Proceedings of The Thirteenth (2003) International Offshore and Polar Engineering Conference

Honolulu, Hawaii, USA, May 25 –30, 2003

Copyright © 2003 by The International Society of Offshore and Polar Engineers
ISBN 1 –880653 -60 –5 (Set); ISSN 1098 –6189 (Set)

Global Performance and Mooring Analysis of a Truss Spar

Tuanjie Liu Xiaohong Chen
Jer-Fang Wu Ken Huang

American Bureau of Shipping

Houston, TX, USA

ABSTRACT
Numerical methods for motion and mooring analysis have been
A coupled analysis tool integrated with industrial experience can developed and improved over the years. Among others, Mekha et al.
accurately predict global motions and mooring/riser tensions. These (1996), Ran and Kim (1998), Ma et al. (2000) developed numerical
responses have a significant impact on hull size, structural design and methods for Spar motion analysis and studied the non-linear effects of
overall costs. This paper discusses key technical issues for the global Spar motions. Tahar et al. (2002) performed nonlinear
performance response and mooring analysis of a truss Spar in hull/mooring/riser coupled dynamic analyses of a classic Spar designed
deepwater. Detailed discussions are given to 1) the influence of wave for 6000 ft water depth. As field development reaches deeper, the
and wind spectra on slow motions of a truss Spar, 2) the distribution of mooring line and riser mass become larger in relation to the hull mass.
extreme responses using different statistical methods, 3) Morison As a result, inertia and damping effects from mooring lines and risers
coefficients for motion analysis and heave plate hydrodynamics, and 4) will no longer be negligible. Current forces acting on mooring lines and
the effects of mooring/riser added mass and damping on global risers are also significant. These effects can only be accurately
responses. The discussions are based on applications of the Coupled accounted for by a coupled dynamic analysis. Thus, coupled dynamic
Analysis Program (CAP), jointly developed by ABS and CSO Aker. analysis is preferable for deepwater and ultra-deepwater floating
The results include 1) the effects of API and NPD wind spectra on the production systems.
Spar slow motions and tension, 2) the sensitivity of tension to wave
spectrum peak enhancement parameter, 3) comparisons of Spar Cooperating with Deep Offshore Technology, the successor of the
extreme motion responses and extreme mooring line tension based on MULTISIM program, ABS developed a dynamic Coupled Analysis
different statistical methods, 4) the effects of heave plate drag Program (CAP version 7) in recent years. The most current version of
coefficients on platform motions and 5) a comparison of responses this program is CAP version 8. CAP is an integration of MULTISIM
obtained from uncoupled and coupled simulations. Understanding the and time-domain line dynamics program CABLE3D (Ma and Webster,
key technical issues for the global performance response presented in 1994). MULTSIM is a quasi-static analysis program written by J. R.
this paper will benefit engineers/designers in the design of safe, reliable Paulling (1995). This program has been used for motion and mooring
and cost-effective platforms. analysis of Spars and has been verified against model test results
(Kristofferen and Weaver, 2001, Datta, et. al., 1999). CABLE3D is a
KEY WORDS cable dynamic analysis program, also verified against model tests
(Chen, et. al., 2000, 2001). The MULTISIM program employs a tension
Truss Spar, Global performance, Coupled dynamic analysis table at each time step to obtain the forces exerted on the platform by a
catenary mooring. In formulating the tension tables, it is assumed that
INTRODUCTION the line experiences no dynamic effects and fluid forces other than
static buoyancy are negligible. The only forces acting are the net
Since the first classic spar was deployed in the Gulf of Mexico in 1996, weight, i.e., the dry weight minus the buoyancy, and the tension in the
extensive studies have been given to the motion behavior of Spar line. On the other hand, CABLE3D deploys a finite-element model for
platforms. A Joint Industry Project involving two representative Spar the dynamics of mooring lines and risers. This model accounts for
platforms was carried out at the Offshore Technology Research Center hydrodynamic loads resulting from the relative motion of the lines and
(OTRC) at Texas A&M University. It was observed in the experiment the exterior fluid, the effect of pressure gradients in both the external
that the Spar responses were characterized by low frequency surge and and internal fluid, and the weight of the lines. The integration of the
pitch motions. The results indicate that an analysis of Spar motion programs makes it possible to accurately predict the motion of the
responses needs to take into account second-order wave loads, which floating structures and the dynamics of the mooring lines and risers.
can be evaluated using second-order diffraction theory (Molin, 1979,
Kim and Yue, 1990, Chau and Eatock Taylor, 1992). In addition to the The objective of this paper is to present recent developments and
steady and wave frequency components, responses of line tension also validations of the nonlinear time-domain dynamic coupled analysis
consist of a significant low frequency component. Heave motions of a program (CAP) and to investigate the key technical issues for global
Spar are relatively small, which is an advantage for riser system design. performance and mooring analysis of a truss Spar in deep water. The
In a classic Spar, vertical wave exciting forces are low due to its deep investigations were conducted systematically by case studies using
draft. In a truss Spar, heave motions are also small due to high damping CAP. These case studies and results are presented in detail in the
introduced by heave plates in the mid-section (Magee, et. al., 2000). following pages.

256
DESCRIPTION OF THE MOORING/RISER SYSTEM OF A Table 2. Environmental conditions
TRUSS SPAR Sig. Wave Height 41.7 ft
Peak Spectral Period 14.9 sec
The truss Spar is designed for 5400 ft water depth in the Gulf of Wave Spectra JONSWAP
Mexico. It is designed with a 9-point permanent spread mooring system Peak enhancement γ 1.0, 2.4, 3.3
that is divided into three groups: Group 1: Lines 1-3, Group 2: Lines 4- Surface Current 4.2 ft/s
6, and Group 3: Lines 7-9. The mooring line configuration and 1-hour Wind Speed at 10m 72.4 knots
properties are equal for the lines grouped together. The horizontal Wind Spectra API, NPD
spread angle between lines in each group is 7.5 degrees for Group 1 and
Group 2, and 5 degrees for Group 3. In the CAP program, the principal x-axis of the platform and its
mooring is assumed from the platform South to the platform North with
Each mooring line consists of three segments: top chain, middle wire the y-axis upward. The origin of the coordinate system is located at the
rope and bottom chain. Studless K4 chain is chosen for the top and mean water surface. In the simulation, wind, wave and current are
bottom chains. Spiral strand wire is chosen for the middle wire rope. assumed collinear and are measured 30 degrees clockwise from the
Mooring line properties are shown in Table 1. positive x as shown in Figure 2.

Table 1. Mooring line properties

Diameter Length Pre-tension
Group# Segment Type [in] [ft] [kips]
1 Bottom K4 5.75 150
Middle Spiral Strand 5.04 7000
Top K4 5.75 200 579
2 Bottom K4 5.75 150
Middle Spiral Strand 5.04 7000
Top K4 5.75 260 634
3 Bottom K4 5.75 150
Middle Spiral Strand 5.04 7000
Top K4 5.75 484 612

The truss Spar has a riser system of 10 steel catenary risers (SCR) and
14 top tension risers (TTR). Diameters of SCRs range from 5 in. to 12
in. The riser pattern of the Spar is presented in Figure 1.

3 4
9 x
2
1
Figure 2. Environmental direction and mooring pattern

The wave system in CAP is represented by a linear superposition of

elementary sinusoidal progressive waves. A realistic random wave is,
therefore, approximated by the sum of several such simple waves, each
z of which differs from the others in frequency, amplitude, phase and
direction of motion. The flow properties, such as velocity, acceleration,
and pressure at a given point in space, are then obtained by a simple
5 summation of the properties computed for each of the component
waves.
7 The wind force is calculated based on API or NPD wind spectra using a
6 one-hour wind speed at 10m. The effects of the second order wind
8 force are included. The current force on the platform is calculated based
9
on Morison’s equation, and is calibrated with wind tunnel test data.

Figure 1. Riser pattern of GOM truss Spar COMPUTER MODELING OF THE TRUSS SPAR

Panel Models
ENVIRONMENTAL CONDITIONS
The Froude-Krylov force is computed by summing up the pressure on
The mooring system of a floating structure is typically designed to the panelized surface of the platform using triangular panels, and the
survive in extreme wind, wave and current environmental conditions force is computed up to the instantaneous wetted surface. For this
having a 100-year recurrence interval. A 100-year hurricane wave reason, the entire surface of the hard tank (including both parts below
condition with associated wind and current conditions in the Gulf of and above the mean waterline up to the under deck steel) is modeled.
Mexico is selected, and is shown in Table 2.

257
The diffraction panel model is different from the Froude-Krolov panel Mooring and Riser System Modeling
model. It models not only the hard tank but also the heave plates and
soft tank. In addition, it is panelized only up to the mean waterline. Each line of the mooring system and steel catenary risers is modeled
Both triangular and quadratic panels can be used for the diffraction and analyzed individually. The 14 top tensioning risers are collapsed to
model in most cases. The diffraction panel model in this case uses an equivalent riser. The analysis of mooring lines, steel caternary risers,
quadratic panels as shown in Figure 3. and top tensioning risers can be based on a quasi-static approach
(implemented in MULTISIM), a finite element dynamic approach
The diffraction panel model is used by diffraction program MORA to (implemented in CAP), or a mixed approach with the combination of
compute hydrodynamic coefficients including diffraction forces, the above two (implemented in CAP as well).
radiation damping coefficients, and mean wave drift force coefficients.
The coefficients from MORA are then imported into CAP to compute According to previous experience, the number of elements for each line
the associated forces. based on the finite element approach is devised differently. In this
study, 20 elements are used for each of the mooring lines, and 14
elements are used for each of the steel catenary and top tensioning
risers.

CASE STUDIES

Based on the application of MULTISIM and CAP, the simulation cases

belonging to the following four tasks are conducted to provide some
insights on the design review and independent analysis of the global
performance responses of the truss Spar:

• Task A - Influence of wave and wind spectra on slow motions and

tensions of the Spar,
• Task B - Extreme responses based on different statistical methods,
• Task C - Morison coefficients for motions analysis and heave plate
hydrodynamics, and
• Task D - Effects of mooring and riser added mass and damping on
Spar motions and mooring/riser dynamics.

The JONSWAP wave spectrum is used and all cases under these four
tasks are 3-hour simulations. The wind spectrum is either the API wind
spectrum or the NPD wind spectrum. For tasks A, B and C, only quasi-
static coupled program MULTISIM is used for the simulations. While
for task D, both MULTISIM and fully dynamic coupled program CAP
are used to run the simulations.
Figure 3. Diffraction panel model
To investigate the influence of wave and wind spectra on Spar slow
motions and tensions, four simulation cases are carried out using
Morison Elements MULTISIM. The parameters of wave and wind spectra used are γ =
1.0, 2.4 and 3.3 for the API wind spectrum and γ = 3.3 for the NPD
The Morison formula is used to compute the drag force on the hard wind spectrum (numbered as cases A0, A1, A2 and A3, respectively).
tank, truss legs, heave plates and braces of the Spar platform. The drag
force is induced by wave, current and motion of the platform, and is In order to obtain the effects of statistical methods on extreme
included in the governing equation of motion. responses, twenty 3-hour simulations with randomly selected seeds
under the 100-year GOM hurricane condition are performed. The first
It may be necessary to use a few integration points along the Morison ten simulations are based on randomly selected seeds for the wave
elements to correctly capture the current force. This is because the spectrum while the wind spectrum uses the same seed (cases B0-B9).
current profile changes along the length of the platform. The current The second ten simulations are based on randomly selected seeds for
force based on the Morison formula has been calibrated with the model the wind spectrum while the wave spectrum uses the same seed (cases
test data. B10-B19).

In addition to the current force on the platform, the heave motion of the Once finished, results of these twenty simulations are used to study the
platform itself induces vertical exciting force on the bottom of the hard most probable extreme response values based on four different
tank and heave plates. The vertical exciting force on the bottom of the statistical methods:
hard tank is modeled by applying the axial drag coefficient on the
Morison members used to model the hard tank. The vertical exciting • Average of the observed extreme values from simulations
force on the heave plates is modeled by a set of Morison thin disks to • Extreme value based on Ochi’s formulation
capture the drag effects. The added mass effects of the heave plates are • Extreme value based on Weibull distribution
modeled in the diffraction model as shown in Figure 3. The drag effects • Extreme value based on Gumbel distribution
are computed by assigning proper drag coefficients to the Morison
members. The hydrodynamic forces acting on the truss legs and heave plates are
calculated using the Morison equation, which is a function of drag
coefficient Cd and added mass coefficient Cm. The drag and added

258
mass coefficients for the truss members are well documented. The
added mass coefficients for the heave plates are modeled in the Apparently, the influence of the peak enhancement parameter γ on the
hydrodynamic diffraction calculations. The uncertainty remains on the Spar slow motions and tensions is very marginal. For example, the
drag coefficients of the heave plates. A systematic variation of drag maximum difference of the horizontal slow translational motion (surge
coefficients Cd of the heave plates around the baseline case C0 (Cd + sway) RMS values among the three loads based on different peak
increased/decreased by 20%, case C1 and C2) is performed. enhancement parameters (γ =1.0, 2.4, and 3.3) is less than 4%. A
comparison of the slow motions and tensions based on the API and
The coupling effects due to mooring and riser added mass and damping NDP wind spectra also reaches the same conclusion.
can be studied based on the application of MULTISIM and CAP,
because MULTISIM and CAP share the same hydrodynamic code for Table 5. Wave and wind spectra on the Spar slow motions and tensions
the motion calculation but differ from the modeling of the mooring and γ Wind Mean LF WF Total Observed Extreme
riser response. MULTISIM is not able to capture the added mass, spectrum RMS RMS RMS Max Min
damping and dynamics induced by the mooring lines and risers. Surge 1.0 API 193.04 5.95 3.86 7.09 221.42 168.83
+ 2.4 API 189.20 6.17 3.96 7.33 224.98 166.81
The study of the coupling effects is performed under the 100-year Sway 3.3 API 187.52 5.93 4.03 7.17 223.38 162.20
GOM hurricane condition, JONSWAP wave spectrum and API wind 3.3 NDP 187.22 5.80 4.03 7.06 213.99 161.18
spectrum. The dynamic coupled analysis program CAP is used.
Heave 1.0 API -1.05 1.01 1.28 1.62 3.80 -5.89
Simulation cases for coupled effects are listed in Table 3.
2.4 API -1.03 0.98 1.12 1.49 3.47 -6.43
3.3 API -1.01 0.95 1.10 1.45 3.31 -5.53
Table 3. Simulation cases for the coupled effects
3.3 NDP -1.01 0.95 1.10 1.45 3.37 -5.36
# Program Mooring SCR TTR
Pitch 1.0 API 1.43 0.54 0.68 0.87 5.39 0.00
D0 MULTISIM Quasi-static coupled
+ 2.4 API 1.42 0.50 0.69 0.85 5.45 0.02
D1 CAP Dynamic Coupled Quasi-static coupled
Roll 3.3 API 1.42 0.49 0.68 0.84 5.83 0.01
D2 CAP Quasi-static coupled Dynamic Quasi-static coupled
coupled 3.3 NDP 1.40 0.49 0.69 0.84 5.65 0.01
D3 CAP Dynamic coupled Quasi-static coupled Yaw 1.0 API 1.11 0.59 0.15 0.61 3.14 -0.81
D4 CAP Dynamic coupled 2.4 API 1.05 0.55 0.15 0.57 3.85 -1.09
3.3 API 1.03 0.57 0.15 0.59 3.12 -0.90
3.3 NDP 1.03 0.55 0.15 0.57 3.02 -0.78
RESULTS AND DISCUSSIONS
Line 1 1.0 API 818 12 15 19 908 754
2.4 API 813 11 15 19 913 760
The results presented and discussed in the following are the mean, root
mean square (RMS) and extreme values of the motion and tension 3.3 API 810 11 15 19 908 760
responses. Because there are resonant frequencies in the low frequency 3.3 NDP 810 11 15 18 904 754
region, it is essential to filter the responses to further explore the Line 2 1.0 API 930 22 27 35 1098 823
coupled effects at different frequency regions as shown in Table 4. 2.4 API 918 21 26 34 1112 832
3.3 API 913 21 26 33 1099 825
Table 4. Frequency region 3.3 NDP 912 20 26 33 1081 822
Frequency ω (rad/s) T (sec)
Low Frequency ≤ 0.31 ≥ 20.3 Table 6. Sensitivity of tensions to spectrum peak enhancement
Wave Frequency 0.31 - 1.6 3.9 – 20.3 parameters
Simulation JONSWAP Peak Enhancement
The motions presented are the horizontal translational motion (surge + γ =1.0 γ =2.4 γ =3.3
sway), vertical translational motion (heave), horizontal rotational 0 1098 1112 1081
motion (yaw), and vertical rotational motion (pitch + roll). The tensions 1 1111 1066 1081
presented are the tension responses of the most and second most loaded 2 1089 1136 1071
mooring lines. The contributions of the riser tensions, added mass and 3 1131 1076 1101
damping are included in the analysis models at different levels, but the 4 1085 1082 1100
riser tensions are not the focus of the study. 5 1091 1078 1060
6 1121 1086 1082
The units used in the following are foot (for motions,1ft=0.3048m) and 7 1080 1067 1090
kips (for tensions, 1kips=4483N). 8 1067 1072 1127
9 1241 1082 1082
A. Influence of wave and wind spectra on Spar slow motions Average 1111 1086 1088
MPEV Gumbel 1089 1076 1079
Wave height time histories generated based on the same random seed MPEV Ochi 1079 1060 1049
are from a JONSWAP wave spectrum with different peak enhancement
MPEV WeiBull 5% 1104 1098 1079
parameter γ ranging from 1.0 to 3.3. Wind velocity time histories with
MPEV WeiBull 10% 1108 1103 1083
the same random seed are generated from an API spectrum with the
MPEV WeiBull 15% 1113 1104 1086
characteristic frequency number equal to 0.025 and a NDP wind
MPEV WeiBull 20% 1108 1101 1088
spectrum.
MPEV WeiBull 25% 1111 1104 1092
Mean, low- and wave-frequency RMS values, and observed extreme MPEV WeiBull 30% 1112 1103 1093
values of the motions and tensions are obtained from each MULTISIM
simulation and are presented in Table 5. The low frequency RMS values of motions and tensions are generally
an indicator of the motion and tension energy in the low frequency

259
region. Since change of the spectrum peak enhancement parameter only The observations found here are inconclusive. In practice, γ and the
changes the wave spectrum shape in the wave frequency region, it has wind spectrum must be selected based on site-specific environmental
limited effects on the low frequency motions and tensions of the Spar. conditions.
These low frequency motions and tensions are more likely to be
affected by the added mass and damping induced by the mooring B. Extreme responses based on different statistical methods
system as discussed below.
The mean value, RMS values, and extreme values are the representative
While γ has limited effects on the low frequency motions and tensions, characteristics of a time domain simulation of a floating system. The
it is desirable to know the sensitivity of the wave frequency extreme input is dynamic time history waves, wind and static current. Thehe
motion and tension response values to this parameter? According to output is the time history of motion and tension responses.
Table 5, the horizontal translational wave frequency motion increases
steadily from 3.86ft to 4.03ft as γ increases. However, this trend is not The dynamic waves and wind in time-domain are generated using
found for the other motions. Although the mooring line tensions are random selected seeds. The time histories generated in this way are
computed quasi-statically, and are not expected to accurately reflect the different from simulation to simulation. For example, the maximum
wave frequency tension RMS values, they can reflect the degree of the wave height from the following ten generations ranges from 35.1ft to
γ effects. 42.7ft – a difference of 20%. The time history responses of a floating
system including motions and tensions under randomly generated wind
Table 7. Spar motion and tension responses (10 random selected wave and waves are thus different. The mean and root mean square values of
seeds) the responses, which reflect the total dynamic energy of the responses,
# Surge + Heave Pitch + Yaw Line1 Line2 are generally insensitive to the random seeds. The remaining task is to
Sway Roll determine the extreme values of the responses. For each time history
Max Min Max Min Max Min Max Min Max Min Max Min response, extreme values can be obtained directly from the simulations.
B0 214 161 3.4 -5.4 5.7 0.0 3.0 -0.8 904 754 1081 822 These values vary from simulation to simulation, and shall not be used
B1 216 167 3.4 -6.2 5.8 0.0 2.7 -0.7 902 755 1081 823 as the representative characteristics of the time history response. The
B2 215 167 3.2 -5.7 5.2 0.0 2.7 -0.8 897 753 1071 820 extreme values must thus be based on multiple simulations of a three-
B3 217 165 3.0 -5.9 5.0 0.0 2.8 -0.8 914 754 1101 823 hour critical sea state. Four methods are applied to calculate the
extreme response values including the average of the extreme observed
B4 217 164 3.8 -5.7 5.6 0.0 3.2 -0.7 913 756 1100 827
values, Ochi’s formulation, Gumbel fitting method, and Weibull fitting
B5 215 167 3.3 -6.0 5.2 0.0 3.0 -0.6 891 760 1060 829
method.
B6 213 164 3.5 -6.3 5.0 0.0 3.1 -0.8 904 751 1082 815
B7 215 165 2.9 -5.6 5.4 0.0 2.8 -0.6 907 752 1090 822 Table 8. Extreme response values (10 randomly selected wind seeds)
B8 226 167 3.5 -5.7 6.7 0.0 2.9 -0.8 925 757 1127 826 Surge+S Heave Pitch+ Yaw Line1 Line2
B9 216 164 3.6 -5.9 4.9 0.0 3.0 -0.6 904 757 1082 822 # way Roll
Average 217 165 3.4 -5.8 5.4 0.0 2.9 -0.7 906 755 1088 823 MaxMin MaxMin MaxMin MaxMin MaxMin Max Min
Uncertainty 1.7 1.2 7.8 -4.7 9.5 - 6.2 -10.4 1.0 0.3 1.7 0.5 B10 214 161 3.4 -5.4 5.7 0.0 3.0 -0.8 904 754 1081 822
(a) Average of the extreme observed values B11 216 165 3.4 -5.4 5.5 0.0 3.1 -0.9 899 758 1079 826
B12 216 164 3.3 -5.4 5.5 0.0 3.0 -0.6 905 755 1087 824
Surge+ Heave Pitch+ Yaw Line1 Line2 B13 215 164 3.3 -5.5 5.6 0.0 2.8 -0.6 894 755 1063 821
Statistical Sway Roll B14 214 166 3.4 -5.3 5.4 0.0 3.0 -0.8 899 757 1071 827
methods Max Min Max Min Max Min Max Min Max Min Max Min
B15 216 164 3.3 -5.2 5.4 0.0 3.4 -1.3 914 755 1099 822
Average 217 165 3.4 -5.8 5.4 0.0 2.9 -0.7 906 755 1088 823 B16 218 166 3.3 -5.4 5.6 0.0 3.5 -1.2 913 747 1100 811
Gumbel 215 166 3.2 -5.7 5.2 0.0 2.8 -0.7 902 756 1079 825 B17 216 167 3.3 -5.4 5.3 0.0 3.6 -1.1 908 755 1089 821
Ochi 217 158 4.8 -6.8 4.9 -2.1 3.2 -1.1 886 734 1049 775 B18 215 165 3.3 -5.4 5.7 0.0 3.1 -0.7 900 754 1072 824
Weibull 5% 216 163 3.6 -5.9 5.3 0.0 3.0 -1.0 905 748 1079 812 B19 217 162 3.4 -5.3 5.4 0.0 3.1 -0.6 907 755 1090 818
Weibull 10% 218 163 4.0 -6.1 5.3 -0.1 3.2 -1.1 905 747 1083 810
Average 216 164 3.3 -5.4 5.5 0.0 3.2 -0.9 904 754 1083 822
Weibull 15% 218 161 4.1 -6.2 5.7 -0.1 3.2 -1.1 907 747 1086 811
Uncertainty 0.5 1.0 1.3 -1.3 2.2 - 7.7 -29.1 0.7 0.4 1.1 0.5
Weibull 20% 220 160 4.2 -6.4 5.6 -0.1 3.3 -1.1 911 745 1088 808
(a) Average of the extreme observed values
Weibull 25% 220 159 4.2 -6.4 5.7 -0.2 3.3 -1.1 913 745 1092 808
Weibull 30% 219 159 4.3 -6.5 5.6 -0.2 3.3 -1.1 914 743 1093 803
Surge+ Heave Pitch+ Yaw Line1 Line2
(b) Predicted extreme values Statistical Sway Roll
methods Max Min Max Min MaxMin MaxMin MaxMin Max Min
As the extreme motion and tension response values vary from Average 216 164 3.3 -5.4 5.5 0.0 3.2 -0.9 904 754 1083 822
simulation to simulation, these values can only be obtained statistically
Gumbel 215 165 3.3 -5.3 5.5 0.0 3.1 -0.8 902 756 1078 824
by running a number of simulations for each of the γ's. Ten simulations
Ochi 217 158 4.8 -6.9 4.9 -2.1 3.2 -1.2 886 733 1050 775
are run for each γ. These additional simulations show that higher
Weibull 5% 217 163 3.9 -5.6 5.6 0.0 3.4 -1.0 907 751 1079 815
enhancement parameters do not necessarily yield larger extreme motion
Weibull 10% 222 162 4.3 -6.3 5.8 -0.1 3.4 -1.1 909 748 1088 814
and tension response values. Table 6 shows the sensitivity of the
Weibull 15% 221 161 4.6 -6.6 5.9 -0.1 3.4 -1.2 908 746 1086 809
extreme tension of the most loaded line to γ. The extreme tensions are
Weibull 20% 222 160 4.7 -6.7 5.9 -0.1 3.4 -1.1 910 744 1089 810
calculated by averaging the maximum observed tensions from each of
Weibull 25% 222 159 5.0 -6.7 5.9 -0.2 3.4 -1.2 911 745 1090 807
the ten simulations with the same parameter γ. The predicted extreme
Weibull 30% 221 159 4.8 -6.7 5.9 -0.2 3.4 -1.2 911 743 1090 805
values based on other estimate methods are also included.
(b) Predicted extreme values

The simplest method is the average of the extreme observed values.

This is done by averaging the extreme observed values from a number

260
of simulations. The uncertainty of this method can be calculated easily. Alternatively, Ochi’s formulation, the Gumbel and Weibull
The averaged extreme response values of the Spar motions and tensions distributions can be used to predict the extreme response values. These
from ten simulations generated by randomly selected wave seeds and values are the most probable extreme values, and have probability of
ten simulations generated by randomly selected wind seeds are being exceeded equal to 0.632. The extreme response values of the
calculated and presented in Table 7 and Table 8. The uncertainty of the Spar motions and tensions are presented in Table 7 and Table 8. The
response in the tables is the percentage of the standard deviation with Gumbel distribution is obtained by fitting the distribution to the
respect to the averaged response value. According to the tables and extreme observed values of the ten simulations. The Weibull
excluding the extremely small responses, the uncertainty based on the distribution is obtained by fitting the distribution to the peaks of the
10 simulations is relatively small. simulations. This is done by first combining the peaks from each
simulation. The combined peaks are then sorted ascendingly.

When fitting the Weibull distribution to the peaks, only the top portion
of the sorted peaks is used. If all the peaks are used, the distribution
tends to be heavily biased towards the most frequently occurring values
and may not provide a good estimate of the extreme values.The upper
tailing peaks are far more important than the lower tailing peaks for the
purpose of the extreme value predictions, but the number of peaks to be
extracted is hard to establish. For this reason, the Weibull fitting based
on the upper portion of the peaks is performed, and sensitivity of the
predicted extremes to the number of peaks extracted is investigated.

According to the predictions, the extreme values based on the average

(1) Second most loaded line with 10 wave seeds of the extreme observed values differ slightly from those based on the
Gumbel fitting method. Both of them are marginally smaller than those
based on the Weibull fitting method. Extreme values generated by
Ochi’s formula are about 3 percent lower than those generated by the
other three methods.

The extreme values based on the Weibull fitting method are generally
insensitive to the number of peaks extracted. However, Figure 4
suggests that a sufficient number of peaks (10%-30%) are to be
extracted to provide an unbiased and stable estimate.

Table 9. Effects of Heave Plate Drag Coefficients on Platform Motion

# Mean LF RMS WF RMS Total RMS Observed Extreme
(2) Most loaded line with 10 wave seeds
Max Min
Surge C0 193.04 5.95 3.86 7.09 221.42 168.83
+ C1 187.17 5.78 4.03 7.05 213.88 161.21
Sway C2 187.27 5.82 4.02 7.07 214.12 161.13
Heave C0 -1.05 1.01 1.28 1.62 3.80 -5.89
C1 -1.00 0.88 1.09 1.40 3.13 -5.11
C2 -1.03 1.05 1.11 1.53 3.68 -5.73
Pitch C0 1.43 0.54 0.68 0.87 5.39 0.00
+ C1 1.40 0.48 0.69 0.84 5.63 0.01
Roll C2 1.40 0.49 0.68 0.84 5.68 0.00
Yaw C0 1.11 0.59 0.15 0.61 3.14 -0.81
C1 1.02 0.54 0.15 0.56 3.00 -0.76
(3) Second most loaded line with 10 wind seeds C2 1.03 0.55 0.15 0.57 3.03 -0.80
Line 1 C0 818 12 15 19 908 754
C1 810 11 15 18 906 753
C2 810 11 15 18 902 754
Line 2 C0 930 22 27 35 1098 823
C1 912 20 26 33 1083 821
C2 913 20 26 33 1078 823

C. Morison coefficients for motion analysis and heave plate

hydrodynamics

The hydrodynamic forces acting on the truss legs and heave plates are
(4) Most loaded line with 10 wind seeds calculated using the Morison equation, which is a function of drag
coefficient Cd and added mass coefficient Cm. The drag and added
Figure 4. Extreme mooring line tensions mass coefficients for the truss legs are well documented. The added
mass coefficients for the heave plates are modeled in the hydrodynamic
diffraction calculation. A systematic variation of drag coefficients Cd

261
of the heave plates around the baseline case is performed to investigate
the sensitivity of Spar responses to heave plate drag coefficients. This is Wave-Frequency Responses. The RMS values of the wave frequency
done by increasing the drag coefficient by 20% in one case (C1) and by motion and line tension (most loaded and second most loaded) are
decreasing the drag coefficient by 20% in the other case (C2). Results presented for the Spar in the 100-year hurricane condition . The wave
of the simulation cases are presented in Table 9. frequency motion RMS values are about the same regardless of the
programs used and the variations of the coupled levels. The mooring
The increased and decreased drag coefficients have little effect on the line tensions, on the other hand, differ significantly. The wave
Spar motions and mooring line tensions. frequency RMS tension values by CAP are significantly larger (30% to
50%) than those by MULTSIM. In MULTSIM, the dynamic tensions
D. Coupling effects of mooring and riser added mass and damping cannot be modeled. This results in smaller wave frequency line tensions
on Spar motions and mooring/riser dynamics by MULTISIM. In practice, the line dynamic tensions are not directly
obtained from MULTISIM, but through a line dynamic program. From
Simulations are performed for quasi-static coupled and dynamic the global motion point of view, the truss Spar has larger line dynamics
coupled motion and tension responses of the Spar under the 100-yr- due to the fact that there is no large mass of entrapped water associated
hurricane environment condition as mentioned above. Simulation time with the traditional cylindrical Spar.
for each run is 3 hours with time steps of 0.1 second for the dynamic
coupled analysis and 1 second for the quasi-static coupled analysis. The Table 11. Results of the coupled analysis.
simulation time includes an additional 1,000-second ramp to minimize # Mean LF WF Total Observed Extreme
the starting transient effect. RMS RMS RMS Max Min
Surge D0 193.04 5.95 3.86 7.09 221.42 168.83
The mooring lines and risers in each platform configuration are + D1 180.24 3.87 4.14 5.66 210.79 157.47
modeled individually. The application of CAP and investigation of the Sway D2 183.21 4.27 4.08 5.91 217.16 159.72
coupled effects are carried out through a comparison between results of D3 179.32 3.75 4.14 5.58 209.22 156.81
the quasi-static coupled (or dynamic uncoupled in the following) and D4 169.76 3.50 3.89 5.24 197.73 148.12
the dynamic coupled simulations, and can be categorized into the Heave D0 -1.05 1.01 1.28 1.62 3.80 -5.89
simulation cases shown in Table 3: D1 -1.01 0.69 0.97 1.19 2.86 -5.00
D2 -0.98 0.87 1.07 1.38 3.12 -5.32
The mean and RMS values of the motion and tension responses are
D3 -1.04 0.67 0.94 1.16 2.75 -4.92
computed to investigate the coupled effects. Table 11 shows motion
and tension responses of the Spar in the 100-year hurricane condition. D4 -1.04 0.65 0.87 1.08 2.52 -4.68
Detailed comparisons of the mean and RMS values in different Pitch D0 1.43 0.54 0.68 0.87 5.39 0.00
frequency regions are given in the following sections in terms of mean + D1 1.44 0.50 0.79 0.93 6.53 0.00
responses, wave-frequency responses, low-frequency responses. Roll D2 1.46 0.48 0.79 0.93 6.43 0.00
D3 1.43 0.47 0.80 0.93 6.45 0.00
Mean Responses. Coupled contributions from inertia force and current D4 1.47 0.42 0.87 0.97 6.62 0.05
loads on the mooring lines and risers can be shown by studying the Yaw D0 1.11 0.59 0.15 0.61 3.14 -0.81
mean horizontal translational motion responses. D1 0.95 0.43 0.30 0.52 3.27 -0.59
D2 1.01 0.53 0.20 0.57 3.07 -0.93
Table 10. Motion and tension responses of the Spar in 100-year D3 0.97 0.36 0.29 0.46 2.86 -0.42
hurricane condition D4 1.10 0.24 0.28 0.37 2.96 -0.02
# Status Extreme Extreme Extreme Sway Line 1 D0 818 12 15 19 908 754
Surge+Sway Surge D1 808 25 105 108 1286 438
D0 Uncoupled 193 146 117 D2 806 9 15 18 902 751
D1 Mooring coupled 180 145 108 D3 808 24 104 107 1279 443
D2 SCRs coupled 183 144 114 D4 810 24 108 111 1306 431
D3 Mooring+SCRs 179 144 106 Line 2 D0 930 22 27 35 1098 823
coupled D1 901 33 131 135 1500 470
D4 All coupled 170 134 105 D2 904 16 26 31 1084 816
D3 901 32 131 134 1491 471
In Table 10, the mean horizontal translational motion by MULTISIM D4 901 31 136 140 1527 456
differs from that by CAP where the mooring lines, steel catenary risers,
and top tensioning risers are dynamically coupled at different levels. Wind and Wave Induced Low-Frequency Responses. Resonant low
There is a difference in the sway motion of the Spar by the two frequency motions will be significantly affected by the damping level
methods. MULTISIM is unable to accurately predict the sway motion when exciting low frequency wind and wave loads are around the
because the direction of the sway motion is not in the planes where the natural frequencies of the Spar.
majority of the mooring lines are defined.
RMS values of the low-frequency motions of the Spar in the 100-year
Another source of discrepancy is the computation of the current forces hurricane condition are presented. The RMS values in the low-
on the mooring lines and risers. In MULTSIM, the mean current forces frequency region by CAP are smaller than those by MULTISIM. The
on the lines can be computed and applied at the associated fairlead reduction in the RMS values is attributed to the added mass and
positions. This may differ from the actual current forces exactly damping of the moorings and risers modeled in CAP but not in
calculated and applied at the element level in the coupled analysis. In MULTISIM, which is in line with the results reported by Webster
the 100-year hurricane condition, the current force on the mooring lines (1995) that the damping induced by the mooring lines are substantial in
and risers is insignificant, and hence, is ignored in the current reducing the low frequency motions of a moored structure.
calculation.

262
On the other hand, the corresponding low frequency tension RMS Chen, XH, Zhang, J., Johnson, P and Irani, M. “Studies on the
values of the most loaded line and second most loaded line of the Spar Dynamics of Truncated Mooring Line,” ISOPE (2000)
by CAP are larger than those by MULTISIM. This is due to the
dynamic effects included in CAP but not in MULTISIM. Chen, XH, Zhang, J, Johnson, P and Irani, M (2001). “Dynamic
Analysis of Mooring Lines by Using Three Different Methods,” ISOPE
CONCLUSIONS (2001)

A series of parametric studies have been conducted to evaluate the Datta, I, Prislin, I, Halkyard, JE, Greiner, WL, Bhat, S,
coupling effects on a truss Spar. In addition, the influence of wave and Perryman, S and Beynet, PA (1999). “Comaprison of Truss Spar
wind spectrum on slow motions and Morison coefficients for motion Model Test Results with Numerical Predictions”, OMAE99.
analysis are also investigated. The conclusions are as follows:
Kim, MH and Yue, DKP (1990). “The complete second-order
• In the cases studied, peak enhancement parameter γ has very diffraction solution for a axisymmetric body, part 2, bichromatic
marginal influence on the Spar slow motions and tensions. A incident waves and body motions”, J. Fluid Mech., Vol. 211, pp. 571-
higher enhancement parameter does not necessarily yield larger 593
extreme motion and tension response values.
Kristoffersen, G and Weaver, T (2001). “Mooring Analysis of a Truss
• According to the predictions, the extreme values based on the Spar”, OTC 13271, 2001
average of the extreme observed values differ slightly from that
based on the Gumbel fitting method. Both of them are marginally Ma, W and Webster, WC (1994). “An analytical approach to cable
smaller than the values based on the Weibull fitting method. dynamics: theory and user manual”, Sea Grant Project R/OE-26.
Ochi’s formula generally produces the lowest predictions among
the four methods investigated. Ma, W, Zou, J, Lee, MY and Huang, E (2000). “Deepwater nonlinear
coupled analysis toll”, OTC 12085, 2000
• Based on the parametric studies, insignificant changes in drag
coefficients of heave plates have limited effects on the Spar Magee, A, Sablok, A, Maher, J, Halkyard, J and Finn, L (2000).
motions and mooring line tensions. “Heave Plate Effectiveness in the Performance of Truss Spars”,
OMAE2000, OSU OFT-4230
• Finally, as to the coupling effects on the Spar responses, the wave
frequency motion RMS values are about the same regardless of the Mekha, BB, Johnson, CP and Roesset, JM. (1995). “Nonlinear response
coupling levels. The mooring line tensions, on the other hand, of a Spar in deep water: different hydrodynamic and structural models”,
differ significantly. The RMS values in low-frequency motion by Proc. 5th ISOPE conf., Vol. 3, pp. 462-469.
CAP are smaller than those by MULTISIM. However, the
corresponding low frequency tension RMS values of the most Molin, B (1979). “Second-oder diffraction loads uponthree dimensional
loaded line and second most loaded line of the Spar by CAP are bodies”, Appl. Ocean Res., Vol. 302, pp. 203-229.
larger than that by MULTISIM due to the dynamic effects
included in CAP but not in MULTISIM. Paulling, JR (1995). “MULTISIM Time Domain Platform Motion
Simulation for Floatinh Platform Consisting of Multiple Interconnected
ACKNOWLEDGEMENT Modules,” Theoretical Manual, Third Edition, November 2, 1995.

The authors are grateful to James Card, Senior Vice President, and John Ran, Z and Kim, MH (1998). “Coupled dynamic analysis of a moored
Spencer, Vice President of ABS for their valuable comments and Spar in random waves andcurrents (time-domain vs. frequency domain
reviews. The authors are indebted to ABS librarian Deborah analysis)”, 17th OMAE International Conference, OMAE98-0604
Feuerbacher for editing the manuscript.
Tahar, A, Ran, Z and Kim, MH(2002). “Hull/mooring/riser coupled
REFERENCES: spar motion analysis with buoyancy-can effect”, Proc. Of 12th Intl.
Offshore and Polar Engn. Conf..
Chau, FP. and Eatock Taylor, R. (1992) “Second-order wave
diffraction by a vertical cylinder,” J. Fluid Mech., Vol. 242, pp.571- Webster, WC(1995) “Mooring induced damping”, J. Ocean
599. Engineering , vol 22, no.6, pp. 571-591.

263