually decrease: compared
with the surface state (S0), the maximum value of surge, heave, pitch, and mooring tension decrease by 94.43%,
66.63%, 91.09%, and 64.91%, respectively, for diving 20 m (SIV) (see Fig. 10).
On
ly
6
fo
rR 20
20.11
S0
SⅣ
ea 17.07
Peak value of response

din 15
g
94.43%
91.09%
at
10
W
7.21
64.91%
CF
5
3.33
2.53
66.63% 1.52
1.12 1.11
0
S
20
Surge (m) Heave (m) Pitch (deg) Tension (MN)
24
Fig. 9. Submerge for floating units. Fig. 10. Comparison of response before and after diving.
4 Conclusions
In this study, a new concrete technology – compression-cast concrete (CCC) is introduced, and the comprehensive
properties of CCC are investigated, including mechanical properties, microstructures, durability, and carbon re-
duction benefits. Compared with normal concrete, CCC has much higher strength, better mechanical performance,
better environmental impact, and better durability in terms of reduced carbonization depth, less damage under
freeze-thaw cycles, smaller chloride penetration depth, as well as greater fire resistance. In addition, this novel
technology has been applied to floating structures with a biomimetic shape. The required issues for the design of
the floating structures have been investigated including bionic honeycomb floating units, flexible connectors for
units, new mooring systems, ocean energy harvesting systems, and methods for surviving in harsh sea conditions.
These new technologies provide effective approaches for the sustainable development of floating structures.
References
1. Tang B, Wu H, Wu Y F. Evaluation of the carbon reduction benefits of adopting the compression cast technology in
concrete components production based on LCA. Resources, Conservation and Recycling 208, 107733 (2024).
2. Yang K H, Jung Y B, Cho M S, et al. Effect of supplementary cementitious materials on reduction of CO2 emissions
from concrete. Journal of Cleaner Production 103, 774-783 (2015).
3. Scrivener K, Martirena F, Bishnoi S, et al. Calcined clay limestone cements (LC3). Cement and concrete research, 114,
49-56 (2018).
4. Li C, Sun H, Li L. A review: The comparison between alkali-activated slag (Si+ Ca) and metakaolin (Si+ Al) cements.
Cement and concrete research 40(9), 1341-1349 (2010).
5. Wu Y F, Kazmi S M S, Munir M J, et al. Effect of compression casting method on the compressive strength, elastic
modulus and microstructure of rubber concrete. Journal of cleaner production 264, 121746 (2020).
6. Kazmi S M S, Munir M J, Wu Y F. Application of waste tire rubber and recycled aggregates in concrete products: A new
compression casting approach. Resources, Conservation and Recycling 167, 105353 (2021).
7. Wang X, Wang J, Kazmi S M S, et al. Development of new layered compression casting approach for concrete. Cement
and Concrete Composites 134, 104738 (2022).
8. Yuan F, Li H, Wu Y. Axial Compressive Behavior of FRP-Confined Compression-Cast Recycled Aggregate Concrete.
Journal of Composites for Construction 28(3), 04024009 (2024).
9. Wang X, Munir M J, Wu Y F. Effect of mix design parameters on the behavior of compression cast concrete. Journal of
Building Engineering 76, 107399 (2023).
10. Liu J, Bao B, Chen J, et al. Marine energy harvesting from tidal currents and offshore winds: A 2-DOF system based on
flow-induced vibrations. Nano Energy 114,108664 (2023).
On
ly
fo
rR
Design methodology of bamboo-raft-type floating structure (BRT-FS)
ea
din
Chongfeng Xie and Peng Feng*
g
MOE Key Lab of Civil Engineering Safety and Durability, Department of Civil Engineering, Tsinghua University, Beijing,
at
100084, China
fengpeng@tsinghua.edu.cn
W
CF
S
Abstract. The authors have developed a novel floating structure—bamboo-raft-type floating structure (BRT-
FS). However, the mechanical properties of BRT-FS differ from that of traditional VLFS. The load-bearing
20
capacity and failure mode of BRT-FS are primarily controlled by longitudinal properties, exhibiting strong
anisotropy. Facing these unique mechanical properties, it is urgent to establish a new design methodology to
24
guide the engineering application of BRT-FS. This paper surveys the load combinations for concrete floating
structures recommended by various standards and establishes load case combinations suitable for the BRT-FS.
Based on these three limit states, serviceability limit state (SLS), ultimate limit state (ULS) and accidental limit
state (ALS), the design methodologies of BRT-FS under different limit states are proposed.
Keywords: BRT-FS, Load combinations, Design methodology.
1 Introduction
The demand for floating structures is rapidly increasing, but many experiences from land-based infrastructure
construction are not applicable to the more challenging offshore environments. Compared to onshore construc-
tion, offshore construction faces unique challenges, such as harsh corrosive environments, extreme load condi-
tions, restricted construction conditions, and high difficulty in maintenance and repair[1-3]. Therefore, it is nec-
essary to explore and study new construction materials and structural configurations applicable to the offshore
environments.
The authors have developed a novel floating structure—bamboo-raft-type floating structure (BRT-FS)[4]. This
novel floating structure employs lightweight aggregate concrete and fiber reinforced polymer (FRP) as the pri-
mary materials for load-bearing components. It includes four main components: concrete tubes, prestressed car-
bon fiber reinforced polymer (CFRP) cables, holding beams, and bellows, as shown in Fig. 1. Compared to tra-
ditional very large floating structures (VLFS), this structure offers advantages such as high design flexibility,
outstanding durability, reduced cracking risk, high construction and transportation efficiency, and lower overall
lifecycle costs.
Fig. 1. Main components of BRT-FS.
BRT-FS exhibits completely different mechanical properties in the longitudinal and transverse directions, indi-
cating strong anisotropy. Due to the unique mechanical property, it is urgent to establish a new design methodol-
ogy to guide the engineering application of BRT-FS. This paper proposes design methodology for BRT-FS
On
ly
2
fo
rR
under three limit states: serviceability limit State (SLS), ultimate limit state (ULS), and accidental limit state
(ALS).
ea
2 Design methodology
din
2.1 Load condition combination selection
g
at
An investigation was conducted into the standards established by various domestic and international organiza-
W
tions and associations, including China Classification Society (CCS), the International Organization for Stand-
CF
ardization (ISO), Det Norske Veritas (DNV), and the European Union (EN). It was found that CCS standards do
not explicitly specify the load partial factors for floating platforms. Instead, they provide recommendations pri-
S
marily on load types and load combinations. In contrast, international standards, based on ISO standards, offer
20
detailed suggestions for load combinations under different limit states. Based on the aforementioned standards,
the fundamental load types selected for BRT-FS design include hydrostatic pressure, wave load, current load,
topside load, and ice load. The load combinations were considered according to Table 1.
Table 1. Load combinations and partial factors.

24
Ultimate state Permanent load Variable load Environmental load
(hydrostatic pressure) (topside load) (wave load, current load, and ice load)
SLS 1.0 1.0 1.0
ULS-1 1.3 1.3 0.7
ULS-2 1.0 1.0 1.3
ALS 1.0 1.0 1.0
2.2 SLS design
The design methodology under serviceability limit state (SLS) is illustrated in Fig. 2. The hydroelastic response
of the structure is analyzed based on the common wave conditions. The acquired vertical deflection amplitude is
used to assess comfort levels. The acquired stress range in the CFRP plates is used to evaluate the long-term
performance. The purpose of this step is to ensure the comfort meet personnel requirements and to guarantee that
the mechanical performance of CFRP plates does not degrade excessively during service.
Fig. 2. Design methodology of SLS.
2.3 ULS design

The design methodology under ultimate limit state (ULS) is illustrated in Fig. 3. Hydroelastic analysis is con-
ducted using wave conditions with a recurrence interval of 100 years. The acquired bending moments and shear
responses are used to verify the overall strength of the concrete tubes. When lateral loads from current and ice act
on the structure, it becomes a critical state requiring verification of the overall strength of the edge concrete tubes.
Hydrostatic pressure and topside load have minimal impact on the overall structural performance but affect sec-
tional behavior, thus requiring verification of sectional load-bearing capacity. Concrete experiences maximum
principal stresses in different locations under different conditions. Hence, their coupled effects are disregarded.
Environmental loads are considered in all overall strength verifications by applying the partial factors of ULS-2.
Permanent and variable loads are considered in sectional load-bearing capacity verifications by applying partial
On
ly
3
fo
rR
factors of ULS-1. The purpose of this step is to ensure that concrete does not crack and that the structure maintains
its mechanical integrity under permanent, variable, and environmental loads.
ea
din
g
at
W
CF
S
20
24
Fig. 3. Design methodology of ULS.
2.4 ALS design

The design methodology under accidental limit state (ALS) is illustrated in Fig. 4. The wave conditions under
ALS are more severe than those under ULS. In this state, the concrete tubes crack. However, the speed of crack
propagation must be limited, meaning the cracking process must be controllable. A specific wave period is se-
lected, and the cracking wave height Hcr is determined based on the hydroelastic response. At this moment, the
bending moment amplitude of each concrete tube is Mcr. The historical maximum bending moment amplitude of
each concrete tube is denoted as Mmax. Initially, the wave height is set as H=Hcr and Mmax=Mcr. H is increased by
0.2 m for each step, and the bending moment response M is calculated for each wave height. If the bending
moment amplitude exceeds the flexural cracking load of concrete tubes, the shear amplitude exceeds the shear
cracking load, or the joint bending moment exceeds the debonding bending moment, the structure is considered
to have failed and the calculation is stopped. Otherwise, M is used to update Mmax. The bending stiffness of the
concrete tubes is corrected based on the stiffness degradation formula for cracked concrete tubes. The corrected
stiffness is used for the hydroelastic response calculation of the next wave height. Each concrete tube has two
joint bending moments. Since the overall bending stiffness of the concrete tube is controlled by the minimum
stiffness, the larger joint bending moment is used for the stiffness correction of the cracked concrete tubes. The
purpose of this step is to prevent progressive damage under extreme wave conditions, ensuring the safety of life
and property. After enduring harsh conditions, the structure can continue to be used following partial repair and
replacement.
Fig. 4. Design methodology of ALS.

On
ly
4
fo
3 Conclusion
rR
ea
This paper first conducted a survey of existing standards and found that the standard of CCS does not specify
din
detailed partial factors for floating platforms. Instead, it provides recommendations only for load types and load
combination methods. In contrast, the international standards offer detailed recommendations for load combina-
g
tions under different limit states based on ISO guidelines. Based on the survey results, suitable load case combi-
at
nations for the BRT-FS are selected. Considering the unique mechanical property of BRT-FS, design methodol-
ogies for the three limit states (SLS, ULS, and ALS) are proposed. Under SLS, verification for comfort and CFRP
W
plate durability are performed. Under ULS, overall strength verification is conducted for hydroelastic response,
CF
ocean current loads, and ice loads, while sectional load-bearing capacity verification is performed for hydrostatic
pressure and topside loads. Under ALS, the overall strength verification is conducted based on the post-cracking
S
bending stiffness of the concrete tubes, aiming to avoid progressive damage under extreme wave conditions.
20
References
24
[1] Adedipe, O., Brennan, F., Kolios, A.: Review of corrosion fatigue in offshore structures: Present status and
challenges in the offshore wind sector. Renewable and Sustainable Energy Reviews, 61, 141-154 (2016).
[2] Li, Y., Hou, B., Li, H., Zhang, J.: Corrosion behavior of steel in Chengdao offshore oil exploitation area.
Materials and Corrosion, 55, 305-310 (2004).
[3] Amaechi, C. V., Reda, A., Butler, H. O., Ja’e, I. A., An, C.: Review on Fixed and Floating Offshore
Structures. Part I: Types of Platforms with Some Applications. Journal of Marine Science and Engineering, 10,
(2022).
[4] Xie, C., Feng, P.: Novel bamboo-raft-type floating structure (BRT-FS) assembled by FRP reinforced
concrete tubes: Conceptual design and analysis. Engineering Structures, 318, 118705 (2024).
On
ly
fo
rR
Durability evaluation of maritime concrete served in ocean environment
ea
din
Qi Zhao1[0000-0003-3452-4948], Keitai Iwama2, Xiao-Ling Zhao1
g
1 Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hong Kong
at
2 Institute for Multidisciplinary Sciences, Yokohama National University, Yokohama, Japan
qi-cn.zhao@polyu.edu.hk
W
CF
S
Abstract. Marine concrete is an economical construction material for durable floating structures. However,
the marine concrete may experience severer damage during service in the ocean than inland environment. This
20
paper assessed the durability of marine concrete exposed to actual ocean environment after 20 years of service.
24
The chloride ions distribution and CO2 transport in concrete with varying grades were theoretically simulated
using the past 22 years of weather records. The chloride penetration depth and concrete carbonation front were
then analyzed and validated by field data. The results indicates that the chloride content and carbonation are
highly linked to the microstructures of concrete and the weather conditions at service site. As W/C ratios
decrease, concrete usually becomes denser and their pore structures are changed. Subsequently, moisture, CO2
and chlorides from ocean environment become more difficult to penetrate concrete than ever before. The re-
sults could be used to determine the critical values of harmful ions and the corrosion initiation time for steel-
reinforced concrete structures in service under ocean environment.
Keywords: Marine Concrete, Durability, Chloride Penetration, Carbonation, Ocean Environment.
1 Introduction
Compare to metallic materials, marine concrete is a preferable choice for massive construction of floating struc-
tures due to its low cost and wide availability. Unlike steel, concrete is chemically stable and durable when ex-
posed to marine environments. However, in contrast to inland environments, concrete structures served in ocean
environment, usually undergo more coupled interactions from harmful ions, high relative humidity (RH) and fre-
quent wet-dry cycling, of which the chloride salts are the most dangerous for the steel reinforcements in concrete.
Current research has mainly focused on the chloride penetration into concrete through rapid chloride migration
(RCM) testing in the laboratory [1]. However, evaluating the chloride content and carbonation process in concrete
under actual environments is of significance to assess the long-term durability of maritime concrete structures.
Predicting the chloride profile in concrete after decades of exposure remains a challenge. This paper is a part of
our recent work to address this knowledge gap.
2 Materials and dataset
2.1 Concrete
Four concrete grades were used to conduct the simulations. The micro-structures in different concrete vary, and
they have a significant influence on the chemical resistance to the environmental attack. Detailed information on
concrete mix is listed in Table 1.
Table 1. Concrete mix with varying grades.
Concrete grades W/C Cement (kg/m3) Sand/Gravel ratio

C35 0.39 370 0.67
C40 0.37 400 0.64
C45 0.35 435 0.61
C50 0.33 470 0.56
Note: W/C denotes the water-cement ratio.
2.2 Dataset
The field data from the dry-dock at Pearl Harbor after 18 years of service, including the chloride penetration depth
and carbonation front were obtained [2]. The test chloride depth from extracted cores were 39.4±12.5 mm, the
On
ly
2
fo
rR
carbonation depth was 5.8±3.6 mm. The weather records at Oahu from 2001 to 2023 (i.e., temperature, relative
ea
humidity (RH) and gas CO2), were collected to perform the simulation. We tried to use the true weather data to
conduct the analysis and validate by the experimental results.
din
g
3 Methodology and simulation
at
The moisture, chloride and CO2 coupled interaction and transport in concrete can be calculated according to the
W
multi-scale mass transport theory, as shown in Table 2. More detailed information can referred to Maekawa et al.
CF
[3,4]. The mass transport was calculated using multiscale physico-chemo-mechanical coupling platform.
S
Table 2. Governing equations regarding concrete [3,4].
20
Items Governing equations Eq. Number
24
𝐸𝑖 1 1
Hydration 𝐻𝑖 = 𝐶𝑖 ∙ 𝐻𝑖,𝑇𝑜 (𝑄𝑖 ) exp [− ( − )] (1)
𝑅 𝑇 𝑇0
Micro-pore structures 𝜙(𝑟) = 𝜙𝑙𝑟 + 𝜙𝑔𝑙 [1 − 𝑒𝑥𝑝(−𝐵𝑔𝑙 𝑟)] + 𝜙𝑐𝑝 [1 − 𝑒𝑥𝑝(𝐵𝑐𝑝 𝑟)] (2)
𝜕(𝜌𝑙 𝜙𝑆)
Moisture transport + 𝑑𝑖𝑣𝐽(𝐾∇𝑃𝑙 ) + 𝑄 = 0 (3)
𝜕𝑡
𝑟𝑐 2
𝜌𝑙 𝜙 2
𝐾𝑙 = (∫ 𝑟𝑑𝑉 ) (4)
50𝜂
0
𝜙𝜌𝑣 𝐷0 𝑀𝑤 ℎ
𝐾𝑣 = [(1 − 𝑆)𝐾(ℎ)] ( ) (5)
2.5 𝜌𝑙 𝑅𝑇
𝜕
CO2 transport [𝜙{(1 − 𝑆)𝜌𝑔𝐶𝑂2 } + 𝑆𝜌𝑑𝐶𝑂2 ] + 𝑑𝑖𝑣𝐽𝐶𝑂2 − 𝑄𝐶𝑂2 = 0 (6)
𝜕𝑡
𝑔𝐶𝑂2
𝜙𝐷0 (1 − 𝑆)𝑛
𝐷𝑔𝐶𝑂2 = (7)
Ω (1 − 𝑙𝑚 )⁄{2(𝑟𝑚 − 𝑡𝑚 )}
𝑑𝐶𝑂2
𝜙𝐷0
𝐷𝑑𝐶𝑂2 = 𝑆𝑛 (8)
Ω
𝜕
Chloride transport (𝜙𝑆𝐶𝐶𝑙 ) + 𝑑𝑖𝑣𝐽𝐶𝑙 − 𝑄𝐶𝑙 = 0 (9)
𝜕𝑡
∅𝑆
𝐽𝐶𝑙 = − 𝐷 ∇𝐶 + ∅𝑆𝒖𝐶𝐶𝑙 (10)
Ω 𝐶𝑙 𝐶𝑙
𝐾∇𝑃
𝒖= (11)
ρ∅S
Concrete strength 𝑓𝑐′ = 𝑓∞′ {1 − 𝑒𝑥𝑝(−𝛼𝐷ℎ𝑦𝑑.𝑜𝑢𝑡 𝛽 )} (12)
Concrete shrinkage 𝜎𝑠 = 𝜎𝑠𝑐 + 𝜎𝑠𝑑 (13)
Notes: Please refer to [3,4] for more information on the multi-scale transport theory.
4.1 CO2 ingress and carbonation development
Carbon dioxide (CO2) from the atmosphere comes into the micro-connected pores of concrete, and the dissolved
CO2 react with the hydrated cement products and produce calcium carbonates, decreasing the alkalinity of con-
crete pore solution. This process is called carbonation, which makes the imbedded steels fragile to corrosion.
In general, when the pH values of concrete pore solutions decrease below 9, the concrete block is usually
thought to be carbonated. The pH evolutions in different concrete C35, C40, C45 and C50 are shown in Fig. 1.
On
ly
3
fo
rR
From Fig. 1(a), the C35 carbonation depth was between 20 and 25 mm after 18 years. Only after 1 year of
ea
service, the carbonated C35 concrete reached 10 mm in depth. As service time increased, for example, the car-
bonation depth increased to 20.1 mm after 15.6 years. It can be predicted that a 25 mm-in-depth of C35 concrete
din (b)
will be carbonated.
g
(a)
at
W
CF
S
20
24
(c) (d)
Fig. 1. Concrete carbonation evolution: (a) C35, (b) C40, (c) C45 and (d) C50.
The carbonation depth in C50 is less than 10 mm after 18 years, validated by the field data from extracted
samples. Compared to the results above in Fig. 1(a)-1(d), it can be concluded that concrete with higher concrete
grade, has better resistance to carbonation. As concrete grade increases, the carbonation depth of concrete with
greater grade becomes smaller than concrete with lower grade after the same exposure period. This is because the
higher concrete grades were adopted, the denser concrete and smaller pore structures became.
4.2 Chloride penetration
The chloride distributions in concrete C35, C40, C45 and C50 during service were simulated, as shown in Fig. 2.
From Fig. 2, the chloride contents in concrete at 5 mm depth changed rapidly as the external environment, which
can be found in all cases. The chloride value of C35 concrete ranged from 0.02% to 2.92% during the 23 years of
service. However, as the depth increased, the fluctuations decreased, which are consistent for the four cases.
It should be noted that C35 concrete had the maximum chloride content near the surface when compared with
other three concrete, as illustrated in Fig. 2(a)-2(d). In general, the maximum chloride content at 5 mm depth
decreased with the increasing of concrete grades. This implies that the chemical resistance is significantly affected
by the density and pore structures of concrete.
According to the measured chloride penetration depth using silver nitrate (NT Build 492), the average chloride
depth from 16 cores is 39.4 mm with a COV of 31.7%. The free chloride concentration at which the color changes
in NT Build 492 is assumed as 0.07 mol/L in pore solution for OPC. However, the single value rather than actual
concentrations will lead to significant error [5]. There is a no consensus on the chloride content by cement weight
On
ly
4
fo
rR
at color change [6], In practice, the suggested total chloride range is between 0.5 - 1.0% by the mass of binder [5].
ea
The total chloride concentration of 0.73% by cement weight was adopted in this paper, which agreed well with
test results.
(a) din (b)

g
at
W
CF
S
20
24
(c) (d)
Fig. 2. Chloride content in concrete: (a) C35, (b) C40, (c) C45 and (d) C50.
4.3 Moisture transport

The moisture distributions in difference concrete are summarized in Fig. 3. The RH in pore structures of the near-
surface concrete is periodically changing with the atmospheric environment. As concrete depth increases, the RH
fluctuations gradually decrease. Furthermore, the RH changes also decreased with the increasing of concrete
grades, validated by the data at 98 mm in four cases.
(a) (b)
On
ly
5
fo
(c) rR (d)
ea
din
g
at
W
CF
S
20
24
Fig. 3. RH changes in concrete: (a) C35, (b) C40, (c) C45 and (d) C50.
5 Conclusion remarks
This paper assessed the mass transport in several grades of concrete exposed to actual ocean environment for more
than 20 years, including chloride ions, CO2 and RH variations. Some valuable conclusions can be drawn as fol-
lows.
⚫ The chloride content and carbonation are highly linked to the microstructures of concrete and the weather
conditions at service site.
⚫ As W/C ratios decrease, concrete usually becomes denser and their pore structures are changed. Subse-
quently, moisture, CO2 and chlorides from ocean environment become more difficult to penetrate concrete
than ever before.
⚫ To increase the durability of maritime concrete structures, marine concrete with higher grade should be
adopted in construction.
⚫ The chloride profile and carbonation front in concrete can be used to determine the corrosion initiation time
for steel-reinforced concrete structures in service under ocean environments. Further analyses are needed to
obtain more accurate results.
6 Acknowledgement
The authors thank Dr. Sandra Rayne for her kind guidance during weather data collections. The first author is
sponsored by the Innovation and Technology Fund (ITF) Research Talent Hub of Hong Kong.
References
1. Zhao, Q., Iwama, K., Dai, J.G., Liu, J., Zhang, D., Maekawa, K., Zhao, X.L.: Deterioration modelling of GFRP-
reinforced cement-based concrete infrastructure in service under the natural inland atmospheric environment, Constr.
Build. Mater. 447, 138005 (2024).
2. Ramanathan, S., Benzecry, V., Suraneni, P., Nanni, A.: Condition assessment of concrete and glass fiber reinforced
polymer (GFRP) rebar after 18 years of service life, Case Stud. Constr. Mater. 14 (2021).
3. Maekawa, K., Ishida, T., Kishi, T.: Multi-scale Modeling of Concrete Performance, J. Adv. Concr. Technol. pp. 91–126
(2003).
4. Maekawa, K., Ishida, T., Kishi, T.: Multi-Scale Modeling of Structural Concrete. CRC Press (2008).
5. He, F., Shi, C., Yuan, Q., An, X., Tong, B.: Calculation of chloride concentration at color change boundary of AgNO3
colorimetric measurement, Cem. Concr. Res. 41, 1095–1103 (2011).
6. Drenkard, H., Fischer, C., Sauer, V., Gehlen, C.: On the Chloride Distribution in Concrete and Mortar Samples after an
RCM Test, Materials (Basel). 16 (2023).
On
ly
fo
rR
Fracture phase field method based crack propagation prediction for
ea
circumferential surface cracks in subsea pipelines
din
g
Yangyang Qu1[0000-0003-0428-2220] and Nian-Zhong Chen1,*2[0000-0002-7661-5269]
1
at
School of Civil Engineering, Tianjin University, Tianjin, 300350, China
W
2 State Key Laboratory of Hydraulic Engineering Intelligent Construction and Operation, Tianjin University, Tianjin,
CF
300350, China
nzchen2018@hotmail.com
S
20
Abstract. Subsea pipelines are normally subjected to complex environmental loads and structural fracture has
24
been one of predominant failure modes. In recent years, the fracture phase field theory, based on Griffith's
variational theory of fracture, has proven to be a valuable tool for simulating material fracture behavior, provid-
ing robust capabilities for simulating crack propagation in subsea pipelines. Therefore, within the framework
of the fracture phase field method, this study aims to investigate the influence of circumferential surface crack
defects on subsea pipelines under axial loads. In this paper, a crack propagation prediction is performed for
circumferential surface cracks in subsea pipelines. The corresponding finite element models are then developed
using ABAQUS user subroutines UEL and UMAT. A staggered algorithm is employed in the finite element
analysis to predict the path of crack propagation in subsea pipelines with circumferential surface cracks. Fur-
thermore, a comprehensive study is conducted to investigate the influence of parameters associated with the
initial sizes of circumferential surface cracks on crack propagation and fracture behavior in subsea pipelines.
In addition, critical factors affecting the crack propagation performance of subsea pipelines are explored, and
insights are gained to improve the structural integrity.
Keywords: Subsea pipelines, fracture phase field, crack propagation.
1 Introduction
In recent years, as the global demand for energy has continued to escalate, there has been a corresponding increase
in the number of pipelines installed [1]. Statistical data reveal that defects in pipe materials and welds are often
cited as the primary causes of pipeline incidents [2]. Particularly, the challenges associated with quality control in
onsite welding and the complexity of defect detection and rectification frequently lead to the development of
cracks during pipeline operation, sometimes resulting in severe incidents such as leaks or ruptures. Such events
not only threaten the functionality of the pipelines but may also precipitate significant environmental and eco-
nomic repercussions. Consequently, a thorough understanding of the crack propagation process is essential for
ensuring the safety and reliability of marine engineering structures.
The advancement of computer technology has provided robust technical support for solving numerical simu-
lation issues related to crack propagation prediction, enabling numerous researchers to utilize various numerical
analysis or computational mechanics techniques to study the crack propagation process in pipelines. For instance,
He et al. [3] employed the Extended Finite Element Method (XFEM) to conduct numerical simulations of pitting
crack propagation for surface cracks in monopile foundation; meanwhile, Song and Chen [4] developed an im-
proved cyclic cohesive zone model and predicted the initiation of low-cycle fatigue cracks in submarine pipelines.
Currently, the primary methods adopted for numerical calculations of fracture in practical engineering applica-
tions include the CZM [5] and the XFEM [6]. The cohesive zone model can simulate the continuous process from
crack initiation, evolution to final material fracture [7], however, its application is limited under complex loading
conditions and in handling multidirectional crack propagation because it requires the predefined fracture paths
using cohesive elements. On the other hand, the XFEM, an improvement on the traditional finite element method,
is quite effective in predicting crack propagation issues. Nonetheless, its approach involves treating cracks as
mathematical discontinuities, which may encounter difficulties in simulating the intersection of cracks or their
propagation in complex materials [8].
Based on the theories of Griffith [9] and Irwin [10], the phase field method conceptualizes crack propagation
as an issue of energy balance stability, occurring when the energy release rate reaches a specific threshold. The
essence of this method lies in the use of a continuous field function, represented by a phase-field parameter, to
depict the material's transition from an undamaged state to complete fracture. This parameter illustrates the degree
of internal damage within the material. Compared to traditional approaches, the phase field method can
On
ly
2
fo
rR
automatically track crack paths through the evolution of the phase-field parameter, thus efficiently simulating
complex crack propagation behaviors with reduced computational complexity. This method provides a novel and
ea
powerful tool for the fracture analysis of engineering structures.
din
Consequently, this paper, based on the phase field fracture model undertakes a secondary development within
the Abaqus and the Fortran compiler. Utilizing a staggered algorithm for solving [11], it investigates the behavior
g
of crack propagation in pipelines under the influence of axial loads.
at
W
2 Fracture phase field model
2.1 Foundation of phase field model

CF
S
The phase field method operates by introducing an additional degree of freedom, denoted as 𝑑, which represents
20
a continuum ranging from undamaged to completely fractured material. The non-smooth topology of a crack is
24
typically approximated using an exponential function.
| |/
𝑑 𝑥 𝑒 (1)
where, 𝑙 represents the length scale parameter, and 𝑑 𝑥 characterizes the regularization of the crack topology.
Eq. (1) which satisfies 𝑑 𝑥 1, and 𝑑 ∞ 0 at the limits, is the solution to the homogeneous differential
equation within the domain 𝛺:
𝑑 𝑥 𝑙 𝑑 ′′ 𝑥 0 in Ω (2)
The expression for the crack surface within the phase field model can be formalized as follows:
𝛤 𝑑 1/𝑙 ⋅ 𝐼 𝑑 𝛾 𝑑, 𝑑 ′ 𝑑𝑉 (3)
The one-dimensional crack surface density function, denoted as 𝛾 𝑑, 𝑑 ′ , characterizes the localized damage
within a linear framework. In multiple dimensions, this function is generalized to accommodate the complexity
of the spatial domain 𝛺, and is expressed as follows:
𝛾 𝑑, 𝛻𝑑 1/ 2𝑙 ⋅𝑑 𝑙 /2|𝛻𝑑| (4)
To integrate the fracture phase field with the deformation problem, Francfort and Marigo [12] proposed an
internal potential energy functional that incorporates both strain energy and crack surface energy.
𝛱int 𝜓 𝜺 𝒖 , 𝑑 𝑑𝑉 𝑔 𝛾 𝑑, 𝛻𝑑 𝑑𝑉 (5)
where 𝜴 𝝍 𝜺 𝒖 , 𝒅 𝒅𝑽 the strain energy, 𝝍 𝜺 𝒖 , 𝒅 the strain energy density, 𝑔 𝛾 𝑑, 𝛻𝑑 𝑑𝑉 the fracture
energy and 𝑔 the critical energy release rate.
The strain energy density is typically expressed in the following isotropic form:
𝜓 𝜺, 𝑑 𝑔 𝑑 ⋅𝜓 𝜺 (6)
where 𝑔 𝑑 1 𝑑 𝑘 the parabolic degenerate function, k a small number (10-7) responsible for the sta-
bility of the solution.
The elastic energy density can be calculated as:
𝜓 𝜺 𝜺 𝑪𝟎 𝜺 (7)
where 𝑪𝟎 the linear elastic stiffness matrix of the material, 𝜺 the strain.
𝜺 𝛻𝒖 𝛻𝒖 (8)
where 𝒖 the displacement vector.

Due to damage, elastic energy deteriorates with damage. The true stress tensor in the phase field model con-
sidering damage is as follows:
𝝈 𝑔 𝑑 𝝈 1 𝑑 𝑘 𝑪𝟎 𝜺 (9)
For stiffness there is
On
ly
3
fo
rR
𝑪 𝑔 𝑑 ⋅ 𝑪𝟎 (10)
ea
It is evident that the damage variables, as represented by the phase field, exert a direct influence on the stress
din
and stiffness of the material. When the value of the damage variable reaches 1, it signifies complete material
failure at that location, resulting in the stress or stiffness of the element reducing to zero.
g
The external potential energy is expressed as:
𝛱ext 𝑃 𝒖 𝜸 ⋅ 𝒖 𝑑𝑉
at
𝒕 ⋅ 𝒖 𝑑𝐴 (11)
where  and t represent volume force and boundary force respectively.

W
CF
The Staggered Solution approach [11] was employed to address the phase field fracture problem. This method
S
involves first achieving equilibrium in the displacement field through iterative solutions, followed by iterative
solutions of the phase field. Both the displacement field and the phase field utilized the Newton-Raphson algo-
rithm for their iterative solutions.
20
24
The function for solving the crack topology is:
𝛱 𝑔 𝛾 𝑑, 𝛻𝑑 1 𝑑 𝐻 𝑑𝑉 (12)
Introduce history variables as part of the model to capture the evolving damage state:
𝜓 𝜺 𝜓 𝜺 𝐻
𝐻 (13)
𝐻
where 𝐻 represents the energy history at the nth step, calculated as the maximum strain energy density, experi-
enced by the material up to that step. The solution to the phase field fracture problem must satisfy the Karush-
Kuhn-Tucker (KKT) conditions:
𝜓 𝐻 0, 𝐻 0, 𝐻 𝜓 𝐻 0 (14)
With the damage variable 𝑑 held constant, the displacement field is computed using the equilibrium equations
of the mechanical system:
𝛱 𝜓 𝒖, 𝑑 𝜸𝒖 𝑑𝑉 𝒕 ⋅ 𝒖𝑑𝐴 (15)
Once equilibrium in the displacement field is achieved, the phase field is updated iteratively. The iterative
solution process for the phase field involves calculating a new phase field, 𝑑, at time 𝑡 based on its state at
time 𝑡 .
𝑑 𝐴𝑟𝑔 𝑖𝑛𝑓 𝑔 𝛾 𝑑, 𝛻𝑑 1 𝑑 𝐻 𝑑𝑉 (166)
The linear approximate solution for the phase field can be expressed as:
𝑲 𝑑 𝒓 (17)
where 𝑑 represents the unknown variable that contains the new value of the phase field at each integration
point, 𝒓 the residual vector, 𝑲 the tangential stiffness at time 𝑡 .
In order to calculate the displacement field at time 𝑡 , the phase field value at time 𝑡 is used:
𝒖 𝐴𝑟𝑔 𝑖𝑛𝑓 𝜓 𝒖, 𝑑 𝜸 ⋅ 𝒖 𝑑𝑉 𝒕 ⋅ 𝒖 𝑑𝐴 (18)
Similar to the phase field, the approximate linearization solution of the displacement field is:
𝑲 𝒖 𝒓 (19)
To implement the scheme in ABAQUS, two element types are used in a hierarchical manner: phase field ele-
ment and displacement field element, each layer connected at the same node but with different degrees of freedom.
The phase field element has only one degree of freedom, the phase field variable d. The degree of freedom of the
displacement field element depends on the dimension of the element. Newton-Raphson algorithm was used for
iterative solution, and the tangential stiffness matrix and residual vector were iteratively updated:
On
ly
4
fo
rR 𝑲
0
0
𝑲
𝑑
𝒖
𝒓
𝒓
(20)
e a
3 Numerical simulation
din
g
at
3.1 Modeling
W
To explore the impact of initial circumferential surface cracks on pipeline safety under identical loading condi-
tions, a simplified pipeline model featuring initial circumferential surface cracks was developed using ABAQUS.
CF
The phase field fracture model discussed in this study does not account for the influence of welding residual
S
stresses and is exclusively solved within the elastic range. Therefore, it is assumed that the weld seam and the
pipeline possess equal strength matching, and no specific model incorporating the weld seam is established. Ad-
20
ditionally, material plasticity is disregarded, and the pipeline material is treated as a linear elastic material. The
24
specific parameters are detailed in Table 1. Surface cracks are modeled as semi-elliptical and semi-circular surface
cracks [13], as illustrated in Fig.1.
(a)
(b)
Fig. 1. Regularization of surface defects. (a) For h < l/2, the surface defect is represented as a semi-elliptical surface crack.
(b) For h≥l/2 (pertinent to fracture considerations), the surface defect is modeled as a semi-circular surface crack.
Due to the symmetry of the pipeline, a quarter-section model was utilized for the finite element analysis to
streamline computations without sacrificing accuracy. The diameter D of the pipeline was 2m, the wall thickness
t was 0.1m, the length of the pipeline was l=2m, and the length of the calculation model was 2D. The initial crack
model was placed on the inner surface at the midpoint of the pipeline. The orientation of the initial crack was
circumferential, as shown in Fig. 2 (a), where t was the pipe wall thickness, a was the crack depth, and c was the
crack half-length.
Table 1. Model Attributes
Elastic modulus E/MPa Poisson's ratio Critical energy release rate gc/Nꞏmm-1
200 000 0.3 150
One end of the pipeline model is subjected to fixed constraints, encompassing all translational and rotational
motions in three directions. At the opposite end from the fixed constraints, displacement loads are applied, which
increase uniformly along the pipeline's axial direction. The symmetric plane of the model is equipped with sym-
metric boundary conditions, as shown in Fig.2 (b). The Newton-Raphson algorithm is utilized for the iterative
solution of the model. The model is solved in fixed incremental steps within ABAQUS, with a maximum of 200
incremental steps allowed. Each step is sized at 0.01, facilitating detailed tracking of response changes over the
loading process. Standard convergence criteria for both force and displacement are employed to ensure that each
simulation step accurately reflects the physical behavior of the pipeline under load. The rate of axial displacement
applied to the pipeline is set at 𝛥𝑢 = 0.002m per incremental step, mimicking realistic load application rates
observed in pipeline operations.
To achieve an optimal balance between computational accuracy and efficiency, this study employs a strategic
mesh refinement approach that focuses on regions prone to high stress concentrations, such as areas with cracks.
The global mesh seed size for the entire pipeline is set at 0.1 meters. For regions around the surface cracks, a
much finer local seed size of 0.005 meters is implemented. The C3D8 element, a linear brick element with eight
On
ly
5
fo
rR
nodes, is utilized throughout the model. As depicted in Fig. 2(c), the mesh division demonstrates the application
of different seed sizes.
ea
din
g
at
W
(a) (b) CF (c)
Fig. 2. Pipeline model specification. (a) Schematic diagram of cracks on the inner surface of the pipeline. (b) Boundary con-
S
ditions. (c) Mesh division.
20
3.2 Numerical examples 24
To investigate the influence of initial surface cracks on the axial reaction of pipelines under consistent loading
conditions, simulations were conducted with pipelines modeled with varying sizes of surface cracks. The models
were all subjected to the same loading conditions to isolate the effects of crack size on the pipeline's performance.
The criteria for defining failure included the material reaching its fracture toughness limit, indicating a critical
threshold where the material can no longer resist crack propagation, and the stress levels in the simulation exceed-
ing the material's maximum stress capacity.
In the study of semi-elliptical surface cracks, the crack depth was set to the specification limit value of a=0.7t,
where a>0.7t would categorize the crack as a penetration crack. The lengths of these cracks, denoted as 2c, were
varied at 0.1m, 0.12m, 0.14m, 0.16m, and 0.18m. For semi-circular surface cracks, the lengths 2c were 0.1m,
0.12m, and 0.14m. The effects of these initial defects on pipeline behavior under axial loads were documented in
Fig.3 (a) and (b), which illustrate the reaction and displacement curves at different pipeline boundaries. The results
clearly demonstrate that as the length of the initial surface cracks increases, the pipeline's ability to resist axial
loads diminishes, leading to a gradual decrease in the maximum reaction force. This trend highlights the critical
impact of crack size on the structural integrity and load-bearing capacity of pipelines.
2c=0.10m 2c=0.10m
5500 2c=0.12m 5500 2c=0.12m
2c=0.14m 2c=0.14m
5200 Semi-elliptical crack (a = 0.07m)
2c=0.16m
Semi-circular crack (a = c)
4400 2c=0.18m 4400
Maximum reaction load/N
5100
Reaction load/N
Reaction load/N
3300 3300
5000
2200 2200
4900
1100 1100 4800
0 0 4700
0.00 0.07 0.14 0.21 0.28 0.35 0.00 0.07 0.14 0.21 0.28 0.35 0.10 0.12 0.14 0.16 0.18
Crack length (2c)/m

Displacement/m Displacement/m
(a) (b) (c)

Fig. 3. Numerical simulation result. (a)Reaction-displacement curve of pipe with semi-elliptical crack; (b) Reaction-dis-
placement curve of pipe with semi-circular crack;(c) Maximum reaction-crack length curve.
(a) (b) (c)

Fig. 4. a=0.07m, 2c=0.1m, local evolution of circumferential surface cracks in the pipeline. (a) Initiation of crack. (b) Expan-
sion along the circumferential direction. (c) Expansion through the wall thickness.
As discussed earlier, the phase field variable d in the model quantifies the state of the material: a value of d=0
indicates that the material is intact, whereas d=1 signifies complete material failure. It is generally accepted that
when d≥0.9, cracks are considered to have formed in that region; these areas are highlighted in red in Fig. 4 and
Fig.5 to denote the crack zones.
On
ly
6
fo
rR
For the local crack growth, the simulation results reveal that with a fixed crack depth at a=0.7t, different be-
haviors emerge based on the relationship between a and c. When a≥c, the crack initially propagates in the cir-
ea
cumferential direction of the pipeline to form a crack with a length equal to half the crack depth. Following this,
din
the crack propagates along the wall thickness direction until it penetrates the wall. Conversely, when a≤c, the
crack spreads uniformly in all directions. The penetration occurs first in the direction of wall thickness, followed
g
by rapid expansion along the pipeline's circumferential direction.
at
W
CF
(a) (b)
S (c)
20
Fig. 5. a=0.07m, 2c=0.2m, local evolution process of circumferential surface crack of the pipeline. (a) Crack initiation. (b)
24
Expansion through the wall thickness to penetration. (c) Post-penetration expansion along the circumferential direction.
4 Conclusions
Based on the fracture phase field method, the behavior of crack propagation in a pipeline with circumferential
surface cracks under axial forces was investigated. Numerical simulation was conducted on pipelines with semi-
elliptical and semi-circular initial circumferential surface cracks under axial loads. The study focused on the crack
propagation and mechanical response of the pipeline with different initial crack sizes. Based on the numerical
results, the applicability of the fracture phase field model for predicting crack growth in offshore structures was
verified. This study shows:
1) The fracture process and the crack propagation path of surface cracks of pipelines can be effectively esti-
mated by the fracture phase field method.
2) As depth of the surface crack increases, the pipeline’s resistance to tensile forces is gradually weakened, and
the maximum reaction force correspondingly is decreased.
3) When the depth of the surface crack exceeds its half-length, stress concentration primarily occurs at the
circumferential edge of the crack so that the crack tends to propagate along the circumferential direction.
References
1. Zhang, Y., Zheng, M., An, C., Seo, J. K., Pasqualino, I. P., Lim, F., Duan, M.: A review of the integrity management of
subsea production systems: Inspection and monitoring methods. Ships and Offshore Structures 14(8), 789-803 (2019).
2. Vishnuvardhan, S., Murthy, A. R., Choudhary, A.: A review on pipeline failures, defects in pipelines and their assessment
and fatigue life prediction methods. International Journal of Pressure Vessels and Piping 201, 104853 (2023).
3. He, M., Pang, M., Chen, N. Z.: Engineering critical assessment (ECA) for monopile foundation of an offshore wind
turbine subjected to pitting. Ocean Engineering 285, 115400 (2023).
4. Song, Z. H., Chen, N. Z.: A modified cyclic cohesive zone model for low-cycle fatigue crack initiation prediction for
subsea pipelines under mode I loading. Ocean Engineering 276: 114200 (2023).
5. Wciślik, W., Pała, T.: Selected aspects of cohesive zone modeling in fracture mechanics. Metals 11(2), 302. (2021).
6. Cervera, M., Barbat, G. B., Chiumenti, M., Wu, J. Y.: A comparative review of XFEM, mixed FEM and phase-field
models for quasi-brittle cracking. Archives of Computational Methods in Engineering 29(2), 1009-1083 (2022).
7. Song, Z. H., Chen, N. Z.: A modified cyclic cohesive zone model for fatigue crack initiation prediction for mooring chain
steels. Engineering Fracture Mechanics 300, 109985 (2024).
8. Wu, J. Y., Nguyen, V. P., Nguyen, C. T., Sutula, D., Sinaie, S., Bordas, S. P.: Phase-field modeling of fracture. Advances
in applied mechanics, 53, 1-183.K (2020).
9. Griffith, A. A.: VI. The phenomena of rupture and flow in solids. Philosophical transactions of the royal society of london.
Series A, containing papers of a mathematical or physical character 221(582-593), 163-198 (1921).
10. Irwin, G. R.: Fracture//Elasticity and Plasticity/Ed. by S. Flügge (1958).
11. Molnár, G., Gravouil, A.: 2D and 3D Abaqus implementation of a robust staggered phase-field solution for modeling
brittle fracture. Finite Elements in Analysis and Design 130, 27-38 (2017).
12. Francfort, G. A., & Marigo, J. J.: Revisiting brittle fracture as an energy minimization problem. Journal of the Mechanics
and Physics of Solids 46(8), 1319-1342 (1998).
13. The British Standards Institution.: Guide to methods for assessing the acceptability of flaws in metallic structures: BS
7910: 2013. London, UK (2013).
On
ly
fo
rR
Hydrogen-assisted Dislocation Pinning and Depinning on Fatigue Crack
ea
Propagation in Subsea Pipeline Steels
din
g
Tingsen Zheng1[0000-0003-2868-9623] and Nian-Zhong Chen1,2[0000-0002-7661-5269]
1
at
School of Civil Engineering, Tianjin University, Tianjin, China
W
2 State Key Laboratory of Hydraulic Engineering Intelligent Construction and Operation, Tianjin University, Tianjin, China
CF
nzchen2018@hotmail.com
S
20
Abstract. The resolution of hydrogen embrittlement (HE) and hydrogen-assisted fatigue assessment in subsea
pipelines is a critical technology for ensuring the large-scale development of offshore hydrogen industry.
24
Hydrogen-assisted dislocation pinning near the crack tip is normally regarded as one of the key mechanisms
of HE. However, there is a lack of a methodology to predict the hydrogen-assisted fatigue for pipeline steel
influenced by dislocation pinning. Therefore, a piecewise linear elastic-plastic constitutive equation for
pipeline steel in gaseous hydrogen environment is developed, in which the increases of yield strength due to
both hydrogen-assisted dislocation pinning and the hardening of the pipeline after depinning are taken into
account. To demonstrate the capability of the proposed model, it is introduced into a cyclic cohesive zone
model (CCZM) for predicting hydrogen-assisted fatigue crack propagation of pipeline steel. The results shows
that the proposed constitutive equation can well reflect the localization of plastic deformation and fatigue crack
propagation acceleration caused by hydrogen-assisted HE.
Keywords: Hydrogen embrittlement, Fatigue crack propagation, Dislocation pinning, Cyclic cohesive zone
model.
1 Introduction
The permeation of hydrogen atoms has a significant detrimental effect on the mechanical properties of pipeline
steel, which is commonly referred to as hydrogen embrittlement (HE). The initiation and rapid propagation of
cracks will occur under loads well below the material's design strength due to HE. Particularly in subsea hydrogen
pipelines or oil and gas pipelines, hydrogen-assisted fatigue crack stands as one of the primary failure modes [1].
Consequently, the assessment for hydrogen-assisted fatigue crack in subsea pipeline steel has become a critical
issue in the subsea oil and gas industry and the development of hydrogen energy.
Nowadays, many theories for HE have been proposed. However, due to the complex mechanisms of HE,
significant controversies remain regarding its interpretation [2]. Among the existing HE theories, the widely
accepted ones include the hydrogen-enhanced de-cohesion (HEDE) theory and hydrogen-enhanced localized
plasticity (HELP) theory [3]. Significant progress has been made in hydrogen-assisted fatigue crack propagation
prediction models based on fracture mechanics theory and these HE theories. A series of studies by Cheng and
Chen [4-6] proposed corrosion-crack interaction models, which is successfully applied to predict fatigue crack
propagation behavior in API pipeline steel under hydrogen and corrosion conditions. In recent years, the cyclic
cohesive zone model (CCZM) exhibits good compatibility with the HEDE theory, which provides a new solution
for simulating hydrogen-assisted fatigue crack. Song and Chen [7,8] developed a CCZM for predicting fatigue
crack initiation in steel, which is successfully applied to predict fatigue crack initiation life of subsea pipelines
and mooring chains. Moriconi et al. [9] introduced an improved hydrogen diffusion model into the cohesive zone
model and conducted simulations and experimental comparisons on high-strength steel in hydrogen environments.
The results showed good agreement between the model predictions and experimental results in low-pressure
hydrogen environments. However, the model struggled to predict changes in fatigue crack propagation rates under
high hydrogen concentrations, indicating limitations in the current cohesive zone models for predicting HIFCP.
To effetely reflect the mechanisms of hydrogen-assisted fatigue crack propagation, a key phenomenon is the
influence of HE on the plastic zone at the crack tip and the dislocation mobility may be a critical parameter [10].
Nagumo [11] reviewed various HE theories in steel and found that the strain-induced evolution of microdefects
such as vacancies and dislocations is affected by hydrogen, which is based on energy stabilization and localized
plastic deformation. Xie et al. [12] conducted quantitative mechanical tests under environmental transmission
electron microscopy. The results show that strong dislocation pinning occurred in aluminum metal under
conditions of excessive vacancy. As seen from the literature review, the pinning-depinning of dislocations has a
On
ly
2
fo
rR
universal impact on hydrogen-assisted fatigue crack propagation across various metals. However, the pinning-
depinning mechanism has not been widely considered in fatigue damage prediction using CCZM.
ea
Therefore, this paper establishes a piecewise elastoplastic constitutive model for pipeline steel in gaseous
din
hydrogen environments, where the hardening of pipeline steel caused by hydrogen-assisted pinning and depinning
of dislocation is considered. To validate the effectiveness of proposed model, it is combined with the CCZM
g
developed by Zheng and Chen [13], in which the degradation of cohesive strength and accumulated cohesive
at
length is taken into account. A comparison study between predicated hydrogen-assisted fatigue crack propagation
rate of X52 pipeline steel under hydrogen conditions, which are achieve by CCZM with different elastic-plastic
W
constitutive models, is then performed. The results show that the established dislocation pinning-depinning model
CF
effectively reflects plastic localization and accelerated fatigue crack propagation under high stress levels induced
by hydrogen embrittlement.
S
2 Methodology
20
24
The CCZM used for predicting fatigue crack propagation of pipeline steel, as well as the dislocation pinning-
depinning model proposed in this paper, are introduced in this section.
2.1 Hydrogen diffusion model

In metallic crystals, the total concentration of hydrogen atoms CH, can be expressed as the sum of the lattice
hydrogen concentration CL, and the trapped hydrogen concentration CT, within the metallic crystal [14, 15]:
𝐶 𝐶 𝐶 𝜃 𝑁 (1)
where θL is the lattice hydrogen coverage, NL the lattice hydrogen density, NT the trap hydrogen density, EB is the
trap binding energy with a value of 35 kJ/mol in this paper, R the universal gas constant with a value of 8.3145
J/(molꞏK), T the absolute temperature. Therefore, the mass transport equation for hydrogen atoms in the metallic
crystal can be transformed into a form containing only CL [16]:
𝐷∇ 𝐶 ∇ ∇𝜎 (2)
where DL is the lattice diffusion coefficient of hydrogen in the metal, De the effective diffusion coefficient of
hydrogen in the metal, 𝑉 is the average molar volume of hydrogen, ∇ is the Hamiltonian, and σh is the
hydrostatic stress. The determination of the diffusion state of hydrogen within the metal under the influence of
internal stress fields can be achieved by solved Eq. (2).
2.2 Cyclic cohesive zone model with HE considering

The cyclic cohesive zone model (CCZM) indicates that the formation of fracture is a progressive process, where
the separation of crack surfaces occurs within a cohesive zone at the crack tip. The traction force at the crack tip
in this region can be expressed as a function of the crack separation displacement, which is so-call the traction-
separation law (TSL). A typical exponential TSL for pure mode I cracks is given as [17]:
exp 1 (3)
,
where Tn is the normal traction force, σmax,0 the initial cohesive strength, which is the cohesive strength of the
metal before degradation of material, δn the normal separation displacement, δ0 the normal characteristic cohesive.
Fig. 1 shows the relationship between the cohesive zone and TSL at the crack tip.
Fig. 1. Diagram of cohesive zone and traction separation law at crack tip.
On
ly
3
fo
rR
The damage evolution equation for CCZM in which impact of HE for fatigue crack is accounted for can be given
as [13]:
ea
𝛿n,𝑡 𝛿n,𝑡 ∆𝑡
din
𝐷 𝐻 𝛿n,𝑡 𝛿n,𝑡 ∆𝑡 d𝑡 𝛿 d𝑡 (4)
, , ,
g
where D is the total damage of the material, δn,t the normal separation at time step t, δn,t-∆t the normal separation
at
at time step t-∆t, ∆t the increment of time step, δΣ,C is the degraded accumulated cohesive length, δΣ,0 the initial
accumulated cohesive length, σmax,C is the degraded cohesive strength, σf the endurance limit of the cohesive zone,
W
H the Heaviside function. Failure of the material is deemed to occur when the total damage D is equal to 1.
CF
Degradation of cohesive strength σmax and accumulated cohesive length δΣ due to hydrogen under gaseous
hydrogen conditions is considered in the CCZM to capture the HE on the material properties of pipeline steel. The
S
degradation of σmax is described using the quadratic relationship between σmax and hydrogen coverage θH in terms
20
of first principles, which is given as [18]:
24
𝜎 , 0.1687𝜃 1.0467𝜃 1 𝜎 , (5)
in which
𝜃 ∆ (6)
where ΔG is the Gibbs free energy difference between any microstructural interface and the surrounding material.
Generally, once the hydrogen concentration exceeds a certain critical value, further increases in hydrogen
concentration no longer significantly affect the degradation of metal material properties, indicating a saturation
effect in HE [19]. The degradation pattern of pipeline steel due to HE is similar to the process described by the
Logistic formula, which typically characterizes a process that begins with slow growth, followed by rapid growth,
and eventually approaches a plateau. The inverse Logistic-based model for of degradation of δΣ established by
Zheng and Chen [13] is used in this paper, which is given as:
𝛿0
.
, 𝛿Σ,0
𝛿0 0.1 (7)
,
𝛿Σ,0
where, , s the material parameter that is determined based on the hydrogen concentration at which the saturation
effect occurs, set at s = 107.62 for pipeline steel.
2.3 Pinning-depinning model

Hydrogen atoms will occupy vacancies in the metallic crystal, which impedes dislocation movement and leads to
dislocation pinning. Unfortunately, there is no such constitutive model capable of effectively describing the
influence of hydrogen-induced dislocation pinning-depinning on material properties. Therefore, this paper
proposed a simplified dislocation pinning-depinning model for describing elastic-plastic constitutive behavior of
pipeline steel under hydrogen environment. Fig. 2 shows the comparison between proposed pinning-depinning
model, elastic-perfectly plastic model and Hollomon hardening model [13]. The pinning-depinning model is in
terms of the stress-strain relationship of hydrogen-charged metals from a series nanoindentation experiments [20].
Fig. 2. Dislocation pinning-depinning model for describing elastic-plastic constitutive behavior of X52 pipeline steel.
As shown in Fig. 2, the hydrogen-induced dislocation pinning-depinning model indicates a material elastic-plastic
behavior consisting of four stages: pinning hardening, plateau, depinning, and depinning hardening. This is
On
ly
4
fo
rR
consistent with the experimental results of Li et al [20]. To effectively describe this proposed constitutive model,
three key parameters are defined: depinning stress σdp, increment of depinning strain Δ ε dp, and depinning
ea
hardening modulus Edp, which can be given as:
din 𝜎 2𝜎 (8)
Δ𝜀
g 𝜎 /𝐸 (9)
at
W
𝐸 𝐸/10 (10)
CF
where 𝜎 the yield strength of pipeline steel, 𝐸 the Young’s modulus of pipeline steel.
S
20
3 Finite element analysis
24
A finite element model of a compact tension (CT) specimen made of X52 pipeline steel (as shown in Fig. 3) is
established to predict hydrogen-induced fatigue crack propagation. Due to the symmetry of the CT specimen, only
half of the model is constructed. The dimensions of the CT specimen are in terms of ASTM E647 standard [21],
with a characteristic dimension W = 25.4 mm. The CPE4RT element is used to mesh the steel, and a layer of
COH2D4 elements is applied along the predetermined fatigue crack growth path. The parameters for the cyclic
cohesive zone model (CCZM) are set as follows: δ0 = 0.2 mm, δΣ,0 = 3 mm, σmax,0 = 4σy, and σf = σy.
Fig. 3. Finite element model of compact tension specimen that is made of X52 pipeline steel.
Different stress intensity factor ranges ∆K at crack tip is selected which is maintained constant during analysis.
The load ratio is set to 0.5, and cyclic loading is applied to the upper edge of the pin hole in the CT specimen. The
initial hydrogen concentration CL,0 is equal to 0.055 mg/kg. The crack growth simulation is performed until the
crack propagated beyond the mesh refinement zone. The ∆K values and corresponding hydrogen-induced fatigue
crack growth rates da/dN are recorded, where a is the crack length and N the number of load cycles. The material
properties for X52 steel are presented in Table 1.
Table 1. Material properties of X52 pipeline steel.
Properties value
Ultimate Strength (MPa) 526
Density (kg/m3) 7850
Yield strength (MPa) 325
Young’s modulus 𝐸 (GPa) 206
Poisson’s ratio 𝜈 0.3
Fig. 4 shows the distribution of equivalent plastic strain at a certain moment during the analysis when different
elastic-plastic constitutive models are employed. As shown in Fig. 4, when the proposed dislocation pinning-
depinning model is used to simulate the elastic-plastic behavior of X52 pipeline steel, the plastic zone area near
the crack tip is smaller than that obtained using the Hollomon hardening model. At the same time, the maximum
equivalent plastic strain at the crack tip is increased. This is due to the significant hardening of X52 pipeline steel
caused by hydrogen-induced dislocation pinning.
On
ly
5
fo
rR
ea
din
g
at
W
CF
(a) (b) (c)
S
Fig. 4. Distribution of equivalent plastic strain at crack tip:(a) pinning-depinning model; (b) elastic-perfectly plastic model;(c)
Hollomon hardening model.
20
24
However, the hardening effect of the dislocation pinning-depinning model after depinning is lower than that of
the Hollomon hardening model. Consequently, the pinning-depinning model predicts a higher maximum plastic
strain than the Hollomon model. Due to the lack of consideration for material hardening, the elastic-perfectly
plastic model predicts a higher maximum equivalent plastic strain compared to the results predicted by other two
models. It can be seen that the proposed hydrogen-induced dislocation pinning-depinning model effectively
reflects the hydrogen-induced plastic localization in pipeline steel, which is consistent with the HELP theory.
Fig. 5 shows a comparison of predicted hydrogen-induced fatigue crack growth rates using different elastic-
plastic constitutive models. As shown in Fig. 5, generally, the elastic-perfectly plastic model provides a relatively
conservative prediction due to the lack of consideration for material hardening. The results achieved by using
Hollomon hardening model mostly agrees well with the experimental results. However, the predictions of the
Hollomon hardening model are slightly lower than the lowest experimental values when there is a higher ∆K,
which may be unfavorable for structural safety assessments of subsea pipeline. When the proposed dislocation
pinning-depinning model is applied, the predicted hydrogen-induced fatigue crack growth rates are more
consistent with the experimental results compared to that predicted by the elastic-perfectly plastic model.
Additionally, different from the results of Hollomon hardening model, the predictions of dislocation pinning-
depinning model does not yield excessively low predictions at higher ∆K values. In summary, the proposed
dislocation pinning-depinning model will better reflect the mechanism of hydrogen-induced fatigue crack growth
in X52 pipeline steel.
Fig. 5. Comparison between model prediction and experimental data of hydrogen induced fatigue crack propagation.
5 Conclusions
A piecewise elastoplastic constitutive model for pipeline steel in gaseous hydrogen environment was established
where the hardening of pipeline steel induced by hydrogen-assisted pinning and depinning of dislocation is
considered. The proposed pinning-depinning model was combined with the CCZM for predicting hydrogen-
induced fatigue crack propagation. A comparison study between rates of hydrogen-assisted fatigue crack
propagation of X52 pipeline steel achieved by CCZM with the use of proposed pinning-depinning model, elastic-
perfectly plastic model and Hollomon hardening model was then performed. The results show that the proposed
On
ly
6
fo
rR
dislocation pinning-depinning model can effectively capture the plastic localization and acceleration of fatigue
crack propagation under high stress levels due to hydrogen embrittlement.
ea
References
din
g
at
1. Kushida, T., Nose, K., Asahi, H., Kimura, M., Yamane, Y., Endo, S., Kawano, H.: Effects of metallurgical factors and
test conditions on near neutral pH SCC of pipeline steels. In NACE CORROSION, pp. NACE-01213. Houston, Texas
W
(2001).
2. Djukic, M. B., Bakic, G. M., Zeravcic, V. S., Sedmak, A., Rajicic, B.: The synergistic action and interplay of hydrogen
CF
embrittlement mechanisms in steels and iron: Localized plasticity and decohesion. Engineering Fracture Mechanics, 216,
106528 (2019).
S
3. Nagao, A., Smith, C. D., Dadfarnia, M., Sofronis, P., Robertson, I. M.: The role of hydrogen in hydrogen embrittlement
20
fracture of lath martensitic steel. Acta Materialia, 60(13-14), 5182-5189 (2012).
4. Cheng, A., Chen, N. Z. Fatigue crack propagation modelling for pipeline carbon steels under gaseous hydrogen conditions.
International Journal of Fatigue, 96, 152-161 (2017).
24
5. Cheng, A., Chen, N. Z.: Corrosion fatigue crack propagation modelling for subsea pipeline steels. Ocean Engineering,
142, 10-19 (2017).
6. Cheng, A., Chen, N. Z.: An extended engineering critical assessment for corrosion fatigue of subsea pipeline steels.
Engineering Failure Analysis, 84, 262-275 (2018).
7. Song, Z. H., Chen, N. Z.: A modified cyclic cohesive zone model for low-cycle fatigue crack initiation prediction for
subsea pipelines under mode I loading. Ocean Engineering, 276, 114200 (2023).
8. Song, Z. H., Chen, N. Z.: A modified cyclic cohesive zone model for fatigue crack initiation prediction for mooring chain
steels. Engineering Fracture Mechanics, 300, 109985 (2024).
9. Moriconi, C., Hénaff, G., Halm, D. Cohesive zone modeling of fatigue crack propagation assisted by gaseous hydrogen
in metals. International journal of fatigue, 68, 56-66 (2014).
10. Barnoush, A., Kheradmand, N., Hajilou, T.: Correlation between the hydrogen chemical potential and pop-in load during
in situ electrochemical nanoindentation. Scripta Materialia, 108, 76-79 (2015).
11. Nagumo, M.: Function of hydrogen in embrittlement of high-strength steels. Iron and Steel Institute of Japan International,
41(6), 590-598 (2001).
12. Xie, D., Li, S., Li, M., Wang, Z., Gumbsch, P., Sun, J., Ma, E., Li, J., Shan, Z.: Hydrogenated vacancies lock dislocations
in aluminium. Nature Communications, 7(1), 13341 (2016).
13. Zheng, T., Chen, N. Z.: A cyclic cohesive zone model for predicting hydrogen assisted fatigue crack propagation (FCG)
of subsea pipeline steels. International Journal of Fatigue, 173, 107707 (2023).
14. Sofronis, P., & McMeeking, R. M.: Numerical analysis of hydrogen transport near a blunting crack tip. Journal of the
Mechanics and Physics of Solids, 37(3), 317-350 (1989).
15. Oriani, R. A., & Josephic, P. H.: Equilibrium aspects of hydrogen-assisted cracking of steels. Acta metallurgica, 22(9),
1065-1074 (1974).
16. Krom, A. H., Koers, R. W., & Bakker, A. D.: Hydrogen transport near a blunting crack tip. Journal of the Mechanics and
Physics of Solids, 47(4), 971-992 (1999).
17. Xu, X. P., & Needleman, A.: Void nucleation by inclusion debonding in a crystal matrix. Modelling and Simulation in
Materials Science and engineering, 1(2), 111 (1993).
18. Serebrinsky, S., Carter, E. A., & Ortiz, M.: A quantum-mechanically informed continuum model of hydrogen
embrittlement. Journal of the Mechanics and Physics of Solids, 52(10), 2403-2430 (2004).
19. Wang, M., Akiyama, E., & Tsuzaki, K.: Effect of hydrogen on the fracture behavior of high strength steel during slow
strain rate test. Corrosion science, 49(11), 4081-4097 (2007).
20. Li, Y., Morgan, T. W., Terentyev, D., Ryelandt, S., Favache, A., Wang, S., Wirtz, M., Hoefnagels, J.P.M., van Dommelen,
J.A.W., De Temmerman, G., Verbeken, K., Geers, M. G. D. Three mechanisms of hydrogen-induced dislocation pinning
in tungsten. Nuclear Fusion, 60(8), 086015 (2020).
21. ASTM E647. Standard test method for measurement of fatigue crack growth rates. American Society for Testing and
Materials Standards (2016).
22. Slifka, A. J., Drexler, E. S., Stalheim, D. G., Amaro, R. L., Lauria, D. S., Stevenson, A. E., & Hayden, L. E.: The effect
of microstructure on the hydrogen-assisted fatigue of pipeline steels. In Pressure Vessels and Piping Conference, 55713:
V06BT06A009, Paris, France, (2013).
On
ly
fo
rR
Nanoscale Insight on FRP-soil Interfacial Mechanical Properties
ea
din
Pengchang Wei [0000-0001-8023-6977] and Zhen-Yu Yin [0000-0003-4154-7304]
g
Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hong Kong, China
at
zhenyu.yin@polyu.edu.hk
W
CF
Abstract. Fiber-reinforced polymer (FRP) is extensively utilized across various engineering disciplines for its
S
superior properties, including its roles in protecting, reinforcing, and repairing structural components. In ma-
rine geotechnical engineering, the interfacial interactions between FRP and soil are crucial, yet they remain
20
poorly understood at the micro- and nano-scale. Molecular dynamics (MD) simulation method has been em-
ployed to explore the interfacial mechanical behavior of the epoxy-quartz interface within the context of FRP-
24
soil systems. The MD models were validated through uniaxial tension tests conducted on bulk epoxy. Subse-
quent MD nanoindentation simulations examined the epoxy-quartz interface, considering the effects of loading
rate and indentation depth. An abrasion model for the epoxy-quartz system was proposed based on the rela-
tionship between force and displacement. The simulation results indicated that increased loading rates lead to
higher indenter forces and greater indentation hardness of the epoxy substrate, with a power-law relationship
observed between indenter force and indentation depth. Additionally, the study determined the overall Young's
modulus of the bulk epoxy to be 2.075 GPa, with an average Poisson’s ratio of 0.377. For the epoxy substrate,
the reduced modulus was 2.419 GPa, and the indentation hardness ranged from 1.492 to 2.174 GPa.
Keywords: Sand; FRP-soil interface; Molecular Dynamics; Nanoindentation; Viscosity
1 Introduction
Fibre-reinforced polymer (FRP) composites are extensively utilized across diverse fields such as marine engineer-
ing, building structures, aerospace, and civil and hydraulic engineering [1]. This widespread adoption is attributed
to their exceptional properties, including high strength, low weight, corrosion resistance, electromagnetic trans-
parency, and ease of handling. In civil engineering, FRP composites are effectively employed to protect, reinforce,
and repair structures. For example, FRP piles are increasingly preferred over traditional concrete and steel piles
in corrosive environments, such as those encountered in marine geotechnical engineering, due to their superior
resistance to corrosion and chemicals [2]. As FRP piles become more common in geotechnical engineering, un-
derstanding the interfacial properties between FRP composites and soil is essential for maintaining the integrity
and stability of the structure-soil system [2, 3]. Shaia et al. [4] indicated that the friction coefficient between FRP
composites and sand was higher than that between conventional mild steel and sand. However, the stiffness and
surface hardness of FRP composites are notably lower compared to those of concrete and steel structures. Thus,
investigating the interfacial mechanical properties of FRP-soil interactions is crucial.
Current research predominantly addresses the macroscopic mechanical properties of FRP and soil/sand mix-
tures through experimental and numerical analyses [5]. These studies largely emphasized the properties of the
mixtures while frequently neglecting the nuances of their interfacial interactions [6]. Consequently, the microscale
interfacial behavior between FRP and soil remains insufficiently understood. Additionally, the macroscopic prop-
erties of materials are profoundly influenced by their micro- and nano-scale characteristics, where factors such as
composition degradation, failure initiation, and microscopic mechanical behavior significantly impact overall per-
formance. Hence, investigating materials at the microscale is crucial for a thorough understanding of their mac-
roscale mechanical properties.
The molecular dynamics (MD) simulation method has proven effective in investigating the interfacial mechan-
ical behavior of materials, offering fundamental insights that are challenging to obtain through experiments. In
MD simulations, the atomistic model of the epoxy-quartz system is frequently utilized to represent the FRP-soil
interface [2, 3], given that epoxy resin is a critical component of FRP and quartz is a primary constituent of the
soil fraction. Research to date has explored various aspects of the epoxy-quartz interface, including adhesion [7],
interfacial friction [2], and creep [3] behavior through MD simulations. However, there remains a paucity of
studies focused on their interfacial interactions at the microscale.
Nanoindentation is a powerful technique for examining the mechanical properties of materials at the microscale
(~ 10-9 m). MD simulations and experimental studies utilizing nanoindentation have predominantly focused on
pure epoxy resin, quartz, and mixtures of epoxy composites with various modifications (e.g., nanoclay, fumed
On
ly
2
fo
rR
silica). These investigations have demonstrated that the interfacial mechanical properties revealed by nanoinden-
tation are significantly influenced by factors such as loading rate, temperature, indenter shape and size, and in-
ea
dentation depth. However, to date, limited research has addressed the interfacial mechanical behavior between
din
epoxy and quartz using MD nanoindentation methods.
In this study, MD nanoindentation simulations were conducted to investigate the interfacial properties of the
g
FRP-soil system at the microscale, specifically focusing on the mechanical behavior between epoxy and quartz.
at
A cross-linked epoxy resin was employed as the substrate, while quartz was the indenter. Initially, uniaxial tension
tests were performed on the bulk epoxy to obtain its tensile stress-strain curve, Young's modulus, and Poisson’s
W
ratio. The mechanical parameters derived from these tests were compared with existing experimental and simula-
CF
tion data to validate the model's accuracy. Subsequently, MD nanoindentation simulations of the epoxy-quartz
system were carried out, considering the effects of loading rate and indentation depth to elucidate the underlying
S
mechanical mechanisms. Moreover, an abrasion model was proposed based on the indenter force Fz and indenta-
20
tion hardness H during the nanoindentation process.
2 Simulation details 24
2.1 System setup
As shown in Figure 1(a), a pyramid-shaped quartz model with dimensions of 35.99 Å × 33.71 Å × 34.05 Å and
comprising 1,380 atoms was created using VMD software. Figure 1(b) shows an epoxy resin model with dimen-
sions of 72.4612 Å × 72.4612 Å × 55.8246 Å, consisting of 500 diglycidyl ether of bisphenol A (DGEBA) and
250 m-phenylenediamine (mPDA). DGEBA serves as the organic component of the epoxy, while mPDA acts as
a curing agent. The resulting epoxy resin achieved a conversion percentage of 84% through the cross-linking
reaction, indicating the ratio of reacted carbon atoms to the initial number of epoxy groups.
Figure 1(c) shows the epoxy resin-quartz system comprising 115,380 atoms. In this system, a pyramid-shaped
quartz model (Figure 1(a)) functions as the indenter, while a 2a × 2b × 1c supercell of epoxy resin (Figure 1(b))
acts as the substrate. The quartz indenter was set as a rigid body during whole simulation. The epoxy resin sub-
strate was segmented into three distinct layers: the Newtonian layer, the thermostat layer, and the boundary layer.
Figure 1. Epoxy-quartz model: (a) pyramid type quartz model, (b) epoxy resin [3], (c) epoxy resin-quartz system for
nanoindentation simulation, where the quartz model is set as an indenter, and the epoxy as a substrate (Note that d1, d2, l1, l2,
φ, θ are corresponding length and angle in the simulation box of quartz (see Figure 1(a)), where φ and θ are set as 27.87° and
26.35°, respectively. hc, hz, ho are indentation depth (i.e., contact depth between the epoxy substrate and the quartz indenter),
displacement of quartz indenter along the z-direction, initial distance between epoxy and quartz, respectively.)
2.2 Force fields

The PCFF and CVFF force fields have been effectively used for organic compounds (e.g., epoxy system) and
inorganic materials (e.g., quartz) [3], respectively. Thus, the PCFF and CVFF force fields were employed for the
epoxy substrate and the quartz indenter, respectively, and their formulation are displayed in Eqs. 1-2.
Using the 9-6 Lennard-Jones (LJ) parameters for each atom in the epoxy system, the corresponding 12-6 LJ
potential parameters were derived through the fitting method we proposed. Subsequently, the non-bonded inter-
actions between the epoxy and quartz were evaluated using the same 12-6 LJ potential, where σij and εij between
atom i and atom j were calculated by Mixing Lorentz-Berthelot's law [8] (i.e., Eqs. 3-4).
On
ly
3
fo
𝐸 𝑘 𝑏 𝑏
rR 𝑘 𝑏 𝑏 𝑘 𝑏 𝑏 𝐻 𝜃 𝜃 𝐻 𝜃 𝜃 𝐻 𝜃 𝜃
𝑉 1
ea𝑐𝑜𝑠 𝜙 𝜙 𝑉 1 𝑐𝑜𝑠 𝜙 𝜙 𝑉 1 𝑐𝑜𝑠 𝜙 𝜙 𝑘 𝜒
𝑏 𝑏 𝑏′ 𝑏′
din 𝜃 𝜃 𝜃′ 𝜃′ 𝑏 𝑏 𝜃 𝜃
g
′ ′
𝑏 𝑏 𝑉 𝑐𝑜𝑠 𝜙 𝑉 𝑐𝑜𝑠 2 𝜙 𝑉 𝑐𝑜𝑠 3 𝜙

(1)
𝑏 ′
𝑏′ 𝑉 𝑐𝑜𝑠 𝜙 𝑉 𝑐𝑜𝑠 2 𝜙
at
𝑉 𝑐𝑜𝑠 3 𝜙
W
′
CF
𝜃 𝜃 𝑉 𝑐𝑜𝑠 𝜙 𝑉 𝑐𝑜𝑠 2 𝜙 𝑉 𝑐𝑜𝑠 3 𝜙
𝑒 𝑞𝑞 𝜎 𝜎
S
𝑘 ′ 𝑐𝑜𝑠 𝜙 𝜃 𝜃 𝜃′ 𝜃′ 𝜀 2 3
4𝜋𝜀 𝑟 𝑟 𝑟
20
′
𝐸 𝐷 1-e 𝐻 𝜃 𝜃 𝐻 1 𝑠 𝑐𝑜𝑠 𝑛𝜙 𝑘 𝜒 𝐹 ′ 𝑏 𝑏 𝑏′ 𝑏′
24
′
𝐹 𝜃 𝜃 𝜃′ 𝜃′ 𝐹 𝑏 𝑏 𝜃 𝜃 𝐹 𝑐𝑜𝑠 𝜙 𝜃 𝜃 𝜃′ 𝜃′
′
′ ′
(2)
𝑒 𝑞𝑞 𝜎 𝜎
𝐹 ′ 𝜒𝜒′ 4𝜀
4𝜋𝜀 𝑟 𝑟 𝑟
′
𝜎 𝜎
𝜎 (3)
2
𝜀 𝜀𝜀 (4)
2.3 Uniaxial traction and MD nanoindentation simulations

All MD simulations were conducted using the LAMMPS code [9]. The simulations employed three-dimensional
periodic boundary conditions and utilized the velocity-Verlet algorithm to integrate the motion equations. The
cut-off radius for van der Waals forces was set at 10 Å, while the cut-off for long-range electrostatic interactions,
calculated using the Particle-Particle Particle-Mesh (PPPM) method, was 8.5 Å. A time step of 1.0 fs was used,
and all simulations were conducted at a temperature of 300 K.
The bulk epoxy system, comprising 114,000 atoms (see Figure 2), was first relaxed using the NPT ensemble
for 1.0 ns at the temperature of 300 K and pressure of 1.0 atm, respectively. After relaxation, the tensile mechanical
properties of the epoxy system were assessed along three principal directions by applying uniaxial traction at 300
K. This was achieved through a strain-controlled loading approach with a tensile strain rate of 5 × 10-7/fs.
In the MD nanoindentation simulation of the epoxy-quartz system, as shown in Figure 1(c), the process was
divided into two stages: Relaxation Process: The stage was performed with Langevin thermostat at 300 K and
NVE ensemble for 2 ns, the quartz indenter kept still during this process. 2) Loading Process: In this stage, the
quartz indenter was moved in a displacement-controlled mode with a loading rate ranging from 0.1 to 2.0 Å/ps.
Initially, the indenter was positioned 20 Å (ho) above the epoxy substrate surface (see Figure 1(c)). During the
loading process, the indenter gradually approached and indented into the epoxy substrate. The final displacement
of the indenter along the z-direction was set to 45 Å, resulting in a final indentation depth of 25 Å.
3 Results and discussions
3.1 Uniaxial traction on bulk epoxy

The Young’s modulus of the epoxy resin was calculated via linear regression of the tensile stress-strain curve for
a strain range of 0 to 0.04 (as shown in Figure 2), where the overall Young's modulus (𝐸 ) was reported in Table
1 as 2.075 GPa. Table 2 presents the Poisson's ratio of the epoxy resin alongside values from similar materials in
previous studies, with an average Poisson’s ratio (𝜐̅ ) of 0.377. These values aligned well with prior experimental
and simulation results, indicating that the model developed in this work was effective for simulating the mechan-
ical properties of epoxy resin. Variations in data across Table 1 and Table 2 were attributed to differences in
epoxy resin compositions, curing agents, test conditions, and simulation parameters such as model size and cross-
linking conversion.
Moreover, as shown in Figure 2, Table 1, and Table 2, the tensile stress, Young's modulus, and Poisson’s ratio
of the bulk epoxy system are relatively close across the three principal directions. This uniformity suggested that
the epoxy resin exhibited isotropic behavior, meaning its mechanical properties were similar in all directions. This
On
ly
4
fo
rR
was supported by previous studies (see Table 1, and Table 2), which confirmed that epoxy resins could generally
be considered isotropic materials.
ea
din
g
at
W
CF
S
20
Figure 2. Evolution of tensile stress of the bulk epoxy system with tensile strain along three directions.
24
Table 1. Young's modulus of epoxy resin in this work and similar materials in previous studies at room temperature. The
overall Young's modulus (𝐸 ) is the square average value of Young's modulus along three directions (i.e., Ex, Ey, Ez). Er is
reduced modulus.
Methods Young's modulus of epoxy (GPa) Ex Ey Ez 𝐸 Er

MD simula- Epoxy resin (containing 2000 DGEBA + 1000 mPDA) 2.203 1.906 2.105 2.075 2.419
tions through uniaxial tension using MD simulations in this work
Epoxy resin (containing 200 DGEBA + 100 mPDA) via uni- 2.61 ± 0.28
axial tension using MD simulation [3]
Various ex- DGEBA epoxy networks (containing DGEBA/D230, 2.21 ~ 2.72
periments in DGEBA/TETA, DGEBA/TEPA) through dynamic mechani-
previous cal analyses experiments [10]
works Matrix epoxy Epon 862 [11] 2.83
Notes: DGEBA: Diglycidyl Ether of Bisphenol A; mPDA: m-phenylenediamine; DETDA: Diethyltoluenediamine; EPON862:
Bisphenol F diglycidyl ether; TETA: triethylenetetramine; TEPA: tetraethylenepentamine; D230: Jeffamine D230.
Table 2. Poisson’s ratio of epoxy resin in this work and similar materials in previous studies at room temperature. The aver-
age Poisson’s ratio (𝜐) is the geometric average value of the six components of the matrix.
Poisson’s ratio of epoxy vxy vxz vyx vyz vzx vzy 𝜐̅

Uniaxial tension using MD simulation in this work 0.351 0.407 0.338 0.420 0.368 0.378 0.377
Epoxy resin composed of EPON862 and curing agent 0.27 ~ 0.475
DETDA (conversion degrees of 0% ~ 86%) [12]
Epoxy resin (containing EPON862 + DETDA) [13] 0.353 (experimental), 0.448 (MD simulation)
Matrix epoxy Epon 862 [11] 0.3
3.2 MD nanoindentation behavior

Figure 3(a, b) illustrates the evolution of the indentation force (Fz) with the displacement of the quartz indenter
(hz) and the indentation depth (hc) at various loading rates. The indentation force gradually increased with hz and
the loading rate. Figure 3(b) shows the relationship between Fz and hc adheres to a power-law function. This
power-law relationship is well-established and commonly used in both molecular dynamics (MD) simulations and
experimental nanoindentation studies [14, 15]:
𝐹 αh (5)
where Fz and hc are the indentation force of the quartz indenter and the indentation depth, respectively. 𝛼 and m
are constants.
During the MD nanoindentation process, as shown in Figure 3(b), the quartz indenter penetrates the epoxy
substrate, causing significant surface deformation and the formation of pits, while no bond or angle breakage
occurs throughout the process. The surface wear of the epoxy substrate was characterized by an evident "groove".
Figure 4 demonstrates that at a loading rate of 1.0 Å/ps, the displacement magnitude of the atoms within the
system increases consistently with the increase in hz. Concurrently, the surface wear, or deformation area, ex-
panded as hz increased. This results in a greater number of epoxy atoms being displaced by the quartz indenter,
which could lead to increased resistance faced by the indenter, thereby causing the indentation force to rise with
h z.
On
ly
5
fo
rR
At a loading rate of 0.1 Å/ps, the indentation force showed considerable dispersion, as illustrated in Figure 3(a,
b), with a fitting coefficient R2 of 0.78. This value was notably lower compared to the coefficients obtained at
ea
other loading rates (see Figure 3(b)). Consequently, the loading rate of 0.1 Å/ps was excluded from the calculation
din
of the indentation hardness of the epoxy substrate.
As shown in Table 3, the indentation hardness of the epoxy substrate increased with the loading rate from 0.25
g
to 0.75 Å/ps. Beyond this point, the hardness tended to stabilize within the range of 0.75 to 2.0 Å/ps. These
at
observations were consistent with previous nanoindentation studies [16, 17], which also reported that the hardness
of epoxy resin increased with the loading rate.
W
CF
S
20
24
Figure 3. (a, b) Evolution of the indentation force (Fz) of the whole system with the displacement of the indenter (hz) and the
indentation depth (hc) at different loading rates. (c) Epoxy-quartz structure at different hz and hc. (d) Displacement magnitude
of atoms within the whole system at different loading rates and hc = 25 Å.
Figure 4. Displacement magnitude of atoms for the whole system at different displacements of the indenter (hz) under two
different slices: one slice along x-direction from 0 to 74 Å, and another one from 64 to 74 Å. (The above figures refer to the
case of a loading rate of 1.0 Å/ps. The indenter is hidden for better visual clarity.)
Table 3. The indentation hardness (H) of the epoxy resin in MD nanoindentation simulation at different loading rates and in
previous experimental works.
Methods Conditions and other notes H (GPa)

MD nanoindentation 1.492 ~ 2.266
Loading rates of 0.25 ~ 2.0 Å/ps
simulation in this work
Nanoindentation experi- Epoxy reinforced nanocomposites with fumed silica of 0% ~ 3% [18] 0.256 ~ 0.318
ments in previous works RTM6 Epoxy resin with strain rate of 0.01 ~ 0.25 and 25 ℃ [17] 0.23 ~ 0.35
Epoxy resin at different loading rates of 0.25 ~ 6.0 mN/s [16] 0.26 ~ 0.276
Notes: RTM: resin transfer moulding.
On
ly
6
fo
4 Conclusions
rR
ea
Molecular dynamics (MD) simulation method was used to investigate the interfacial mechanical behavior of
din
epoxy and quartz, considering the effect of loading rate. The uniaxial tension tests of bulk epoxy system were
conducted for validation of the MD models. The main conclusions of this study are as follows:
g
1) The overall Young's modulus of the bulk epoxy and average Poisson’s ratio were 2.075 GPa and 0.377,
at
respectively. The reduced modulus of epoxy substrate was 2.419 GPa, and its indentation hardness ranged from
1.492 to 2.174 GPa. These simulation results were consistent with previous simulation and experimental works,
validating the accuracy of the epoxy model.
W
CF
2) During the nanoindentation process, the higher the loading rate, the smaller the distance of surface connect-
ing atoms between quartz and epoxy, and the higher the indentation force Fz and hardness H of the epoxy substrate.
S
3) The relationship between the indentation force Fz and the indentation depth hc followed a power-law for the
20
nanoindentation process, and the surface wear of the epoxy substrate was an obvious “groove”.
This work could provide molecular insight into the interfacial mechanical behavior of epoxy and quartz, re-
24
flecting their local deformation behavior. Furthermore, the nanoscratching behavior of epoxy-quartz, the effect of
indenter shape and size, and their adhesion effects are worth developing in the future.
References:
1. Aliasghar-Mamaghani, M., Khaloo, A.: Effective Flexural Stiffness of Beams Reinforced with FRP Bars in Reinforced
Concrete Moment Frames. Journal of Composites for Construction 25(1), (2021).
2. Wan, J., et al.: Molecular dynamics simulations of the interface friction behavior between fiber-reinforced polymer pile
and sand. Tribology international 192, 109288 (2024).
3. Xu, W.Q., Yin, Z.Y., Zheng, Y.Y.: FRP–soil interfacial mechanical properties with molecular dynamics simulations:
Insights into friction and creep behavior. International Journal for Numerical and Analytical Methods in Geomechanics, (2023).
4. Shaia, H.A., Al-Asadi, A.K., Ramadan, S.H.: Evaluation of the Interface Friction between Fiber-Reinforced Polymers and
Granular Materials Using Modified Shear Apparatus. International Journal of Civil Engineering and Technology 9(13), 1010-
1016 (2018).
5. Abyaneh, M.J., El Naggar, H., Sadeghian, P.: Numerical Modeling of the Lateral Behavior of Concrete-Filled FRP Tube
Piles in Sand. International Journal of Geomechanics 20(8), (2020).
6. Zappalorto, M., et al.: Mechanical behaviour of epoxy/silica nanocomposites: Experiments and modelling. Composites Part
A: Applied Science and Manufacturing 72, 58-64 (2015).
7. Yaphary, Y.L., et al.: Molecular dynamics simulations on adhesion of epoxy-silica interface in salt environment. Compo-
sites Part B: Engineering 131, 165-172 (2017).
8. Frenkel, D., Smit, B., Understanding molecular simulation: From algorithms to applications. 2001: Academic Press.
9. Plimpton, S.: Fast Paraller Algorithms for Short-Range Molecular Dynamics. journal of computational physics 117, 1-19
(1995).
10.Amaral, C.R., et al.: Impact of aliphatic amine comonomers on DGEBA epoxy network properties. Polymer engineering
and science 54(9), 2132-2138 (2014).
11.Muller, A., et al., Influence of particle diameter on coefficient of thermal expansion of SiO2/epoxy particulate composites.
in Process SAMPE Fall Technology Conference, Salt Lake City (2010).
12.Li, C., Strachan, A.: Molecular dynamics predictions of thermal and mechanical properties of thermoset polymer
EPON862/DETDA. Polymer 52(13), 2920-2928 (2011).
13.Tack, J., Thermodynamic and Mechanical Properties of Epon 862 With Curing Agent Detda by Molecular Simulation.
Texas A&M University (2006).
14.Ma, Z., Gamage, R.P., Zhang, C.: Mechanical properties of α-quartz using nanoindentation tests and molecular dynamics
simulations. International Journal of Rock Mechanics and Mining Sciences 147, 104878 (2021).
15.Oliver, W.C., Pharr, G.M.: An improved technique for determining hardness and elastic modulus using load and displace-
ment sensing indentation experiments. Journal of Materials Research 7(6), 1564-1583 (1992).
16.Fu, K., et al.: Effect of loading rate on the creep behaviour of epoxy resin insulators by nanoindentation. Journal of Materials
Science: Materials in Electronics 25(8), 3552-3558 (2014).
17.Frontini, P., et al.: High Temperature Nanoindentation Response of RTM6 Epoxy Resin at Different Strain Rates. Experi-
mental Mechanics 55(5), 851-862 (2015).
18.Tzetzis, D., et al.: Nanoindentation measurements of fumed silica epoxy reinforced nanocomposites. Journal of Reinforced
Plastics and Composites 32(3), 160-173 (2013).
On
ly
fo
rR
Evaluation of bond strength between FRP bars and concrete under
ea
seawater immersion using explainable machine learning models
din
g
Heng Cao1[0000-0002-2828-2199], Xiao Lin Zhao [0000-0001-9659-3995], Mudassir Iqbal1[0000-00002-4317-3978], Daxu Zhang1[0000-
at
0002-6586-5238] 1[0009-0001-0699-3651]
, Pei-Fu Zhang
W
1 School of Ocean and Civil Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
CF
mudassiriqbal29@sjtu.edu.cn
S
20
Abstract. Floating structures are required to mitigate the issues of overcoming the land shortage in Hong Kong
and other densely populated coastal cities. For this purpose, conventional steel pontoons need to be replaced
24
by fiber-reinforced polymer (FRP) from the standpoint of durability and economy. Among various issues of
FRP and concrete in coastal areas, the bond performance between FRP bars and concrete is critical in structural
design and application. This study compiled 219 real pull-out test records with seven input parameters to in-
vestigate the bond performance between FRP bars and concrete under seawater immersion. Based on the da-
taset, three machine learning algorithms, namely Decision Tree, Random Forest, and XGBoost, are employed
to construct predictive models for the bond strength between FRP bars and concrete under seawater immersion.
Evaluation metrics, including R2, MAE, and RMSE, are exploited to calculate the performance of the devel-
oped models. Due to the superior predictive performance of the Random Forest model, this study further em-
ploys SHAP analysis to investigate the effects of influencing parameters on the bond strength. The developed
models and the SHAP analysis provide a reference for assessing and designing bond performance between
FRP bars and concrete under seawater immersion conditions.
Keywords: FRP bars, bond strength, seawater immersion, machine learning, SHAP analysis
1 Introduction
Introducing floating structures as a complementary approach to land reclamation is necessary to overcome the
land shortage challenge and sustainable growth of densely populated areas [1]. The use of conventional steel is
prone to durability issues on account of the corrosion of steel bars, resulting in substantial economic losses. In-
stead, fiber-reinforced polymer (FRP) is a high-performance material characterized by its lightweight, high
strength, and excellent corrosion resistance. Substituting FRP bars for steel bars in concrete is considered a profi-
cient solution to mitigate the corrosion issues of traditional steel-reinforced concrete, particularly in harsh envi-
ronments.
The bond performance between FRP bars and concrete is crucial in their synergistic operation, and the bond
strength between them is a significant attribute to consider in structural design and application. While researchers
have generally agreed on bond strength under ordinary environmental conditions [2], the bond performance in
aggressive environments, such as seawater immersion conditions, remains inadequately understood. Despite sev-
eral experimental studies [3]–[6], the bond performance between FRP bars and concrete in such environments
remains to be comprehensively explicated. Notably, machine learning, an artificial intelligence method with a
distinct advantage in solving complex and non-linear problems, has been increasingly employed in civil engineer-
ing research, demonstrating high feasibility and effectiveness [7],[8]. In light of the limitations of traditional re-
search methods and the excellence of machine learning approaches, this study leverages three explainable machine
learning models in the domain of FRP bars and concrete to introduce a novel method for evaluating bond strength
between fiber-reinforced polymer (FRP)bars and concrete under seawater immersion.
2 Methodology
2.1 Machine Learning Algorithms

Machine learning is a subfield of artificial intelligence wherein a computational system processes a given dataset
to generate a model capable of addressing specific problems [9]. In this study, three tree-based machine learning
algorithms, namely Decision Tree (DT), Random Forest (RF) and XGBoost, are employed to construct regression
models to predict the bond strength between FRP bars and concrete subjected to seawater immersion solutions.
On
ly
2
fo
rR
Among them, a Decision Tree is a typical supervised learning algorithm that utilizes a hierarchical structure to
classify data or predict outcomes. It operates through a recursive division of the dataset into more diminutive
ea
subsets based on the input features’ values and then uses the majority vote of the instances in each subset to make
din
a prediction. Random Forest is an ensemble learning technique that augments the precision and robustness of
predictions by amalgamating an array of decision trees. Each tree is generated from a stochastic selection of the
g
training dataset and the feature set. XGBoost is an extreme gradient boosting algorithm that iteratively trains
at
decision trees on the residuals of the preceding trees to correct their errors, thereby improving model performance.
2.2 Evaluation Criterion
W
CF
For a thorough assessment of the machine learning regression models' efficacy, three metrics are utilized for
evaluation: Mean Absolute Error (MAE), Root Mean Squared Error (RMSE) and the coefficient of determination
S
(R2). The equations (1)-(3) detail the calculation methods for these criteria, where 𝑦 represents the predicted
value, 𝑦 denotes the true value and n signifies the total number of the samples. In model evaluation, lower values
20
of MAE and RMSE indicate better predictive performance, while an R2 value closer to 1 signifies a higher degree
24
of prediction accuracy. Moreover, it is also crucial to consider the discrepancy in evaluation criteria between the
training and testing datasets, as substantial discrepancies typically suggest model overfitting, which needs to be
addressed through hyperparameter optimization to enhance generalizability.
1
𝑀𝐴𝐸 |𝑦 𝑦| 1
𝑛
1
𝑅𝑀𝑆𝐸 𝑦 𝑦 2
𝑛
∑ 𝑦 𝑦
𝑅 1 3
∑ 𝑦 𝑦
2.3 SHAP Analysis
SHAP (SHapley Additive exPlanations) analysis is a robust method for interpreting machine learning models
[10]. In SHAP analysis, the machine learning model generates the corresponding predicted value for each data
containing all input parameters. The SHAP value is calculated according to Equation (4), where 𝑦 denotes the
predicted value of the ith sample, 𝑦 is the base value (typically the average value of the predicted variable),
and 𝑓 𝑥 , represents the SHAP value, which measures the impact that the 𝑗th parameter exerts on the forecasted
value for the 𝑖th sample. A positive 𝑓 𝑥 , implies that the jth parameter increases the predicted value of the
ith sample, while a negative 𝑓 𝑥 , indicates a decrease. The absolute value of 𝑓 𝑥 , signifies the magnitude
of the impact. SHAP analysis offers a unique advantage over other interpretability methods like Permuta-
tion Importance and tree-based Feature Importance by not only ranking input parameters' impact but also
visualizing their effects on predictions through SHAP value calculations
𝑦 𝑦 𝑓 𝑥, 𝑓 𝑥, ⋅⋅⋅ 𝑓 𝑥 , ⋅⋅⋅ 𝑓 𝑥 , 4
3 Datasets
A comprehensive dataset is paramount in machine learning, as the accuracy and versatility of the developed mod-
els are contingent upon the quality of the dataset [9]. To ensure a robust and accurate dataset of bond performance
between FRP bars and concrete subjected to seawater immersion, this study extensively reviews relevant literature
and establishes four rules for data collection:
(1) Only tests conducted under seawater immersion, including natural seawater immersion and artificial sea-
water immersion according to standard configuration, are selected.
(2) Only tests conducted through cube direct pull-out tests are selected, while tests carried out through beam
or cylinder pull-out tests are filtered out.
(3) Only tests whose failure mode was pull-out failure are selected, while tests with failure modes of splitting
failure are filtered out.
(4) Only tests with sufficient parameters, including test materials, procedures, and results, are selected, while
tests with missing parameters are filtered out.
According to the mentioned conditions, a total of 219 experimental datasets are collected from relevant research
papers [3]–[6],[11]–[14]. The statistical overview of the parameters is encapsulated in Table 1. Given that the
parameters “FRP bar type” and “Surface treatment” are categorical variables that cannot be directly received by
the machine learning models, these two parameters are processed by sequential coding detailed in Table 2. The
Pearson coefficient is calculated to gain a preliminary understanding of the internal relationships between the
parameters, as shown in Fig. 1. The Pearson coefficient measures the correlation between two parameters, ranging
On
ly
3
fo
rR
from -1 to 1. The closer the absolute value is to 1, the stronger the linear correlation between the parameters. As
observed in Fig. 2, all the parameters in this study exhibit weak linear correlations, indicating a complex internal
ea
nonlinearity and underscoring the necessity to develop advanced machine learning models.
din
Table 1. Statistical overview of the parameters
Parameters Type Max Min Average Std
FRP bar type input
g 3 1 1.6 0.6
at
Surface treatment input 6 1 2.9 2.0
FRP bar tensile modulus (GPa) input 152.1 42.8 61.0 27.1
FRP bar diameter (mm) input 12.0
W 8.0 10.0 1.6
CF
Concrete compressive strength (MPa) input 66.9 26.6 50.4 8.7
Corrosion age (d) input 360 0 85.3 95.9
S
Corrosion temperature (℃) input 60 25 35.7 21.2
20
Bond strength (MPa) output 30.51 10.21 18.98 4.21
Table 2. Sequential coding of categorical parameters
FRP bar type
Coding
BFRP
1
GFRP
2
CFRP
3 24
Helically Ribbed Sand coated Helically Ribbed and sand Helically wrapped and
Surface treatment
wrapped (HW) (RB) (SC) grooved (HG) coated (RBSC) sand coated (HWSC)
Coding 1 2 3 4 5 6
Fig. 1. Pearson correlation coefficient heatmap of the parameters
4 Model Development and Parameter Analysis
4.1 Model Development

Following the determination of datasets, the data is randomly partitioned to two subsets: 80% for training the
model and 20% for testing. The input parameters are standardized to mitigate the influence of the magnitude and
dimension of the input parameters on model training. Equation (5) illustrates the method for standardization.
𝑋 𝜇
𝑋 5
√𝜎
Table 3. Hyperparameters for model tunning in GridSearchCV
Models Hyperparameters Default [Searching Range] (increment) Optimal
max_depth None [1,100] (1) 8
DT
max_features None [1,7] (1) 6
n_estimators 100 [10, 20, 50, 100, 200] 200
RF max_features auto [1,7] (1) 6
max-depth None [5, 7, 10, 15, 20] 7
n_estimators 100 [20,200] (1) 21
XGBoost min_child_weight 1 [2,10] (1) 7
gamma 0.0 [0.0, 0.1, 0.2, 0.3, 0.4, 0.5] 0.2
On
ly
4
fo
rR
Based on the standardized datasets, DT, RF, and XGBoost models are trained. It is noteworthy that these models
possess numerous built-in hyperparameters that require manual tuning, which significantly impacts model perfor-
ea
mance. Machine learning models using default hyperparameters often underperform in accuracy or generalization.
din
Therefore, GridSearchCV is employed to enhance model performance. It is a hyperparameter optimization method
for machine learning models that systematically uses cross-validation to evaluate all possible combinations of
g
selected hyperparameters to determine the optimal combination. Table 3 presents the selected hyperparameters,
at
their default values, the searching range, and the optimal combinations for the three models.
The evaluation results of the optimized models are graphically represented in Fig. 2. The outcomes reveal that,
W
compared to DT and XGBoost, RF demonstrates the highest R2, lowest RMSE, and lowest MAE. Considering the
CF
discrepancy in performance metrics between the training and testing datasets, the RF model also prevails over the
other two models in R2 and RMSE. These two aspects suggest that the RF model exhibits superior predictive
S
accuracy and the least overfitting, indicating excellent overall performance in prediction and generalization.
20
24
Fig. 2. Performance of the models

4.2 Parameter Analysis
To comprehensively evaluate the bond performance between FRP bars and concrete subjected to seawater im-
mersion conditions, parameter analysis on the strength of the SHAP analytical approach is conducted to assess
the impact of each input parameter on the output variable, namely, bond strength. Given the superior performance
of the RF model, SHAP analysis is carried out on the predictive outcomes yielded by the RF model.
The summary plot of SHAP values is depicted in Fig. 3 (a), where each point represents a sample, the color of
the point indicates the value of the parameters, and the abscissa represents the magnitude of the SHAP value. Fig.
3 (b) shows the ranking of each parameter based on feature importance calculated by average SHAP values. It can
be seen that, among all the seven input parameters, FRP bar tensile modulus, concrete strength, and surface treat-
ment exert a pronounced effect on the bond strength between FRP bars and concrete under seawater immersion,
while the impact from FRP bar type, corrosion temperature, and FRP bar diameter is relatively insignificant.
(a) Summary plot of SHAP values (b) Feature importance of input parameters
Fig. 3. Results of SHAP analysis
The effect of each input parameter on bond strength is detailed in Fig. 4. Concerning FRP bar tensile modulus
and concrete strength, SHAP values gradually increase as their values increase, which can be seen in Fig. 4 (a)
and (b), showing higher FRP bar tensile modulus and concrete strength lead to higher bond strength between FRP
On
ly
5
fo
rR
bars and concrete subjected to seawater immersion. An explanation for this phenomenon could be that concrete
with high compressive strength often has denser internal microstructure and better compactness, which can inhibit
ea
the generation of cracks, thereby reducing the corrosion effect of seawater on FRP bars [8].
din
g
at
W
CF
S
20
24
Fig. 4. SHAP value distribution of each input parameter

For corrosion age and temperature, Fig. 4 (d) and (f) show that as their values increase, SHAP values de-
crease, indicating a negative effect on the bond strength. On one hand, longer immersion time leads to more mi-
cro-damage caused by seawater corrosion. On the other hand, the coefficients of thermal expansion for FRP bars
and concrete are discrepant. Upon temperature fluctuations, the rates of thermal expansion and contraction of
FRP bars exceed those of concrete [15]. This phenomenon generates radial stress and produces micro-cracks
and voids at the interface between FRP bars and concrete, decreasing the bond strength [16]. Additionally, Fig.
4 (c) and (e) show that the surface treatment of 5, i.e., ribbed and sand-coated, along with the FRP type of 3, i.e.,
CFRP bars help improve the adhesive strength between FRP bars and concrete subjected to seawater immersion,
On
ly
6
fo
rR
while Fig. 4 (g) shows the impact pattern of FRP bar diameter is not significant, corresponding to the minimum
influence in Fig. 3.
ea
5 Conclusions
din
g
at
In this study, a comprehensive dataset of bond performance between FRP bars and concrete in seawater immersion
solutions is established, and based on the dataset, three explainable machine learning models, namely DT, RF,
W
and XGBoost, are developed for predicting the bond strength between FRP bars and concrete. Additionally, the
CF
SHAP analysis method is employed to assess the impact of different input parameters on bond strength. The
subsequent specific conclusions can be delineated:
S
(1) A dataset comprising seven input parameters and one output parameter is established. The dataset comprises
219 real experimental data points from direct cube pull-out tests collected from previously published research
20
papers. This accurate and comprehensive dataset provides a solid foundation for training machine learning mod-
24
els.
(2) The evaluation metrics, including R2, MAE, and RMSE, are utilized to assess the performance of the three
models. The developed models demonstrate efficacy in prediction of the bond strength between FRP bars and
concrete under seawater immersion conditions. The RF model outperforms the other two models, exhibiting su-
perior predictive performance.
(3) A comprehensive parameter analysis based on predictive outcomes yield by the RF model is conducted
using the SHAP analysis method. The impact of all seven parameters is visualized and discussed. It is concluded
that the FRP bar modulus, concrete strength, surface treatment, and corrosion age are the four key parameters
significantly influencing bond strength.
References
1. Zhao, X. L., Dai, J. G., Ding, X. L., et al.: Developing a sustainable and smart floating structure solution for enhancing
liveability in Hong Kong’s crowded built environment. In: Shigeru TABETA. The 3rd World Conference on Floating
Solutions (WCFS2023). Springer, Japan (2023).
2. Yin, S., Hua, Y. & Xu, S.: A review on research progress and application of concrete structures internally reinforced with
FRP bars. Jianzhu Jiegou Xuebao/Journal of Building Structures 42, 134–150 (2021).
3. El Refai, A., Abed, F. & Altalmas, A.: Bond Durability of Basalt Fiber–Reinforced Polymer Bars Embedded in Concrete
under Direct Pullout Conditions. Journal of Composites for Construction 19, 04014078 (2015).
4. Ruiz Emparanza, A., Kampmann, R., De Caso, F., et al.: Durability assessment of GFRP rebars in marine environments.
Construction and Building Materials 329, 127028 (2022).
5. Ruiz Emparanza, A., De Caso Y Basalo, F., Kampmann, R., et al.: Evaluation of the Bond-to-Concrete Properties of
GFRP Rebars in Marine Environments. Infrastructures 3, 44 (2018).
6. Lu, Z., Su, L., Lai, J., et al.: Bond durability of BFRP bars embedded in concrete with fly ash in aggressive environments.
Composite Structures 271, 114121 (2021).
7. Ganaie, M. A., Hu, M., Malik, A. K., et al.: Ensemble deep learning: A review. Engineering Applications of Artificial
Intelligence 115, 105151 (2022).
8. Zhang, P.-F., Iqbal, M., Zhang, D., et al.: Bond strength prediction of FRP bars to seawater sea sand concrete based on
ensemble learning models. Engineering Structures 302, 117382 (2024).
9. Baduge, S. K., Thilakarathna, S., Perera, J. S., et al.: Artificial intelligence and smart vision for building and construction
4.0: Machine and deep learning methods and applications. Automation in Construction 141, 104440 (2022).
10. Lundberg, S. & Lee, S.-I.: A Unified Approach to Interpreting Model Predictions. 4–7 Preprint at
https://doi.org/10.48550/arXiv.1705.07874 (2017).
11. Taha, A. & Alnahhal, W.: Bond durability and service life prediction of BFRP bars to steel FRC under aggressive envi-
ronmental conditions. Composite Structures 269, 114034 (2021).
12. Wu, G., Dong, Z., Xu, B., et al.: Bond performance and calculation method for the basic anchorage length of BFRP bar
with concrete in marine environment. Tumu Gongcheng Xuebao/China Civil Engineering Journal 49, 89–99 (2016).
13. Xu B.: Study on bond behavior between FRP bar and concrete in marine environment. Ph.D. Thesis, Southeast Univer-
sity, Ch.
14. Altalmas, A., El Refai, A. & Abed, F.: Bond degradation of basalt fiber-reinforced polymer (BFRP) bars exposed to
accelerated aging conditions. Construction and Building Materials 81, 162–171 (2015).
15. Belarbi, A. & Wang, H.: Bond Durability of FRP Bars Embedded in Fiber-Reinforced Concrete. Journal of Composites
for Construction 16, 371–380 (2012).
16. Masmoudi, R., Masmoudi, A., Ouezdou, M. B., et al.: Long-term bond performance of GFRP bars in concrete under
temperature ranging from 20°C to 80°C. Construction and Building Materials 25, 486–493 (2011).
On
ly
fo
rR
Hydroelastic Analysis of a Novel Offshore Fish Cage Design
ea
din
Xinyu Zhang1, Hong Zhang1,2, Chien Ming Wang3,4, Yunil Chu3,4, Mingyuan Ma1,2
g
1 School of Engineering and Built Environment, Griffith University, Gold Coast, Queensland 4222, Australia
at
2 Cities Research Institutes, Griffith University, Gold Coast, Queensland 4222, Australia
3 School of Civil Engineering, The University of Queensland, St Lucia, Queensland 4072, Australia
W
4Blue Economy Cooperative Research Centre, Launceston, Tasmania, Australia
CF
xinyu.zhang6@griffithuni.edu.au
S
Abstract: A flexible, modular, submersible fish cage is an attractive design concept for offshore fish farming
20
due to its cost-effectiveness in construction and survivability during extreme weather events when submerged
at an appropriate water depth. However, under some intensive circumstances, the structure would experience
24
extreme weather conditions at the water surface, which leads to potential failure. This paper investigates the
hydroelastic behaviour of a simplified design of a newly proposed modular fish cage, named the SeaFisher,
under various wave conditions, including waves in severe weather conditions at sea level. The software Aqua-
Sim is employed to model the cage frame, nets, and mooring line system and perform the hydroelastic analysis
to assess the strength, stiffness, and stability of the structure in floating conditions. The study highlights the
need for an optimal design to prevent large deformations and potential failure of the fish cage. As wave height
and period increase, the compressive axial forces within the structural members also rise.
Keywords: offshore fish farming, flexible frame, hydroelastic behaviour analysis.
1 INTRODUCTION
Moving fish farms offshore offers several advantages: access to larger sea space for large-scale farming, better
water quality with lower temperatures, fewer parasites and pathogens, better dispersal of fish wastes and less
interaction with other human activities than nearshore fish farming. However, the most significant challenge faced
by offshore fish farming is maintaining structural integrity and fish well-being under rough environmental condi-
tions, including strong waves and faster currents [1].
Existing offshore fish cages such as Shenlan 1 and Havfarm 1 have huge steel frames to resist the high envi-
ronmental loads on the structure. However, such designs come with a considerable construction cost of over 100
million US dollars [2]. A flexible, smaller, submersible fish cage design reduces construction costs as intense
surface wave loads can be avoided.
Arising from a recent Blue Economy CRC project in Australia, the team developed a novel offshore fish cage
called the SeaFisher [3]. This submersible modular cage is constructed using bundled high-density polyethylene
(HDPE) pipes. Compared with the aforementioned offshore steel cages, the high-density polyethylene (HDPE)
frame has a lower construction cost and is easier to transport and construct. Although HDPE is durable in seawater
environments and an excellent anti-biofouling material, it has some drawbacks, such as its lightness and flexibil-
ity. Due to its lower stiffness, the flexible HDPE structure is more likely to fail under severe environmental con-
ditions. Under some circumstances, the fish pen structure is possibly exposed to severe environmental load con-
ditions at the water surface. When the wave frequency is closer to the structure's natural vibration frequency,
resonance may cause the structure to deform excessively, leading to structural failure.
The present study evaluates the new HDPE structure fish cage design by examining the hydroelastic behaviour
of a simplified SeaFisher design under different wave heights and wave periods while the structure is under float-
ing conditions.
2 HYDROELASTIC ANALYSIS
In this study, a simplified design of the SeaFisher is used. A single equivalent HDPE pipe was used for the struc-
tural member. The equivalent pipe has the same material properties and stiffness. The front bow structure was
also modified by removing the shield mat to decrease the computational time and resources required. Since this
study aims to evaluate the relationship between hydroelastic deformation and various environmental load
On
ly
2
fo
rR
conditions, the simplification has minor effects on the hydroelastic behaviour of the structure. Fig. 1 presents the
plan and side views of the simplified fish cage model. HDPE material was used for the main structure frames of
ea
the model, including top, bottom, vertical and bow frames. HDPE material has a density of 958 kg/m3, Young’s
din
modulus of 1 GPa, and Poisson’s ratio of 0.4. A multiple-point mooring system is adopted to capture better the
interaction between the mooring line and the flow. The detailed design parameters of the model are given in Table
g
1 and Table 2. In this study, the model floats at a water depth of 70m, with the free water surface as the reference
at
for the Z coordinate. The Z value indicates the depth of each component from its centre to the free water surface.
Submerged states of the fish cage at various depths were not considered.
W
CF
S
20
Fig. 1. Side view (a) and Top view (b) of the fish cage model
Table 1. Model fame structure
24
Outer diameter (m) Thickness (m) Length (m) ballast Element type
Top longitudinal frame 0.956 0.095 120 Yes Beam

Top Transverse frame 0.956 0.120 20 per cubic cage No Beam
Vertical Frame 0.956 0.095 20 per cubic cage No Beam
Bottom Frame (including longitudi- 0.956 0.095 20 per cubic cage No Beam
nal and transverse members)
Bow Frame 0.600 0.085 10 – 22m No Beam
Table 2. Model mooring line and net structures
Thickness (m) Thread Diameter (m) Young’s modulus (GPa) Density Poisson Element Type Area (m2)
(kg/m3) Ratio
Net 0.002 0.002 1.0 1025 - Membrane -
Hawser - - 2.7 788 - Truss 0.022
3 Methodology
The time-domain finite element software AquaSim [4,5] was used for the hydroelastic analysis of the fish cage
under wave action. The software adopted the hydrodynamic force option (diffraction) and the Morison equation
from wave and current for environmental conditions. The Morison force includes drag force, Froude-Krylov dif-
fraction force, and added mass, and it is given by
𝐹 𝑢 𝑣 𝑢 𝑣 𝑢 𝑣 𝜌 1 𝐶𝑎 𝑉 𝐿 𝑎 𝜌 𝐶𝑎 𝑉 𝐿 𝑣 (1)
where
 𝐶 is the drag coefficient in the local y-direction;
 𝑢 𝑣 𝑢 𝑣 is the relative velocity between the element and the fluid in the cross-section plane;
 𝑢 𝑢 𝑢 , where 𝑢 is the fluid velocity due to wave, 𝑢 is the current velocity in
the local y-direction;
 𝑣 is the velocity of the element in the local y-direction;
 𝑎 is the fluid acceleration in the local y-direction;
 𝐶 is the added mass coefficient in the local y-direction;
 𝐷𝑖𝑎𝑚 is the cross-section diameter in the direction of the relative velocity;
 𝜌 is the seawater density;
 𝑉 is the 2D volume.
We have selected the time-domain finite element software AquaSim because the small diameters of the structural
and mooring elements and the smooth HDPE frame have less effect on the surrounding flow field. In contrast, the
analytical method makes it hard to capture the elastic behaviour of the fish pen. AquaSim provides suitable ele-
ments for modelling various structural components: membrane elements, beam elements, truss elements, and
springs. Beam elements were used for the HDPE pipes forming the main structural frame and bow structure to
account for bending moments and shear forces. Truss elements were used for the hawser and anchor lines, ex-
cluding compressive forces. The Kikkonet was modelled with a membrane-type element, automatically generating
On
ly
3
fo
rR
the net and allowing adjustments to the net's solidity by changing the thread diameter and mesh width. Biofouling
effects were excluded. The drag coefficients for the hawser, net, and frame were assumed to be 1.2 per the default
setup. [4,5].
ea
din
g
4 Environmental conditions
at
Several offshore environmental load cases were considered to capture the maximum compressive axial force dis-
W
tribution across the cage structure with the view of identifying the most critical member that is most likely to fail.
CF
Table 3 shows the load cases used to determine the critical element. The wave profile was generated according to
de Hauteclocque et al. [6]. Based on the identified critical element, the relationship between the compressive axial
S
force in each segment under various wave heights and wave periods was observed, and the combinations of wave
height and period used are listed in Table 4. For all the environmental load cases considered, a 0.5m/s current
speed was assumed. The current and wave are assumed to be heading in the same direction toward the model's
20
24
front end. The water depth is taken as 70m.
Table 3. Load cases used to identify the critical element
Load case Wave height (m) Wave frequency (Hz)
1 3 0.125
2 4.5 0.100
3 6.5 0.083
4 8 0.071
5 8.5 0.063
Table 4. Load cases are used for divided elements.
Wave height (m) Wave frequency (Hz)
J1 = 0.125 J2 = 0.100 J3 = 0.083 J4 = 0.071 J5 = 0.063
C1 =2 (C1 , J1) (C1 , J2) (C1 , J3) (C1 , J4) (C1 , J5)
C2 =3 (C2 , J1) (C2 , J2) (C2 , J3) (C2 , J4) (C2 , J5)
C3 =4 (C3 , J1) (C3 , J2) (C3 , J3) (C3 , J4) (C3 , J5)
C4 =5 (C4 , J1) (C4 , J2) (C4 , J3) (C4 , J4) (C4 , J5)
C5 =6 (C5 , J1) (C5 , J2) (C5 , J3) (C5 , J4) (C5 , J5)
5.1 Identifying the critical elements

During the simulated time period, the computed maximum compressive axial force distribution of the cage struc-
ture is presented in Fig. 2. The compressive axial force distribution scale is consistent across all load cases, with
red indicating the highest compressive axial force and blue indicating the lowest. In AquaSim, the maximum axial
compression force is displayed as a positive value. The distribution shows that the bottom structural members
experience higher compressive forces than other components, with the bottom middle longitudinal element having
the highest compressive force. The maximum compressive axial force of the entire structure increases under more
severe environmental conditions, making the bottom middle longitudinal element the critical component with the
highest risk of failure.
Element 1 Element 3 Element 5
Element 2 Element 4 Element 6
(a) (b)
Fig.2. Maximum compressive axial force distribution for Load Case 1 (a), Divided element number and observation points
marked red dots (Top view of the structure). (b)
The bottom middle longitudinal element was divided into six segments for further analysis. The maximum com-
pressive axial force observation point along each divided element was numbered to obtain a detailed compressive
axial force distribution along the whole element, as shown in Fig. 2.
Fig. 3 shows the maximum compressive axial force F divided by the cross-sectional area of the HDPE pipe A
and elastic modulus E at the observation points as the compressive axial strain for the five extreme environmental
On
ly
4
fo
rR
conditions listed in Table 3. The Elastic modulus used for the HDPE Pipe is 1GPa, and the diameter of the HDPE
pipe can be found in Table 1.
ea
din
In each case, the middle part of the cage shows higher compressive axial strain distribution, while the front and
back have lower compressive axial strain, which presents the elastic behaviour of the HDPE structure. Case 1
g
exhibits the lowest overall compressive axial strain distribution, whereas Case 4 has the highest. The compressive
at
axial strain distributions for Load Case 4 are the highest. The middle elements, specifically elements 3 or 4, have
the highest compressive axial strain. However, the axial compressive strain distribution pattern across elements
varies with different environmental conditions, indicating a need for further study.
W
CF
8 case1
case2
S
case3
6 case4
20
case5
4
2
24
0
1 2 3 4 5 6
Element
Fig.3. Axial compressive strain (F/EA) along the bottom middle longitudinal pipe
The maximum compressive axial strain distribution of each divided segment along the bottom middle element
has been plotted using selected environmental condition combinations (Table 4) to investigate the relationship
between the compressive axial strain distribution of each segment with wave height and wave frequency, with
either the same wave height at different wave frequencies or the same wave frequency at different wave heights—
Fig. 4 and 5 display the results for elements 2, 3, 4, and 5. Element 2 is at the front of the cage, element 3, which
has the highest compressive axial force, is in the middle, and element 5 is toward the end of the cage. Fig. 4
illustrates the relationship between F/EA and wavelength normalised by structure length. Fig. 5 shows the
relationship between F/EA and wave height, demonstrating the relationship between HDPE pipe compressive
strain and various wave heights and frequencies.
10-4 10-4
7 H = 2m 8
H = 3m
6 H = 4m
H = 5m 6
5 H = 6m
4 4
3
2
2
1 0
0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5
Wavelength/structure length Wavelength/structure length
(a) (b)
10-4 10-4
7 8
6
6
5
4 4
3
2
2
1 0
0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5
Wavelength/structure length Wavelength/structure length
(c) (d)
Fig.4. Relationship between axial compressive strain and wavelength normalised by structure length for (a) Element 2, (b)
Element 3,(c) Element 4 and (d) Element 5.
10-4 10-4
7 8
f = 0.125 Hz
6 f = 0.1 Hz
f = 0.09 Hz 6
5 f = 0.08 Hz
f = 0.07 Hz
4 4
3
2
2
1 0
2 3 4 5 6 2 3 4 5 6
Wave height(m) Wave height(m)
(a) (b)
On
ly
5
fo
7
10 -4
rR 8
10 -4
ea
6
6
5
din 4
g
3
2
at
2
1 0
W
2 3 4 5 6 2 3 4 5 6
Wave height(m) Wave height(m)
(c) (d)
CF
Fig.5. Relationship between axial compressive strain and wave height for (a) Element 2, (b) Element 3,(c) Ele-
ment 4 and (d) Element 5.
S
Fig. 4 illustrates the divided segments under the same frequency and current speed, and Fig. 5 demonstrates that
for each divided segment under the same wave height and current speed. The maximum axial compressive strain 20
in each element increases with higher wave heights, which indicates that the structure is more likely to fail. Under
consistent wave height and current speed, the compressive axial strain distribution increases with higher frequen- 24
cies. However, elements 2 and 3 at the front show a different pattern under high-frequency conditions. Moreover,
when comparing different wave heights under the same frequency, axial compressive strain generally shows an
approximately linear increase with wave height. However, under the same wave height, axial compressive strain
within the element varies with increasing frequency. Therefore, further investigation is needed to understand the
relationship between frequency and elastic deformation of the flexible structure.
From the comparison result above between axial strain and various wave heights for elements 2,3,4 and 5, it
can be seen that axial strain increases linearly with higher wave heights under the same wave frequency. For this
study, the Morison equation was used to calculate the hydrodynamic force acting on the structure; it can be sig-
nificant to identify which part of the Morison equation has more effect on the result. Based on Sumer and Fredsøe
[7], the 𝐾 number can be used to determine if the drag or inertia parts govern the Morison equation with equation
2 below.
(2)
where
 𝐹 is the inertia force part of Morison force;
 𝐹 is the drag force part of Morison force;
 𝐶 is the inertia coefficient, which is equal to 1 𝐶 ;
 𝐶 is the added mass coefficient;
 𝐶 is the drag coefficient from the Morison equation;
 𝑑 is the characteristic length;
 𝑈 is the maximum flow velocity calculated using linear wave theory;
 𝐾 is the Keulegan-Carpenter number.
Where 𝐾 number is calculated by using the following equation 3:
𝐾 (3)
Where 𝑈 is the maximum flow velocity, 𝑇 is the wave period; 𝐷 is the characteristic length.
Fig. 6 shows the resulting inertia and drag force ratio for the HDPE pipe, net, and mooring line under different
wave conditions.
Front barrier
Main frame HDPE pipe
2.5 1.5
H=2m
H=3m
H=4m
2 H=5m
H=6m
1
1.5
1
0.5
0.5
0 100 200 300 400 500 0 100 200 300 400 500
L (m) L (m)
On
ly
6
fo
rR Mooring -3 Net
10
0.5 6
0.4
ea 5
0.3 din 4
g
3
at
0.2
2
0.1 1
W
CF
0 100 200 300 400 500 0 100 200 300 400 500
L (m) L (m)
S
Fig.6. Inertia force and drag force ratio with different wavelength (L) and wave height for each component.
The above Fig. 6 shows that the inertia force of the Morison equation governs the main structure and front barrier.
20
Where for the net and mooring of the fish pen, the drag force governs the force on the component. Under the same
24
wave height, the inertia and drag force ratio have a non-linear descending trend, which clearly shows that with
wavelength longer than the structure length, the Morison equation is more governed by drag force, while for short
waves less than the structure length the inertia and drag ratio does not vary clearly with increasing wavelength.
Also, for the main frame HDPE Pipe, under higher wave heights such as 5m and 6m, the ratio between inertia and
drag force is lower than one, which could indicate that the drag force part is governing the Morison equation,
whereas under lower wave height such as 4m, 3m and 2m, the inertia force part of the Morison equation will be
dominant the equation.
6 Conclusion
This study examined the hydroelastic behaviour of a simplified HDPE offshore fish cage design at sea level under
various extreme weather conditions using the software AquaSim, aiming to investigate the structural internal
forces caused by waves and changes in wave conditions when the structure is floating on the water surface. It is
found that the compressive axial strain in structural members increases with higher wave height and wave fre-
quency. Under the same wave frequency, the compressive axial strain increased linearly with increasing wave
height. With elastic behaviour and high compressive axial force, the structure is also likely to buckle under higher
wave heights.
Future research could explore the relationship between wave frequency and hydroelastic buckling, including
detailed studies on hydroelastic buckling deformation and structural failure, where submerge conditions can also
be included. More parametric analyses should be conducted to better understand how different design options,
such as adding reinforcement cables, changing cage length, and different material stiffness, would affect structure
response under various environmental conditions. The relationship between various current speeds and hydroe-
lastic buckling could also be examined at different operating depths.
Acknowledgements
The authors acknowledge the financial support of the Blue Economy Cooperative Research Centre, established
and supported under the Australian Government’s Cooperative Research Centres Program, grant number CRC-
20180101. Also, the authors would like to thank Aquastructures AS for granting access to their software AquaSim
(https://aquasim.no).
References
1. Scott, D. and Muir, J.: Offshore cage systems: a practical overview. In: Option Mediterranean’s-International Centre for
Advanced Mediterranean Agronomic Studies, pp.79–89, (2000).
2. Wang C.M., Chu Y.I., Baumeister J., Zhang H., Jeng D.S., and Abdussamie N.: Offshore fish farming: challenges and
developments in fish pen designs, Glob. Blue Econ. Anal. Dev. Challenges, pp. 87–128, (2022).
3. Wang, C.-M. et al.: ‘Seafisher—a submersible high-density polyethylene offshore fish pen’, Journal of Marine Science
and Engineering, 11(9), p. 1795, (2023).
4. Berstad, A.: The AquaSim Package Theory User Manual, (2013).
5. Aquastructures AS.: AquaEdit user manual, (2022).
6. de Hauteclocque G., Maretic, N.V. and Derbanne, Q.: ‘Hindcast based global wave statistics’, Applied Ocean Research,
130, p. 103438, (2023).
7. Sumer, B.M., Fredsøe, J.: Hydrodynamics Around Cylindrical Structures. World Scientific, Singapore (1997).
On
ly
fo
rR
Submerged Floating Tunnel High Potential Energy System
ea
din
Wei Lin1, Mark Cassidy1, Yinghui Tian1, Ming Lin2
g
1Faculty of Engineering and IT, University of Melbourne, VIC, Australia
at
2 China Communications Construction Company, Ltd., Beijing 100088, China
linwl2@student.unimelb.edu.au
W
CF
S
Abstract. Although proposals to build submerged floating tunnels (SFTs) have existed for over a century, their
history is littered with well-meaning suggestions of form and type, but very few have been built. More recently,
20
however, their potential as a highly cost-effective and sustainable crossing has rejuvenated industry interest.
The main types of SFTs are pier-, tether-, pontoon-supported and free-spanned. With the introduction of a
24
novel high-potential energy system, the benefit of providing axial tension along a tether-supported SFT is
studied. The system imposes permanent axial tension in the tube by pre-tensioning the tube from the shore
during construction. The system enables a positive adjustment of the stiffness and natural vibration frequency.
This novel energy system has three main advantages: (i) expensive and challenging construction in the offshore
and submerged environment can be dramatically reduced; (ii) introducing design flexibility can ensure a glob-
ally optimised design; and (iii) the risks of system failure due to changing environments or uncertain system
behaviours can be eliminated. Design comparisons are made of the SFT’s effectiveness against wave loading,
with and without the high potential energy system. An optimisation approach allows for a desk-top study that
minimises system cost while satisfying multiple design criteria, such as tube deflection and acceleration, tether
breaking strength, no slack-and-snap in the tethers, and SFT weight balance equilibrium.
Keywords: submerged floating tunnel; design; construction; optimisation
1 Introduction
Submerged floating tunnels (SFT) have the potential to cross broader and deeper waterbodies than the longest
cable bridges in the world because they use natural buoyancy to balance their own weight. Historically, ambitious
efforts have been put forth to cross some of the world’s most famous waterways [1-12]. Many extreme concepts
have been proposed, including a futuristic TransAtlantic tunnel with a nuclear-powered thruster system to keep
the tunnel at the desired location (Østlid [13]).
An SFT generally consists of three subsystems: a tube, a support (for the tube’s main body) and a shore connection
(for the tube’s ends). Based on the different approaches for support, SFTs can be classified into four types: tether-
supported, pontoon-supported, pier (column)-supported and unsupported free-spanned. To date, only two pier-
supported SFTs have been constructed [6]. Challenges in building other types of SFTs include uncertainties of
hydrodynamic loadings and the highly complex system behaviours between the loading, tube, supports and soils.
To address the challenges, and to allow an SFT to positively adapt to the environment and improve its structural
behaviour, this paper investigates a novel SFT structure [14]. It allows the SFT to self-tune the structural stiffness
(and thus natural vibration frequency) before or in the operational stage after construction. This also shifts some
high-risk and expensive offshore construction to relatively cost-effective and more accessible land works. This
paper studies its effectiveness when subjected to wave loading (response versus design criteria) using a recently
published SFT design optimisation framework [19].
2 SFT high potential energy system
2.1 Concept
Potential (internal) energy can be stored on an SFT by imposing permanent axial tension on the tube’s body
through its ends. This tension increases the tube’s geometric bending and torsional stiffness. Any change in stiff-
ness changes the system’s natural vibration frequency, and a more optimised natural vibration frequency can be
obtained by adjusting the magnitude of axial tension.
On
ly
2
fo
rR
Axial tension can be applied into the SFT in several ways. For instance, even a simple change in temperature from
the installed conditions and the operating conditions can change the axial tension. If the SFT is installed in summer
ea
and both ends fixed axially, by winter the temperature drop will induce an axial tension [9, 25] and the SFT will
din
keep some potential energy.
g
However, a more active system may be required. Fig. 1 shows two scaled SFT models with axial tension imposed
at
by hydraulic cylinders. Fig. 1a is a scaled tethered support SFT model in a wave basin. The SFT tube was tensioned
by 0, 20,000, 40,000, 80,000 newtons of force and then excited by knocking, waves or currents, during which the
W
tube’s strains, deflections, accelerations and tethers’ tension were recorded. As expected, and it was even observ-
CF
able by eye that the more tension imposed the less response of the model. Fig. 1b is a free-spanning SFT tube
model axially tensioned by a maximum of 100,000 newtons of force [14]. The free-decay vibration was produced
S
by enforcing an initial displacement at the mid-point of the tube and then releasing it suddenly, both in air and in
20
still water. It was observed that as the axial tension increases the model’s vibration frequency increases. Further
results from these tests are not shown in this paper for brevity but will be published in an upcoming PhD thesis.
24
(a) Tethered SFT test in basin [14,30,31] (b) SFT tube free vibration decay test [14]
Fig. 1. SFT scaled models subjected to axial tension
Fig. 2 Shore connection details

On
ly
3
fo
2.2 Design Considerations
rR
ea
Tension and anchor system of the shore connections. Fig. 2 shows the details of a possible shore connection to
tension the SFT tube to obtain permanent potential energy. One possible option is tensioning the tube by both
din
ends and anchoring it to the shore. Specifically, tension rods or cables are connected to the tube on one end and
g
to rock or precast concrete blocks on the other, as shown in Fig. 2a. Tension works can reference proven technol-
ogies, such as tendon cables used in cable bridges or external prestressing used in girder bridges. If the tube is
at
very long, or the temperature fluctuation is large, the length of the tube may be subject to cyclic elongation and
W
shortening. An expansion joint, as shown in Fig. 2a, can be set to accommodate the large axial movement of the
CF
tube. The large-diameter expansion joint is a proven technology used in the pipelines of China’s Three Gorges
project [17], for instance, and the connection between bored tunnel and immersed tunnel an example from the
S
Bosphorus strait [4]. Fig. 2b shows a cross-sectional view where the tension rod or cables are evenly set along the
circumferential outline of the tube. Another possible form tensions the tube from one end, with the other set as a
20
passive end, as shown in Fig. 2c. The tube is anchored into rock or concrete blocks that bear the axial tension.
24
Fig. 2d shows an alternative for soft ground, where the potential energy for an SFT can be obtained by building
an open caisson in advance [25]. The caisson ensures a strong anchorage for the tube and facilitates tube installa-
tion works to the shoreline.
Tube wall configurations. The tube wall can adopt composite structures, e.g., concrete-steel with shear studs,
concrete-steel allow-sliding, steel-concrete-steel for double waterproof, or concrete-rubber-steel for anti-collision.
3 Optimisation
3.1 Method
The SFT tube was modelled as Euler-Bernoulli beam in Eq. 1. The modelling of the tether accounted for geometry
and elastic stiffness using [22] but disregarded mass and fluid drag damping (this ensures that the response pre-
diction is conservative). The tether material assumed steel wires and their elastic stiffness, submerged weight and
Maximum-Breaking-Strength (MBS) was estimated as a function of tether’s nominal diameter using [23]. With
these assumptions the system’s stiffness, mass and damping matrices were assembled in Rayleigh-Ritz form [15].
The random waves used a JONSWAP spectrum [24] with the wave load estimated by the linearised Morison
equation. The response computation was in the frequency domain [15, 27] with three hours peak responses derived
[24]. The response calculation was verified by the free vibration decay model test (i.e., the test shown in Fig.1b),
the analytical solution [16], and the finite element time domain programming [28].
For the system verification, the SFT with and without High Potential Energy (HPE) system was optimised with
the objective function of system cost of tube, tether, and anchor. The SFT response was constrained to design
criteria when subject to wave loading, an extension of the methodology and work of [19]. The constraint functions
are (i) that the tube’s deflections and accelerations remain less than 0.3m and 0.5m/s2 respectively under wave
load [29], (ii) the tube’s cross-section structural capacity [20] derived from concrete and rebar steel’s stress-strain
relation [21] must be larger than the detrimental combination of extreme value of the moment and axial tension
of the tube, (iii) all tethers have no overstressing or slack (or in other words the sign-convention of the computed
tether angle at the anchor must remain the same), and (iv) the SFT is buoyant (i.e. the system buoyancy is greater
than or equal to the weight. Further, the interaction between the tube’s residual buoyancy and the tethers’ pre-
tension was considered in the optimisation framework. The open-sourced software OpenMDAO was used as a
highly efficient, easy-coding, and gradient-based design optimisation tool [18].
3.2 Results
Results are presented for the following design case. The SFT must be 200 m long, crossing 50m water depth,
submerged under water at 5 m depth and having a circular (inner diameter) cross-section of 5 m and a composite
wall configuration, as shown in Fig.5a. Three pair of vertical tethers connect the tube on one end and the gravity
anchor on the other end. The significant wave height is 2m and the wave mean period is 10s. The design variables
being optimised for were (i) the SFT tube concrete wall thickness and (ii) steel shell thickness, (iii) axial tension
(zero, or value if HPE activated) and (iv) nominal diameters for each pair of tethers.
The results for this base-case design are provided in the first two columns of Table 1. The first column is the
optimised design with no HPE and the second column the optimised design with HPE. The first row is inner
On
ly
4
fo
rR
diameter of the tube; default setting is 5.00m as the HPE design because the stiffness can be ensured via axial
tension. However, in the no HPE design, the stiffness can only be ensured by increasing second moment of area
ea
and thus inner diameter of tube needs to increase to 6.72m. The second and third rows are the wall thicknesses of
din
inner concrete layer and outer steel shell. No HPE design chooses thicker concrete wall and thinner steel shell
while HPE design chooses the opposite; these show different design strategy: (i) the no HPE design uses as much
g
concrete as possible because the unit cost of concrete is cheaper than that of steel in this model setting, and sec-
at
ondly, large volume of concrete is used to compensate the large buoyancy resulted from the enlarged tube, so as
to suppress the tether’s pretension and the anchor design cost. (ii) the HPE design uses as little concrete wall as
W
possible because the tube’s outer diameter can be further reduced to decrease wave load, and at the same time, the
CF
lightweight structure with more steel and less concrete yields a high Buoyancy-Weight-Ratio (BWR) design that
allows the tethers and anchors contributing more to the system stiffness from two sources, the tether’s elastic
S
stiffness of larger nominal diameter tether, and the tether’s geometric stiffness of the high pretension resulted
20
from high residual buoyancy. The fifth row is axial tension only concerned with HPE model. This value is opti-
mised to 0.34 10 N. The sixth to eleventh rows are design constraints relevant indexes. As seen, both designs
24
reach tube deflection limit but neither reach acceleration limit; HPE design reaches the limits of tube cross-section
structural capacity and tethers’ MBS while no HPE design has redundancy safety for both. And both designs have
a positive BWR: it means a residual buoyancy is reserved for pretension the tethers, and thus implies that the
tether is not abandoned by the optimiser, in another word, the free-spanning design is not the optimum option for
the two base cases. The last three rows are costs: as seen the no HPE design has higher cost and most cost came
from tube and shore, on the contrary the HPE design has higher cost in tether an anchor; these cost distribution
patterns are in line with the design strategy discussed.
Table 1. Results summary of six optimised design of different tethers arrangement and with or without HPE
Base cases Sensitivity study of tethers arrangement
Tether A — 0, 0, 0 [° Tether B —45, 45, 45 [° Tether-C — 0, 45, 0 [°
No HPE* HPE No HPE HPE No HPE HPE
Inner diameter [m] 6.71 5.00 5.00 5.00 5.00 5.00
Concrete wall thickness[m] 0.90 0.20 0.63 0.58 0.20 0.20
Steel shell thickness [m] 0.02 0.59 0.02 0.02 0.02 0.02
Tether diameters [mm] 27, 29, 27 55, 52, 55 34, 74, 34 10, 20, 10 69, 90, 69 69, 86, 69
Tube axial tension [×109N] 0.00 0.34 0.00 0.30 0.00 0.13
Maximum tube deflection [m] 0.30 0.30 0.30 0.30 0.27 0.23
Maximum tube acceleration [m/s2] 0.25 0.25 0.26 0.26 0.22 0.19
Tube cross-section safety factor 2.51 1.00 3.74 1.00 6.15 1.00
Tether MBS safety factor 1.02 1.00 1.00 1.00 1.00 1.00
Minimum force of tether’s anchor end 0.00 0.00 0.00 0.00 0.00 0.00
(slack-and-snap check) [N]
Buoyancy-Weight-Ratio (BWR) 1.03 1.51 1.05 1.00 2.20 2.20
Tube and shore cost [×106 USD] 6.55 3.34 3.66 4.38 1.64 1.64
Tether and anchor cost [×106 USD] 0.61 1.00 1.51 0.93 2.87 2.71
Total (system) cost [×106 USD] 7.15 4.34 5.17 4.51 4.51 4.35
The results for the optimised case are shown in Fig.3 for both the no HPE and HPE design. In the frequency
domain Fig.3a shows the norm of response transfer function matrices versus wave surface power density spec-
trum. Both designs avoid the peak frequency of waves by setting the SFT’s natural vibration frequency several
times higher. However, no HPE design has a smaller value of the norm of response transfer matrices than that of
the HPE design, because it endures larger wave forces due to a larger tube. On the other hand, HPE design has
twin spectral peaks, indicating a more flexible design, but at the same time the primary vibration mode has a
higher frequency than that of the no HPE design. In summary, the optimised HPE design shows the potential
advantage of higher flexibility in re-distributing internal forces and higher ‘rigidness’ in avoiding resonance.
Fig.3b shows the tube cross-section capacity envelop versus the maximum internal forces. The envelop for the no
HPE design is much larger particularly in compression, which is typical for a reinforced concrete tube. HPE design
has a much smaller capacity envelop in compression, but the envelop is nearly identical in combined compression
and tension. This is because the steel’s yielding stress and stress-strain relation is equal in both compression and
tension [21]. In terms of internal forces, in HPE design, though the pre-imposed larger axial force drives the
scatters of internal forces towards the envelop, the moment decreases significantly. The reason for the significant
decrease in moment internal forces for HPE design are not only the smaller tube size (that attracts less wave load)
On
ly
5
fo
rR
and stiffer tethers (that convey more load to mooring system then to the shore end), but also the system ‘flexibility’
and higher natural vibration frequency (as shown in Fig.3a).
ea
din
g
at
W
CF
S
20
(a) Norm of response transfer function matrix of SFT sys-
24
(b) SFT tube cross-section structural capacity versus maxi-
tem versus power density spectrum of wave mum internal forces
Fig. 3 Structural characteristics of the optimised design of vertically tethered SFT the base cases
Two design sensitivities were also studied, with results presented in the third to sixth columns of Table 1 and in
Fig. 4. In the first the layout of the tethers was changed and in the second the severity of the design sea-state was
increased.
The base-case design (Tether A) assumed all tethers for the mooring system were vertical. This was compared to
tethers inclined at 45 (Tether B) and a design where the mid-span tether was inclined with the rest vertical (Tether
C), with the design variables set the same as the base case. As shown in Table 1, HPE always yielded lower costs
than those without HPE, irrespective of the tether arrangements. However, Tether B, the design with inclined
tethers, has a lower system cost than design with vertical tethers. This is because the tethers can advantageously
restrain the tube’s lateral response.
As a second sensitivity case the wave period was varied and 4, 7, 10, 13, 16s were studied. All other base case
conditions were maintained (including Tether A and HPE, as in the second column of Table 1) with design vari-
ables being the tube’s inner diameter, wall thicknesses of concrete and steel shell, and the axial tension (when
using HPE). Fig. 4 shows the results. SFT designs with HPE still yield lower system cost than those without HPE.
For wave periods of 4 and 7s the SFT without HPE was unable to find a valid design. Designs with HPE could be
optimised because the axial tension could suppress the response to be within the allowed criteria. From engineer-
ing design optimisation view [26], adding axial tension as a design variable indicates that a new dimension for
design space exploration is added. Therefore, a better design can be achieved.
Fig. 4 Sensitivity study against wave period

On
ly
6
fo
4 Conclusions
rR
ea
In this paper use of a new high potential energy system to optimise the design of a Submerged Floating Tunnel
din
(SFT) was presented. Further, how optimisation analysis could be used to select an appropriate design under
different wave loading conditions and different tether designs was presented. It was shown that axial tension helps
g
the system better distribute structural inner member forces, increase system stiffness and alter natural vibration
at
frequency to prevent resonance again wave action. Those benefits reduce the outer dimension of the tube and thus
attract less wave load — a virtuous cycle. Design optimisation results show that the SFT design with HPE has a
W
higher chance of yielding a possible design to survive in unfavourable wave conditions, and always yields a more
cost-effective scheme.
CF
S
The optimisation framework does not represent a detailed design, but as a desk-top study provides confidence to
20
construction of a very long SFT in deep and open sea environments. The approach has potential for engineering
contractors to quickly and reliably develop a potential design concept, on which detailed assessment. The results
also show the potential of adding a HPE system to unlocking the economics of SFTs.
24
References
1. Ingerslev, C.: Immersed and floating tunnels. Procedia Eng 4, 51–59 (2010).
2. Andrew, C.E.: Floating tunnel for long wave crossing. ASCE 76(8), 1–13 (1950).
3. Grant, A.B.: Floating submerged tunnel for the strait of Messina. TBILISI, 53–60 (1972).
4. Gursoy, A.: Seismic joints and transition tunnels for the Messina straits fixed-link crossing. In: Ozdemir, L. North Amer-
ican Tunnelling ’96, pp. 203–209, Balkema, Rotterdam (1996).
5. Ahrens, D.: SFT — a concept whose time has arrived. Tunn. Undergr. Space Technol., 12(2), 317–336 (1997).
6. Minoretti, A., Peiretti H.C., Fjeld, A. et al.: Guidelines for submerged floating tube bridge. fib Bulletin 96 (2020).
7. Hiroshi, S. et al.: Underwater tunnel research and investigation submarine tunnel manual vol 2 case studies. Underwater
tunnel research investigation Legal Person (1995)
8. Kanie, S.: Feasibility studies on various SFT in Japan and their technological evaluation. Procedia Eng 4, 13–20 (2010)
9. Tveit, P.: Ideas on downward arched…underwater concrete tunnels. Tunn. Undergr. Space Technol., 15(1), 69–78 (2000).
10. Mazzolani, F.M., Landolfo, R., Faggiano, B., et al.: Structural analyses of the submerged floating tunnel prototype in
Qiandao lake (PR of China). Advances in Structural Engineering 11(4), 439–454 (2008).
11. Fjeld, A.: Feasibility study for crossing sognefjorden submerged floating tunnel. Reinertsen & Dr.techn.OlavOlsen (2012)
12. Project Team, Soreide T.H., Haugerud, S.A.: Bjørnafjord submerged floating tube bridge K3/K4 technical report.
Reinertsen, Dr.techn.OlavOlsen & Norconsult (2016)
13. Østlid, H.: When Is SFT competitive. Procedia Eng 4, 3–11 (2010)
14. CCCC Submerged Floating Tunnel Engineering Technical Joint Research Team: SFT build methodology, novel structure,
mechanism test, and engineer’s dialogues, Science Press, Beijing (2021) (in Chinese)
15. Clough, R.W., Penzien, J.: Dynamics of structures. 2nd edn. McGraw-Hill College, New York. (1993).
16. Duan, W., Li, Z., Zhang, J., et al.: Study on mathematical model and exact solution of overall structural response and
natural frequency of SFT. 30th Int Conf on Ocean, Offshore & Arctic Eng, ISOPE, Shanghai. (2020)
17. Cui, J.: Private Conversation (2019)
18. Gray, J.S., Hwang, J.T., Martins J.R.R.A., et al.: OpenMDAO: An open-source framework for multidisciplinary design,
analysis, and optimisation. Structural and Multidisciplinary Optimisation 59(4), 1075–1104 (2019).
19. Lin, W., Tian, Y., Cassidy, M.J.: A preliminary study on design optimisation of submerged floating tunnel. 42nd Interna-
tional Conference on Ocean, Offshore & Arctic Engineering, pp. 1–9, ISOPE, Melbourne (2022)
20. Zhao G.: Advanced reinforced concrete structure. Mechanical Industry Press (2012) (in Chinese)
21. MHURDPRC: Code for design of concrete structures. GB 50010-2015 (in Chinese)
22. Thai, H-T., Kim, S-E.: Nonlinear static … analysis of cable structures. Finite Elem. Anal. Des. 47(3), 237-246 (2011)
23. Ma, K-T., Luo, Y., Kwan, T., Wu, Y.: Mooring system engineering for offshore structures. Elsevier (2019)
24. Faltinsen, O.M.: Sea Loads on ships and offshore structures. Cambridge University Press (1998)
25. Lin, W., Liu, L., Tian, Y., Cassidy, M.J.: Building submerged floating tunnel in Kanas Lake. 43rd International Confer-
ence on Ocean, Offshore & Arctic Engineering, pp. 1–9, ISOPE, Rhodes. (2023)
26. Martin, J.R.R.A., Ning, A.: Engineering design optimisation. Cambridge University Press. Michigan. (2022)
27. Remseth, S., Leira, B.J., Okstad, K.M., et al.: Dynamic response and fluid/structure interaction of submerged Floating
tunnels. Computers and Structures, 72, 659-685 (1999)
28. Newmark, N.M.: A method of computation for structural dynamics. J. Eng. Mech. Div. Proc. ASCE, EM3, 67-94 (1959)
29. Utsunomiya, T., Watanabe, E.: Analysis and design of floating bridges. Prog. Struct. Engng. Mater., 5, 127-144 (2003)
30. Pan, W., Zhou Z., Gu, Q., et al.: Research on the Manufacturing method of 3D hydroelastic submerged floating tunnel
model, Dalian Marine Time University press, Dalian (2024) (in Chinese)
31. Zhang, H., Yang, Z., Li, J., et al.: A global review for the hydrodynamic response… Ocean Eng. 225, 108825. (2021)
On
ly
fo
rR
Hydroelastic analysis of a modular very large floating structure in a
ea
focused wave group
din
g
Zhiqiang Li1, Xingya Feng1, Dengshuo Chen1, and Jian-Fei Chen1
1
at
Department of Ocean Science and Engineering, Southern University of Science and Technology, Shenzhen, 518055, China
W
fengxy@sustech.edu.cn
CF
S
Abstract. This paper investigates the hydroelastic behavior of modular floating structures (MFS) made from
20
ultra-high performance concrete (UHPC) reinforced with prestressed fiber-reinforced polymer (FRP) for float-
ing photovoltaic (FPV) systems. The study analyzes an equivalent plate model to assess the impact of peak
24
frequency and focused location on structural response. Results indicate a limited variation in structural re-
sponse with peak frequency, except for notable peaks at 0.78 rad/s and 1.38 rad/s, corresponding to the struc-
ture's first two natural frequencies, signifying resonance. Higher natural frequencies have a negligible effect
on the overall response variation. Furthermore, the variation of the focused location does not significantly
alter the internal force response throughout the focusing wave generation process. Energy distribution across
the structure and transfer between modules mitigate disparities in force loading, making the MFS design adapt-
able and resilient to varying wave conditions.
Keywords: hydroelastic analysis, modular floating structure (MFS), Ultra-high performance concrete
(UHPC), focused wave.
1 Introduction
With the increasing global focus on renewable energy, solar power has emerged as a popular option due to its
abundance and sustainability. Land-based photovoltaic systems, especially in China, have made significant ad-
vancements [1]. However, limited land resources restrict further expansion, thereby highlighting the potential of
Floating Photovoltaic (FPV) systems, which install solar panels on water bodies like lakes, reservoirs, and offshore
environments. The development of FPV technology has driven the rise of Very Large Floating Structures (VLFS)
[2,3]. Selecting suitable materials for these structures is crucial for ensuring safety and cost-effectiveness. High-
density polyethylene (HDPE) is low-cost but lacks strength and durability [4], while steel, though robust, is ex-
pensive [5]. Concrete floating structures offer a balanced solution, combining cost-effectiveness and safety.
Concrete has been widely used in marine structures, such as floating bridges, platforms, and breakwaters [6].
Prestressed concrete addresses cracking issues in traditional reinforced concrete, allowing for the design of longer,
thinner, and more cost-effective structures [7]. In recent years, Ultra-High Performance Concrete (UHPC) has
increased the attractiveness of floating structures due to its superior mechanical properties, durability, and corro-
sion resistance [8]. UHPC has been effectively utilized in bridge structures, reducing size and weight while main-
taining strength and lowering life-cycle costs [9]. These benefits make UHPC an appealing option for FPV sys-
tems.
While practical engineering applications of UHPC in floating structures are not found, some studies explored
design concepts utilizing UHPC as a fundamental material in FPV systems. One design concept of double-seg-
ment floating sandwich structures composed of two steel panels and a honeycomb UHPC core layer was proposed,
and the mechanical performance optimization of the floating structure has been conducted by structure-fluid cou-
pling method [2]. Another lightweight and high-stiffness floating structure composed of UHPC surface panel was
designed as the upper support for solar panels [10]. A hydrodynamic-structural-material approach has been pro-
posed for designing these large floating structures.
Our previous research proposed a modular box-pontoon structure using UHPC and prestressed FRP bars, fea-
turing thin plates that enhance flexural performance and durability [11,12], balancing cost and crack resistance.
This modular design, combined with the use of UHPC and FRP, leads to a structure that is not only cost-effective
but also highly resilient, stable, and environmentally friendly. These studies examined the hydroelastic response
of the proposed MFS in regular waves. Building upon this work, this paper further investigates the hydroelastic
analysis of the MFS in focused waves. An equivalent plate model is first employed to conduct the hydroelastic
analysis. Then the effects of the focused position and the peak frequency of the focused wave are then discussed.
discussed.
On
ly
2
fo
2 Equivalent plate model of MFS
rR
ea
In this study, the proposed MFS designed for FPV systems is composed of multiple box-pontoon modules
din
connected by hinge connectors. Each module is constructed using UHPC-CFRP composites, as shown in Fig. 1.
The deck, bottom plate, and side walls are made of UHPC plates reinforced with prestressed CFRP bars. To
g
complete the hydroelastic analysis, an equivalent plate model of the box-pontoon structure must first be devel-
at
oped.
W
CF
S
20
24
Fig. 1. Modular floating structures applying UHPC-CFRP composites.
Based on the principle of modal equivalence, a comparative analysis of the vibration frequencies and mode
shapes of the equivalent plate model and the box-pontoon model was conducted to ensure consistency in the
vibration responses across all modes. Fig.2 shows the comparison of mode shapes between the box-pontoon model
and the equivalent plate model from the 7th to the 10th mode shapes. It can be observed that the mode shapes of
the box-pontoon structure model closely match those of the equivalent plate model, and the frequency values of
each mode are also similar, indicating that the equivalence principle is satisfied (see Fig. 2).
(a) (b) (c) (d)
(e) (f) (g) (h)
Fig. 2. The 7th, 8th, 9th, and 10th modes of equivalent plate and actual structural models. (a), (b), (c), and (d) represent the
7th, 8th, 9th, and 10th modes of the equivalent plate model, respectively; (e), (f), (g), and (h) correspond to the 7th, 8th, 9th,
and 10th modes of the actual structure model.
3 Hydroelastic analysis
To investigate the hydroelastic behavior of the proposed MFS in focused waves, a numerical model HN4, was
investigated based on the equivalence principle, as shown in Fig.3. This model consisting of 16 interconnected
modules, each module possesses dimensions of 15 m by 15 m with a 2 m depth, resulting in an overall structure
spanning 60 meters in length (L) and 60 meters in width (B). The structure is designed with a draft of 0.8 m.
On
ly
3
fo
rR
ea
din
g
at
W
CF
S
20
Fig. 3. HN4 model.
24
3.1 Focused pressure model
Based on a random wave spectrum, a linear NewWave model was proposed to generate a focused wave group
[13]. This model has been widely used in the ocean engineering field to represent extreme waves. For a given
wave energy spectrum, the focused wave elevation is given by:
N
 ( x, y, z)  An cos[kn ( x  xc ) cos   kn ( y  yc )sin   n (t  tc )] (1)
n 1
where An , k n and ωn are the wave amplitude, wave number and wave circular frequency of the nth wave com-
ponent, respectively. are the coordinates of the focal position and 𝑡𝑐 is the focal time. The amplitude is represented
as
S ( )
An  N n (2)
n1 S (n ) 
where the total number of wave components N is supposed to be large enough to regenerate the desired wave spectrum. The
random waves are generated by 3-parameter JONSWAP wave spectrum which can be expressed as
  p
5  4
5  exp( 0.5(
 p
)2 )
(3)
S ( )  (1  0.287ln  ) H s2 p5 ( ( )4 )
16  4 p
where γ is the peak enhancement factor, Hs the significant wave height, and ωp the peak frequency. The peak
shape parameter is defined as
0.07,    p
  (4)
0.09,    p
In this paper, a modified pressure focused wave model is proposed to represent the action of focused wave
group.
P( x, y, t )  nN1 An Pn ( x, y)cos(nt  n ) (5)
where Pn (x,y,t)and 𝜑𝑛 are the amplitude and phase of the hydrodynamic pressure induced by the nth wave com-
ponent. The former can be obtained by the procedure for regular waves proposed in our previous studies [10,11]
is expressed as φn =-ωn t c -k n (x-xc )-k n (y-yc ).
3.2 Effect of peak frequency
In this study, the focused waves are generated using the focused pressure model, which modulates the phases
of the hydrodynamic pressures caused by individual wave components under random sea conditions so that they
converge at a designated location and time. The random sea conditions are described using the three-parameter
JONSWAP spectrum, where different sea states have varying significant wave heights and peak frequencies,
which are crucial factors in analyzing structural response. Since the model assumes a linear relationship, the wave
On
ly
4
fo
rR
height is proportional to the amplitude of the structural response. Therefore, this study primarily examines the
ea
impact of peak frequency on structural response under different sea conditions. As shown in Tab.1, shows the
range of peak frequency (ωp ) from 0.55-1.49 rad/s, corresponding to peak period ranging from 4.20s-11.39s. The
din
significant wave height (Hs ) is set as 2.5m. The ωlow and ωhigh represent the lower and higher frequency limits
of the selected wave components.
g
at
Table. 1. The frequency range of numerical cases.
W
Case Hs (m) Hs (m)(s) ωp (rad/s) ωlow (rad/s) ωhigh (rad/s)
H2.5ω1.49 2.5 4.20 1.49 0.75 3.72
H2.5ω1.38 2.5
2.5
4.56
5.04
1.38
1.25
0.69
0.63CF 3.45
3.13
S
H2.5ω1.25
H2.5ω1.14 2.5 5.52 1.14 0.57 2.85
20
H2.5ω1.02 2.5 6.12 1.02 0.51 2.55
24
H2.5ω0.92 2.5 6.83 0.92 0.46 2.30
H2.5ω0.78 2.5 8.03 0.78 0.39 1.95
H2.5ω0.66 2.5 9.47 0.66 0.33 1.65
H2.5ω0.59 2.5 10.55 0.59 0.30 1.48
H2.5ω0.55 2.5 11.39 0.55 0.27 1.38
As shown in Fig.4, the maximum values of the internal force responses of the structure vary with the peak
frequency. The most significant responses are the maximum longitudinal bending moment 𝑀𝑦 and the maximum
shear force 𝐹𝑧 , which are analyzed as the variables of internal force response with respect to peak frequency
changes. It can be observed that the maximum values of the internal force responses of the structure do not sig-
nificantly vary with the peak frequency. However, in the frequency ranges of 0.78 rad/s and 1.38 rad/s, the internal
force responses reach relatively higher peaks.
By analyzing the natural frequencies of the structure, it is found that the first six elastic modes of the structure
correspond to natural frequencies of 0.80 rad/s, 1.48 rad/s, 2.13 rad/s, 3.52 rad/s, 3.93 rad/s, and 4.00 rad/s, re-
spectively. It can be seen that the frequencies of the first two elastic modes (0.80 rad/s, 1.48 rad/s) are very close
to the peak frequencies (0.78 rad/s, 1.38 rad/s) at which the structure’s responses show significant peaks. Thus, it
can be inferred that the significant peaks in the internal force responses are closely related to the natural frequen-
cies of the structure. At these frequencies, the structure is prone to resonance, leading to a significant increase in
internal force responses. Additionally, apart from the first two modes, the other natural frequencies of the structure
(2.13 rad/s, 3.52 rad/s, 3.93 rad/s, and 4.00 rad/s) are significantly higher than the range of wave peak frequencies
(0.55 rad/s to 1.49 rad/s). Hence, considering the contributions of other modes to the structural response, the
overall trend of structural response variation with peak frequency is not obvious.
Fig.4. The structural responses with the peak frequency. (a) 𝑀𝑦 ; (b) 𝐹𝑧
3.3 Effect of focused location

This section explores the influence of different focusing locations on the structure's internal forces. To achieve
this, we systematically positioned four focusing points (P1, P2, P3, P4) along the x-axis at equal intervals, effec-
tively dividing the axis into three sections. Fig. 5 and 6 illustrate the contour plots of internal force response
amplitudes (My and Fz) for each of the 16 modules under these varied focusing conditions. Fig.5 reveals that the
largest bending moments are concentrated at the modules positioned at the bow and stern. In contrast, the bending
On
ly
5
fo
rR
moment amplitudes in the middle modules are significantly lower. This observation suggests that the bow and
ea
stern modules act as "breakwaters," shielding the interior modules from the brunt of the bending forces.
Further analysis of the shear forces (Fig.6) indicates that the areas with the highest shear force intensities pri-
din
marily reside near the hinge positions of the modules at the bow and stern. This phenomenon aligns with the
observed concentration of bending moments at these locations. The combined effects of wave diffraction and
g
radiation, intensified at the bow and stern, generate greater wave forces in these regions. This leads to higher
at
internal forces within the bow and stern modules, while the internal forces in the middle modules remain compar-
atively lower.
W
Furthermore, comparing the structural response diagrams at different focus locations, it can be observed that
CF
the variation in focus location does not significantly impact the internal force response throughout the entire fo-
cusing wave generation process. During the focusing wave's interaction with the structure, wave energy might be
S
transferred and dispersed across different components, leading to minimal overall differences in the internal force
20
response. This effect is particularly pronounced in larger or multi-modular structures, where energy distribution
is more widespread. The energy transfer among modules can further mitigate disparities in force loading at spe-
cific locations.
24
(a) P1 (b) P2 (c) P3 (d) P4

(a) P1 (b) P2 (c) P3 (d) P4

4 Conclusion
This study explored the hydroelastic analysis of modular floating structures (MFS) constructed from UHPC
reinforced with prestressed FRP, specifically designed for FPV systems. By analyzing an equivalent plate model,
the influence of peak frequency and focused wave location on structural response was examined.
The analysis revealed that the structural response exhibits limited variation with peak frequency, except for
notable peaks coinciding with the first two natural frequencies, indicating resonance phenomena. Higher natural
frequencies had a negligible impact on the overall response.
Furthermore, the variation in focused location did not significantly alter the structural response. This is at-
tributed to the energy distribution and transfer within the multi-modular structure, effectively mitigating variations
in force loading across different sections. These findings highlight the inherent adaptability and resilience of the
proposed MFS design, offering a promising solution for FPV systems exposed to various wave conditions. Further
research could explore the optimization of structural parameters and material properties to enhance the perfor-
mance and stability of these systems in engineering applications.
On
ly
6
fo
References
rR
ea
1. Gorjian S., Sharon H., Ebadi H., et al.: Recent Technical Advancements, Economics and Environmental Impacts of Float-
din
ing Photovoltaic Solar Energy Conversion Systems. Journal of Cleaner Production, 2021, 278:124285.
2. Chen Y, Ma H, Li A, et al.: Hydroelastic analysis of double-segment floating sandwich structures under wave action.
g
Ocean Engineering, 2022.
at
3. Karpouzoglou T., Vlaswinkel B., Molen J V D.: Effects of large-scale floating (solar photovoltaic) platforms on hydro-
dynamics and primary production in a coastal sea from a water column model. Ocean Science, 2020, 16(1):195-208.
W
4. Dai J, Zhang C, Lim H V, et al.: Design and construction of floating modular photovoltaic system for water reservoirs.
CF
Energy, 2020, 191(Jan.15):116549.1-116549.15.
5. Claus R, López, M.: Key issues in the design of floating photovoltaic structures for the marine environment. Renewable
S
and Sustainable Energy Reviews, 2022, 164.
6. Wang C M, Tay Z Y: Very Large Floating Structures: Applications, Research and Development. Procedia Engineering,
20
2011, 14(3):62-72.
24
7. Meng G., Jia J., Wang J.: Study on flexural behavior of prestressed ultra-high strength concrete beams. Journal of Harbin
Engineering University, 2013, 34(5):575-580.
8. Bajaber M A., Hakeem I Y.: UHPC Evolution, Development, and Utilization in Construction: A Review. Journal of
Materials Research and Technology, 2020, 10(3).
9. Li C, Feng Z, Pan R, et al.: Experimental and Numerical Investigation on the Anchorage Zone of Prestressed UHPC Box-
Girder Bridge. Journal of Bridge Engineering, 2020.
10. Jin Z, Liu Y, Zhu M, Fang H, Li A.: A novel analytical model coupling hydrodynamic-structural-material scales for very
large floating photovoltaic support structures, Ocean Engineering, 2023,275, 114113
11. Li Z Q, Chen D S, Feng X, Chen J F.: Hydroelastic analysis and structural design of a modular floating structure applying
ultra-high performance fiber-reinforced concrete. Ocean Engineering, 2023, 277, 114266.
12. Li Z, Chen D, Feng X.: Hydroelastic and expansibility analysis of a modular floating photovoltaic system with multi-
directional hinge connections. Ocean Engineering, 2023, 340289, 116218.
13. Chen D, Feng X, Hou C, et al.: A coupled frequency and time domain approach for hydroelastic analysis of very large
floating structures under focused wave groups. Ocean Engineering, 2022, 111393
On
ly
fo
rR
Stability and Hydrodynamic Analysis of a Cylindrical Floating Planter
ea
din
Senqin Zhang1, Jinghua Wang1,2 and Xiaolin Zhao1,3
g
1 Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hong Kong, China
at
2 Shenzhen Research Institute, The Hong Kong Polytechnic University, Shenzhen, China
3 Research Institute of Land and Space, The Hong Kong Polytechnic University, Hong Kong, China
jinghua.wang@polyu.edu.hk
W
CF
S
Abstract. Floating coastal structures, considered as an alternative to land reclamation for expanding livable
20
space, are regarded as cost-effective and sustainable solutions for land-scarce cities. Among these structures,
floating planters stand out for their ability to enhance the aesthetics of water bodies, provide habitats for aquatic
24
organisms, and reduce land usage. Despite their advantages, there is a dearth of research and engineering
guidelines for the design and construction of floating planters. In this research, a preliminary design of a cy-
lindrical floating planter is proposed, and its stability and hydrodynamics in still water is analyzed. The planter
consists of an inner steel skeleton and an outer UHPC skin. The cylinder is 5 m in diameter and 5 m in height.
The overall weight including ballast, soil and trees is about 699.5kN. For the stability analysis, the wind load-
ings on the coaxial branched trees are considered. With appropriate weight distribution and ballast, the stability
of the structure is analyzed and verified under designed wind condition in still water. The Runge-Kutta Method
is used to obtain the motion response of the floating planter by solving the six-degree-of-freedom motion
equation. The results showed that the natural periods are 4.2611s for the heave, 11.0291s for the sway and
5.5655s for the roll, which agree well with the theoretical predictions. Further study is still in demand to ana-
lyze its motion under design wave conditions.
Keywords: Floating Structure, Response Amplitude Operator, Free decay tests.
1 Introduction
Floating planter is a type of auxiliary coastal infrastructure that can be beneficial for the sustainable development
of the region [1]. Despite their potential benefits, the design and construction of floating planters, lack adequate
research and engineering codes. Particularly, when considering the application of floating planters in locations
like Hong Kong, several challenges arise due to the unique environmental conditions. Hong Kong's shallow water
depth coupled with significant tidal range (about 1.54 meters [4]) presents a hurdle for the deployment of floating
structures. Additionally, the region is prone to typhoons, which impose an enormous challenge on the stability
and safety of floating planters. To date, guidelines and specific engineering codes for the design and construction
of floating planters are in still demand.
This study aims to answer this open question about how to design a floating planter with sufficient stability in
still water based on the current knowledge of design practice for floating offshore structures. The paper is struc-
tured in the following way: Section 2 introduces the environment conditions and describes the design of the
planter; Section 3 reports the results of stability analysis and free decay tests; Section 4 makes concluding remarks
with a discussion on the limitations of the present study.
2 Floating planter design
2.1 Environment conditions

The site to be studied is located in the vicinity of Kau Yi Chau, where a land reclamation project is under consid-
eration to meet the increasing need of land space (see Fig. 1 [6]). The mean water depth in this area is about 9m.
The maximum gust wind speed in Hong Kong water is 44 m/s under normal environmental conditions [4]. Under
extreme environmental conditions, the maximum gust wind speed in Hong Kong water is expected to be 70m/s
with a return period of about 50 years [4].
2.2 Planter Design

The pontoon of the planter is in cylindrical shape of 2.5 m in radius and 5 m in height. The wall is made of UHPC
with a thickness of 100 mm, and the skeleton is made of steel with square hollow section (80 × 80 × 8 mm). To
On
ly
2
fo
rR
enhance stability, the stone ballast is adopted. The bottom of the pontoon is filled with stones, and the top of the
ballast is secured with a spacer. The opening at the top of the pontoon provides space for storing soil and plants
ea
with a dimension of 1.5 m in radius and 1.2 m in height. The overall design of this floating planter is illustrated
din
in Fig. 2.
g
at
W
CF
S
20
24
Fig. 1. Location of the land reclamation.
We consider a commonly used landscape tree species, the Coaxial Branched Tree, which typically has a height
ranging from 2.5m to 5m and a crown width between 1.5m and 3.0m [8]. In this study, we take the windward area
of the tree to be 4m2. In the unmoored state, we need to ensure that the floating platform will remain stable under
normal wind loads (refer to section 3.1). Whereas in extreme wind conditions, floating planters will be reallocated
and anchored in a sheltered harbor for safety.
(a) UHPC wall and steel skeleton (b) Side view of the floating planter
Fig. 2. An illustration of the floating planter.
2.3 Mooring
For safety purpose, it is often necessary to use moorings to limit their range of motion in ocean environment.
Commonly used mooring systems include mooring piles, catenary mooring system and taut leg mooring system.
Mooring piles reduce the lateral movement of the floating body and minimize the risk of capsizing and are often
used for near-shore structures. Catenary moorings limit the movement of the float by means of chains under their
own weight, which are not in a straight line between the float and the anchor since they do not exhibit much
tension or sag, thus allowing the float to move within a certain range. Taut leg mooring system use tensioned
mooring ropes to hold the floats in place, resulting in a smaller mooring footprint and a mooring system that uses
less cable material [11].
In this paper, a tensioned mooring system is used, which is made of 20 mm diameter polyester fiber with a
material stiffness coefficient λ = 653.5 kN/m [18]. The mooring line length L = 5.99 m. The layout of the mooring
is shown in Fig. 3.
On
ly
3
fo
rR
ea
din
g
at
W
Fig. 3. Top view (left) and side view (right) of the mooring system.
3 Stability and free decay tests CF

S
3.1 Stability
20
24
Stability is the ability of a floating structure to return to its original equilibrium position after displacement by
external forces. This property ensures that the structure will return to equilibrium once these forces cease to act.
When the floating structure tilts at a small angle, the waterline changed from WL in still water to W1L1 (see Fig.
4). Despite the tilt, the structure's weight remains constant, and its center of gravity does not shift. Consequently,
the displacement volume of the structure stays unchanged. However, the shift in the waterline alters the shape of
the submerged volume, causing the center of buoyancy to move from point B to B’. Buoyancy and gravity form a
torque that rotates the structure back to its original equilibrium position.
Fig. 4. Cross section and dimension of the planter.
The center of gravity G lies below the metacenter M, and the direction of the moment produced by gravity is
opposite to the direction of tilt. When the external force is removed, this configuration helps the structure return
to its original equilibrium state, signifying a stable equilibrium. Conversely, if the structure cannot return to its
original state, it is considered to be in an unstable equilibrium.
The metacentric radius BM is a crucial factor in determining stability [14]. It is mathematically expressed as:
BM = I/∇ , (1)
where I represents the moment of inertia of the waterplane area about the axis of inclination, ∇ denotes the volume
of displacement of the floating structure.
Fig. 5. Righting arm and wind moment in different heel angles.

On
ly
4
fo
rR
According to [4], the design wind speed under normal environmental conditions is specified as 44 m/s. This
value is used for stability checks. The wind load can be calculated using equation as below [10]:
ea 𝐹 𝜌 𝐶 𝑈 𝐴 (2)
din
which is approximately 13,187 N, where Atr is windward area of trees, ρa is the air density, Cd = 0.769 is the
g
drag coefficient [12], and U is the incoming wind speed.
at
The steel skeleton, UHPC shell and soil weighs 28.34 kN, 196.34 kN and 152.68 kN, respectively. A total of
292.12 kN of stones is added at the bottom to enhance the stability. The total mass of the floating planter after
W
counterweighting is 699.5 kN, and the height of the center of gravity KM is 2.21m. The draft is calculated to be
CF
3.55m, and center of gravity KB = draft/2 = 1.77 m, center of buoyancy KG = 1.99m and BM = 0.44m. The
recovery moment and the overturning moment curve under wind load are calculated, and result is shown in Fig.
S
5.
20
As shown in the Fig. 5, 3 regions are identified, i.e., A, B and C. Note that, the center of gravity is lower than
the center of rotation (i.e., KM>KG in Fig. 4). According to [15], dynamic stability can be satisfied when area A-
24
area +B-area > 1.4 (B-area +C-area). In the case shown in Fig. 5, A-area+B-area = 94.75kNꞏrad is greater than
1.4 (B-area+C-area) = 41.016kNꞏrad. Hence, the floating planter satisfies the dynamic stability criterion.
3.2 Free decay tests
Governing equation and numerical scheme

For hydrodynamic analysis, a floating body can be regarded as rigid. Its six motional degrees-of-freedom (DOF)
can be described by the equation below [2]:
𝑴 𝐀 η 𝑩η 𝐂η 𝐹 𝐹 (3)
where
η, η and η are the displacement, speed and acceleration,
Frs represents reaction forces from the mooring system,
Fw contains the wind-excited forces on the fixed structure,
M are components of the generalized mass matrix for the structure,
A are hydrodynamic mass or added mass,
B is the hydrodynamic damping or radiation damping,
C is the hydrostatic stiffness,
and
𝐶 0 (4)
𝐶 𝜌𝑔𝐴 (5)
𝐶 𝜌𝑔 ∬ 𝑦 𝑑𝑠 𝜌𝑔𝑉 𝐾𝐵 𝐾𝐺 (6)
𝐹 𝐤η (7)
where Awp is the waterplane area, V is the displaced volume of water, ρ is the water density, g is the gravitational
acceleration, S is the floating structure submerged surface. The coordinate axes 0xyz are shown in the Fig. 3.
The idealized stiffness k is derived by assuming that each mooring cable is weightless and perfectly elastic
with a known tension and elastic stiffness [13]. Each cable has a constant tension Tm an elastic stiffness λ and is
strung between coordinates (x1, y1, z1) on the seabed and (x2, y2, z2) at the toe (see Fig. 3). L is the mooring line
length. The stiffness factor for each cable can be expressed as:
𝑘 𝜆 𝑐𝑜𝑠 𝛽 𝑠𝑖𝑛 𝛽 (8)
𝑘 𝜆 𝑐𝑜𝑠 𝛾 𝑠𝑖𝑛 𝛾 (9)
𝑘 𝑘 z 2𝑘 y z 𝑘 y (10)
where
cos 𝛼 , cos 𝛽 , cos 𝛾 (11)
The cable stiffness matrix is computed for each cable in a mooring system and summed to yield a total mooring
stiffness matrix. The fourth-order Runge-Kutta Method is used to solve the equation (3) in time domain with a
time step size of 0.005 second, which provides high accuracy.
On
ly
5
fo
Verifications
rR
In this case, the maximum gust wind speed of 70 m/s under extreme conditions (return period 50 years [4]) is
ea
considered for analysis. The results are summarized in Fig. 6 including displacements in heave, sway and roll for
din
different damping factors c. Damping factors typically ranging from 0.01 to 0.3 [16].
g
at
W
CF
S
20
(a) Heave (b) Sway
24 (c) Roll
Fig. 6. Time history of displacement in heave (a), sway (b) and roll (c) under various damping factors.
When the damping coefficient c is 0, the motions are periodic, and the corresponding natural frequency can be
estimated. By performing a Fourier transform on the time signal, the frequency spectra can be obtained as shown
in (a) . The natural period can be easily identified and extracted from the figures, i.e., T33 = 4.08 s for heave, T22
= 14.76s for sway, and T44 = 5.56 s for roll.
(a) Heave (b) Sway (c) Roll

Fig. 7. FFT Amplitude Spectrum in heave (a), sway (b) and roll (c).
To verify the accuracy of the calculations, the intrinsic frequency of the floating body can also be determined
analytically through the formula [7]:
ω 𝑘 /𝑚 (12)
where ktotal is the stiffness including mooring line stiffness and hydrostatic stiffness, mtotal is the generalized total
mass including generalized mass and added mass. The natural frequencies obtained are T33 = 4.0788s, T22 =
14.7606s, T44 = 5.5655s. Comparing with the previous calculations obtained from Eq. (3), the results are consistent
with a maximum difference less than 1.5%.
3.3 Discussions
In general, floating installations with a length/width ratio of less than 3 should not tilt at an angle of more than
15° [3]. The maximum tilt of the towing structure should not exceed 5° in wind conditions of 16 m/s [7]. In this
floating planter design, the maximum inclination angle was 0.075 rad, or 4.30°, which shows that the float is quite
stable under mooring.
In designing the floating body, it is crucial to avoid having motion from resonating with the ocean surface wave.
While in Hong Kong, the mean peak period ranges from 4.90 ~ 6.74 seconds (0.148 ~ 0.204 Hz) [17]. In further
research, to ensure the safety of the floating planter, adjustments to the parameters of the mooring cable (such as
diameter, material, etc.) can be made so that the vibration frequency of the floating planter differs from the wave
frequency, thereby preventing resonance. This will be carried out in a future study.
On
ly
6
fo
4 Conclusions
rR
ea
This research focused on designing a cylindrical floating planter and analyzing its stability and 6 DOF motions in
din
still water. Fourier analysis was carried out to verify the calculations of the natural frequencies. From this study,
the following primary conclusions can be drawn:
g
1. It is feasible to design stable cylindrical floating planters through careful mass distribution.
at
2. Under the mooring condition, the motion of the floating planter in sway and roll is considerably within a
small range, indicating its safe deployment at designed location.
W
3. By performing Fourier analysis on the time domain simulation data, the natural frequency can be accurately
calculated, which is useful for the optimization of the design in the future.
CF
The limitation of this study is the neglection of the incident ocean surface waves. A more comprehensive study
will be conducted in the follow-up study.
S
20
References
24
1. Barltrop, N.: Floating structures: a guide for design and analysis. 1st edn. Oilfield Pubns Inc, England (1998).
2. Bergdahl, L.: Wave-induced loads and ship motions. Technical report, Chalmers University of Technology, Sweden
(2009).
3. China Classification Society: Classification specification for offshore mobile platforms, China (2023).
4. Civil Engineering Office, Civil Engineering and Development Department.: Port Works Design Manual Part 1: General
Design Considerations for Marine Works. Hong Kong, China (2023).
5. Civil Engineering Office, Civil Engineering and Development Department.: Port Works Design Manual Part 2: Guide to
Design of Piers and Dolphins. HKSAR Government, China (2023).
6. Development Bureau, Civil Engineering and Development Department, Planning Department.: Study on the artificial
islands in the central waters. Hong Kong, China (2023).
7. El-Reedy, M. A.: Offshore structures: design, construction and maintenance, 2nd edn. Gulf Professional Publishing,
United States (2019).
8. Guangdong Landscape Architecture and Ecological Landscape Association, The Hong Kong Institute of Landscape Ar-
chitects Association for the Promotion of Landscaping and Greening in Macao.: Standard Specification of Nursery Plant
Materials for Guangdong Hong Kong and Macau. Guangdong Landscape Architecture and Ecological Landscape Asso-
ciation, China (2021).
9. Lewis, E.V.: Principles of Naval Architecture: Motions in Waves and Controllability. 2nd edn, Society of Naval Archi-
tects, United States (1989).
10. Lin, P., Hu, G., Tse, K. T., Song, J., Leung, A. K., Jiang, Z., & Zou, L.: Characterizing wind-induced reconfiguration of
coaxial branched tree. Agricultural and Forest Meteorology, 340 (2023).
11. Ma, K. T., Luo, Y., Kwan, C. T. T., Wu, Y..: Mooring system engineering for offshore structures. 1st edn. Gulf Profes-
sional Publishing, United States (2019).
12. Manickathan, L., Defraeye, T., Allegrini, J., Derome, D. and Carmeliet, J.: Comparative study of flow field and drag
coefficient of model and small natural trees in a wind tunnel. Urban Forestry & Urban Greening, 35, 230-239 (2018).
13. Patel, M.H., Lynch, E.J.: Coupled dynamics of tensioned buoyant platforms and mooring tethers. Engineering Structures,
5(4), 299–308 (1983).
14. Sheng, Z.B., Liu Y.Z.: Ship Design. 2nd edn. Shanghai Jiao Tong University Press, China (2003).
15. Technical Committee CSB/17.: Code of practice for Maritime structures —Part 6: Design of inshore moorings and float-
ing structures, United Kingdom (1989).
16. Wang, S.: Analytical solutions for the dynamic analysis of a modular floating structure for urban expansion. Ocean En-
gineering, 266, 112878 (2022).
17. Wave Measurement ， https://www.cedd.gov.hk/eng/public-services-forms/port-and-marine/wave-measurement/in-
dex.html, last accessed 2024/05/03.
18. Yu, X., Lei, H. & Wang, Y.: Research on the applicability of multi-points mooring system for floating platform in shallow
waters. Naval Science and Technology, 42(1), 105-110 (2020).
On
ly
fo
rR
Effect of mooring pretension force of a soft-connected floating photovoltaic
ea
system under regular waves.
din
Jenish Baniya 1, Jian Dai 1, Zhiyu Jiang 2, Zhang Chi 3 and Arvind Keprate 4
g
at
1Department of Built Environment, Oslo Metropolitan University, Oslo, Norway
2
Department of Engineering Sciences, University of Agder, Grimstad, Norway
W
3 Technology Centre for Offshore and Marine Singapore, Singapore
CF
4 Green Energy Lab, Department of Mechanical, Electrical and Chemical Engineering, Oslo Metropolitan University, Oslo,
Norway
S
Email: jeban4151@oslomet.no
20
Abstract. The rise in global temperature has accelerated the demand for replacing fossil fuels with sustainable
renewable energy. Alongside wind, waves and current energies in the oceans, there has been huge interest in
24
developing large floating photovoltaic (FPV) plants in offshore environments. Applying FPV structures in
offshore environments remains challenging due to harsh environmental conditions. This paper numerically
studies a recently developed soft-connected FPV concept for harsh offshore conditions. The case of a small
array consisting of 6 modules interconnected by soft ropes that were experimentally examined is considered.
A numerical model for the time domain analysis is built using the commercial solver SIMO. The numerical
model is calibrated and validated according to the available experimental results. Next, a parametric study is
conducted to examine the effect of different mooring line pretensions on the global responses of the FPV array
under regular waves. The study will provide a better understanding of the effect of mooring pretension on the
global FPV design. Moreover, recommendations are presented for further optimisation and development of
such FPV systems.
Keywords: Floating structure, offshore floating photovoltaic, hydrodynamic analysis, numerical calibration
1 Introduction
There has been an increased interest in harvesting sustainable and efficient energy sources to replace fossil fuels.
In the past decades, efficient alternatives to fossil fuels have been substantially developed, such as hydropower,
solar and wind. Among these alternatives, solar energy generates electricity by directly converting solar radiation
through photovoltaic (PV) cells, and energy production is sustainable and stable because of the abundant amount
of solar radiation. The advancement in solar technology has also significantly reduced the cost of developing solar
farms, a large cluster of PV panels. Many large solar farms have been installed on land. Installation of such solar
farms on land requires a large amount of open space as land space is scarce to sustain the growing population.
Large-scale solar project such as Sarnia in Canada occupies a land area of 365 hectares [1]. Therefore, there has
been a huge interest in developing floating photovoltaic (FPV) for land-scarce countries and regions with high
land prices. Installing such plants over water bodies frees spaces for other purposes such as agriculture and hous-
ing.
A typical FPV system consists of a pontoon with mounted solar panels, connection systems, mooring and
anchoring, spreading over a huge waterplane area. Developing FPV farms has numerous benefits compared to
land-based PV farms, e.g., the natural cooling effect which increases power generation, reduces water evaporation,
reduces eutrophication, and improves water quality. Most FPV farm projects are installed in lacustrine environ-
ments in lakes and reservoirs with calm and self-contained water. Compared to the in-land water body, the ocean
provides significantly more space for FPV farms. However, there are many challenges in developing FPV plants
for offshore conditions, where the structures are exposed to saltwater, marine growth, and UV radiation. These
can reduce the life span due to material degradation. Moreover, FPVs in marine environments are constantly
exposed to complex excitations from wind, waves, tides, and currents. For these reasons, despite a few studies on
FPV, the application is still limited to calm [2] and relatively mild conditions [3]. However, these studies were
conducted for FPV developed for calm and benign water conditions. There is still a need for research on FPV
farms in harsh environments. There are two challenges for developing FPV farms in harsh environments. First,
the dynamic motions of large FPV structures induced by strong wind and waves can stress the mechanical struc-
tures and fatigue compromising the structural integrity. Second, the FPV farm is a complex multi-body system,
and the multi-body interactions with waves introduce complexities to the system's behaviour. This study will
investigate the FPV systems in harsh marine conditions for safe installation and operation of the FPV.
In this study, we investigate the hydrodynamic response of a novel FPV system through numerical studies,
focusing on the effect of mooring systems. The experimental results are used to calibrate the numerical model
On
ly
2
fo
rR
which are presented previously in the study [4]. The numerical model is built in the time domain. Based on the
calibrated numerical model, a parametric study is conducted to examine the impact of different mooring pretension
ea
forces on the global structural response of an FPV array under regular waves. The results from the numerical study
din
provide a better understanding of the effect of mooring pretension on the global design of such FPV systems.
2 Conceptual design of the FPV system

g
at
The FPV system studied in this study consists of multiple interconnected single FPV modules proposed in the
W
article [4]. The single FPV module consists of a lightweight porous pontoon to support four PV panels and the
CF
buoyancy for the module is provided with six cylinders with a diameter of 0.6 m. Both the porous pontoon and
the cylinders are hollow inside to reduce the overall weight of the system. The single modules are interconnected
S
with soft connecting ropes to form a large array which is anchored with a mooring system. The conceptual design
20
for this single FPV module can be found in Fig. 1(a) and an array of interconnected single modules can be found
in Fig. 1(b). The concept has been extensively tested through a laboratory experiment in the towing tank at the
24
Canal de Ensayos Hidrodinamicos of the Universidad Politecnica de Madrid [4]. The experimental study was
conducted for six pontoons arranged in a small array of 3 × 2. For the model test, the single float is idealised to
reduce the complexity of the proposed FPV system. In addition, the model test was conducted at a 1:60 Froude
scale in the towing tank whereas the numerical model was established at full scale. The full-scale numerical model
is based on the model test for a valid comparison between experimental results and the numerical model. Fig 2
presents the model test setup for the numerical study. The numerical model has a spacing of 1 m between pontoons
connected with connection ropes and a mooring system. The full-scale single module has a dimension of 4.7 m in
length, 2.9 m in width and 0.6 m in height. In addition, the floating modules are submerged with a draft of 0.2 m.
The surge direction for both the numerical and experimental model is along the wave direction which is propa-
gated in the longitudinal direction of the pontoon. The wave height and period for the experiment were conducted
for nine wave cases typical of wind-generated ocean waves for coastal and nearshore areas. A detailed summary
of the experimental study is presented in [4].
Fig. 1. Conceptual design of FPV module (a) and an array of interconnected single modules with connection ropes (b) [4].
Fig. 2. Idealised full-scale numerical model with pontoons, connection ropes and mooring lines.
On
ly
3
fo
3 Methodology
rR
3.1 Dynamics of global model
ea
din
As explained in Section 2, the FPV array comprises 6 pontoons interconnected by connection ropes. The corner
g
pontoons are restrained by mooring lines. In the numerical modelling, the pontoons are treated as rigid bodies
at
interacting with the fluid. The connection ropes and mooring lines are modelled as bilinear springs that can only
withstand tensile forces. For the pontoons, the dynamics of fluid motions are governed by the opposing actions of
W
different forces and moments exerted on the structures [5]. The forces on floating structures are mainly exerted
CF
by hydrostatic forces, hydrodynamic forces, and mooring forces. Such forces are influenced by fluid inertia,
weight, viscous stresses, and secondary effects such as surface tension. According to Newton’s second law of
S
motion, the governing equations of motion of the floating structure are expressed by:
20 (1)
24
Where is body mass and the frequency-dependent added mass matrix, is the frequency-dependent potential
damping matrix, is the linear damping matrix, is the quadratic damping matrix, and is the stiffness matrix
due to mooring stiffness, connection stiffness and hydrostatic restoring stiffness. The denotes the pretension
force applied on the mooring system. Additionally, the wave excitation forces are expressed as,
(2)
where and are the first and second-order wave excitation forces respectively.
3.2 Potential flow theory

The hydrodynamic analysis model was created based on the Panel model. The frequency domain analysis was
conducted with Wadam to calculate the wave-structure interaction of the single FPV panels. The method based
on Potential flow theory is used to calculate the system's response. The potential flow theory assumes the fluid is
ideal, namely, inviscid, irrotational and incompressible. The velocity field of an irrotational fluid can be repre-
sented as a scalar function or as a velocity potential. Thus, the potential is expressed by [5],
(3)
The vector field needs to satisfy the conservation of mass and momentum for the Laplace equation. Addition-
ally, the velocity potential of the harmonic wave must fulfil the criteria for the seabed boundary condition, free
surface dynamic boundary condition and free surface kinematic boundary condition. In addition to wave velocity
potential, the flow must incorporate the reflected waves. Thus, the wave potential can be expressed with a more
general velocity potential [6],
(4)
following, ϕr is the radiation potential from the motion of the body, ϕw is the undisturbed wave potential and ϕd
is the diffracted wave potential.
3.3 Morison element

The potential flow theory simplifies the laws of fluid mechanics. As previously mentioned, the fluid is assumed
to be inviscid where the effect of the viscosity is not considered by the potential flow theory. However, the vis-
cosity has a significant influence on the fluid behaviour of small and slender structures. The calculation based on
the potential flow method underestimates the wave drift forces on the structures [7]. Similar outcomes were also
observed in other studies conducted by [8] and [9]. The discrepancies provided by idealised flows will be accom-
modated by introducing the Morison element in the numerical model by applying the viscous drag force [10]. The
quadratic viscous drag force F(t) is given by
(5)
where 𝐶 denotes the dimensionless quadratic drag coefficient, is the fluid density, is the volume and 𝐷 is the
diameter of the slender elements. and is the flow velocity of the fluid and slender elements respectively.
While and is the acceleration of the fluid and slender elements.
On
ly
4
fo
3.4 Numerical procedure
rR
ea
The numerical simulation of the FPV system is conducted with a combination of different software packages. The
geometry of the FPV modules and discretised meshes of the wet surface were generated in the GeniE, a prepro-
din
cessing and FE analysis software package. Then, a frequency-domain boundary element-based hydrodynamic
g
analysis was conducted in Wadam, a wave diffraction and radiation analysis solver based on 3D linear potential
flow theory. The hydrodynamic analysis was conducted for a single pontoon. The effect of hydrodynamic cou-
at
pling between pontoons is neglected in this study. Frequency-dependent hydrodynamic coefficients such as added
W
mass, potential damping, and wave excitation forces are obtained and imported into SIMO, which is a time-do-
CF
main analysis of the entire array with connection ropes and the mooring system. This model is used to evaluate
the global response, especially the interactions among the structural components and incident waves.
4 Results and discussion S

20
4.1 Calibration and validation of numerical model
24
The numerical model is calibrated with the responses obtained from an experimental model test [4]. The calibra-
tion is made according to the measure characteristics of the mooring system, free decay test, and the heave and
surge response amplitude operators (RAOs) of all six pontoons under regular waves. The detailed calibration and
validation process for the numerical model is presented below.
Mooring and connection systems. The mooring and connection systems significantly affect the dynamic re-
sponse of the FPV system. A four-point above-water-level mooring line connected with a spring to a load cell was
used for the experimental setup. During the experimental study, the spring stiffness of 3.43 kN/m was determined
to obtain restoring properties of the mooring lines. The spring length for the mooring line was determined to avoid
slacking of mooring lines during the experimental study. Therefore, the mooring system for the numerical model
in SIMO is defined as a fixed elongation coupling with a spring element. The spring element is defined as a
specified length-force relation with the stiffness obtained from the numerical study. The defined mooring system
act as a coupled element between the floats and anchored points. In addition, the mooring system is applied with
a pretension of 53 kN as a station-keeping force, consistent with the model test setup. However, the damping of
the mooring lines was neglected for the numerical study due to the limitation of experimental study. The damping
of the mooring lines was not measured during the model test. The mooring system in the numerical model was
verified by examining the applied tension forces during the static and dynamic simulation. The connection system
between the FPV panels for the numerical model was defined as simple wire coupling. However, the stiffness of
the connection ropes was not measured during the model test. For the numerical model, the axial stiffness is
reasonably set to 3.9 GPa, which is in accordance with a new generation of high-performance fibre ropes for
marine and offshore applications [11].
Decay test. The decay tests were first conducted using the time-domain coupled model. The decay test was con-
ducted for a free-floating single pontoon in calm water. The natural periods and damping ratios of the single
pontoon in heave, roll, and pitch were compared to the experimental results. Note that the damping was calculated
according to the logarithmic damping decrement method [4]. According to the decay test results, the numerical
model based on the potential flow theory does not require additional damping in pitch and heave. However, a
linear damping of 700 Nsm in the roll was required for the numerical model to match the model test.
Motion responses under regular waves. The RAO of pontoons for the numerical model in SIMA and the exper-
imental model test were generated for the nine regular wave cases presented in the article [4]. The RAOs in surge
and heave were obtained from the model test. Fig 3 presents the RAOs in heave and surge generated for nine
regular wave cases reported in the model test and the numerical simulations. Note that in the numerical model,
the motion RAOs were generated for motion response after a simulation time of 400 s when the transient response
is damped out. Also note that the RAOs presented are the average values of all six pontoons, which is identical to
the experimental study. The average response amplitude of individual six pontoons depicts the response of the
whole 3x2 array. It can be found from Fig. 3(a) that the heave RAOs from the numerical model have good agree-
ment with the model test results with a discrepancy between 0.3 % and 7.2 %. The surge motions were calibrated
by introducing additional viscous drag via the employment of slender elements in SIMO. The drag coefficient for
the slender elements was calibrated iteratively until reasonable surge motion was obtained compared to experi-
mental results. As can be seen in Fig. 3(b), a good agreement is also observed for surge RAOs between the nu-
merical and experimental results.
On
ly
5
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 3. Comparison of average RAO of six pontoons between the experimental and numerical results: (a) heave and (b)
surge.
4.2 Effect of mooring pretension on motion responses

The mooring system is a critical component of the FPV farm to withstand wave-induced forces. During the ex-
perimental model test, the mooring lines in the experiment were pretensioned to simulate the behaviour of a real-
istic catenary mooring system. The pretension force of 53 kN for each mooring line was determined after a nu-
merical screening study carried out in OrcaFlex [4]. In this study, the effect of mooring pretension force on the
motion responses of the FPV farm is based on a parametric study based on the calibrated numerical model in
SIMA-Riflex. The parametric study is conducted with two cases. In the first case, the pretension force is reduced
by 37.8 % to a value of 30 kN and in the second case, the pretension force is increased by 37.8 % to a value of 70
kN. The wave cases for the parametric study are conducted for periods between 3 s to 12 s with a unit amplitude
of 1 m for all cases.
Fig 4 shows the heave and surge RAOs of FPV farms from the parametric study. According, to the parametric
study the pretension force influences the motion responses of the FPV system. Generally, the motion response in
both heave and surge decreases with a higher pretension force. The surge RAOs are increased by 40 % when the
pretension force is reduced by 37.8 % to 30 kN at a period of 6 s. But when the pretension force increased by
37.8 % to a value of 70 kN, the surge RAOs were reduced by 24% at a period of 6 s. When the pretension force
was increased by 37.8 % at a period of 6 s the RAOs increased by 3 % in heave. However, the RAOs are reduced
by 2 % when the pretension force decreases by 37.8 %. In general, the surge RAOs are more sensitive to the
change of the pretension forces as compared to heave. This is because heave motion is mainly dominated by
hydrostatic stiffness for long waves, while surge motion relies on the mooring system to provide restoring forces.
When the pretension force increases, the mooring lines experience high stresses which makes the FPV system
more rigid to withstand wave-induced forces. Therefore, the mooring pretension force should be carefully deter-
mined to ensure safe design and operation. Additionally, it is found that the effect of the pretension force is greater
for shorter waves compared to longer waves on the heave motion.
In addition to the pretension force, the motion responses of the FPV system are significantly influenced by the
wave period. In the heave direction, the minimum RAOs were observed for waves with the shortest and maximum
RAOs for the longest waves. The short wave exhibits high-frequency forces on the FPV system compared to
longer waves. The numerical response shows that the FPV system can withstand rapid and small-scale oscillations
from short waves. The heave motions under high-frequency forces from the short wave are reduced due to the
inertia of the FPV system and increased pretension in the mooring lines. However, the maximum RAOs in the
surge are observed around the resonance period due to wave excitation. The surge motion of the FPV system is
amplified due to constructive interference between the wave-induced force and the natural oscillation of the struc-
ture. The RAOs in surge increase drastically with a maximum horizontal displacement observed at the resonance
period. The motion response of the FPV system is reduced gradually after the resonance period according to the
numerical simulation. The maximum RAOs in the surge direction is at a wave period of 4.5 seconds for a model
with a pretension force of 53 kN and 70 kN. The maximum surge motion for the model with 30 kN pretension
force is observed around the wave period of 5.5 seconds. The natural period of the FPV system is influenced by
the pretension force. The resonance period of the FPV system is found to increase when the pretension force in
the mooring lines is reduced.
On
ly
6
fo
rR
ea
din
g
at
W
CF
S
20
24
Fig. 4. The Response Amplitude Operators (RAO) of the calibrated numerical model with unit wave amplitude. Figure (a)
shows the RAO in heave and figure (b) in the surge.
5 Concluding remarks
In this paper, a numerical study of a floating photovoltaic (FPV) system in regular waves is conducted. The nu-
merical model is developed based on frequency domain analysis in HydroD and subsequent time domain analysis
in SIMO. The study is benchmarked with the available results from a model test. Through numerical calibration,
the numerical model is found to provide accurate predictions of the motion response in both the heave and surge
of the pontoons in a moored array. A parametric study is next conducted, and the results show that the mooring
pretension force affects the motion response in both heave and surge directions. The difference in the heave motion
response due to different pretension forces is small when compared with the surge motion. The latter is found to
be more sensitive to the wave frequency and mooring pretension. In addition, the FPV system exhibits signifi-
cantly reduced heave motions under shorter waves that are commonly seen in coastal regions, which is desirable
from the engineering design perspective. For the surge, the large RAO amplitudes of the pontoons are observed
near the resonance period. This needs to be taken into account in the design practice.
The numerical model for this study was based on the simplified model without the effect of hydrodynamic
coupling between pontoons. Additionally, the parametric study is limited to the effect of mooring pretension force
under regular waves. In the future, a more detailed study is fundamental to obtain insight into the effect of different
parameters, including the configuration and stiffness of connection ropes as well as the gap between neighbouring
pontoons, on the response of the FPV system. Besides, the numerical study is based on a small array comprising
6 pontoons. There is also a necessity to develop an efficient numerical model for larger arrays.
References
1. de Castro, C., et al.: Global solar electric potential: A review of their technical and sustainable limits. Renewable and
Sustainable Energy Reviews, 28: p. 824-835 (2013).
2. Dai, J., et al.: Design and construction of a floating modular photovoltaic system for water reservoirs. Energy, 191: p.
116549 (2020).
3. Zhang, C., et al.: Development of compliant modular floating photovoltaic farm for coastal conditions. Renewable and
Sustainable Energy Reviews, 190: p. 114084 (2024).
4. Jiang, Z., et al.: Design and model test of a soft-connected lattice-structured floating solar photovoltaic concept for harsh
offshore conditions. Marine Structures, 90: p. 103426 (2023).
5. Newman, J.N.: Marine Hydrodynamics. The MIT Press, (1977)
6. Journée, J.M.J., et al.: Offshore Hydromechanics. TU Delft, (2000).
7. Fonseca, N. and C.T. Stansberg: Calibration of a Time-Domain Numerical Hydrodynamic Model for Mooring Analysis
of a Semi-Submersible. In: ASME 2016 37th International Conference on Ocean, Offshore and Arctic Engineering
(OMAE2018), Spain (2018).
8. Ma, S., et al.: The numerical study of viscous drag force influence on low-frequency surge motion of a semi-submersible
in storm sea states. Ocean Engineering, 213: p. 107511. (2020).
9. Stansberg, C.T., et al.: Challenges in Wave Force Modelling for Mooring Design in High Seas. in Offshore Technology
Conference. (2015).
10. Morison, J.R., J.W. Johnson, and S.A. Schaaf: The Force Exerted by Surface Waves on Piles. Journal of Petroleum Tech-
nology, 2(05): p. 149-154. (1950).
11. Leech, C.M., et al.: Modelling Tension And Torque Properties of Fibre Ropes And Splices. In: The Third International
Offshore and Polar Engineering Conference. (1993).
On
ly
fo
rR
A baseline study of the biodiversity enhancement of the Tung Chung East New Town
ea
Extension (TCE) eco-shoreline installation at its pioneer stage
din
g
XIE Yao1, Cheryl K.C. Chu1, Kenneth M.Y. Leung1* and Jay J. Minuti1
1
at
State Key Laboratory of Marine Pollution, City University of Hong Kong, Hong Kong SAR, China
W
kmyleung@cityu.edu.hk
CF
S
Abstract. Hong Kong's extensive urbanisation and land reclamation efforts have significantly altered its
20
coastal environments, resulting in approximately 15% of its coastlines being replaced by artificial structures
like concrete and rubble mound seawalls. While these structures are effective in preventing erosion and flood-
24
ing, they have also led to a considerable decline in biodiversity and the disruption of essential ecosystem ser-
vices. To address these ecological challenges, the Hong Kong government has launched the Tung Chung East
(TCE) eco-shoreline project. This ambitious initiative spans 3.8 km and integrates both soft and hard engineer-
ing solutions to create a more ecologically sustainable shoreline. The project features three distinct sections: a
1.8 km vertical shoreline with intricate eco-patterns, a 0.8 km rocky shoreline with diverse micro-habitats, and
a 1.2 km artificial mangrove area. This study aims to evaluate the initial ecological performance and biodiver-
sity enhancement of these three sections. Through comprehensive field surveys, we assessed species richness,
community structure, and habitat usage, comparing the results to control sites with traditional artificial shore-
lines. Preliminary findings indicate that the eco-shoreline supports higher species richness and more complex
community structures than the control sites. These results suggest that the TCE eco-shoreline project is effec-
tive in promoting biodiversity and ecological resilience. This study not only demonstrates the potential benefits
of eco-shoreline development but also provides a valuable framework for future coastal projects around the
world
Keywords: Biodiversity, Eco-shoreline, Coastal Development, Habitat Enhancement, Ecological Resilience.
1 Introduction
Hong Kong has experienced significant coastal transformation, with approximately 15% of its coastlines now
consisting of artificial shorelines, primarily in the form of concrete or rubble mound seawalls (Leung, 2013).
These structures, constructed to combat land erosion and flooding, have led to a reduction in biodiversity and the
associated ecosystem functions (Perkins et al., 2015). As a result, natural ecosystem services crucial to human
society, such as fisheries production (Al-dulrazzak & Pauly, 2014), biofiltration of seawater (Zu Ermgassen et al.,
2013), carbon sequestration (Duarte et al., 2013), and natural coastal protection (Borsje et al., 2011; Duarte et al.,
2013), have been compromised. Despite these challenges, the Hong Kong government has undertaken the ambi-
tious Tung Chung East (TCE) eco-shoreline project, which spans an impressive 3.8 km (Fig. 1(A)). This ecolog-
ically engineered shoreline employs a combination of soft and hard engineering approaches to address the limita-
tions of traditional seawalls. The TCE eco-shoreline is composed of three distinct sections, including a 1.8 km
vertical shoreline, a 0.8 km rocky shoreline, and a 1.2 km artificial mangrove, each designed to provide a range
of unique habitats for promoting biodiversity enhancement and ecological resilience. Here we conducted a field
study to review the progress of biodiversity enhancement on these three key sections of the eco-shoreline, to assess
the ecological performance of this pioneering eco-shoreline installation at its early stage.
The vertical shoreline is designed with intricate eco-patterns (Fig. 1(B)), incorporating pre-cast wave-curve-
shaped horizontal ridges that increase surface area and provide shading (Firth et al., 2014). This design aims to
mimic natural coastal features and create a more hospitable environment for marine organisms. The rocky shore-
line provides diverse micro-habitats, including eco-holes, micro pools, and macro pools (Fig. 1(C)), which offer
refuge and breeding grounds for a variety of species (Firth et al., 2014). By incorporating diverse types of micro-
habitats, the TCE eco-shoreline aims to support a greater range of marine life and enhance overall biodiversity.
The project also includes an artificial mangrove section (Fig. 1(D)), serving as a fully natural-based solution for
coastal defence and biodiversity enhancement. Mangroves are known for their ability to stabilize shorelines, pro-
tect the coast against erosion, and provide critical habitats for a wide array of species (Alongi, 2008). By integrat-
ing an artificial mangrove into the eco-shoreline, the project aims to recreate the ecological functions of natural
mangrove ecosystems, contributing to the overall resilience of the shoreline.
On
ly
2
fo
rR
The TCE eco-shoreline project represents a significant step towards mitigating the negative impacts of artificial
shorelines and promoting sustainable coastal development in Hong Kong. This large-scale construction provides
ea
a unique opportunity to study the benefits and effectiveness of eco-shoreline development on a substantial scale,
din
potentially serving as a model for future coastal projects worldwide.
g
at
W
CF
S
20
24
Fig. 1. (A) Map of TCE eco-shorelines; (B) Photo of the vertical eco-shoreline; (C) Photo of the rocky eco-shoreline; (D)
Photo of the artificial mangrove.
On
ly
3
fo
2 Methodology
rR
2.1 Vertical eco-shoreline
ea
din
In this study, the north-east vertical eco-shoreline of TCE (1.6 km) was surveyed at two tidal levels: high-shore
g
(1.8-2.3 m above C.D.) and mid-shore (1.0-1.5 m above C.D.), excluding the underwater section below 1.0 m
at
C.D. Large quadrats (50 cm * 50 cm) were randomly placed along a 100 m transect on the 1.5 km seawall, with
a total of 15 quadrats surveyed for both high-shoreline and mid-shore by presence/absence method. Additionally,
W
sessile species coverage was analysed using photos of small quadrats (15 cm * 15 cm) in the laboratory, and
CF
species colonising eco-holes and eco-pots (Fig. 1(B)) were surveyed by presence/absence method with 15 eco-
holes and 15 eco-pots randomly selected at mid-shore. A traditional vertical seawall with no eco-features was also
S
surveyed using the same method as a control (Fig. 2(A)).
2.2 Rocky eco-shoreline

20
24
The study identified five distinct eco-features along the rocky eco-shoreline: bare rock, eco-holes, micro pool 1,
micro pool 2, and macro pool. (Fig. 1(C)). TCE's rocky shoreline was divided into sections 1 (550 m) and 2 (250
m). Species colonizing these features were surveyed using the same method as the vertical seawall eco-pattern,
with 10 replicates at the high, mid and low tidal levels. Additionally, the traditionally designed riprap (~ 210 m)
served as a control (Fig. 2(B)), with boulders surveyed at the three tidal levels using the same method with 10
replicates.
2.3 Mangrove eco-shoreline

The study site encompassed two 550-meter sections and one 100-meter section of the mangrove shoreline in TCE.
Within each 550-meter section, 6 transects were established, each approximately 90 meters long. Surveys included
mangrove epifauna, infauna sampling, and fast-moving species (crabs) using video recording. Additionally, a 60-
meter section of a nearby natural mangrove in Tung Chung (Fig. 2(C)) was surveyed with the same method as a
reference.
Fig. 2. Control sites of (A) vertical seawall (B) riprap seawall and (C) natural mangrove reference.
3 Preliminary results
3.1 Vertical eco-shoreline

The eco-pattern had the highest species richness at the high and mid-shore, whereas the eco-hole and eco-pot had
the lowest intertidal species at the mid-intertidal (Fig. 3). The vertical seawall community was more similar to the
reference site than the control, with the eco-pattern having a distinctly different community structure than both
the eco-hole and eco-pot.
On
ly
4
fo
rR
ea
din
g
at
W
CF
S
20
Fig. 3. Species richness within taxon groups on vertical eco-shoreline
3.2 Rocky eco-shoreline 24

The Macro pool, Micro 1 and Micro 2 had the highest species richness across all levels, with substantially higher
species richness than the control site (Fig. 4). The community structure is more similar to the natural rocky shore
than the control site.
Fig. 4. Total number of species on rocky eco-shoreline features, control riprap, and natural rocky shore.
4 Discussion
The Tung Chung East eco-shoreline project has showcased promising results in terms of biodiversity enhance-
ment, despite being in its early stages. Across various habitats and micro-habitats, there was a noticeable variation
in baseline biodiversity, with the enhanced shoreline exhibiting a greater number of species and enhanced com-
munity structure compared to the control sites. These findings highlight the positive impact of the Tung Chung
East New Town Extension (TCE) eco-shoreline project, indicating its potential for further colonisation and con-
tinued biodiversity enhancement. This large-scale and unique installation holds great promise for the future, as it
demonstrates the effectiveness of eco-shoreline development in promoting ecological resilience and fostering
thriving ecosystems.
References
1. Al-Abdulrazzak, D., Pauly, D. (2014). Managing fisheries from space: Google Earth improves estimates of distant fish
catches. ICES Journal of Marine Science 71, 450-454.
2. Borsje, B.W., van Wesenbeeck, B.K., Dekker, F., Paalvast, P., Bouma, T.J., van Katwijk, M.M., de Vries, M.B. (2011).
How ecological engineering can serve in coastal protection. Ecological Engineering 37, 113-122.
3. Duarte, C.M., Losada, I.J., Hendriks, I.E., Mazarrasa, I., Marba, N. (2013). The role of coastal plant communities for
climate change mitigation and adaptation. Nat. Clim. Change. 3, 961-968.
4. Firth, Louise B, Scho, M., White, F.J., Skov, M.W., Hawkins, S.J., 2014. Biodiversity in intertidal rock pools : Informing
engineering criteria for artificial habitat enhancement in the built environment 102, 122–130.
https://doi.org/10.1016/j.marenvres.2014.03.016
On
ly
5
fo
rR
5. Leung, T., 2013. Seminar on land supply series talk 3: Shoreline development.
6. Ove Arup & Partners Hong Kong Limited, 18 May 2013.
ea
7. Perkins, M.J., Ng, T.P.T., Dudgeon, D., Bonebrake, T.C., Leung, K.M.Y. (2015). Conserving intertidal habitats: What is
din
the potential of ecological engineering to mitigate impacts of coastal structures? Estuarine, Coastal and Shelf Science.
167, 504-515.
g
8. zu Ermgassen, P.S.E., Spalding, M.D., Grizzle, R.E., Brumbaugh, R.D. (2013). Quantifying the loss of a marine ecosys-
tem service: filtration by the Eastern Oyster in US estuaries. Estuaries Coasts 36, 36-43.
at
W
CF
S
20
24
On
ly
fo
rR
A data-driven approach for urban transitions in delta regions towards
ea
living with water through floating and mobile infrastructure
din
g
Sridhar Subramani23[0009-0002-7879-4229], Mathilde Marengo1, Iacopo Neri1, Koen Olthuis23
1
at
Institute for Advanced Architecture of Catalonia, C. de Pujades, 102, Sant Marti, 08005 Barcelona, Spain
W
2 Delft University of Technology (TU Delft), Mekelweg 5, 2628 CD, Delft, Netherlands
CF
3 Waterstudio.nl, Generaal Berenschotlaan 211, 2283 JM, Rijswijk, Netherlands
s.subramani@tudelft.nl
S
20
Abstract. Delta cities are complex systems characterized by continuous change and unpredictable behavior
24
due to global (e.g. sea level rise) and local (e.g. flooding and urban densification) agents. The "living with
water" paradigm offers a framework to address these challenges by deploying modular floating infrastructure
within urban planning strategies. These solutions boast inherent adaptability and flexibility, materialized by
dynamic platforms, allowing for spatial reconfiguration and functional transformation in response to evolving
user needs and environmental pressures.
The paper discusses the research developments of a planning methodology to facilitate the transition towards
"living with water" in delta cities. The methodology leverages scenario creation to enable cities to respond and
coexist with uncertainty. The three-stage process begins with designing a planning system based on user ac-
tivities within functional spaces. Rotterdam serves as a testbed due to its strategic advantages and knowledge
of floating solutions for adapting to rising sea level threats. Time-use surveys inform the allocation of func-
tional types on floating platforms, with their distribution determined by frequency of use. Overlaying user
flows with accessibility identifies the demand for a specific function over time.
In the second stage, the planning system uses reinforcement learning to simulate responses of floating infra-
structure to varying demand locations. Then, the system measures de-densification, proximity, and functional
object availability over time. This evaluation tool redefines thresholds for functional distribution and demand
locations by incorporating scenarios such as seasonal flooding. Such a data-driven performance evaluation
facilitates the identification of implementation strategies for cities to develop on water, optimizing urban plan-
ning in delta regions to foster resilience, adaptation, and long-term sustainability.
Keywords: Living with Water, Data-Driven Design, Mobile Floating Solutions, User Activity.
1 Introduction
1.1 Delta Urbanism

Balancing urbanization and natural environments, Delta Landscapes house about 10% of the world's population,
including major cities like Amsterdam, London, Miami, New York, Mumbai, Bangkok, and Shanghai [1]. Climate
emergencies, particularly sea-level rise, pose significant threats. From 2006 to 2022, sea levels have risen at an
average rate of 3.7mm/year, marking a concerning trend. Greenhouse gas emissions and uncertainties in ice-sheet
processes may cause sea levels to rise by up to 5m by 2150 [2]. These vast uncertainties in the projection signifi-
cantly increase the challenges for investment planning in coastal management strategies in low-lying countries
like the Netherlands. Since these coastal defense projects lack flexibility for adaptation during their development,
they will need to be redone periodically once they reach their limits. These strategies demand an adaptive approach
with short-term and long-term implementation options depending on how the future unfolds [3]. Urban planning
in delta regions must address these complexities flexibly, requiring new relationships between design, engineer-
ing, science, and governance.
The Netherlands exemplifies these challenges, being a low-lying country in the Rhine, Meuse, Scheldt, and
Ems river deltas. About 26% of its territory is below sea level, and 60% is vulnerable to flooding. Major urban
areas, including Amsterdam, The Hague, and Rotterdam, rely on an extensive flood protection system. This sys-
tem includes dams, seawalls, storm surge barriers, dikes, and dunes, costing about 1.2 billion euros annually for
maintenance [4]. The Delta Programme, initiated in 2010, aims at protecting the Netherlands from flooding, secure
fresh water supplies, and make the country climate-proof [5]. With a growing population expected to reach 18
million by 2025, the Netherlands faces increased pressure on urban areas and infrastructure. Sea levels are pro-
jected to rise by 1.5 meters by mid-century, with annual rainfall between 900 to 1000 mm affecting major urban
regions. These threats impact civic infrastructure and water infiltration in urban areas. The Dutch Delta
On
ly
2
fo
rR
Programme has developed strategies to address sea-level rise, focusing on protection and adaptation, categorized
ea
as 'Protect-Closed,' 'Protect-Open,' 'Advance,' and 'Accommodate' [6]. 'Protect-Closed' involves shielding inland
areas with dams, navigation locks, and sluices to prevent sea-level rise and salinization. 'Protect-Open' maintains
din
a sea connection, requiring extensive dike improvements to manage flooding. 'Advance' extends the coastline
seaward to build a stronger coastal flood defense. 'Accommodate' adapts to rising water levels by raising buildings,
g
creating mounds, constructing ring dikes, or developing floating structures. The 'living with water' concept, de-
at
rived from 'Accommodate,' promotes floating infrastructures as an adaptive response.
W
CF
1.2 Urban System
Rotterdam, one of the Netherlands' largest ports and fast-growing urban regions, has integrated climate change
S
adaptation approaches, such as Rotterdam Climate Proof (2008) and Rotterdam Climate Change Adaptation Strat-
20
egy (2013). These strategies strengthen flood defenses, adapt urban spaces, and increase city resilience through
measures like water squares, infiltration zones, dikes, coastal protection, evacuation routes, water-resistant build-
24
ings, and floating structures. Projects like the Stadshavens (City-Ports) district 2004 are exemplary of the city’s
intention to showcase ‘urban delta technology’ by focusing on adaptive building and floating construction [7][8].
This research by design explores how a dynamic planning system can address urban uncertainties with floating
solutions, how mobile elements adapt to human behavior and scenarios, and whether data-driven methods can
enhance our understanding of urban dynamics. The goal is to develop a tool for strategizing resilient urban devel-
opments in climate sensitive delta regions, using Rotterdam as a testbed.
To begin, a systems study of Rotterdam illustrates relationships between key elements, with nodes representing
system elements and networks depicting influences. This research evaluates resilience based on adaptability and
growth under uncertain conditions. In this framework, key nodes include not only urban growth predictions, but
also lead time to address changes in demand, water authority, accessibility to facilities, and user distribution.
Urban growth is constrained not only by space availability within the city but also by the potential for expansion
beyond city limits, negatively impacted by sea-level rise, and heavily utilized port activities. The water authority
faces challenges in expanding and maintaining defense systems. Rapid population growth strains space and re-
sponse times, while increased densification affects accessibility. Translating research objectives into tangible
nodes, the proposed systems map evaluates the impact of the floating infrastructure on the availability and adapt-
ability of the surrounding space as they enhance response time and functional proximity.
2 Methodology
The methodology is classified into three essential topics that together define a data-driven evaluation tool: plan-
ning system, scenario testing, and evaluation criteria for impact assessment (see Fig. 1).
Fig. 1. The methodology for determining transition phases through the evaluation process.
Initially, a planning system is established based on the current frequency of use (see Fig. 2). This system trans-
lates user demand for particular functions into objects that can move based on time. It shows how many people
will engage in a particular activity and the locations of people shifting from one activity to another at a given time,
helping identify the movement of floating structures. Overlaying the Time Use Survey data of the Netherlands
with land use and population density distribution provides a guideline for the flow and distribution of mobile
infrastructures over time. Furthermore, in order to validate the system's ability to accommodate different scenar-
ios, a computational model based on Reinforcement Learning is built to predict the most efficient movement
pattern of these structures. Based on the inferred behavior, the number of floating elements can be tuned and
On
ly
3
fo
rR
evaluated to improve performance. The final step is the impact analysis, which assesses the availability of spaces
ea
for different functions and how much demand is satisfied based on some floating elements. Thus, the performance
evaluation defines the transition phases for moving from land to water.
din
g
at
W
CF
S
20
24
Fig. 2. An overview of the planning system based on frequency of use (translation of user demands to object movement).
2.1 Planning System: Translation of User Demands to Object Movement
Defining User Location Based on Land-Use. The planning system is established through several critical steps:
defining user location based on land use, determining user position for object distribution, and scheduling user
flow. The initial step involves the spatial harmonization of all data sets to a 100m x 100m grid to ensure data
interoperability. Each grid cell collects land use information correlated with activities defined in Time Use Sur-
veys. Lastly, activities are clustered into categories such as personal care corresponding to residential areas and
paid work to industrial areas and offices, creating a detailed activity map.
In parallel, population density data is also converted into 100m grids to identify activity based hot spots
throughout a typical working day. The study visualizes user distribution and activity patterns by plotting the per-
centage of the population engaged in different activities at ten-minute intervals over 24 hours. This analysis uses
a typical working day (Tuesday) to show how population density shifts across the urban landscape.
National statistical data is overlaid on the standardized grid to evaluate proximity to specific amenities e.g.
household necessities, with a 300m threshold considered based on national standards. A circular packing system
arranges circles within a given space to optimize the use of a particular functional space, minimizing gaps and
maximizing efficiency. This system determines intervention locations on land and water, satisfying these services'
300m proximity requirement. For research purposes, only water-based locations are analyzed. This approach iden-
tifies zones where new infrastructure can meet demand, such as satisfying 20,200 people across 32 locations for
household activities on a typical day.
Defining User Location to Distribution. The next step involves verifying user positions at various time steps to
predict user flow between locations for specific activities. The study identifies shifting activity patterns by ana-
lyzing the proportion of the population engaged in different activities over time. For example, every three-hour
interval is examined to determine population movements, helping to pinpoint where interventions are necessary
at different times of the day. This process involves calculating the proportion of people moving from one activity
to another, using this data to define the distribution of intervention locations based on time steps. By overlaying
this information with the demand zones identified in the previous step, the planning system determines where and
when floating infrastructures should be deployed to meet changing user demands.
On
ly
4
fo
rR
Defining User Schedule to Flow. The final step involves defining a schedule of demand locations and the number
ea
of people shifting to these activities. Using data on population movements, the study calculates the number of
people near these demand locations at specific times. This information is combined with previous results to de-
din
termine the placement and timing of movable floating infrastructures, ensuring that demand is met efficiently
throughout the day. For instance, at different times of the day, the number of demand locations and people in-
g
volved can vary significantly. One scenario may show a single-demand location with 825 people, while another
at
might involve 15 locations with 4,474 people. This variability demonstrates the necessity for flexible, mobile
infrastructures that can adapt to changing urban dynamics.
W
2.2 Testing for Different Scenarios
CF
S
To evaluate the performance of the planning system, an agent-based model is used to understand the behavioral
20
patterns of floating infrastructures under different scenarios. This model employs Reinforcement Learning to train
agents (i.e. the floating infrastructures) to respond to varying demand locations over time. Reinforcement Learning
24
is a type of machine learning where an agent learns to make decisions by performing actions in an environment
to maximize cumulative rewards through trial and error. The agents' actions are guided by three primary parame-
ters: context (current status), actions (agent behavior), and new states (outcomes based on reward functions).
The context is defined by the current state of demand and supply, using a matrix of travel times between loca-
tions and a demand table for each location over time. Agents then determine their following location based on
higher demand and minimal travel time, balancing the need to satisfy demand with efficient movement. Reward
functions categorize demand satisfaction and distance traveled, ensuring agents prioritize high-demand locations
while minimizing travel distances.
Weekdays Scenario. The initial iteration begins with six platforms (mean value of the identified spots) acting as
agents to understand their movement patterns over time. The agents start at the highest recurring demand locations
and move according to the calculated time steps. Evaluating the capacity of these floating infrastructures against
the number of people in each location reveals the dynamics of demand satisfaction. In the first iteration, agents
can meet up to 33.43% of demand, with extra space availability reaching 83.11% during specific intervals (see
Fig. 3). Further iterations with increased agents show improved performance, with twelve agents satisfying
56.24% of demand and maintaining 72.78% extra space availability. These results highlight the effectiveness of
deploying multiple agents to meet urban demands dynamically.
Fig. 3. Agents (floating infrastructure) behavior at a time interval on a weekday [Iteration – 1].
Seasonal Hinderance Scenario. The tool is also tested under seasonal flooding scenarios to evaluate its robust-
ness. Flood zones are converted into 100m grids to identify new demand locations. This data reveals how seasonal
changes impact infrastructure needs within existing demand zones. In this scenario, agents' movement patterns
shift significantly, with a higher demand satisfaction rate of 93.6% and 26.23% extra space availability during
On
ly
5
fo
rR
peak flooding periods. These results underscore the tool's adaptability to extreme conditions, demonstrating its
ea
potential to manage urban infrastructure efficiently under various scenarios.
3 Evaluation Criteria and Impact Assessment

din
g
at
The evaluation process focuses on the capacity of the system to reduce infrastructure needs by dynamically re-
sponding to population flows based on the concepts of de-densification and proximity. Performance is measured
W
by demand satisfaction and extra space availability, providing insights into the tool's effectiveness.
CF
Initial calculations show that the dynamic response to demand variations over time can significantly reduce the
required infrastructure. The tool validates the need for mobile infrastructures that adapt to changing urban condi-
S
tions by ensuring efficient use of functional spaces throughout the day. The planning tool's performance is evalu-
20
ated by comparing demand satisfaction and extra space availability across different scenarios. Iterations with var-
ying numbers of agents reveal an optimal balance between infrastructure deployment and demand fulfillment. For
24
example, the second iteration with twelve agents shows a marked performance improvement, with higher demand
satisfaction and efficient use of space. Testing under seasonal hindrance conditions further validates the tool's
adaptability, with agents successfully meeting increased demand despite environmental challenges (see Fig. 4).
Fig. 4. Agents' movement patterns and performance comparisons between iteration-2 with 12 agents and testing under scenario
– seasonal hindrance.
4 Conclusion
4.1 Transitional Phases
The defined planning process and performance evaluation identify four transition phases for developing urban
areas into the water (see Fig. 5). Initially, floating infrastructures are deployed to dynamically respond to city
demands based on flow patterns and needs. In the second phase, the increased spatial capacity produced by the
dynamic system allows refurbishment for immediate concerns on land, such as housing in Rotterdam. The third
phase involves developing floating infrastructures to support additional activities, increasing urban capacity, and
freeing land space. In the final phase, a comprehensive toolbox of functional spaces on modular floating platforms
integrates with urban areas, reducing land dependency and accommodating sea-level rise, eventually allowing
certain land regions to accommodate sea-level rise and rejuvenate natural resources. These phases enable urban
planning systems to shift from land to water, preparing for uncertain future conditions.
4.2 Discussions and Future Steps

This research defines delta urbanism resilience through three core attributes: reflective, learning from experiences
to avoid fighting against water; robust, managing systems through behavioral/activity changes; and flexible,
adapting to disruptions and evolving circumstances via scenario-based planning.
The study validates the receptivity of floating infrastructures in urban areas to address rising sea levels. The
ability to dynamically respond to uncertain situations demonstrates flexibility in urban planning by incorporating
On
ly
6
fo
rR
mobile floating infrastructures. The performance system enables decision-makers to strategically plan the transi-
ea
tion of infrastructures from land to water, simulating various scenarios. This methodology establishes a foundation
for a planning system with a coefficient of floating infrastructure.
din
The primary goal was to develop a planning tool based on user activity frequency, integrating floating solutions
to adapt and grow amidst sea-level rise uncertainties. Currently, the method relies on the Time Use Survey to
g
understand human flows. Expanding this to include GPS data and other data collection methods could refine the
at
data used in the research. The urban infrastructure systems are not integrated into this study, by incorporating
other city systems, the transitional phases could be enriched. Each section of the methodology can be further
W
studied. For instance, parameters such as the life expectancy of systems, economic factors, and life cycle are yet
CF
to be tackled. Also, future analysis could compare the proposed approach with existing defense systems, as well
as research into legality, stakeholder involvement, and public receptivity could enhance its urban strategies. Fi-
S
nally, the introduced methodology highlights the potential of floating solutions, providing a leading case study
20
for the ongoing research on urban development on water as an adaptive response to sea-level rise.
24
Fig. 5. Four transitional phases for developing urban functions from land to water.
Acknowledgment. This paper is a result of Frequencity – a data-driven tool for resilience of delta urbanism, a
project of IAAC, Institute for Advanced Architecture of Catalonia, developed at Master in City & Technology in
2021/22 by student: Sridhar Subramani and faculty: Mathilde Marengo, Gonzalo Delacámara & Iacopo Neri.
References
1. Global Center on Adaptation, World Resources Institute. Adapt Now: A Global Call for Leadership on Climate Resili-
ence. Rotterdam, The Netherlands; Washington, DC, USA (2019).
2. IPCC. Summary for Policymakers. In: Climate Change 2021: The Physical Science Basis. Cambridge, UK (2021).
3. IPCC. Summary for Policymakers. In: Climate Change 2022: Impacts, Adaptation, and Vulnerability. Cambridge, UK
(2022).
4. Bloemen, P.J.T.M., Hammer, F., van der Vlist, M.J., Grinwis, P., van Alphen, J. DMDU into Practice: Adaptive Delta
Management in The Netherlands. In: Marchau, V., Walker, W., Bloemen, P., Popper, S. (eds.) Decision Making under
Deep Uncertainty, pp. 321–351, Springer, Cham (2019).
5. Deltacommissaris. The 2022 Delta Programme, Every New Development Climate Proof. Ministry of Infrastructure and
Water Management, The Hague, The Netherlands (2021).
6. Deltacommissaris. The 2011 Delta Programme, Working on the Delta. Ministry of Transport, Public Works and Water
Management, The Hague, The Netherlands (2010).
7. Rotterdam Climate Programme. Adaptation Programme (2009).
8. City of Rotterdam. Rotterdam Climate Change Adaptation Strategy (2013).

Wcfs2024 Day 1

Uploaded by

Copyright:

Available Formats

Wcfs2024 Day 1

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Wcfs2024 Day 1

Uploaded by

Copyright:

Available Formats

This document contains the final draft of papers to be

presented on Day 1 of WCFS 2024. The papers in this

This is NOT a publication. It is solely for participants

Table 1. Cause of structural failures and risk reduction measures 2.

2.4 Structural integrity management of floating structures in the emerging sectors

2.5 Structural reliability and risk analysis

3 Risk-informed design of floating bridges against ship impacts

3.3 Risk estimation and mitigation

Fig. 1. Diagram of different layouts with numbered tendons

Table 3. Parameters of load cases

Table 4. Failure scenarios of tendon failure simulations

Failure scenarios Mooring configurations Number of failed tendons Abbreviated as

3 Results and discussion

3.1 Response analysis of different mooring configurations

Fig. 2. Statistics of six degrees of freedom motion response under LC1

Mean Max Min Std Rms

Fig. 3. Statistics of six degrees of freedom motion response under LC2

Mean Max Min Std Rms

Fig. 4. Statistics of six degrees of freedom motion response under LC3

Mean Max Min Std Rms

2 STUDE AREA AND EMPLOYED DATA

Fig. 1. Study site of fish farms in Storm Bay, Tasmania, Australia.

Keywords: hydroelasticity, very large floating structure, offshore wind farm

Goo’s method Present method

Structure domain Fluid domain Structure domain Fluid domain

STRUCTURE FLOATING BODIES STRUCTURE GROUP BODIES

PANELS SUB-STRUCTURES FLOATING BODIES

Fig. 3. Hierarchy of the present method. [5]

2.1 Structural design

Table 1. Key parameters of semisub-large floating structure.

Item Symbol Unit Value

Fig. 5. Model - 2 Fig. 6. Model - 3

2.3 Analysis Case

Table 2. Crane placement conditions in Model-2.

Crane No. Case - 1 Case - 2 Case – 3 Case - 4 Case - 5

Angular displacement [deg.]

a) Vertical displacement distribution. b) Angular displacement.

Angular displacement [deg.]

a) Vertical displacement distribution. b) Angular displacement.

Angular displacement [deg.]

0.5 0.2 Case - 1 Case - 2 Case - 3 Case - 4 Case - 5

a) Vertical displacement distribution. b) Angular displacement.

3.3 Fully loaded

a) Only self-weight. b) All payloads.

Keywords: Tension leg platform; floating wind turbine; transient effect.

2.3 Lumped mass method

3 TLP wind turbine parameters and designed cases

3.1 OC4 and Tendon parameters

(a) Panel model (b) TLP FOWT

Table 1. Environmental conditions

4 Results and discussions

4.1 Tensions of the tendon

4.2 Motions of FOWT

Table 1. Parameters of beams and pontoons.

Components Parameter Unit Value

Pitch inertia, Itp t·m2 1.105×105

Fig. 2. Moving vehicle-floating bridge system model.

(a) Vertical displacement (b) Lateral displacement

3.2 Vehicle-induced dynamic amplification effect

(d) Third span (e) Fourth span (f) Fifth span