Laboratoire d’Analyses et Modélisation de Systèmes pour

1’Aide à la Décision
UMR 7243

CAHIER DU LAMSADE
399
Revised version June 2022

Integrated Laycan and Berth Allocation Problem

with ship stability and conveyor routing
constraints in bulk ports

Hamza Bouzekri, Gülgün Alpan and Vincent Giard

 Integrated Laycan and Berth Allocation Problem
with ship stability and conveyor routing constraints
in bulk ports

In this paper, we study the integrated Laycan and Berth Allocation Problem
(LBAP) in the context of bulk ports, which considers two problems in an
integrated way: the tactical Laycan Allocation Problem and the dynamic hybrid
case of the operational Berth Allocation Problem. To make the LBAP closer to
reality, we consider tidal bulk ports with conveyor routing constraints between
storage hangars and berthing positions, preventive maintenance activities,
multiple quays with different water depths and fixed heterogeneous bulk-
handling cranes, navigation channel restrictions, vessels with multiple cargo
types, charter party clauses, non-working periods, and ship stability
considerations during loading operations. The proposed integer programming
model aims to define an efficient schedule for berthing chartered vessels and
optimal laycans for new vessels to charter. The model is formulated with
predicates that guarantee maximum flexibility in the implementation and greatly
improve the computational performance. Finally, the model is tested and
validated through a small set of relevant case studies inspired by the operations of
OCP Group at the bulk port of Jorf Lasfar in Morocco in very reasonable
computational time using commercial Software.

Keywords: laycan allocation; berth allocation; tidal bulk ports; ship stability;
conveyors; preventive maintenance; integer programming

1. Introduction:
With an estimated 80 per cent of the volume of world merchandise trade by sea,
international shipping and ports provide crucial linkages in global supply chains and are
essential to enable all countries to access global markets UNCTAD (2019). Although
containerization has revolutionized the shipping industry, bulk cargoes are still the
fundamental and enduring trades that support the dynamism of maritime transport. Five
cargo types can be distinguished: container cargo, liquid bulk, dry bulk, break-bulk
cargo, and ro-ro. In container terminals, all cargo is packed into standard containers, and
thus there is no need for any specialized equipment to handle any particular type of
cargo. In contrast, cargo is not packaged in bulk ports, and a wide variety of
loading/unloading equipment and means of transport is used depending on the vessel
requirements and cargo properties. For example, dry bulk goods are handled using fixed
bulk-handling cranes and are transferred using conveyors between storage hangars and
bulkers considering conveyor routing constraints, while liquid bulk goods need
pipelines to be handled and transferred between storage tanks and tankers. On the other
hand, containers are handled using mobile cranes and are transferred between storage
areas and container ships using internal vehicles. Despite their importance in maritime
logistics, bulk ports have received less attention than container terminals in the
scientific literature.
Container and bulk port management share many common characteristics, but
some specificities prevent applying RO models treating containerships to bulkers. To
show the originality of our approach, we must point out the shared and distinctive
characteristics of these two classes of problems, mainly since our paper uses several
proposals on the treatment of common characteristics proposed in (Bouzekri et al.
2021). Figure 1 compares the characteristics of container and bulk port models by
focusing mainly on the relationship between vessels and the port and the relationship
between vessels and warehousing. A third axis treats general characteristics of time and
space referencing as well as the used optimization criterion; these shared characteristics
cannot be included in the sets defining the two-precedent axis. Characteristics are
numbered, and a solid line links the characteristics retained in our modelling, whereas a
dashed line links the characteristics considered in the model previously proposed for
container vessels. Many of the common characteristics were never considered by
models dealing with bulk ports; there is no reason not to introduce them in our
modelling, which has its own scientific originality.
 As shown in Figure 1, the main difference between container and bulk port
problems deals with the interaction between vessel and warehousing, underlined by the
characteristic ❽, essential to consider in case of a significant variety of goods to load
(if this variety does not exist, the bulk problem is not very different from the container
problem). A second difference is the consideration of ship stability during the loading
(characteristic ④). In the case of bulkers, cargoes are loaded in several holds, each hold
receiving a unique kind of goods to keep product integrity. Loading operations are
processed from a conveyor along the vessel, considering the possibility of transporting
batches of different goods to store in different holds. This detailed scheduling problem,
close to real-time, is correctly treated by commercial Software which does not consider
conveyor constraints. At the studied operational level, our concern is the expedition of
batches from hangars to vessels compatible with loading constraints considered in detail
by commercial Software.
 The used economic criterion (characteristic (a)) conciliates two important
decision problems in port management: the tactical Laycan Allocation Problem (LAP)
and the operational Berth Allocation Problem (BAP). The LAP assigns berthing time
windows to new vessels to charter within a medium-term planning horizon (three to
four weeks), considering the availability of cargo and port resources (berthing positions,
handling equipment, etc.). Hence, the LAP has a clear interaction with one of the most
important operational problems in the seaside area of ports: the BAP. The latter assigns
berthing positions and times to every vessel projected to be served within a short-term
planning horizon (one to two weeks) such that a given objective function is optimized.
To easily manage the integration between these two problems that have different
decision levels, we consider a modular decision time-interval inside the planning
horizon (characteristic ⑰). This approach was first proposed in Bouzekri et al. (2021)
to integrate the LAP with the integrated Berth Allocation and Quay Crane Assignment
Problem (BACAP) in the context of container terminals. To the best of our knowledge,
the current paper is the first research to integrate the LAP and the BAP in the context of
bulk ports, considering ship stability and conveyor routing constraints with preventive
maintenance activities between storage hangars and berthing positions.
This study also considers all common constraints of port management listed
above (Figure 1), never considered altogether in bulk ports modeling. Finally, we use
predicates in the proposed integer linear programming model to define the feasibility
zone of decision variables. This approach permits reducing the number of variables and
constraints and hence makes it possible to solve real size problems using commercial
Software. This type of formulation has successfully been used in Bouzekri et al. (2021).
The current paper confirms its efficiency.
The remainder of this paper is organized as follows. In Section 2, we present a
literature review of the BAP and the LAP in the context of bulk ports. Section 3 is
dedicated to the description of the LBAP, while Section 4 is dedicated to its
mathematical formulation. In Section 5, we present a case study with an illustrative
example, then we discuss the results. Finally, in Section 6, we draw some conclusions
and indicate future research directions.

2. Literature review:
In this section, we review the academic literature on the BAP and the LAP in the
context of bulk ports.
 2.1.BAP Literature:
The BAP in bulk ports has received little attention in Operations Research literature
compared to container terminals until recently. A list of papers that propose new models
for the BAP in the context of bulk ports, as an individual problem or using an integrated
approach, is described below.
The berth layout can be either discrete, continuous, or hybrid. Barros et al.
(2011) propose an integer linear programming model for the discrete BAP considering
homogeneous berthing positions with tide and stock level constraints, prioritizing
vessels related to the most critical mineral stock level. The authors then propose a
Simulated Annealing-based algorithm as a valid alternative to the commercial solver to
find good and fast solutions for hard instances. Ribeiro et al. (2016) also solve the
discrete BAP by proposing a mixed-integer linear programming model considering
maintenance activities. The authors model each maintenance activity as a dummy vessel
which must be handled at a precise time by a specific berthing position, which means
that this berthing position cannot receive vessels during that time. They then develop an
adaptive large neighborhood search heuristic that finds good solutions within low
computational times on all instances.
Ernst et al. (2017) solve the continuous BAP with tidal constraints that limit the
departure of fully loaded vessels from dry bulk terminals using a commercial solver.
The authors propose two new mixed-integer linear programming models, and then they
provide several valid inequalities for both models, which improve both their solution
quality and run time. To solve efficiently medium to large-sized instances of Ernst et al.
(2017), Cheimanoff et al. (2020) develop a metaheuristic approach based on the
Reduced Variable Neighborhood Search. The authors also develop a machine learning
algorithm to tune the metaheuristic's hyper parameters.
Umang et al. (2013) study the hybrid BAP by proposing two exact methods
based on mixed-integer programming and generalized set partitioning and a heuristic
method based on squeaky wheel optimization. The authors consider the fixed equipment
facilities, such as conveyors and pipelines, which are installed at only certain sections
along the quay, the cargo type on the vessel and its draft. They also consider the time
taken to transfer cargo between its location on the yard and the berthing position of the
vessel. de León et al. (2017) propose a Machine Learning-based system to select the
best algorithm for solving the BAP model proposed by Umang et al. (2013) in each
particular case. The latter depends on factors such as the percentage of vessels that need
specialized handling equipment, and the congestion level, which is influenced by the
distribution of the estimated time of arrival of vessels and their workload.
Most authors consider dynamic vessel arrivals, while Tang et al. (2016) consider
static vessel arrivals. The authors implement a multi-phase particle swarm optimization
algorithm to minimize the total service time of vessels or their makespan.
Since the operational problems observed in port terminals are often interrelated,
some authors study the BAP using an integrated approach. Indeed, Robenek et al.
(2014) extend the work of Umang et al. (2013) by integrating berth allocation and yard
assignment problems. The authors propose an exact solution algorithm based on a
branch and price framework and a metaheuristic approach based on a critical-shaking
neighborhood to solve this integrated problem. Al-Hammadi and Diabat (2017) apply
the model proposed by Robenek et al. (2014) for Mina Zayed Port in Abu Dhabi in
order to test different scenarios as a means of sensitivity analysis, with respect to certain
factors such as the congestion level, in terms of the relative arrival time of vessels, the
unavailability of certain resources and the addition of new resources.
In the same logic of integrating problems, Unsal and Oguz (2019) propose an
exact solution procedure for an integrated problem that consists of three operations:
berth allocation, reclaimer (a large machine used to recover bulk material from
a stockpile) scheduling and stockyard allocation, considering tide and reclaimers non-
crossing constraints. The authors develop a novel logic-based Benders decomposition
algorithm in which a master problem and a subproblem are modeled using mixed-
integer programming and constraint programming, respectively. The subproblem's role
is either to find a feasible schedule for reclaimer schedules and yard allocations given
mooring and departure times of vessels or to prove that the problem instance is
infeasible.
Note that few papers consider tide with navigation channel restrictions, such as
the maximum number of vessels to pass simultaneously through the navigation channel
and the vessels' incapability to pass in opposing directions. These restrictions are
considered by Zhen et al. (2017) and Corry and Bierwirth (2019) in the context of
container terminals. In the context of bulk ports, Pratap et al. (2017) develop a decision
support system to solve the integrated problem of berth and ship unloader allocation,
under the condition that the channel allows only one vessel to pass at a time, using
metaheuristics. The authors consider two different approaches: either solving the
problem sequentially as a two-phase optimization model, berth allocation and ship
unloader allocation or integrating the two phases in a single-phase problem. The
integrated approach gives a better result than the sequential approach, but the latter is
useful for the port authorities to revise their contract with their clients. Liu et al. (2021)
propose a mixed-integer linear programming model for integrated planning of berth
allocation and vessel sequencing in tidal seaports with one-way navigation channel,
which obliges vessels to queue up to enter or leave the port alternately. The authors also
develop a tailored adaptive large neighborhood search algorithm to solve the integrated
problem within a reasonable time.
Krimi et al. (2019, 2020) study the integrated Berth allocation and Quay Crane
Assignment Problem in tidal bulk ports with multiple quays, vessels with multiple cargo
types, and unavailability constraints due to preventive maintenance of quay cranes and
bad weather conditions. The authors develop a general variable neighborhood search-
based approach to solve instances the commercial solver failed to solve optimally.
We also note that we found no papers that consider the BAP with conveyor
routing constraints between storage hangars and berthing positions. Some authors
consider conveyor routing constraints in other problems. For example, Menezes et al.
(2017) study the production planning and scheduling problem in bulk ports which
defines the amount and destination of each product and simultaneously establishes a set
of feasible routes from storage subareas to vessels, where there is no conflict regarding
equipment allocation. However, the authors consider the berthing positions of vessels as
inputs in the problem.
The journal papers cited above are summarized in Table 1, in which the
following information is presented.
 Port type: either import or/and export ports.
 Spatial attribute: it concerns the berth layout (either discrete, continuous or
hybrid), the number of quays (either a single quay or multiple quays), and it
specifies if the BAP considers the draft of berthing positions and the restrictions
of the navigation channel and the conveyor system when deciding on a vessel's
berthing position.
 Temporal attribute: it describes the arrival process of vessels (either static or
dynamic), and it specifies if the BAP considers tide constraints and non-working
periods (e.g., non-working days and maintenance activities) when deciding on a
vessel's berthing time.
  Handling time attribute: it describes the productivity of handling equipment
(either homogeneous or heterogeneous). In the case of homogeneous handling
equipment, handling time of vessels is fixed, while it is variable in the case of
heterogeneous handling equipment. The attribute also specifies if the BAP
considers the distance between berthing positions and storage locations when
calculating a vessel's handling time.
 Performance measure attribute: it specifies the optimization criteria used in the
objective function (either efficiency or effectiveness). Most models consider
minimizing various times or costs.
 Vessel attributes: it specifies the number of cargo types (if a vessel carries only a
single type or multiple types of cargo).
 Problems integrated with the BAP: it specifies the problems that are integrated
with the BAP when the latter is studied using an integrated approach.
 Resolution approach: either exact methods, heuristics, or/and metaheuristics.
 Modeling choices: it specifies if the conditions of the BAP (berthing time and
space, draft, tide, navigation channel, non-working periods) are modeled as
either constraints or predicates using binary variables.
For detailed reviews of the BAP literature in the context of container terminals,
we refer readers to Bierwirth and Meisel (2010, 2015).

2.2.LAP Literature:
While BAP literature is abundant, only two papers were found that deal with the LAP.

Lorenzoni et al. (2006) develop a mathematical model, based on a multi-mode

resource-constrained scheduling problem improving the attendance of vessels. The
proposed model determines laycans in a way that avoids simultaneous or nearly
simultaneous arrivals of vessels competing with the same port resources (berthing
positions, handling equipment, etc.), under the first come first served regime of
attendance. However, the authors consider only time windows for resources' availability
without considering spatial constraints such as vessel and berth lengths.
Bouzekri et al. (2021) study the integrated Laycan and Berth Allocation and
time-invariant Quay Crane Assignment Problem in tidal ports with multiple quays.
Then, they extend the integrated problem to the Specific Quay Crane Assignment,
which includes the assignment of a set of specific quay cranes to each vessel. This
research is more suitable for container terminals since it does not specify the cargo type,
which is an important point in bulk ports.
As a conclusion to this section, we highlight that in this paper, we propose a new
integer linear programming model for a new integrated problem: the LBAP in the
context of bulk ports. Moreover, we consider numerous conditions related to port
management in the definition of the LBAP, which reduces the gap between the abstract
representation of the studied problem and its applicability in real situations. Indeed, our
model considers tidal bulk ports that have conveyor routing constraints between storage
hangars and berthing positions with preventive maintenance activities, a navigation
channel, multiple quays with different water depths and heterogeneous loading
equipment, vessels with multiple cargo types, charter party clauses, non-working
periods, and ship stability constraints.
Table 1. BAP literature in the context of bulk ports.
Performance
Port Handling time Vessel Problems integrated Resolution
Spatial attribute Temporal attribute measure Modeling choices
type attribute cargo types with the BAP approach
attribute
number berth vessel handling

quay equipment assignment

yard equipment scheduling

heuristics/metaheuristics
of quays layout arrivals equipment

non-working periods

non-working periods
navigation channel

vessel sequencing
conveyor system

laycan allocation

yard assignment

exact methods

berthing space
berthing time
effectiveness
maintenance
Authors

efficiency
distance
heterogeneous
homogeneous
import
export

mono

multi
draft

draft
continuous

tide

tide
dynamic
discrete

hybrid
mono
multi

static
Barros et al. (2011) x x x x x x x V x x C C C
Umang et al. (2013) x x x x x x x x S x x x C C C
Robenek et al. (2014) x x x x x x x x S x x x x C C C
Ribeiro et al. (2016) x x x x x x x V x x C C C
Tang et al. (2016) x x x x x x S, M x x C C
Pratap et al. (2017) x x x x x x W, H, R x x x C C
Al-Hammadi and Diabat (2017) x x x x x x x x S x x x C C C
Ernst et al. (2017) x x x x x x D x x C C C
Unsal and Oguz (2019) x x x x x x x D x x x x C C C
Krimi et al. (2019, 2020) x x x x x x x T x x x C C C C
Cheimanoff et al. (2020) x x x x x x D x x C C C
Liu et al. (2021) x x x x x x x x x S x x x C C P C
Our paper x x x x x x x x x x x V D x x x P P P P P

Performance measure attribute: D: Departure times; T: Tardiness; W: Waiting times; H: Handling times; S: Total service times; M: Makespan; R: Priority deviation; V: Demurrage vs despatch
Modelling choices: P: Predicate; C: Constraint
 3. Problem description:
In this section, we present all the constraints considered in the problem modelling.
These constraints include both general port constraints and specific bulk port
constraints. Then we present some possible optimization criteria related either to
efficiency or effectiveness.

3.1.General port constraints:

We consider a tidal port with multiple quays. Each quay has a hybrid layout where large
vessels may occupy more than one berthing position; however, small vessels cannot
share a berthing position. In Figure 2, the vessel 3 occupies the berthing position 5 that
is the union of berthing positions 3 and 4, however, the vessels 2 and 2′ cannot share the
berthing position 2.

Figure 2. Hybrid berth layout.

Each berthing position is characterized by a length and a minimum water depth.

All the berthing positions of a quay can have the same water depth, or the water depth
increases seaward by berthing positions, as in Figure 3. The indexation of berthing
positions is independent of the quays.
Figure 3. Example of a bulk port.

We consider three types of vessels:

 Already berthed vessels: these vessels have residual handling time and a
predetermined berthing time and position.
 Chartered vessels: the charter party of these vessels is already signed.
Consequently, their expected arrival time is fixed. The decisions remaining to
take are when and where to berth.
 New vessels to charter: the charter party of these vessels is under negotiation.
Consequently, their laycan is not yet fixed. The decisions to take are the first
layday and where to berth.
We assume dynamic vessel arrivals, which means that expected arrival times are
given for chartered vessels. Each vessel is characterized by a length and a draft. A
maximum waiting time in the harbor per vessel is also introduced to circumvent
solutions with very high waiting times. The port manager can fix this parameter based
on what he judges acceptable. Besides this practical relevance, it also plays an
interesting role in the computational performance by limiting the search space. We also
consider the technical constraints of vessels that prohibit their berthing at some berthing
positions or oblige them to berth at a specific berthing position.
In tidal ports, the use of the navigation channel is impacted by the tide cycle. We
assume that large loaded vessels with deep drafts cannot pass through the navigation
channel while leaving the port during low tides and thus have to wait for high tide
cycles where the sea level is superior to their drafts (Figure 4). The detailed calculation
of the tide parameters used in the case study is shown in the appendix (Table 11).
Moreover, a maximum number of vessels must not be exceeded in a given period of
time while passing simultaneously through the navigation channel.

Figure 4. Navigation channel restrictions.

We also consider non-working periods, which can either be included or excluded
in the counting of the laytime. For example, in SSHEX (Saturdays/Sundays/Holidays
Excluded), the time lost in port on Saturdays, Sundays and Holidays does not count as
laytime (from 5 pm on Friday until 8 am on Monday, and on holidays from 5 pm of the
day preceding a holiday until 8 am of the next working day), while in SSHINC
(Saturdays/Sundays/Holidays Included), no exception periods are in effect and the
laytime will count seven days a week as well as during holidays. The used restriction of
the case study is given in appendix (Table 12).

3.2.Specific bulk port constraints:

The following list of specific bulk port constraints is considered in our problem
modelling.
Each berthing position in the port is characterized by one (or two) fixed bulk-
handling crane(s) also characterized by a productivity (see Figure 3). Berthing positions
are linked to storage hangars by a conveyor system, which can be divided into sections
composed of identical parallel conveyors.
Berthing positions are linked to storage hangars by a conveyor system, as shown
in Figure 5. The conveyor system is a set of identical parallel conveyors (represented by
horizontal bars) connected by switches (represented by dots) and identical
feeding/transfer conveyors (represented by vertical bars). The black box represents a
flexible transfer system that connects the upstream conveyors linked to the hangars to
the downstream conveyors linked to each berthing position on the quays. A route is a
collection of interconnected horizontal and vertical conveyors that links a hangar to a
berthing position. The number of possible routes is quite high (e.g., more than 1.3
million combinations in the example of Figure 5). It must be noted that all potential
routes cannot be operational at the same time since a conveyor cannot transport two
different bulk products at a time. This particularity is used to circumvent the
combinatorial nature of the problem by defining compatible routes Menezes et al.
(2017). Two routes are said to be compatible if and only if they do not share a conveyor
(or a switch). This reduces considerably the number of routes to consider in assigning
conveyors. However, it is a delicate task to list all compatible routes without errors.

Figure 5. Sections of the port conveyor system.

Therefore, in this paper, we adopt another approach. Instead of considering a

conveyor assignment problem, we consider conveyor capacity allocation. To this end,
we divide the port conveyor system into sections (for example, two sections, s=1 and 2,
in Figure 5). We define two parameters: the first one gives the number of identical
parallel conveyors in each section that expresses the maximum number of bulk products
that can be transported simultaneously in the given section (e.g., in Figure 5, maximum
3 in section 1 and 4 in section 2). The second one is a Boolean parameter with three
indices (storage hangars, berthing positions and sections) that specifies which section is
needed to transfer a product from a given hangar to a given berthing position (e.g., in
Figure 5, the conveyors of any route between berthing position 1 and storage hangar 3
belong to sections 1 and 2). This allows replacing a list of compatible routes that is hard
to build free of errors, by a simple constraint that limits the number of identical parallel
conveyors to use simultaneously in a section, at a given time period. The solution to our
model will provide which bulk product to be transported from which hangar to which
berthing position at a time, respecting this capacity constraint. Given this solution, the
allocation of conveyors and the maintenance of the model in case of infrastructural
changes in the conveyor system can be done easily a posteriori.
We also consider scheduled preventive maintenance activities to be performed at
the conveyors and berthing positions over a period of time or at a fixed date.
Maintenance activities at conveyors in the same section can overlap with each other,
while they are disjoint at a berthing position. Some berthing positions can also be
discarded to some vessels, either permanently due to some of their characteristics or, for
the next few days, due to weather conditions (tempest, equinoctial tide…). We note that
the conveyor system is the bottleneck of the port since it limits the number of vessels
that can be handled simultaneously (e.g., in Figure 5, only four out of ten vessels can be
handled simultaneously).
Handling times of vessels depend on loading equipment's productivity in the
berthing positions, and a vessel can be served by more than one loading equipment
depending on its length. Each vessel is also characterized by a number of cargo types
with different amounts to load on it. These amounts of cargo types can be expressed as
batches. Each batch is characterized by an availability date and a storage hangar. It has
to be noted that the batches to load on a single vessel can be stored in the same hangar
or different hangars. We assume that only one batch at most can leave a storage hangar
at a time and that two (or more) batches cannot be loaded at the same time on a vessel,
but they can be loaded in any order without downtime. This assumption favors the
waiting of vessels in the harbor until their continuous loading is guaranteed to minimize
their berthing time in the port.
In Figure 6, vessel 3 is a chartered vessel that is berthed in berthing position 5.
The vessel is represented by a large rectangle placed in the area of berthing position 5
starting from its berthing time with a length equal to its handling time. Small rectangles
inside the big rectangle represent the batches to load on the vessel according to a
loading sequence chosen by the optimal solution. Each small rectangle starts from the
loading start time of the batch with a length equal to its handling time. Vertical bars
represent high tide windows.
 Berthing
position p
Planning horizon
T
Loading time Θvp
p=4

p=5 Vessel v=3

p=3 Batch bv=3 Batch bv =1 Batch bv =2

Quay 1
(Hangar h=1) (Hangar h=5) (Hangar h=7)

Loading time Loading time Loading time

p=2 θbvpv

p=1

Berthing time Time t

Figure 6. Representation of a chartered vessel and the batches to load on it.

That list of specific constraints would not be completed without considering
stability issues while loading bulkers. Ship stability can be defined "as its characteristic
or tendency to return to its original state or upright state, when an external force is
applied on or removed from the ship" (Karan, 2021), involving that the center of gravity
remains in the same position. Specific rules must be followed during loading or
unloading operations (Directive 2001/96/EC), and several commercial Software treats
that issue in a real-time perspective with a granularity that is too fine for our modelling.
In addition, they do not consider conveyor constraints which must be considered when
more than one type of bulk is to load, each one into specific holds. The split of large
cargoes of the same product into several reasonably sized batches makes it possible to
define a loading sequence compatible with the use of this kind of Software. Many
predefined loading sequences compatible with the stability search can be defined.
However, the use of one of them in our model is discarded because enforcing the
vessel's sequence in the problem formulation can make it very difficult, if not
impossible, to obtain a solution respecting conveyor routing constraints. The retained
modelling approach, compatible with the stabilization of the gravity center, defines the
sequence to use as a result of the optimization. Stability is considered through
constraints inspired by project scheduling and illustrated in Figure 7. The batches are of
homogeneous composition, numbered starting from the bow towards the stern and
assigned to a hatch that can receive a unique product. The loading sequence starts from
the vessel extremities and progresses towards the vessel center. In that example,
sequencing starts with two batches (1 and 8) which can be placed in any order, before
sequencing two batches (2 and 7) placed in any order, also before sequencing two
batches (3 and 6) placed in any order, and ends by sequencing two batches (4 and 5)
also placed in any order. The retained sequence is a solution given by optimization.

Figure 7. Batch loading constraints used in modelling.

3.3.Optimization criteria:

3.3.1. Efficiency criteria:

The LAP defines some major contractual terms that are found in a maritime contract
between a shipowner and a charterer for the hire of a vessel (called charter party).
Some of these contractual terms are the following: laycan, laydays, laytime, demurrage,
and despatch. All these chartering terms are shown in Figure 8 and will be used for
expressing the objective function and some decision variables in the mathematical
model proposed in this paper.
Figure 8. Contractual vs actual parameters of vessels (adapted from Bouzekri et al.
(2021).
One of the efficiency criteria that could be applied is to find an efficient
schedule for berthing chartered vessels that maximizes the sum of the difference
between the despatch money and the demurrage charges for each vessel (i.e., minimize
the demurrage charges and maximize the despatch money) while proposing optimal
laycans for new vessels to charter considering all the characteristics and constraints
described above.

3.3.2. Effectiveness criteria:

We note that other objective functions based on physical criteria could be considered,
such as minimizing the sum of expected vessel departure times or vessel stay times.
These sums can be weighted to consider vessel priority. The effectiveness point of view
does not require any change in the model, except in the objective function's formula.

4. Model formulation:

4.1.Notation:
The sets are represented by calligraphic letters, the parameters by Greek letters or capital
Latin letters, the variables by italic letters, and the indices by italic lowercase letters. The
latter are always written as subscripts, except for the indices bv , m p and ms , that are

related to the indices v, p and s, which are always written as superscripts.

Index Description

t Index of time periods T  1,...,T .

v Index of vessels V  1,..., V with V V 1 V 2 V 3 and V  V1  V2  V3,

where:
 V 1  1,..., V1 is the set of already berthed vessels.

 V 2  V1  1,..., V1  V2  is the set of chartered vessels.

 V 3  V1 +V2  1,..., V is the set of new vessels to charter.

bv Index of batches to load on vessel v Bv  1,..., Bv  .

n bvv Batch level of batch bv (see Figure 7) used in the batch sequencing in vessel
 v loading to maintain ship stability
p Index of berthing positions P  1,..., P .

mp Index of maintenance activities to be performed at berthing position p


M p  1,..., M p . 
s Index of sections composed of identical parallel conveyors S  1,...,S .

ms Index of maintenance activities to be performed at a conveyor in section s

M s  1,..., M s  .

h Index of storage hangars H  1,...,H .

Parameter Description
Navigation channel
M Maximum number of vessels allowed to pass simultaneously through the
navigation channel.
Time decision restriction
Kt Boolean parameter that equals 1 if a decision of berthing vessels can be
taken during time period t, 0 otherwise (see Section 4.3).
Tide cycle
Ot Boolean parameter that equals 1 if time period t is within a high tide cycle, 0
otherwise.
Berthing positions
Qp Length of berthing position p.

Wp Minimum water depth of berthing position p.

ρp Productivity of berthing position p

E pp
 Boolean parameter that equals 1 if berthing positions p and p share a
berthing position, 0 otherwise (e.g., in Figure 2, berthing positions 3 and 5

share berthing position 3). When p  p, E pp   1.

Sections
Us Number of identical parallel conveyors in section s.

Fsh Boolean parameter that equals 1 if one of the conveyors belonging to the
route that links a berthing position to storage hangar h belongs to section s, 0
otherwise.
Preventive maintenance activities
m Duration of maintenance m p to be performed at berthing position p.
Rpp

m Earliest time to perform maintenance m p at berthing position p.

Rpp

m Latest time to perform maintenance m p at berthing position p.

Rpp

Rm
s
s Duration of maintenance ms to be performed at a conveyor in section s.

Rm
s
s Earliest time to perform maintenance ms at a conveyor in section s.
m Latest time to perform maintenance ms at a conveyor in section s.
Rs s
Vessels
Av Expected arrival time of chartered vessel v and earliest time a new vessel to
charter v can arrive to the port.
Jv Contractual handling time of vessel v, calculated with the contractual

productivity ρ : J v   b B φvbv / ρ, v V , p P ; φvbv is the weight

v v

of batch bv .

δv Contractual finishing time of vessel v, δv  Av  Jv 1, v V .

vp Loading time of vessel v when the latter is berthed at berthing position p,
which equals the sum of loading times of all the batches to load on this
vessel, in any order without downtime:

vp   b B θbvpv , v V , p P ; θ bvpv is the loading time of batch bv .

v v

λv Length of vessel v.

Dv Draft of vessel v when it is fully loaded.

Iv Maximum waiting time in the harbor of vessel v.

ωv Boolean parameter that equals 1 if vessel v is tide-dependent, 0 otherwise.

γv Boolean parameter that equals 1 if the handling of vessel v is restricted to

working periods, 0 otherwise.
Lv Laydays of vessel v.

ηv Contractual demurrage by hour of vessel v. ηv  0, v V 1 and

ηv  1, v V 3.
βv Contractual despatch by hour of vessel v. βv  0, v V 1 and

βv  1, v V 3. Iv  48, v
Gvp 1 if vessel v can berth at berthing position p, 0 otherwise.

Batches
b Hangars where batch bv to load on vessel v is stored
H vv

b Date of availability of batch bv to load on vessel v.

Cvv

φ vbv Weight of batch bv

n bvv Batch level of batch bv (see Figure 7) used in the batch sequencing in vessel

v loading to maintain ship stability

θbvpv Loading time of batch bv on vessel v when the latter is berthed at berthing

position p: θ bvpv  φ vbv / ρ p , v V , p  P , bv Bv .

Time framework
ψvt Boolean parameter that equals 1 if the handling of vessel v should not be
carried out during time period t, 0 otherwise (see Section 4.2).
Γvt Relative period of the absolute time period t of vessel v considering non-
working periods (see Section 4.2).
vt Absolute period of the relative time period t of vessel v considering non-
working periods (see Section 4.2).
Decision Description
variable
1 if vessel v starts berthing at berthing position p in time period t, 0
xvpt
otherwise.
b
v 1 if batch bv stored in hangar h starts loading on vessel v at berthing
yvpth
position p in time period t, 0 otherwise.
m 1 if maintenance m p starts performing at berthing position p in time period
z ptp
t, 0 otherwise.
m 1 if maintenance ms starts performing at a conveyor in section s in time
zst s
period t, 0 otherwise.
uv Delay of vessel v, which is the number of time periods exceeding its laytime,
 u v  Z  (since the planning horizon is divided into equal-sized time

periods).
wv Advance of vessel v, which is the number of time periods saved in its
laytime, wv  Z  .

Intermediary Description
variable
v b
Berthing position of vessel v in the decision variables xvpt and yvpth
v ,  vbv
respectively.
v Berthing time of vessel v.

v Finishing time of vessel v.

vbv Loading start time of batch bv .

 vbv Loading finishing time of batch bv .

4.2.Representation of time
Port operations might be unavailable at some periods for some vessels (e.g. non-
working days). If such periods coincide with the berthing period of the related vessel,
they must be considered to estimate the ending time of berthing for this vessel. This
requires adjusting index t for this vessel in the mathematical model. We use the
approach proposed by Bouzekri et al. (2021) to this end.
We define four parameters; γv to indicate if the handling of vessel v is

restricted to some working periods  γ v  1 or not  γ v  0  . The second one is, ψvt ,

which is used to indicate the non-working periods for the vessels for which γv  1. ψvt

is equal to 1 in this case and will be equal to zero for all other situations. The third
parameter Γvt is a relative time scale that keeps track of the non-working periods and

accounts for them in advancing time. It is written as follows when γv  1:

t t
Γ  t   t 1 ψ vt  , t  T ,v V and Γ Γ , t  T ,v V ,
vt ψvt  0 vt ψvt 1 vt

where t  is the first working period after t, as it is the first t   t ψ vt   0 . Obviously,

Γvt  t, t  T ,v V , when γv  0 . Finally, the fourth parameter vt records the
calendar time t for vessel v.
An example of these parameters is shown in Table 2, where each time period is
one hour long and non-working periods last two hours. A realistic assignment of two
day-long non-working periods is provided in Appendix (Table 10).
Table 2. Calculation of non-working periods parameters.
ψ vt Γ vt  vt
t / γv 0 1 0 1 0 1
1 0 0 1 1 1 1
2 0 0 2 2 2 2
3 0 1 3 3 3 5
4 0 1 4 3 4 6
5 0 0 5 3 5 7
6 0 0 6 4 6 8
7 0 0 7 5 7 9
8 0 0 8 6 8 12
9 0 0 9 7 9 13
10 0 1 10 8 10 14
11 0 1 11 8 11 15
12 0 0 12 8 12 16
13 0 0 13 9 13 19
14 0 0 14 10 14 20
15 0 0 15 11 15 21
16 0 0 16 12 16 22
17 0 1 17 13 17 23
18 0 1 18 13 18 0
19 0 0 19 13 19 0
20 0 0 20 14 20 0
21 0 0 21 15 21 0
22 0 0 22 16 22 0
23 0 0 23 17 23 0

To consider non-working periods in the calculation of the contractual finishing

time of vessel v, v  Av  Jv 1 becomes  v  
v vA  Jv 1
v 
. Indeed,  vA v gives

the relative period of the absolute expected arrival time of vessel v considering non-
working periods, then 

v  vA  J v 1
v  gives the absolute period of the relative

contractual finishing time of vessel v also considering non-working periods. If vessel v


is not restricted to working periods,

v vA  Jv 1
v   Av  Jv  1 , therefore

v  
  is used in both cases.
v vA  Jv 1
v
 4.3.Decision time-interval:
To reduce the computational complexity and consider the increasing uncertainty of
inputs as the length of the planning horizon increases, we follow the approach proposed
by Bouzekri et al. (2021), which modulates decision time-interval through the planning
horizon. So, we define a Boolean parameter K t , that equals 1 if vessels can berth
during time period t (without considering other constraints). Thanks to this parameter,
we are able to restrict berthing decision periods inside the planning horizon and hence
change the decision time interval.
The user of the model is free to define the values of K t . For example, during

the first week, chartered vessels can berth every hour « 1 », hence Kt  1, t ; during the

second week, every four hours « 0 – 0 – 0 – 1», hence K t  1,  t |  t mod 4  0 and

Kt  0 otherwise; during the third week, every eight hours « 0 – 0 – 0 – 0 – 0 – 0 – 0 –

1», hence K t  1,  t |  t mod 8  0 and Kt  0 otherwise. New vessels to charter can be

planned during the second and third week, providing for them an estimated position in
the schedule. Then, as we advance in the planning horizon, the schedule is refined:
some chartered vessels V 2  will become berthed vessels V 1  , and some new vessels

to charter V 3  will become chartered vessels V 2  , and hence their laydays will be

replaced by an expected arrival time. The decisions related to the loading of batches are
constrained by the decisions related to the berthing of vessels, as they must be done
during the vessel's stay, which starts from the berthing time but can be taken at any
period without considering the parameter K t .

Modulating time intervals in this manner helps integrating short-term decisions

(BAP) and medium-term decisions (LAP) in a single model. As the time approaches to
present, the decisions are taken in a finer granularity (every hour), while for decisions
that concern the planning in a few weeks from now, a rough decision is taken (every 8
hours). Besides facilitating the integration of LAP and BAP, this approach also helps
control the number of variables (i.e., the number of variables is lower for medium-term
decisions).
 4.4.Predicates:
A mathematical program is made of a set of variables and a set of constraints made of a
linear or non-linear combination of these variables, one of them being an objective
function to optimize. The variables' validity domain is usually restrained by constraints,
each defined for a set of variables through a universal quantifier. The variables' validity
domain may be narrowed by using an Algebraic Modelling Language (AML, available
in some software like Xpress (used here) or GAMS…; Fourer (2013)), which rests on
the separation of a generic description of the model, and data to use; after what, an
instance of the model combining the generic model and data, which can be submitted to
a solver. AML allows the usage of predicates to drive the creation of an instance of the
problem. A predicate is a logical statement that returns either a value of "True" or
"False", based on the parameter values used in the statement, which in turn binds the
existence of a variable, depending on the parameters' values. Predicates can be used to
restrain
˗ the number of expanded constraints in relation using a universal quantifier.
˗ the validity domain of some variable without using a constraint, decreasing the
number of constraints in a model; for example, the use of the predicate
D v  W p (which enforces the draft of vessel v to not exceed the water depth of

berthing position) in the definition of the validity domain of variable xvpt avoids

creating the constraint D v  xvpt  W p , p, t . The use of this kind of predicates

presents the advantage of preventing the introduction of additional constraints in

modelling a complex problem. It also avoids unnecessary calculations in the
optimization search, as the predicate used in the problem expansion guarantees
the respect of that (unintroduced) constraint.
The extensive use of predicates in the proposed model acts like a pre-treatment
based on the problem data reducing the number of binary variables and constraints.
Consequently, problems of practical sizes can be solved in a reasonable time using off-
the-shelf commercial Software
We will use these logical statements to describe the validity domain of decision
variables. In our model, a decision variable exists only when the associated set of
predicates returns "True". In this section, we present how predicates are implemented in
our mathematical model.
 The decision variable xvpt determines for each chartered vessel  v V 2  , the

berthing time t and berthing position p. Each already berthed vessel  v V 1  has a

residual handling time and a predetermined berthing position. For simplicity, we assume
the berthing time t  1 for these latter vessels.
b
v
The decision variable yvpth determines for each chartered vessel  v V 2 
berthed in berthing position p, the loading start time t of batch bv , stored in hanger h.

The first batch to load on each already berthed vessel  v V 1  has a residual handling

time with a loading start time assumed at t  1.

For each new vessel to charter  v V 3  , the berthing position will be reserved

from its latest berthing time decreased by its laydays, t  Lv 1 , until its latest finish

date, 
 
v  vt vp 1
(Figure 9). Similarly, the conveyors used to transport each batch

bv from storage hangar h to berthing position p will be reserved from the latest loading

start time of the batch decreased by the vessel's laydays, t  Lv 1 , until the latest finish

date of the batch, 

  , depending on the loading sequence chosen by the
v  vt θbvpv 1

optimal solution. This assures that, any time during their laydays, new vessels to charter
can be berthed at the reserved berthing position, and thus all the batches can be loaded,
in any order without downtime, using the reserved conveyors.

Figure 9. Reservation of port resources for new vessels to charter.

The existence of the decision variable xvpt is subject to seven conditions:

(1) Vessel v must be able to berth at berthing position p: G vp  1 .

 (2) The length of vessel v must not exceed the length of berthing position p:
λv  Q p .

(3) The draft of vessel v must not exceed the water depth of berthing position p:
Dv  W p .

(4) Vessel v can berth only after its expected arrival time without exceeding its
maximum waiting time in the harbor: Av  t  Av  Iv . To allow new vessels to

charter  v V 3  to berth at their first layday, t is replaced by t  Lv 1 , then

condition 4 becomes Av  t  Lv  1  Av  Iv . Already berthed and chartered

vessels have fictitious laydays equal to one hour  L v  1 since they have fixed

expected arrival times, and hence they are not concerned by the decision of
fixing laycans as in the case of new vessels to charter. Hence, the new condition
is valid for all types of vessels. The same applies to the following conditions.
This modelling approach allows for the merging of LAP and BAP decisions.

(5) Vessel v can berth only during time periods where a decision of berthing vessels
can be taken: K  t  L 1 =1 .
v

(6) If the handling of vessel v is restricted to working periods  γ v  1 , it can enter

the port only during working periods: ψ v  t  L 1  0 .

v

(7) If vessel v is tide-dependent (ωv  1) , it can leave the port at the time period

t   v    1
 vt vp  if the latter is within a high tide  Ot  1 :
(1-ωv )+ωv  O   1.
 
v vt vp 1

v b
Similarly, the existence of the decision variable yvpth is subject to seven

conditions:
(1) Conditions 1, 2 and 3 of the existence of the decision variable xvpt :

G vp  1  λ v  Q p  D v  W p .

(2) Batch bv can be loaded on vessel v between the expected arrival time of this
vessel and its finishing time as it reaches its maximum waiting time in the
 harbor, minus the loading time of this batch:
A v  t  Lv  1  

v v A  I  vp θbvpv
v v 
(3) Batch bv can be loaded on vessel v only after its date of availability:

t  L v  1  C bvv .

(4) If the handling of vessel v is restricted to working periods  γ v  1 , batches can

start loading only during working periods: ψ v  t  L 1  0 .

v

(5) Batch bv is loaded on vessel v from its storage hangar h: h  H bvv .

m
Regarding the preventive maintenance activities, the decision variables z ptp and

m
zst s determine the starting time t of performing maintenance m p at berthing position p

and maintenance ms at a conveyor in section s, respectively. The existence of these

two decision variables is subject to only one condition, each which states that each
m m
maintenance must be performed between its earliest and latest time: R p p  t  R p p

ms
for berthing positions and R m
s  t  Rs
s for sections. When the maintenance has a

fixed date, the indices of the decision variables are predetermined.

To facilitate the readability of the mathematical model, we represent each
predicate by a simplified notation given in Table 3.

Table 3. Notation of predicates.

Predicate Notation
G vp  1  λ v  Q p  D v  W p Pvp
A v  t  Lv  1  Av  Iv  K t  L 1 =1  ψv t  L 1  0 
v v
Pvpt
(1-ωv )+ωv  O  1

v vt vp 1 
A v  t  Lv  1    t  Lv  1  Cbvv  ψv t L 1  0

v v A  I  vp θbvpv
v v  v
bv
Pvpt

h  H bvv b
Pvhv
m m
Rpp t  Rpp mp
P pt
 ms
Rm
s  t  Rs
s m
Pst s
v , b
The logical conditions of the existence of the decision variables xvpt , yvpth

m ms
z ptp and zst are the following ones. For instance, the variable xvpt exists only when

the predicates Pvp and Pvpt are both "True".

xvpt  0,1 , v V , p  P Pvp , t  T Pvpt

bv bv b
yvpth  0,1 , v V , bv Bv , p P Pvp , t  T Pvpt , h  H Pvhv

m mp
z ptp  0,1 , p P , m p  M p , t  T Ppt

m m
zst s  0,1 , s  S , ms  Ms , t  T Pst s

4.5.Mathematical model:

First, we define the intermediary variables v and  vbv , which give for each vessel v the
b
v , respectively.
berthing position in the decision variables xvpt and yvpth

v   pP P  tT P p  xvpt , v V

vp vpt

 vbv   pP P  tT Pbv  hH Pbv p  yvpth

b
v , v V , b B
v v
vp vpt vh

 v   pP P  tT P t  xvpt , v V

vp vpt

vbv   pP P tT Pbv  hH Pbv t  yvpth

b
v , v V , b B
v v
vp vpt vh

Similarly, we define for each vessel v the berthing and finishing times  v and

 v , by replacing b  xvpt in v by t  xvpt and 

   xvpt ,
v v t L 1 vp 1
v

respectively. Likewise, we define for each batch bv to load on vessel v, the loading start
 b b
and finishing times vbv and  vbv by replacing p  yvpth
v in  bv , respectively, by t  y v
v vpth

b
and   yvpth
 
v .
v v t L 1 θbvpv 1
v

The mathematical model can be formulated as follows:

Max  vV β v  wv  ηv  uv  (1) or Min 
vV
 v ( 1 )

 pP Pvp tT x  1, v V

Pvpt vpt
(2)

 pP Pvp tT hH

b
bv
Pvpt
bv y v
Pvh vpth  1, v V , bv Bv (3)

m
tT m z p  1, p P , m  M
Ppt p pt p p (4)

m
tT m z s  1, s  S , ms  Ms
Pst s st
(5)

 v   vbv ,  v V ,  bv  Bv (6)

  bv   , v V , b B n bv  0
 v v v v v
 (7)
 vbv  vbv , v V , bv Bv n bvv  0, bv Bv n bvv  n bvv  1

 vbv   v ,  v V ,  bv  Bv (8)

b
bvBv  pP Pvp tT bv  +θbv 1Tt t 
Pvpt  vt vp
hH b y v  1,
Pvhv vpth
v vt +θvp
bv 
1 (9)
 
t  T ,v V

 vV  b B P  pP bv
Pvp  tT bv
Pvpt bv
  vt +θ vp 1T t  L v 1t 
bv
yvpt h  1,
v vt +θvp
v v vh bv 
1 (10)
 

t  T ,h  H

 vV  b B  pP Pvp  tT b

Pvptv  vt  +θbvpv 1T t  Lv 1t 
v vt  +θvp
v v bv 
1
 
(11)
 hH Fsh 1 Pvhv
b
bv
ynpt h  U s   m M
s s
 tT t R m s ms
s 1t t  Pst
zstms , t  T ,s  S

 vV  pP E pp 1 Pvp  tT Pvpt   vt  vp 1T t  L v 1t  v
 vt vp 1
xvpt  

m
(12)
1   pP E pp 1 m  tT t  R pp 1t t  Pptp
m m z ptp , t  T ,p P
pM p
 
vV 2 V 3  pP Pvp t T Pvpt t t  Lv 1 xvpt   
 
  (13)
  pP Pvp t T Pvpt vt vp 0t v   1 xvpt   M, t  T
 vV  vt vp  

uv   v  δv , v V (14)

wv  δv   v , v V (15)

uv  wv   v  δv , v V (16)

uv , wv  0, v V (17)

Objective function (1) is based on an efficiency criteria that maximizes the

difference between the despatch money and the demurrage charges of each vessel v
while objective function ( 1 ) is based on an effectiveness criteria that minimizes the
finishing time (departure time) of each vessel v. Equation (2) ensures that each vessel v
starts berthing at a unique berthing position p, and in a unique time period t. Equation
(3) ensures that each batch bv starts loading in a unique vessel v at a unique berthing

position p, in a unique time period t, and is stored in a unique hangar h. Equation (4)
ensures that maintenance m p to be performed at a berthing position p has a unique start

time. Similarly, equation (5) ensures that maintenance ms to be performed at a

conveyor in section s has a unique start time. Equation (6) ensures that berthing position
b
v . Equations (7) ensures that the
p is the same in both decision variables xvtp and yvtph

loading of batch bv can only begin once vessel v has been berthed and all the batches

that must precede bv in the loading sequence have been loaded. Equation (8) ensures

that each vessel v can only leave the port when all batches have been loaded. Equation
(9) ensures that at most one batch can be loaded at the same time on each vessel v.
Equation (10) ensures that only one batch at most can leave at a time from each storage
hangar h. Equation (11) limits the number of identical parallel conveyors used
simultaneously in a section s during the loading time of each batch due to the limit
or/and the maintenance of conveyors. Equation (12) avoids the overlapping of vessels in
each berthing position p, the simultaneous use of berthing positions that share a space of
the quay since the berth layout of each quay is hybrid and the use of berthing positions
where maintenance activities are performed (e.g., in Figure 2, berthing positions 3 and 5
share berthing position 3, consequently, they cannot be used simultaneously. Moreover,
if maintenance is performed at berthing position 3, it will also be formed at berthing
position 5. The opposite is also true.). Equation (13) limits the number of incoming and
outgoing vessels to pass simultaneously through the navigation channel. Equations (14-
17) determine the delay and the advance of each vessel.

5. Case study:
In this section, we describe one test instance of the case study and report the
computational results. An example schedule obtained using the model is also given to
illustrate a typical output (from another test instance). The formulations are written on
Mosel and implemented in Xpress IVE Version 1.24.24, with 64 bits. All the tests are
run on a server with an Intel® Xeon® Gold 6138 processor (8 cores) of 2.00 GHz
processing speed and 32 GB of memory using the Xpress Optimizer Version 33.01.05
with the default options.

5.1.Test instances
Six test instances, based on data obtained from OCP group, a world leader in the
phosphate industry, operating six quays in the bulk port of Jorf Lasfar in Morocco,
recognized as the largest bulk port in Africa. Each set corresponds to the actual
observation of vessels' expected arrivals during four weeks of different months
(expedition range of 250,000t to 660,000t by 13 to 24 boats), and all their
characteristics. The first set is used as a case study; the other ones will be presented in
an extended version which is about to be submitted to a scientific journal. As the port
management problem depends on the hangar configuration, conveyor network and
production, the classical robustness test approach based on factors under control is
problematic and must lay on arbitrary assumptions on upstream characteristics. The
used samples share the same configuration, and the observed dispersion provides a kind
of robustness proof. For all instances:
 We assume a 4-week planning horizon discretized into 1-hour intervals, hence
T=672. We vary the decision time interval inside this planning horizon to handle
term and medium-term decisions as explained in Section 4.3. For the short-term
planning (e.g. during the first week in our instances) the accuracy of the planning is
set to every hour. The decision time interval is set to four, eight and twelve hours for
weeks two, three and four, respectively. Characteristics related to time periods
 (decision time-intervals, high-tide cycles and non-working periods) can be found in
Table 10.
 We consider a navigation channel in which the maximum number of vessels
allowed to pass simultaneously is limited to three vessels. We also consider two
quays with hybrid berth layout, partitioned into five berthing positions each. Table 4
gives respectively, for each berthing position, the length (m), the minimum water
depth (m), the productivity (t/h), the incompatibilities and the number of
maintenance activities with the duration (h), the earliest and latest time (h) to
perform each one.

Table 4. Characteristics of berthing positions.

 m m m
Productivity E pp Rpp Rpp Rpp
p Qp Wp ρp Mp
p'=1 p'=2 p'=3 p'=4 p'=5 p'=6 p'=7 p'=8 p'=9 p'=10 m p =1 m p =2 m p =1 m p =2 m p =1 m p =2
1 180 13.5 2000 1 0 0 0 0 0 0 0 0 0 2 12 10 24 328 48 328
2 255 14.5 2000 0 1 0 0 0 0 0 0 0 0 1 7 0 405 0 450 0
Quay 1

3 150 15.6 1000 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0

4 150 15.6 1000 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0
5 (3 ∪ 4) 300 15.6 2000 0 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0
6 180 13.5 2000 0 0 0 0 0 1 0 0 0 0 2 29 14 266 320 266 320
7 235 14.5 2000 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0
Quay 2

8 125 14.5 2000 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0

9 125 15.6 2000 0 0 0 0 0 0 0 0 1 1 2 10 6 10 360 53 382
10 (8 ∪ 9) 250 14.5 4000 0 0 0 0 0 0 0 1 1 1 1 23 0 300 0 320 0

 We consider nine storage hangars which are linked to all the berthing positions via a
conveyor system (Figure 5). The latter is divided into two sections composed of
different numbers of identical parallel conveyors. Table 5 gives for each section of
the port conveyor system the number of identical parallel conveyors, the conveyors
that belong to it and the number of maintenance activities with the duration ( R m
s ),
s

m
the earliest ( R m s
s ) and latest time ( R s ) to perform each one.
s

Table 5. Characteristics of sections of the port conveyor system.

m
Fsh Rms
s
Rms
s
Rs s
s Us M s
h =1 h =2 h =3 h =4 h =5 h =6 h =7 h =8 h =9 m s =1 m s =2 m s =3 m s =1 m s =2 m s =3 m s =1 m s =2 m s =3
1 1 1 1 1 0 0 0 0 0 3 2 12 8 0 282 284 0 282 286 0
2 1 1 1 1 1 1 1 1 1 4 3 6 10 9 213 290 362 213 290 390

 The case study data on vessels and cargos are given in tables 6 and 7. In the
studied sets, the length of vessels varies between 100 and 300 meters, their draft
varies between 5 and 15 meters; vessels having a draft over 14 meters are tide-
dependent while leaving the port; the number of different fertilizers references to
load on each vessel varies between one and three; the tonnage of fertilizers to
load in a boat varies between 2.000 and 55.000 tones; 10% of vessels have a
 SSHEX clause; the laydays are set arbitrarily at 48 hours for new vessels to
charter, while already berthed and chartered vessels have fictitious laydays equal
to one hour.
 For chartered vessels, the contractual demurrage is chosen randomly from a
Uniform distribution between 50 and 150. The contractual dispatch is assumed
half the demurrage. For new vessels to charter, we assume negligible demurrage
and dispatch per hour. Note that these are fictitious values and are only used in
order not to impact the economic results of already chartered vessels. Finally, for
already berthed vessels, hourly demurrage and dispatch rates are assumed zero
since no decisions need to be made for this group of vessels; they are already
berthed.
 The maximum waiting time in the harbor is set arbitrarily at 72 hours for all the
chartered vessels and one week for all the new vessels to charter. The latter have
high maximum waiting times in the harbor in order not to affect the economic
results of chartered vessels.
 The availability date Cbvv of batch bv to load on vessel v is lower or equal to the

date of the expected arrival time A v of vessel v.

Thanks to these conventions, all the vessels are dealt with together since there is
no need to define specific constraints for each type of vessels (already berthed vessels,
chartered vessels, and new vessels to charter).

5.2.Case study
In this section, we present a typical schedule that the port manager can obtain by using
the proposed model.
It uses the characteristics of time periods, berthing positions, conveyor sections
and vessels presented in the previous section. Table 6 gives the characteristics of 15
chartered vessels and 2 new vessels to charter. Table 7 gives the characteristics of
batches to load, with their storage location and the concerned fertilizer (which is
interesting to know to see the solution impact on conveyor use).
Table 6. Characteristics of vessels.
Expected Contractual
Vessel Vessel Tide Working Demurrage Despatch Nvp vessel v can berth at position p
vessel Arrival Laydays handling
Length Draft Dependent Restriction rate rate p =
Time time
v Av lv Dv wv gv Lv Jv hv bv 1 2 3 4 5 6 7 8 9 10
1 1 194 8 0 0 1 9 101 50.5 1 1 1 1 0 1 1 1 1 0
2 12 197 9 0 0 1 17 59 29.5 1 1 1 1 0 1 1 1 1 0
3 22 188 8 1 0 1 17 111 55.5 1 1 1 1 0 1 1 1 1 0
4 55 190 10 0 1 1 11 105 52.5 1 1 1 1 0 1 1 1 1 0
5 107 182 9 0 0 1 14 97 48.5 1 1 1 1 0 1 1 1 1 0
6 171 128 7 0 0 1 3 101 50.5 1 1 1 1 0 1 1 1 1 0
7 174 129 5 1 0 1 2 64 32.0 1 1 1 1 0 1 1 1 1 0
8 190 191 9 0 0 1 11 94 47.0 1 1 1 1 0 1 1 1 1 0
9 203 131 5 0 1 1 2 61 30.5 1 1 1 1 0 1 1 1 1 0
10 299 201 11 0 0 1 25 143 71.5 1 1 1 1 0 1 1 1 1 1
11 440 207 11 0 0 1 23 111 55.5 1 1 1 1 0 1 1 1 1 1
12 480 158 9 0 1 1 20 130 65 1 1 1 1 0 1 1 1 1 0
13 514 231 13 0 0 1 25 147 73.5 1 1 1 1 0 1 1 1 1 1
14 541 112 5 0 0 1 2 131 65.5 1 1 1 1 0 1 1 1 1 0
15 400 122 7 1 0 48 2 93 46.5 1 1 1 1 0 1 1 1 1 0
16 430 174 10 0 0 48 11 56 28 1 1 1 1 0 1 1 1 1 0
Maximum waiting time Iv  48, v
 Table 7 characteristics of batches

batch sequencing

batch sequencing

batch sequencing
Batch level used

Batch level used

Batch level used

Storage hangar

Storage hangar

Storage hangar
in the loading

in the loading

in the loading
Date of batch

Date of batch

Date of batch
availability

availability

availability
of batch bv

of batch bv

of batch bv
Quantity

Quantity

Quantity
Vessel

Vessel

Vessel
Batch

Batch

Batch
Fertilizer

Fertilizer

Fertilizer
v bv φvbv b
Cvv Hbvv nb
v
v
v bv φ bv v b
Cvv
b
Hvv nb
v
v
v bv φ bv v b
Cvv Hbvv nb
v
v

1 2750 1 1 0 7 2750 1 3 0 1 3967 393 5 0

6 6 2
2 2750 1 1 1 8 2750 22 2 0 2 3967 430 4 0
1 3 3667 1 4 2 7 1 2 3245 28 3 0 3 3967 466 5 0
3
4 3 3667 1 8 1 1 3767 1 12 0 4 3967 503 5 1
5 3667 1 9 0 2 9 3767 1 12 1 5 3967 540 8 2
12
1 4125 1 9 0 3 3767 1 11 2 6 3967 576 6 3
8
2 4125 1 8 1 4 3413 1 1 2 7 3828 600 12 3
3 4125 1 4 2 5 6 3413 1 3 1 8 3828 600 10 3
2
4 4125 1 5 3 6 3413 15 2 0 9 3828 600 10 2
2 11
5 4125 1 4 3 9 1 3 3500 299 5 0 10 3828 600 12 1
6 4125 1 9 2 1 4415 341 2 0 1 4400 326 8 0
7 4125 1 9 1 2 4415 369 1 1 2 4400 367 8 0
8 4125 1 9 0 3 4415 395 1 2 3 4400 407 7 1
1 4450 1 7 0 4 4415 204 1 3 4 4400 448 4 2
11
2 4400 1 9 1 5 4415 232 3 4 5 4400 489 7 3
6
3 3867 1 1 2 10 6 4415 258 2 4 13 6 1 4400 529 9 4
4 6 3867 1 3 3 7 4415 122 2 4 7 4400 570 6 4
3
5 3867 1 2 3 8 4415 150 3 3 8 4400 610 4 4
6 3844 1 8 2 9 4415 176 2 2 9 4400 204 8 3
7 3 3844 1 4 1 10 4415 286 2 1 10 4400 245 5 2
8 3844 1 6 0 11 8 4450 313 3 0 11 4400 285 9 1
1 2750 1 10 0 1 4500 372 4 0 1 4400 537 4 0
3
2 2750 1 12 1 2 4500 413 8 1 2 4400 578 4 0
3 4074 1 8 2 3 4500 124 9 2 14 3 8 4400 618 7 0
4
4 4074 1 8 2 4 4500 166 9 3 4 4400 659 8 1
6
5 4074 1 6 1 5 4500 206 7 4 5 4400 699 6 2
11 8
6 4074 1 8 0 6 4500 248 8 4 15 1 1 3300 548 4 1
1 4375 1 9 0 7 4500 290 7 3 16 1 1 2350 637 2 0
2 4375 31 5 1 8 4500 330 9 2
3 4375 1 5 2 9 4500 455 5 1
5
4
2
4375 31 7 2 10 4500 496 4 0 Cb
v  Av ,v V ,bv Bv
v
5 4375 1 4 1
6 4375 31 4 0

Figure 9 shows an example of Gantt chart of vessel and batch schedule, drawn
from the solution of another dataset. At its bottom, there is a timeline in hours and a
one-week time frame (from Friday of week 2 to Monday of week 3) with high tide
hours. The decision time-intervals are highlighted every four hours for week 2 and
every eight hours for week 3. All constraints are respected by the solution given by the
LBAP model for the case study of Figure 9.
 All vessels are berthed at restricted time periods for which a decision can be
made and do not occupy berthing positions where maintenance is performed.
 The batches of each vessel are loaded in any order without downtime after their
date of availability, and only one batch at most leaves at a time a storage hangar.
  Vessels 2 and 3 have high drafts and thereby are tide-dependent, so they occupy
berthing sections with high water depth and leave the port during high tides.
 The handling of vessel 3 stops during non-working hours because it has a
SSHEX clause.
 The handling time of vessel 2 is shorter than its contractual handling time since
it occupies a berthing position with high productive loading equipment.
 The number of vessels passing through the navigation channel and used parallel
conveyors per section do not exceed their limits. The number of allowed parallel
conveyors per section can decrease due to maintenance.
 The optimal laycan proposed for the new vessel to charter 7 is [276, 323]. This
vessel can berth at any time period during its laydays. We note that, a precise
berthing time will be assigned to this vessel as its status changes from new
vessel to chartered one and as the time progresses from week two towards week
one.
 Table 8 details the vessel berthing schedule (time and location). Table 9 details
the batch loading schedule. Figure 10 gives the batch loading schedule for vessel
3 to ensure ship stability; it can be noticed that the diversity of fertilizers to load
implies switching fertilizers in conveyor transportation. Table 10 details the
maintenance schedule of quay positions and conveyor sections, which is
compatible with the berthing schedule.
Thanks to the integration of the LAP and the BAP, the port managers can
propose laycans for the new vessels to charter considering the allocation of berthing
positions to already chartered vessels and conveyors to batches, thereby avoiding the
payment of demurrage charges, and knowing when to accept or refuse a new vessel to
charter.
Table 8. Vessel berthing schedule
Position Vessel START END Position Vessel START END
2 3 22 39 7 1 1 10
2 4 55 70 7 2 12 35
2 5 107 124 7 6 176 179

Quay 2
2 7 176 177 10 10 424 445

Quay 1
2 8 212 223 6 12 624 643
1 9 324 325 7 14 744 758
2 11 564 593 7 16 672 673
2 13 636 668
1 15 576 577

Table 9. Batch loading schedule

Vessel Batch Vessel Batch Vessel Batch
Start End Start End Start End
v bv v bv v bv

1 24 26 1 636 638 1 424 425

2 27 29 2 642 644 2 430 431
3 34 35 3 651 653 3 434 435
4 36 37 4 654 656 4 438 439
3
5 38 39 5 663 665 5 442 443
6 32 33 13 6 666 668 10 6 444 445
Quay 1

7 30 31 7 660 662 7 440 441

8 22 23 8 657 659 8 436 437
1 55 56 9 648 650 9 432 433
2 63 64 10 645 647 10 428 429
3 68 70 11 639 641 11 426 427
4
4 65 67 15 to [576, [624, 1 624 625
1
5 60 62 charter 623] 577] 2 628 629
Quay 2

6 57 59 1 3 4 3 632 633
1 110 112 2 7 8 4 636 637
1
2 113 115 3 9 10 5 640 641
berthed 12
3 119 121 4 5 6 6 642 643
5
4 122 124 5 1 2 7 638 639
Quay 1

5 116 118 1 12 14 8 634 635

6 107 109 2 18 20 9 630 631
Quay 2

7 1 176 177 3 27 29 10 626 627

1 214 215 4 30 32 1 744 746
2
2 218 219 5 33 35 2 750 752
3 222 223 6 24 26 14 3 756 758
8
4 220 221 7 21 23 4 753 755
5 216 217 8 15 17 5 747 749
6 212 213 1 178 179 16 to [672, [673,
6 1 719] 720]
9 1 324 325 2 176 177 charter
1 567 569
2 570 572
3 579 581
4 585 587
5 591 593
11
6 588 590
7 582 584
8 576 578
9 573 575
10 564 566
Figure 9. Example of Gantt chart of vessel berthing and loading.
OPTIMAL LOADING
BATCH LOADING CONSTRAINTS FERTILIZER
SEQUENCE DATE
1 1 11 2 24
2 2 11 3 27
3 3 6 5 32
4 4 6 7 107
5 5 6 8 110
6 6 3 6 45
7 7 3 4 30
8 8 3 1 22
Fertilizer sequence in the conveyor system:: 3 - 11 – 11 –3 – 6 – 3 – 6 - 6
Figure 10. Batch loading schedule for Vessel 3.

Table 10. Maintenance scheduling.

Maintenance Maintenance
Quay
Start End Conveyors
Position Start End
Section
36 47
Quay1

1
328 337 282 293
1
2 417 423 284 291
266 294 213 218
6
320 333 2 290 299
Quay 2

365 374 374 382

9
20 25
10 305 327
 This instance and the five other ones are solved to optimality in computation
times not exceeding one hour. Hence, we can assume that the integer linear
programming model proposed for the LBAP can easily be used in bulk ports where such
decisions need to be made frequently, with only commercial Software. Thus,
developing a heuristic for the problem is not necessary.

6. Conclusions:
In this paper, we integrate the Laycan Allocation Problem and the dynamic hybrid Berth
Allocation Problem in the context of tidal bulk ports with multiple quays and a
conveyor system between storage hangars and berthing positions. While laycans
concern only vessels for export, a symmetric approach can be applied for berthing
decisions in the context of import ports. Our research is motivated by the bulk port of
Jorf Lasfar, but it is also valid for any other bulk port. A new integer linear
programming model is proposed to solve this integrated problem. The latter integrates
two problems with different decision levels (tactical and operational) thanks to the
modulation of the time-interval between decisions and the introduction of fictitious
laydays for already berthed and chartered vessels.
Several characteristics are addressed simultaneously in the definition of the
LBAP to make it closer to reality, such as the multiplicity of quays, navigation channel
restrictions, conveyor routing constraints with preventive maintenance activities, the
variation of water depth, vessel tide-dependency, the productivity of bulk-handling
cranes, the multiplicity of cargo types on the same vessel, charter party clauses and non-
working periods. Instead of expressing these characteristics by a set of constraints in the
model, we used predicates to formulate them. These characteristics were already present
in a previous model (Bouzekri et al., 2021), and this model also includes two new
characteristics that must be considered for ports that deal with a variety of bulk to load:
conveyor constraints modelled in a more general way and constraints of boat stability
during loading. This approach is quite handy predicates are easy to modify in the model.
Furthermore, they reduce the number of variables and constraints in the model and
improve the computational performance. Moreover, the port conveyor system is
modeled in a new way that does not list each route of conveyors between storage
hangars and berthing positions, which makes easier the formulation of the problem.
This model assumes that the cargo to load on vessels is always available in the
hangars; this assumption is verified in the used data set, but, in practice, it depends on
the upstream supply chain that can work in a pull mode (production-to-stock) or a push
mode (production-to-order) and on the variety of products to manage as the
management difficulty increases with the product variety. An extension to this study
could be searching for an integrated approach of the port, hangars and production
management from a supply chain perspective.

References

Al-Hammadi, Jasem, and Ali Diabat. 2017. "An Integrated Berth Allocation and Yard
Assignment Problem for Bulk Ports: Formulation and Case Study." RAIRO -
Operations Research 51 (1): 267–84. https://doi.org/10.1051/ro/2015048

Barros, Victor Hugo, Tarcísio Souza Costa, Alexandre C.M. Oliveira, and Luiz A.N.
Lorena. 2011. "Model and Heuristic for Berth Allocation in Tidal Bulk Ports
with Stock Level Constraints." Computers & Industrial Engineering 60 (4):
606–13. https://doi.org/10.1016/j.cie.2010.12.018

Bierwirth, Christian, and Frank Meisel. 2010. "A Survey of Berth Allocation and Quay
Crane Scheduling Problems in Container Terminals." European Journal of
Operational Research 202 (3): 615–27.
https://doi.org/10.1016/j.ejor.2009.05.031

Bierwirth, Christian, and Frank Meisel. 2015. "A Follow-up Survey of Berth Allocation
and Quay Crane Scheduling Problems in Container Terminals." European
Journal of Operational Research 244 (3): 675–89.
https://doi.org/10.1016/j.ejor.2014.12.030

Bouzekri, Hamza, Gülgün Alpan, and Vincent Giard. 2019. "Integrated Laycan and
Berth Allocation Problem." In 2019 International Conference on Industrial
Engineering and Systems Management (IESM), 1–6.
https://ieeexplore.ieee.org/document/8948110/

Bouzekri, Hamza, Gülgün Alpan, and Vincent Giard. 2021. "Integrated Laycan and
Berth Allocation and Time-Invariant Quay Crane Assignment Problem in Tidal
Ports with Multiple Quays." European Journal of Operational Research 293 (3):
892–909. https://doi.org/10.1016/j.ejor.2020.12.056

Burdett, Robert, Corry, Paul , Eustace, Colin and Smith Simon, "Scheduling pre-
emptible tasks with flexible resourcing options and auxiliary resource
 requirements". Computers & Industrial Engineering, Volume 151, 2021,
106939, https://doi.org/10.1016/j.cie.2020.106939

Cheimanoff, Nicolas, Frédéric Fontane, Mohamed Nour Kitri, and Nikolay Tchernev.
2021. "A Reduced VNS Based Approach for the Dynamic Continuous Berth
Allocation Problem in Bulk Terminals with Tidal Constraints." Expert Systems
with Applications 168 (April): 114215.
https://doi.org/10.1016/j.eswa.2020.114215

Corry, Paul, and Christian Bierwirth. 2019. "The Berth Allocation Problem with
Channel Restrictions." Transportation Science, April, trsc.2018.0865.
https://doi.org/10.1287/trsc.2018.0865

Directive 2001/96/EC of the European Parliament and of the Council of 4 December

2001 establishing harmonized requirements and procedures for the safe loading
and unloading of bulk carriers. https://eur-lex.europa.eu/eli/dir/2001/96/oj

Ernst, Andreas T., Ceyda Oğuz, Gaurav Singh, and Gita Taherkhani. 2017.
"Mathematical Models for the Berth Allocation Problem in Dry Bulk
Terminals." Journal of Scheduling 20 (5): 459–73.
https://doi.org/10.1007/s10951-017-0510-8

Fourer R. (2013) Algebraic Modeling Languages for Optimization. In: Gass S.I., Fu
M.C. (eds) Encyclopedia of Operations Research and Management Science.
Springer, Boston, MA. https://doi.org/10.1007/978-1-4419-1153-7_25

Karan, Chopra, Understanding Intact & Damage Stability of Ships, in Naval

Architecture, Marine Insight, 2021, https://www.marineinsight.com/naval-
architecture/understanding-intact-damage-stability-of-ships/

Krimi, Issam, Rachid Benmansour, Abdessamad Ait El Cadi, Laurent Deshayes, David
Duvivier, and Nizar Elhachemi. 2019. "A Rolling Horizon Approach for the
Integrated Multi-Quays Berth Allocation and Crane Assignment Problem for
Bulk Ports." International Journal of Industrial Engineering Computations 10
(4), 577–91. https://doi.org/10.5267/j.ijiec.2019.4.003

Krimi, Issam, Raca Todosijević, Rachid Benmansour, Mustapha Ratli, Abdessamad Ait
El Cadi, and Afaf Aloullal. 2020. "Modelling and Solving the Multi-Quays
Berth Allocation and Crane Assignment Problem with Availability Constraints."
 Journal of Global Optimization 78 (2): 349–73. https://doi.org/10.1007/s10898-
020-00884-1

León, Alan Dávila de, Eduardo Lalla-Ruiz, Belén Melián-Batista, and J. Marcos
Moreno-Vega. 2017. "A Machine Learning-Based System for Berth Scheduling
at Bulk Terminals." Expert Systems with Applications 87 (November): 170–82.
https://doi.org/10.1016/j.eswa.2017.06.010

Liu, Baoli, Zhi-Chun Li, Dian Sheng, and Yadong Wang. 2021. "Integrated Planning of
Berth Allocation and Vessel Sequencing in a Seaport with One-Way Navigation
Channel." Transportation Research Part B: Methodological 143 (January): 23–
47. https://doi.org/10.1016/j.trb.2020.10.010

Lorenzoni, Luciano Lessa, Hannu Ahonen, and Arlindo Gomes de Alvarenga. 2006. "A
Multi-Mode Resource-Constrained Scheduling Problem in the Context of Port
Operations." Computers & Industrial Engineering 50 (1–2): 55–65.
https://doi.org/10.1016/j.cie.2005.11.001

Menezes, Gustavo Campos, Geraldo Robson Mateus, and Martín Gómez Ravetti. 2017.
"A Branch and Price Algorithm to Solve the Integrated Production Planning and
Scheduling in Bulk Ports." European Journal of Operational Research 258 (3):
926–37. https://doi.org/10.1016/j.ejor.2016.08.073

Pratap, Saurabh, Ashutosh Nayak, Akhilesh Kumar, Naoufel Cheikhrouhou, and Manoj
Kumar Tiwari. 2017. "An Integrated Decision Support System for Berth and
Ship Unloader Allocation in Bulk Material Handling Port." Computers &
Industrial Engineering 106 (April): 386–99.
https://doi.org/10.1016/j.cie.2016.12.009

Ribeiro, Glaydston Mattos, Geraldo Regis Mauri, Saulo de Castro Beluco, Luiz Antonio
Nogueira Lorena, and Gilbert Laporte. 2016. "Berth Allocation in an Ore
Terminal with Demurrage, Despatch and Maintenance." Computers & Industrial
Engineering 96 (June): 8–15. https://doi.org/10.1016/j.cie.2016.03.005

Robenek, Tomáš, Nitish Umang, Michel Bierlaire, and Stefan Ropke. 2014. "A Branch-
and-Price Algorithm to Solve the Integrated Berth Allocation and Yard
Assignment Problem in Bulk Ports." European Journal of Operational Research
235 (2): 399–411. https://doi.org/10.1016/j.ejor.2013.08.015
Tang, M., D. Gong, S. Liu, and H. Zhang. 2016. "Applying Multi-Phase Particle Swarm
Optimization to Solve Bulk Cargo Port Scheduling Problem." Advances in
Production Engineering & Management 11 (4): 299–310.
https://doi.org/10.14743/apem2016.4.228

Umang, Nitish, Michel Bierlaire, and Ilaria Vacca. 2013. "Exact and Heuristic Methods
to Solve the Berth Allocation Problem in Bulk Ports." Transportation Research
Part E: Logistics and Transportation Review 54 (August): 14–31.
https://doi.org/10.1016/j.tre.2013.03.003

UNCTAD. 2019. Review of Maritime Transport. United Nations Conference on Trade

and Development.
https://unctad.org/system/files/officialdocument/rmt2019_en.pdf

Unsal, Ozgur, and Ceyda Oguz. 2019. "An Exact Algorithm for Integrated Planning of
Operations in Dry Bulk Terminals." Transportation Research Part E: Logistics
and Transportation Review 126 (June): 103–21.
https://doi.org/10.1016/j.tre.2019.03.018

Zhen, Lu, Zhe Liang, Dan Zhuge, Loo Hay Lee, and Ek Peng Chew. 2017. "Daily Berth
Planning in a Tidal Port with Channel Flow Control." Transportation Research
Part B: Methodological 106 (December): 193–217.
https://doi.org/10.1016/j.trb.2017.10.008
 Appendix
Table 11. High tide periods.
Ot HIGH TIDE PERIOD for vessels 3, 7 & 15 (wv =1 for v =3,7 & 15)
DAY
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
1 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0
2 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0
3 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0
4 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0
5 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0
6 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0
7 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0
8 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0
9 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1
10 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1
11 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1
HOUR

12 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1
13 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1
14 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0
15 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0
16 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0
17 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0
18 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0
19 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0
20 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0
21 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1
22 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1
23 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1
24 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1
Table 12. Non-working periods.
y vt NON-WORKING PERIODS for Vessels 4, 9 & 12 (gv =1 for v = 4,9,12)
DAY
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1
2 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1
3 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1
4 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1
5 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1
6 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1
7 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1
8 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1
9 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1
10 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1
11 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1
HOUR

12 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1
13 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1
14 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1
15 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1
16 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1
17 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1
18 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1
19 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1
20 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1
21 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1
22 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1
23 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1
24 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1