Advanced Engineering Informatics 56 (2023) 101989

Contents lists available at ScienceDirect

Advanced Engineering Informatics

journal homepage: www.elsevier.com/locate/aei

Full length article

Disruption management-based coordinated scheduling for vessels and ship

loaders in bulk ports
Jingyun Wang a, b, Xinyu Zhang a, *, Wenqiang Guo a, Zaili Yang c, Nyamatari Anselem Tengecha a
a
Maritime Intelligent Transportation Research Group, Navigation College, Dalian Maritime University, Dalian 116026, China
b
College of Transportation Engineering, Dalian Maritime University, Dalian 116026, China
c
Liverpool Logistics, Offshore and Marine Research Institute, Liverpool John Moores University, Liverpool L3 3AF, UK

A R T I C L E I N F O A B S T R A C T

Keywords: For ensuring the orderly operation of the port, it is vital to coordinately schedule available ship loaders and
Bulk port vessels that plan to enter and exit the port when ship loaders are unable to work due to faults. Therefore, this
Disruption management paper studies the coordination between vessels and ship loaders scheduling problem affected by failed ship
Vessel traffic scheduling
loaders (VSLPB), and proposes a novel disruption management-based method to address this problem. An
Coordinated scheduling
innovative optimization model is developed to reduce the generalized cost with the constraints of disruption
Ship loader failure
Row generation management strategies (DMS), aiming to minimize the impact of failed ship loaders on the coordinated sched­
uling and the bulk cargo handling efficiency. For solving the VSLPB, an effective two-stage row generation
(TSRG) algorithm is developed. In the first stage, the disruption conditions in the model are released to find the
available ship loaders and berths for vessels affected by the failure factors. In the second stage, the optimal
strategy is sought among multiple DMS to minimize the objective function value. Using the proposed method in
Huanghua Coal Port as a case study, the results show that our method can effectively solve the impact of ship
loader failure on the efficiency of bulk cargo handling and the efficiency of vessels entering and leaving the port.
These further highlights the importance of implementing DMS, and show that the proposed method can provide
an efficient and reliable solution for port production and operation to deal with disruption problems. Further­
more, the proposed method in this paper can help improve the ability of the port to resist uncertain factors, thus
improving the ability of the entire supply chain to resist risks.

1. Introduction efficiency of vessels entering and leaving the port appears, which
eventuates in an increased waiting time of vessels at the port and sub­
The worldwide dry bulk shipping industry is increasing, and the sequentially affects the efficiency of the port. Ship loaders are the
Baltic exchange dry index (BDI) hit a 12-year high of 5 650 points in equipment that connects terminals and vessels directly, and ship loader
early October 2021; in 2021, the average value of the BDI was 2 943 allocation is essential to vessel traffic scheduling in decreasing the
points, up 176.1 percent in the corresponding period. Moreover, the waiting time of vessels. Thus, coordinating and optimizing ship loader
worldwide dry bulk cargo market’s transportation capacity reached 945 and vessel traffic organization to generate the initial scheme is an
million deadweight tonnages (DWT) at the end of 2021, up around 3.6 awkward problem [2].
percent [1]. With this background, the throughput of bulk ports in­ However, as one of the crucial ways to improve the efficiency of
creases, which puts forward new requirements and challenges for the vessels entering and leaving the port, vessel traffic scheduling is easily
efficiency of bulk vessels scheduling at bulk ports. To solve this, terminal affected by the ship loader, especially the ship loader failures during the
operators around the world are committed to improving efficiency by scheduling process. Moreover, when executing the initial scheme, op­
updating equipment and enhancing management levels. However, the erations in modern coal terminals are frequently interrupted [3],
efficiency of vessels entering and leaving the port is still constrained by resulting in the initial scheme being inapplicable. In practice, the ter­
various reasons and has not considerably improved. As a result, a sig­ minal’s adjustment configuration for the plan is based mainly on the
nificant mismatch between the efficiency of ship loaders and the unique scenario and prior operational experience, lacking systematic

* Corresponding author.
E-mail address: [email protected] (X. Zhang).

https://doi.org/10.1016/j.aei.2023.101989
Received 2 February 2023; Received in revised form 19 April 2023; Accepted 22 April 2023
Available online 4 May 2023
1474-0346/© 2023 Elsevier Ltd. All rights reserved.
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

adjustment goals. As a result, the whole system will be severely dis­ 2.1. Vessel traffic scheduling
rupted, which could lead to issues like service quality degradation,
decreased efficiency, cost increases, etc. In summary, it is necessary to Vessel traffic scheduling optimization is primarily concerned with
coordinate ship loader allocation and vessel inbound and outbound the orders of vessels passing through the channel, safely and effectively.
process, and study how to adjust the initial scheme to realize vessels’ Most existing research on vessel traffic scheduling was based on deter­
effective entry and exit when ship loaders failures are taken into account ministic conditions and did not refer to disruption factors. Previous
in a bulk cargo port. Therefore, this paper studies the combined opti­ studies have been conducted to carry out relevant research based on
mization problem of available ship loaders and vessels entering and different channel types, Zhang et al. [4,5] studied the integration
leaving the port under the influence of fault factors. The coordination problem of vessel scheduling and berth allocation in a one-way channel.
between vessels and ship loaders is equivalent to the coordination be­ They first studied the problem in a fixed planning period for a discrete
tween the scheduling process of vessels entry and exit the port and the berth bulk cargo port with a single harbor basin, then studied the
allocation process of ship loaders. One of the key issues in determining problem in a multi-harbor basin. Zhang et al. [6] studied the optimi­
the entry and exit process of vessels is to allocate berths. When assigning zation model and algorithm of vessel traffic scheduling in a restricted
berths to a vessel, choosing different berths will result in different entry two-way channel in Huanghua Port, taking the channel as the main
and exit processes, as well as different feasible ship loaders. Therefore, object. Li et al. [7] investigated the traffic scheduling problem of vessels
vessels and ship loaders are connected and coordinated by berth allo­ entering and leaving restricted channels in a multi-harbor basin and
cation process. This problem allocates appropriate time and berths for generated an optimal traffic scheduling scheme for each vessel to ensure
vessels entering and leaving the port, and arranges available ship loaders the safety and efficiency of vessel navigation. Within this context, the
for them to minimize the impact of ship loader failure on cargo handling state-of-the-art is the studies on the coordination and optimization of
efficiency and ship delay. vessel traffic scheduling and terminal equipment. For instance, Li et al.
The new contributions of this paper are stated as follows: (1) [8] studied loading plans and equipment cooperation problems to get an
Development of a novel mixed integer linear program (MILP) model allocation scheme.
based on disruption management strategies to minimize the generalized In addition, many other scholars have researched vessel traffic
cost and solve the VSLPB, in which this paper, for the first time, tackles scheduling, considering different elements involved in the port pro­
the practical challenge of scheduling the vessels and ship loaders duction process. Jia et al. [9,10] considered the characteristics of
considering the disruptive factors. (2) Exploitation of the disruption different channel types and took the vessel entering and leaving port
management strategies and the method of soft and hard constraints to process as the critical factor to study the traffic scheduling problem
prepare for designing a solving algorithm. These approaches could take under the combination of vessel and berth. Niu et al. [11] studied the
advantage of reducing the model size and simplifying the problem. coordination problem of anchorage allocation and vessel traffic sched­
Furthermore, it develops an efficient TSRG algorithm to solve the uling in Shanghai Yangshan deep-water port. Abou Kasm et al. [12]
problem optimally. (3) Demonstration of the good interpretability of the studied the mathematical model of the vessel scheduling problem with
proposed method and provide a case study with deep insights. Our so­ tug and pilotage constraints and channel restrictions; then, they
lution can allocate available ship loaders for inbound and outbound designed an exact solution method based on constraint separation. Liu
vessels under the influence of fault factors and can minimize the impact et al. [13] studied the seaport berth and channel planning problem,
of failed ship loaders on cargo handling efficiency and vessel delay. aiming to minimize the expected total weighted completion times of
In the rest of the paper, a literature review is presented in Section 2. ships. Chen et al. [14] studied how to optimize slot capacity allocation
Section 3 describes the studied problem in detail and provides a novel within a container liner alliance under the slot exchange mode in the
mathematical formulation. The new formulation enables the develop­ containerized maritime logistics industry.
ment of an efficient TSRG algorithm approach for its solution, which is However, there is no evidence showing the existence of any opti­
enhanced by a disruption management strategy and column generation mization research on ship loaders coordinating with vessels, and the
(Section 4). Computational studies are conducted in Section 5 to eval­ primary method of the present research on vessel traffic scheduling
uate the performance of the proposed TSRG based approach. Finally, the optimization focused on deterministic conditions concerning no
research results are summarized in the conclusions in Section 6. disruption factors. Obviously, they could not reflect and fit the high
research demand in today’s bulk shipping industry.
2. Literature review
2.2. Disruption management
It is evident from the current literature that little coordinated
research on vessel traffic scheduling concerning the disruptive effect of Researchers have conducted many methods to lessen the impact of
ship loader failures, despite the increasing number of relevant incidents disruptions, among which disruption management is deemed to be a
that occurred in practice. Within the context of water transportation, the leading position. Disruption management performs local optimization
most relevant references are focusing on berth plan recovery (BPR) and and adjustment of the initial scheme based on the state after the
liner ship schedule recovery (LSSR). However, both BRP and LSSR disruption factor has ended, resulting in an adjusted scheme that re­
overlook the incorporation of cargo loading failures as a disruptive duces the influence of disruption factors on the scheduling system [15].
factor in the vessel scheduling optimization and coordination, which is Disruption management has been successfully applied to production job
evident to be an unavoidable problem to be addressed in urgency to shops coping with disruptions in production and scheme execution. In
ensure the success of the VSLPB. From an applied research perspective, it the study of production scheme recovery problems, Baykasoglu and
strikes the new coordination of vessel scheduling and port operations in Karaslan [16] proposed a new disruption management approach, which
the dry bulk area beyond the dominated container sector. Obviously, includes a disruption management model and a multi-objective opti­
given the difference between container and bulk shipping, the existing mization algorithm that can effectively reduce the deviation. Ning et al.
methods in container shipping cannot be applicable to the bulk sector [17] designed an improve quantum bacterial optimization algorithm
without new developments. In the following subsections, we focus on (IQBFO) based on prospect theory to solve the proposed model, and four
three aspects of vessel traffic scheduling, disruption factors, and response indexes of 16 kinds of rescheduling scenarios are simulated and
disruption management applied to water transportation. analyzed by using the IQBFO and comparing with the existing sched­
uling algorithms. Li et al. [18] proposed a value function metric for the
disruption problem in uncertain job shop scheduling problems to reduce
carbon emissions in the manufacturing process. Fischer et al. [19]

2
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

presented different strategies for handling disruptions in fleet deploy­ To sum up, relatively few articles have studied the problem of
ment in roll-on roll-off liner shipping, which basically consists of disruption management in vessel traffic scheduling. Most of the existing
assigning a fleet of vessels to predefined voyages at minimum cost. Ke theories in the field of disruption management focus on the recovery of
et al. [20] proposed a framework based on optimization and regression production plans; studies on disruption management in waterway
analysis for recovery from random disruptions of rail intermodal ter­ transportation concentrate on container ships. However, the above
minals. At the same time, in light of the idea of gradual optimization for studies did not involve the coordinated optimization of vessel traffic
the target to obtain the job-shop scheduling adjustment scheme with scheduling and ship loaders, nor did they concern the disruption factors
minimum disturbance. Ning and Wang [21] proposed the measurement with the coordinated optimization problem. Therefore, the existing
method of value function based on prospect theory and the disruption models and algorithms could not be used to address the research prob­
management strategy of user’s psychological perception and established lem described below, and it is necessary to explore new models and
a multi-objective optimization model for job-shop scheduling manage­ algorithms according to their unique characteristics.
ment through multi-objective programming. Sun et al. [22] proposed an
improved multi-objectives method to solve the dynamic job-shop 3. Problem description and mathematical formulation
scheduling problem based on disruption management, and a quantum
genetic algorithm for adaptively adjusting the rotation angle. Malik and 3.1. Problem description
Sarkar [23] developed a mathematical model of a multi-item produc­
tion-inventory system to maximize the total profit within a single To realize the adjustment of vessels’ arrival and departure times and
disruption-recovery time window. Sang et al. [24] proposed a new redistribution of ship loaders within a limited range, this paper proposes
disruption management method, that includes the disruption manage­ disruption management strategies by dividing the operation status of
ment model and the many-objective optimization algorithm. vessels, affected by ship loader failures, into different stages based on
In light of the above, the most similar sector to bulking shipping is actual situations. Fig. 1 shows all the possible divided stages.
job shop scheduling, in terms of both theories and applications of Fig. 2 describes the specific disruption management strategies in
disruption management, and hence the relevant papers have been various stages according to the direction of vessels, whether ship loaders
thoroughly reviewed for a cross reference purposes. can be repaired before the vessels inbound or outbound the port,
whether the vessel berths at the initial berth, and whether uses the initial
2.3. Disruption management applied to water transportation ship loader.
When executing the initial scheme, disruption factors prevent the
Holistic research on waterway transportation and disruption man­ initial scheme from having the optimal effect it should. It will be
agement mainly focused on berth plan recovery and liner ship schedule determined whether to manage the factors; if disruption management is
recovery. Cheraghchi et al. [25] concerned with speeding up strategy in not required, then the initial scheme is executed; if disruption man­
vessel schedule recovery problems, modeled S-VSRP as a multi-objective agement is required, a new scheme is generated and executed based on
optimization problem and resorted to several multi-objective evolu­ the disruption management strategies. Fig. 3 shows the flow chart of the
tionary algorithms to approximate the optimal Pareto set, which pro­ disruption management adjustment strategy.
vides vessel route-based speed profiles. Han et al. [26] applied the Since other operations in the port are arranged according to a ves­
dynamic disruption management method to collaboratively plan the sel’s arrival and departure scheme and loading operation scheme, the
resources of container terminals in a cyclical environment, considering temporary adjustment of these two schemes will significantly increase
the uncertainty of vessel arrival time and market demand. Then, Abioye the time and economic costs. The problem studied in this paper is how to
et al. [27] formulated a novel mathematical model for the vessel adjust the initial scheme to generate a new scheme based on less
schedule recovery problem in liner shipping, aiming to minimize the adjustment and redistribution of resources. The new scheme should take
total profit loss, suffered by the liner shipping company due to disrup­ into account the original optimization objectives while minimizing the
tion occurrences at a given liner shipping route. van der Steeg et al. [28] effects caused by disruptions [15]. Therefore, it is essential to choose a
proposed a rolling window strategy to deal with the disruption factors proper objective function to measure the cost of the adjusted scheme.
coping with the early or late arrival of vessels or disruptions requiring To reflect the rationality of the new scheme and the difference be­
longer loading and unloading times, and a real-time disruption man­ tween the new scheme and the initial scheme comprehensively, [33–37]
agement decision model was proposed. De et al. [29] addressed the are referred to use the generalized cost as the objective function,
environmental concerns related to fuel consumption and carbon emis­ including the cost of implementing the new scheme and the penalty cost
sion within shipping operations and simultaneously presents strategies of adjusting the initial scheme. In addition, the hard constraints in the
for countering disruption within the maritime transportation domain. model have to be satisfied, while the soft constraints will be satisfied to
Elmi et al. [30] offered a thorough review of the current liner shipping the greatest possible extent. Given that the vessels studied in the one-
research primarily focusing on two major themes: uncertainties in liner way channel in this paper have the characteristics of entering and
shipping operations; and ship schedule recovery in response to disrup­ leaving the port in a group if the vessel-grouping constraints are met
tive events. They provided representative mathematical models that absolutely, it is easy to encounter the situation where the solution set is
could be used further in future research efforts dealing with liner ship­ empty when solving VSLPB. Therefore, according to the related concepts
ping and ship schedule recovery uncertainties. Chen et al. [31] studied of soft and hard constraints, the vessel-grouping constraints are set as
the co-deployment of liner alliance fleets under the vessel pool operation soft constraints.
with uncertain demand. Then, Chen et al. [32] studied the fleet sched­ The assumptions of establishing the optimization model are defined
uling problem of container liner alliance members in the slot exchange as follows:
mode, with sulfur emission restrictions taken into consideration.
From the above analysis, disruption management research in the (1) The weather conditions and berth depth in the port meet the
shipping field primarily focused on container liner shipping, with less berthing requirements of each vessel.
research on bulk cargo transportation. However, there are some dis­ (2) The maintenance time windows of ship loaders are known.
tinctions between bulk cargo ports and container ports. For example, the (3) The berthing plan has been generated before the vessels’ inbound
loading and unloading equipment in a bulk cargo port often moves or outbound ports.
continually, while the one in the container port moves between bays, (4) The resources of pilots, tugs, and yard storage are sufficient.
which is often seen as a separate activity. As a result, research is required (5) The decision time is short, so the influence of the decision point is
based on the peculiarities of bulk cargo ports. ignored.

3
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

Fig. 1. The Divided Stages.

(6) The initial scheme existed before disruption management started. (continued )
Sets and parameters
Symbol Explanation
3.2. Model formulation pk upper bound of the maintenance time window for ship loader k,
k∈K
This section introduces an innovative model formulation for coor­ pk lower bound of the maintenance time window for ship loader k,
k∈K
dinated re-optimizing vessel traffic scheduling and ship loaders alloca­
vi speed of the vessel i, i ∈ I
tion problem. We considered the disruption management strategies and gi upper bound of tide riding time window of vessel i, i ∈ I
the practical requirement of safety. The notations of the VSLPB model gi lower bound of tide riding time window of vessel i, i ∈ I
are presented as follows. qsu the amount of cargo u, u ∈ U stored in yard s, s ∈ S
Sets and parameters qiu the amount of cargo u, u ∈ U required by the vessel i, i ∈ I
Symbol Explanation qiu
′
the amount of cargo u, u ∈ U that vessel i, i ∈ I still needs to
carry after the occurrence of the disruption
I set of vessels ωk operation efficiency of ship loader k, k ∈ K
J set of berths 1
tijk start time of vessel i, i ∈ I served by ship loader k, k ∈ K at
K set of ship loaders berth j, j ∈ J
N set of time intervals 1
tijk ’ start time of vessel i, i ∈ I served by ship loader k, k ∈ K at
R set of time points berth j, j ∈ J after berth shifting
S set of yards 2
tijk end time of vessel i, i ∈ I served by ship loader k, k ∈ K at
m1 /m2 /m3 /m4 /m5 coefficients of each influence factor in the objective function berth j, j ∈ J
M a sufficiently large positive number tijk
2
’ end time of vessel i, i ∈ I served by ship loader k, k ∈ K at
ai ready time of vessel i, i ∈ I berth j, j ∈ J after berth shifting
ai
′
adjusted ready time of vessel i, i ∈ I tkj operation time of ship loader k, k ∈ K at berth j, j ∈ J
li length of vessel i, i ∈ I tkj ’ operation time of ship loader k, k ∈ K at berth j, j ∈ J during
L length of channel the disruption management phase
h safe time interval of vessels in the same direction Tn upper bound for vessel i, i ∈ I inbound or outbound the port in
h
′
safe time interval of vessels in different directions the n, n ∈ N subgroup
xi the earliest start time of vessel i, i ∈ I Tn lower bound for vessel i, i ∈ I inbound or outbound the port in
xi
′
adjusted the earliest start time of vessel i, i ∈ I the n, n ∈ N subgroup
yi the earliest end time of vessel i, i ∈ I Wi total waiting time for vessel i, i ∈ I
yi
′
adjusted end time of the vessel i, i ∈ I Oi total operation time for ship loader k, k ∈ K

(continued on next column) (continued on next page)

4
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

Fig. 2. Specific disruption management strategies in various stages.

(continued )
Sets and parameters
Symbol Explanation

0–1 Decision variables

Bij equals 1 if vessel i, i ∈ I moors at berth j, j ∈ J and 0, otherwise
Bij
′
equals 1 if vessel i, i ∈ I moors at berth j, j ∈ J after the
occurrence of disruption and 0, otherwise
Cius equals 1 if cargo u, u ∈ U required by vessel i, i ∈ I stored in
yard s, s ∈ S and 0, otherwise
Dijs equals 1 if yard s, s ∈ S is connected to berth j, j ∈ J where
vessel i, i ∈ I berths and 1, otherwise
Ein equals 1 if vessel i, i ∈ I belongs to subgroup n, n ∈ N and 0,
otherwise
Ein
′
equals 1 if vessel i, i ∈ I belongs to subgroup n, n ∈ N after the
occurrence of disruption and 0, otherwise
Fijk equals 1 if ship loader k, k ∈ K serves vessel i, i ∈ I at berth j,
j ∈ J and 1, otherwise
Fijk
′
equals 1 if ship loader k, k ∈ K serves vessel i, i ∈ I at berth j,
j ∈ J after the occurrence of disruption and 1, otherwise
Gkj
′
equals 1 if ship loader k, k ∈ K can be moved and used for
berth j, j ∈ J after the occurrence of disruption and 0,
otherwise
Uikr equals 1 if ship loader k, k ∈ K serves vessel i, i ∈ I at moment
r, r ∈ R and 0, otherwise
IOi equals 1 if vessel i, i ∈ I is entering port and 0, otherwise

Based on the above assumptions and symbols, the mathematical model

of coordinated scheduling for vessels and ship loaders considering
disruption management is formulated as follows:
Objective function
∑( ⃒ ⃒ )
min Wi + Oi + m1 *⃒x’i − xi ⃒ + m2 *Bij ’ + m3 *Fijk
’
+ m4 *G’kj + m5 *Ein’
Fig. 3. Disruption Management Adjustment Strategy Flow Chart.
i,j,k

(1)

5
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

Hard constraints general standard. As per the definition of generalized cost, we take it as a
disruption measure function composed of seven terms. The generalized
xi ⩾ai (2)
cost includes the vessels’ total waiting time (Wi ), the ship loaders’ total
( ) operation time (Oi ), the penalty for altering the vessel’s inbound or
(IOi − IOi′ ) Uikr *xi − Ui′ k(r− 1) *xi′ ⩾h (3) ⃒ ′ ⃒
outbound time (m1 *⃒xi − xi ⃒), the penalty for shifting berth (m2 *Bij ), the
′

( ) ′ penalty for altering ship loaders (m3 *Fijk ), the penalty for moving ship
′
[1 − (IOi − IOi′ ) ] Uikr *xi − Ui′ k(r− 1) *xi ⩾h (4)
loaders (m4 *Gkj ), and the penalty for changing the groups of vessels
′

h = max(li , li )*α1/v ′ (5) (m5 *Ein ). Function (2) ensures that the vessels’ start time cannot be
′

behind
earlier than the arrival time. Functions (3) to (6) state the safe distance
/
h = lbehind *α2 v
′
(6) between vessels measured by time. Function (7) ensures the corre­
before
spondence between berths, vessels, cargos, and yards. Functions (8) to
(12) state the time–space constraints of the ship loaders. Function (8)
Dijs *Cius *Bij *Fijk = 1 (7)
and (9) ensure one ship loader can merely service one vessel at a time,
∑ and one vessel can only be handled by one ship loader at a time.
Uikr ⩽1 (8)
i
Function (10) and (11) ensure the feasible service time of ship loaders.
Function (12) restricts the number of ship loaders serving simulta­
’
Fijk + Fijk =1 (9) neously to no more than the total number. Function (13) states the
weight constraint of the loaded cargo. Functions (14) and (15) ensure
( )
∑ ∑ ∑ the tidal time window for the vessels. This paper studies a one-way
Uikr *xi + M* 1 − Uikr ⩾ Uik(r− 1) *yi (10) channel, which means that vessels can only enter or leave the port
i i i
simultaneously. Additionally, to limit the number of channel direction
changes and ensure an orderly inbound and outbound process, vessels
1
tijk 2
+ tkj ⩽tijk (11)
were usually grouped to pass the channel in a practical process, func­
∑ tions (16) and (17) state the grouping time constraint of the vessels.
Fijk ⩽kmax (12) Function (18) calculates the end of vessels’ inbound or outbound time.
Functions (19) to (23) state the disruption management strategies, and
i,j,k

∑ ∑ the objective function values corresponding to different recovery stra­

qiu ⩽ qsu *(1 − Cius ) (13)
i s
tegies are calculated when ship loader fails. The recovery strategy with
the smallest objective function value is selected. Function (19) states
gi ⩽xi ⩽gi (14) that the berth corresponding to the ship loader is unavailable during the
maintenance time window. Function (20) ensures vessels’ start time
gi ⩽xi + L/v ⩽gi (15) cannot be earlier than their adjusted arrival time. Function (21) calcu­
i
lates the earliest start time of the vessel outbound the port after
adjustment. Function (22) calculates the initial ship loader operation
Tn ⩽Ein *xi ⩽Tn (16)
ending when the vessel still berths at the initial berth and is served by
the initial ship loader. Function (23) calculates the end time of the ship
Tn ⩽Ein *yi ⩽Tn (17)
loader operation when the vessel still berths at the initial berth but is
served by another ship loader. Function (24) ensures that when the
yi ⩾xi + Fijk *tkj (18)
vessel shift berth, the cargo required by the vessel can still be trans­
∑ ( ) ported to this berth. Function (25) calculates the time when the vessel
Uikr pk − pk = 0 (19)
needs to be served by the ship loader on the corresponding berth after
k
shifting berth. Therefore, functions (26) and (27) state the soft con­
′
xi ⩾ai
′
(20) straints, which mean that the disturbed vessel’s adjusted arrival and
departure time meet the initial group as much as possible.
(1 − IOi )xi’ = pk + tijk
1
’ + tkj (21) The existing research and applications of soft constraints are pri­
marily reflected in intelligent optimization algorithms [38–43]. Most of
( ) ( ) ′ ( ) these studies use the method of punishing fitness and adding punish­
= pk + tkj * 1 − Bij *Fijk * 1 − Gkj (22)
2 ′ ′ ′
tijk
ment measures to reduce individual fitness for violating hard and soft
( / ) constraints in the individual genetic algorithms. To simplify the model
∑ ( ) and ensure the effectiveness of the model solution, we add the soft
2
tijk ’= 1
tijk + q’iu ωk * 1 − B’ij *Fijk
’
*G’kj (23)
u
constraints into the objective function based on an interior point penalty
function method. The simplified model is stated as follows.
Bij ’*Aij = 1 (24) Objective function
∑( ⃒ ′ ⃒
( / ) Wi + Oi + m1 *⃒xi − xi ⃒ + m2 *Bij + m3 *Fijk + m4 *Gkj
′ ′ ′

∑ Φ(x, μ) = min
2
tijk ’= 1
tijk ’+ qiu ’ ωk *B’ij *Fijk
’
*Gkj ’ (25) i,j,k
)
+ m5 *Ein + μC− 1 (x)
u ′

Soft constraints (28)

(26)
′
Tn ⩽(1 − Ein )*xi ⩽Tn Functions (2) to (25) are constraints. Moreover, function (28) states
the simplified soft constraint C(x) defined in function (29).
(27)
′
Tn ⩽(1 − Ein )*yi ⩽Tn
(29)
′
C(x) = Tn − (1 − Ein )*yi
The objective function (1) is the disruption measure function, it is
chosen primarily depending on actual demands, but there is currently no

6
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

4. A TSRG algorithm

The solution objective of the algorithm is to obtain the adjusted

scheme at a fast speed when disruption factors occur. According to the
established mathematical model above, the number of constraints
involved in this paper is enormous. Since the problem of the vessel
traffic scheduling process considering disruption factors is complex, it is
difficult to solve the model directly using commercial solving software.
Among all the constraints, only a small number of the constraints play a
decisive role. The above propositions inspire the consideration of a
method to reduce the model size. The row generation method has ad­
vantages in solving problems with multiple constraints. Each interfer­
ence management strategy is independent of the others, and when
solving the constraints corresponding to one interference management
strategy, it does not have an impact on other interference management
strategies. Accordingly, using the row generation method for solving is
suitable. Therefore, an algorithm based on the row generation method is
employed to reduce the number of constraints. By delaying the gener­
ation of optimal solutions, we reduce the number of constraints to speed
up the solution.
To solve the problem effectively, a new TSRG algorithm based on the
row generation algorithm and the disruption management strategies in
Section 3 is designed. The main idea of TSRG is to relax some constraints
of the model formulated in Section 3 to obtain a master problem,
construct the corresponding sub-problem after solving the master
problem, then add the corresponding constraints until finding the
optimal solution. First, all constraints of the disruption management
strategy are relieved, and all available berths and ship loaders are ob­
tained. Then, based on all the berths and ship loaders obtained, different
objective function values corresponding to the various disruption
management strategies are calculated. Suppose the decision variable
corresponding to the optimal objective function value satisfies the
constraint of the integer variable simultaneously. In that case, the
objective function value is the optimal solution, and the corresponding
disruption management strategy is the optimal adjustment strategy
selected. Suppose the decision variable corresponding to the obtained
objective function value does not meet the integer constraint. In that
case, it is necessary to add an integer constraint on this basis to continue
solving the issue and repeating this process until the optimal solution Fig. 4. Flow Chart of TSRG.
meeting the conditions is obtained.
Fig. 4 shows the specific flow of the TSRG algorithm. The first stage To obtain the solution efficiency and accuracy of the TSRG algo­
of the TSRG algorithm is to get the available berths and ship loaders rithm, we used the genetic algorithm (GA), particle swarm optimization
when vessels berth at the adjusted time. In this stage, we relaxed the algorithm (PSO), and differential evolution algorithm (DE) to solve the
integer variables in the model into continuous variables, took function same problem. Fig. 5 presents the basic flow chart of the three afore­
(28) as the objective function and functions (2) to (19) as the initial mentioned intelligent optimization algorithms.
constraints, then treated them as the master problem. The second stage
is to get the minimum objective value and corresponding strategy. In the 5. Computational experiments and results
second stage, we took function (28) as the objective function and add
functions (20) to (25) for further solutions. We have adopted some To verify the effectiveness of the proposed TSRG algorithm, we
heuristic strategies to accelerate the solving process. Firstly, the feasible conducted experiments based on the data from the Huanghua Coal Port.
berths and ship loaders obtained in the first stage are grouped: all so­ The illusion of Huanghua Coal Port is sketched in Fig. 6.
lutions of the same berths are grouped and arranged in ascending order. Generally, in the Huanghua Coal Port, the inbound vessels enter the
In this case, the ascending order is for the corresponding changes of each channel at buoy no.22, and buoy no.46 is the joint of the harbor basins
disruption management strategy, while the strategy with a larger and the channel. Therefore, the buoy no.22 and no.46 are defined as
sequence number refers to the solution with a change in the ship loaders. channel entry and exit, respectively. The incoming vessels successively
In addition, the greedy criterion is used to prioritize the search for so­ pass through the channel entrance and exit from the anchorage, finally
lutions with small changes, i.e. those with higher sequence numbers. arriving at the berth. The outcoming vessels navigate through the
Then, using a rolling search strategy, the intersection of available berth channel from the berth and finally depart at the channel entry.
time windows, available ship loader time windows, formation time Then, we studied the impact of different time windows of ship loader
windows, and unoccupied time windows between two adjacent vessels maintenance and different numbers of failed ship loaders based on the
is searched first. If no solution satisfies the intersection time window, the initial scheme of 25 vessels, which is supported by historical data. The
complement of formation time windows is searched. At the same time, a initial scheme was shown in Table 1, and Table 2 shows the specific ship
rolling search strategy is used during the search process, ensuring the loaders’ failure data.
unoccupied time windows between adjacent vessels are searched one by All the computational experiments were performed on a computer
one. Last but not least, tabu strategy is used to label the found optimal with Intel Core i7-7500U 2.70 GHz CPU and 8 GB RAM. The TSRG al­
solutions and avoid repeated searches in the subsequent search process. gorithm was solved using CPLEX 12.6.3, and each intelligent

7
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

Fig. 5. Basic Flow Chart of Each Algorithm.

Fig. 6. Illusion of Huanghua Coal Port.

optimization algorithm was executed in Matlab with the R2016a moment 0 every day, and the moment is accumulated over time. The
version. Table 3 displays the specific values of each parameter for the time intervals of vessels’ inbound and outbound groups in a day are
three algorithms. [0,360] min, [361,720] min, [721,1 080] min, and [1 081,1 440] min in
order, and the navigation direction of vessel-grouping changes every 6
h.
5.1. Rational verification of the adjusted scheme
Taking experiment 12 in Table 2 as an example, we study which
disruption management strategy should be adopted to adjust the initial
The experiment is conducted based on the known initial scheme with
scheme when SL5 and SL8 are under repair at [600,700] min and
25 vessels, including the vessel arrival and departure scheme and
[800,900] min, respectively. The adjusted scheme of 25 vessels is solved
loading operation scheme. The initial scheme is shown in Table 1. Ac­
by the TSRG algorithm, as shown in Table 4, which includes the adjusted
cording to the actual production of Huanghua Port, 0:00 is recorded as

8
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

Table 1
Initial scheme of 25 vessels.
Vessel In or Tidal time window Inbound moment Outbound moment Time of leaving the Berth Ship The operation time window of
number out (min) (min) (min) channel (min) loader* ship loader (min)

1 out — — 0 92 #2 SL2 [0,0]

3 out — — 80 183 #3 SL3 [0,0]
4 out — — 114 222 #8 SL8 [0,57]
5 out — — 163 259 #10 SL10 [0,128]
7 out — — 206 286 #17 SL13 [0,151]
8 out — — 258 330 #7 SL7 [0,206]
9 out [200,500] — 286 349 #9 SL9 [0,252]
2 in — 389 — 471 #1 SL1 [513,902]
11 in — 413 — 492 #13 SL9 [570,867]
6 in [400,700] 430 — 516 #8 SL8 [640,851]
12 in — 459 — 536 #10 SL10 [884,1 183]
13 in — 483 — 581 #2 SL2 [861,1 052]
15 in — 523 — 605 #4 SL4 [800,1 056]
17 in — 570 — 669 #17 SL13 [822,1 156]
10 in — 591 — 698 #3 SL3 [861,1 253]
16 out [720,820] — 740 812 #4 SL4 [230,662]
18 out — — 775 876 #16 SL12 [232,651]
14 out — — 796 900 #5 SL5 [354,707]
19 out — — 822 929 #6 SL6 [344,768]
20 out — — 880 988 #14 SL10 [350,834]
22 out — — 960 1 068 #11 SL11 [530,920]
23 in — 1 095 — 1 198 #7 SL7 [1 261,1 437]
21 in [1 080,1 280] 1 116 — 1 227 #9 SL9 [1 255,1 461]
24 in — 1 142 — 1 249 #6 SL6 [1 281,1 633]
25 in — 1 266 — 1 408 #1 SL1 [1 428,2 041]
*
the number of ship loaders is from SL1 to SL13.

Table 2
Ship loader failure data.
Experiment Failed ship The time window of ship loader maintenance Experiment Failed ship The time window of ship loader maintenance
number loader (min) number loader (min)

1 SL5 [500,600] 8 SL8 [800,900]

2 SL5 [500,700] 9 SL8 [800,1 000]
3 SL5 [600,700] 10 SL8 [600,700]
4 SL5 [600,800] 11 SL8 [600,800]
5 SL5 [600,900] 12 SL5,SL8 [600,700],[800,900]
6 SL5 [600,1 000] 13 SL5,SL11 [600,700],[600,700]
7 SL5 [600,1 100] 14 SL8,SL13 [800,900],[800,900]

corresponding operation time window for SL5 changed from [640,851]

Table 3
min to [640,799] min ∪ [901,953] min. Meanwhile, vessel 14 adopts the
Parameters of GA, PSO, and DE.
same strategy as vessel 6 but has a different direction from vessel 6,
Algorithm Parameter Value which means stage14 = 2. The corresponding operation time window for
′

GA generation gap 0.9 SL8 changed from [354,707] min to [354,599] min ∪ [701,809] min.
maximum generations 200 Figs. 7 and 8 show the adjusted berthing time and ship loaders’
crossover probability 0.7
operation time, which can clearly reflect the berthing time of vessels and
variation probability 0.002
PSO iterations 200 working time. Table 4, Figs. 7 and 8 show no conflict between vessels,
particle swarm 100 berths, or ship loaders, so the rationale of the adjusted scheme is verified
maximum archive 200 with 25 vessels.
initial inertia weight 0.9
Berth safety assurance verification. According to the berth opera­
particle size 3
the first velocity update parameter 1.5 tion time chart shown in Fig. 7, taking berth 17 as an example, it has two
the second velocity update parameter 2 operation time windows, [0,151] min and [799,1 133] min. There is no
maximum velocity 0.2 overlap between the two yellow bars, which means no conflict between
minimum velocity − 0.2 all berths, and the safety of each berth is ensured. It verifies that the
divided raster 50*50
safety of the rest berths is ensured.
DE iterations 200
population 50 Ship loaders allocation verification. According to the ship loaders’
scaling factor 0.2 operation time shown in Fig. 8, taking ship loader 1 (SL1) as an example,
crossover probability 0.9 it has two operation time windows, [513,902] min, and [1 428,2 041]
min. Because there is no overlap between the two green bars, there will
be no conflicts between the SL1, thereby ensuring its safety. The rest of
vessel arrival and departure scheme and adjusted loading operation
the ship loaders are confirmed to be safe.
scheme. Comparing Table 1 with Table 4, it can be found that the
Tide riding time verification. Table 4 shows that vessel 6, vessel 9,
damage SL5 affects vessel 14, and the damage SL8 affects vessel 6. The
vessel 16, and vessel 21 need to take the tide to enter or leave the port.
disruption management strategy adopted by vessel 6 is: wait at the
Vessel 6 is an inbound vessel with its tide riding time window at
initial berth and use the initial ship loader, that is, stage6 = 2. The
[400,700] min and the adjusted inbound time at [430,516] min. The

9
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

Table 4
Adjusted scheme of 25 vessels.
Vessel In or Tidal time window Inbound Outbound Time of leaving the Berth Ship The operation time window of ship
number out (min) moment moment channel (min) loader loader (min)
(min) (min)

1 out — — 0 92 #2 SL2 [0,0]

3 out — — 80 183 #3 SL3 [0,0]
4 out — — 114 222 #8 SL8 [0, 57]
5 out — — 163 259 #10 SL10 [0,128]
7 out — — 206 286 #17 SL13 [0,151]
8 out — — 258 330 #7 SL7 [0,206]
9 out [200,500] — 286 349 #9 SL9 [0,252]
2 in — 389 — 471 #1 SL1 [513,902]
11 in — 413 — 492 #13 SL9 [570,867]
6 in [400,700] 430 — 516 #8 SL8 [640,799] ∪ [901,953]
12 in — 459 — 536 #10 SL10 [884,1 183]
13 in — 483 — 581 #2 SL2 [861,1 052]
15 in — 523 — 605 #4 SL4 [800,1 056]
17 in — 570 — 669 #17 SL13 [822,1 156]
10 in — 591 — 698 #3 SL3 [861,1 253]
16 out [720,820] — 740 812 #4 SL4 [230,662]
18 out — — 775 876 #16 SL12 [232,651]
19 out — — 822 929 #6 SL6 [344,768]
20 out — — 880 988 #14 SL10 [350,834]
14 out — — 906 1 010 #5 SL5 [354,599] ∪ [701,809]
22 out — — 960 1 068 #11 SL11 [530,920]
23 in — 1 095 — 1 198 #7 SL7 [1 261,1 437]
21 in [1 080,1 280] 1 116 — 1 227 #9 SL9 [1 255,1 461]
24 in — 1 142 — 1 249 #6 SL6 [1 281,1 633]
25 in — 1 266 — 1 408 #1 SL1 [1 428,2 041]

tide riding time window is still satisfied, and the safety of traveling
through the channel is ensured. Vessel 9 is an outbound vessel with the
adjusted outbound time of vessel 9 at [286,349] min, which is falling in
its tide riding time window [200,500] min. It proves that its safety in
traveling through the channel is ensured. It is verified that all vessels’
tide riding time windows are satisfied.
Vessel-grouping verification. In the adjusted scheme shown in
Table 4, the first outbound vessel-grouping number is {1,3,4,5,7,8,9}.
The first vessel in this group starts to leave the port at 0 min, and the last
vessel ends up leaving the port at 349 min. This group of vessels leave
the port at [0,360] min (the first time interval). The first inbound vessel
group number is {2,11,6,12,13,15,17,10}. The first vessel in this group
starts to enter the port at 389 min, and the last vessel enters at 698 min.
This group of vessels enters the port at [361,720] min (the second time
interval). It is verified that the time intervals of other vessel groups are
also ensured.
In summary, the above results show that the adjusted scheme solved
Fig. 7. Adjusted berthing time. by the proposed TSRG algorithm can effectively ensure the safe coor­
dinated optimization of vessels and ship loaders when ship loaders
failed. Therefore, the method for solving VSLPB in this paper can obtain
the valid adjusted scheme.

5.2. Algorithm performance analysis

To verify the performance of the TSRG algorithm, we conduct each

group of experiments 50 times, and the average results of each experi­
mental value are shown in Table 5. The average runtime of each algo­
rithm is shown in Fig. 9.
According to the experimental results in Table 5, columns TSRG, GA,
PSO, and DE, are the solution results of each algorithm, respectively, the
results show that the proposed TSRG algorithm out-performs the other
algorithms in terms of solution results, and can solve all the instances to
an optimal level. Column GAP shows the difference in the solution re­
sults between the TSRG algorithm and other algorithms. This column’s
values fluctuate between 13.1% and 15.6%, which indicates that the
Fig. 8. Adjusted ship loaders’ operation time. TSRG algorithm’s performance is stable. From the running time of each
algorithm in Fig. 9, it can be seen that the TSRG algorithm designed in
this paper has the shortest solution time with an average value of 3.0

10
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

Table 5 suitable for the problem and the model established in this paper, and
Experimental results. better fits the related disruption management techniques at various
Experiment number Objective function value (min) GAP*(%) stages and enables quick revision of the initial plan.
TSRG GA PSO DE

1 10 178 11 114 11 674 11 766 15.6 5.3. Algorithm sensitivity analysis

2 10 290 11 360 11 792 11 700 14.6
3 10 176 10 746 11 417 11 509 13.1 In this section, we study differences caused by the impact of different
4 10 383 11 504 11 889 11 743 14.5
disruption scenes. Sensitivity analysis is performed on the experiments.
5 16 902 18 609 19 150 19 302 14.2
6 17 757 19 621 20 119 20 421 15.0 The sensitivity analysis data is shown in Table 6, where the first set of
7 18 266 20 147 20 860 21 024 15.1 experiments is used to verify the sensitivity of the start time when the
8 10 307 11 327 11 698 11 884 15.3 ship loaders failure occurs. The second set is used to verify the sensitivity
9 12 540 13 970 14 271 14 459 15.3 of the time length of a failure to the ship loader acting on the outbound
10 7 289 8 076 8 186 8 302 13.9
vessels. The third set is used to verify the sensitivity of ship loaders at
11 7 389 8 431 8 431 8 431 14.1
12 15 244 16 677 17 317 17 531 15.0 different stages. The fourth set is used to verify the sensitivity of the time
13 20 076 22 003 22 907 23 107 15.1 length in a failure to the ship loaders acting on the inbound vessels. The
14 13 647 14 875 15 530 15 749 15.4 fifth set is used to verify the sensitivity of the number of failed ship
*
GAP = [(the maximum result of GA、PSO、DE)-(the result of TSRG)]/ (the loaders. The sixth set is used to verify the sensitivity of the number of
result of TSRG)*100%. affected inbound and outbound vessels.
Comparing the results of sensitivity analysis data in Table 6, it re­
veals that in the six experimental cases, the influence of the vessels’
directions inbound and outbound of the port on the initial schemes is the
smallest. However, the failure duration of ship loaders at different stages
of inbound and outbound vessels significantly influences the objective
function. Accordingly, it is necessary to make targeted adjustment
strategies according to the time and number of ship loader failures, to
generate an adjustment scheme with the minimum generalized cost
when disruptive factors occur.

6. Conclusions and further research

This paper studied disruption management in vessel traffic sched­

uling to solve the problem of ship loader failures in bulk ports. It con­
structed a disruption management model based on the disruption
management theory. In addition, a new TSRG algorithm was designed
according to the characteristics of the problem and the model. The
experimental and sensitive analyses were conducted, and the results
showed that the proposed method is effective and helpful in generating
Fig. 9. Runtime of each algorithm. adjusted schemes.
The results of multiple experiments showed the fitness of the TSRG
min. To verify the convergence speed of each algorithm, we take algorithm to the solution of the optimization model, which provided a
experiment 1 in Table 5 as the example depicts the convergence speed of new idea for the design of algorithms to solve problems of the same kind.
each algorithm in Fig. 10, which shows the convergence speed of the Through sensitivity analysis experiments it was found that ship loaders’
TSRG algorithm is superior to traditional algorithms. failure times at different stages had the most significant influence,
To sum up, these above solutions indicate that the TSRG algorithm is meaning that it is necessary to purposefully adjust the initial scheme
according to the specific stages in the actual operation process. In this

Table 6
Sensitivity analysis data.
Number of Number of Objective Number of Results GAP
experimental experiments function controlled (min) *
groups value experiments (%)
(min)

NO.1 1 10 359 3 10 176 1.8

2 10 290 4 10 383 0.9
NO.2 3 10 176 5 16 902 66.1
3 10 176 6 17 757 74.5
3 10 176 7 18 266 79.5
NO.3 8 10 307 9 12 540 21.7
10 7 289 11 7 398 1.5
NO.4 8 10 307 10 7 289 41.4
9 12 540 11 7 398 69.5
NO.5 12 15 244 3 10 176 49.8
12 15 244 8 10 307 47.9
NO.6 13 20 076 12 15 244 31.7
14 13 647 12 15 244 11.7
14 13 647 13 20 076 39.7
*
GAP=|(the result of controlled experiments)-(the result of experiments)| /
Fig. 10. Convergence speed of each algorithm. (the minimum result between experiments and controlled experiments)*100%.

11
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

paper, we proposed a TSRG algorithm. In the first stage, we did not [4] X.Y. Zhang, J. Lin, Z.J. Guo, T.S. Liu, Vessel transportation scheduling optimization
based on channel-berth coordination, Ocean Eng. 112 (2016) 145–152, https://
consider disruption management strategies and rounding constraints to
doi.org/10.1016/j.oceaneng.2015.12.011.
solve to obtain a feasible adjustment scheme quickly. In the second [5] X.Y. Zhang, R.J. Li, X. Chen, J.J. Li, C.B. Wang, Multi-object-based vessel traffic
stage, we add disruption management constraints and rounding con­ scheduling optimisation in a compound waterway of a large harbour, J. Navig. 72
straints to find the optimal adjustment scheme. Therefore, considering (2019) 609–627, https://doi.org/10.1017/S0373463318000863.
[6] X.Y. Zhang, Z.Q. Wang, Z.P. Deng, Ship traffic organization optimization of
the distinctive properties of the TSRG algorithm, the framework of this constrained two-way channel in huanghua port, Navig. China 43 (04) (2020) 1–7,
algorithm can be applied to other similar problems, especially in https://doi.org/10.3969/j.issn.1000-4653.2020.04.001.
adjusting the initial plan after being affected by disruption factors. These [7] J.J. Li, X.Y. Zhang, B.D. Yang, N.N. Wang, Vessel traffic scheduling optimization
for restricted channel in ports, Comput. Ind. Eng. 152 (2021) 107014, https://doi.
main findings provide useful insights for generating adjusted production org/10.1016/j.cie.2020.107014.
schemes in the factor of unexpected disruptions in production [8] J.J. Li, X.Y. Zhang, J.Y. Wang, J. Yan, Ship loading operation scheduling
operations. optimization in bulk cargo export terminals considering operation conflicts of
loading equipments, J. Dalian Marit. University 48 (02) (2022) 66–74, https://doi.
On the basis of directing the safe entry and leave of vessels, these org/10.16411/j.cnki.issn1006-7736.2022.02.008.
discoveries and findings can support the VTS center and increase the [9] S. Jia, L.X. Wu, Q. Meng, Joint scheduling of vessel traffic and pilots in seaport
port’s operational efficiency. In this study, we have proposed a frame­ waters, Transp. Sci. 54 (2020) 1495–1515, https://doi.org/10.1287/
trsc.2020.0990.
work, for solving the VSLPB, including the optimization model and the [10] S. Jia, C.L. Li, Z. Xu, Managing navigation channel traffic and anchorage area
TSRG algorithm. The model proposed in this paper for dealing with utilization of a container port, Transp. Sci. 53 (2019) 728–745, https://doi.org/
VSLPB can be applied to different types of ports and other types of 10.1287/trsc.2018.0879.
[11] M. Niu, Z.Y. Wang, Y.J. Li, H.X. Zheng, Berth allocation and ship scheduling
disruption factors after simple changes to the constraints of operation
integrated optimization considering the priority of berth in use, Chin. High
modes and disruption management strategies. In addition, the TSRG Technol. Letters 30 (09) (2022) 972–981, https://doi.org/10.3772/j.issn.1002-
algorithm framework can also be extended to other types of ports and 0470.2020.09.012.
handle other types of disruption factors. However, it still requires some [12] O. Abou Kasm, A. Diabat, M. Bierlaire, Vessel scheduling with pilotage and tugging
considerations, Transp. Res. Part E Logist. Transp. Rev. 148 (2021) 102231,
detailed modifications to meet different solving objectives for specific https://doi.org/10.1016/j.tre.2021.102231.
problems. Our method thus can be widely applied in many ports that [13] B.L. Liu, Z.C. Li, Y.D. Wang, A two-stage stochastic programming model for seaport
seek an efficient and helpful way to handle disruption factors. Addi­ berth and channel planning with uncertainties in ship arrival and handling times,
Transp. Res. Part E Logist. Transp. Rev. 167 (2022), 102919, https://doi.org/
tionally, it helps the port increase its competitiveness while making a 10.1016/j.tre.2022.102919.
significant contribution to enhancing the port’s capability for emer­ [14] J.H. Chen, Q.J. Xu, H. Zhang, Z. Wan, M.Z. Yu, Bilateral slot exchange and co-
gency response. allocation for liner alliance carriers of containerized maritime logistics, Adv. Eng.
Inf. 51 (2022), 101479, https://doi.org/10.1016/j.aei.2021.101479.
The limitation of this study lies in the fact that the VSLPB is [15] G. Yu, X.T. Qi. Disruption Management: Framework, Models and Applications,
addressed within one cycle. Moreover, this paper only investigates the World Sci. Publ. Co. Pte. Ltd, Singapore, 2004, https://doi.org/10.1142/5632.
impact of failed ship loaders, which is one of the uncertain factors in the [16] A. Baykasoglu, F.S. Karaslan, Solving comprehensive dynamic job shop scheduling
problem by using a GRASP-based approach, Int. J. Prod. Res. 55 (2017)
port production process. Thus, future studies could consider the exten­ 3308–3325, https://doi.org/10.1080/00207543.2017.1306134.
sions model and solution method of this study. In this case, a study on [17] T. Ning, Z. Wang, P. Zhang, T. Guo, Integrated optimization of disruption
the optimization vessel traffic scheduling problem of multiple plan cy­ management and scheduling for reducing carbon emission in manufacturing,
J. Clean Prod. 263 (2020) 121449, https://doi.org/10.1016/j.
cles considering other uncertain factors is of good research value.
jclepro.2020.121449.
Moreover, researching a new solution algorithm to improve its speed [18] N. Li, G. Chen, K. Govindan, Z.H. Jin, Disruption management for truck
and effect on the new problem remains highly valuable in the future appointment system at a container terminal: a green initiative, Transp. Res. Part D
investigation in the associated direction. Transp. Environ. 61 (2018) 261–273, https://doi.org/10.1016/j.trd.2015.12.014.
[19] A. Fischer, H. Nokhart, H. Olsen, K. Fagerholt, J.G. Rakke, M. Stalhane, Robust
planning and disruption management in roll-on roll-off liner shipping, Transp. Res.
Declaration of Competing Interest Part E Logist. Transp. Rev. 91 (2016) 51–67, https://doi.org/10.1016/j.
tre.2016.03.013.
[20] G.Y. Ke, M. Verma, A framework to managing disruption risk in rail-truck
The authors declare that they have no known competing financial intermodal transportation networks, Transp. Res. Part E Logist. Transp. Rev. 153
interests or personal relationships that could have appeared to influence (2021), 102340, https://doi.org/10.1016/j.tre.2021.102340.
the work reported in this paper. [21] T. Ning, X.P. Wang, Study on disruption management strategy of job-shop
scheduling problem based on prospect theory, J. Clean Prod. 194 (2018) 174–178,
https://doi.org/10.1016/j.jclepro.2018.05.139.
Data availability [22] J.H. Sun, L. Xu, Disruption management of multi-objective flexible job-shop
scheduling problem, in: the Poceedings of 2019 IEEE 7TH international conference
on computer science and network technology, 2019, pp. 98–102.
Data will be made available on request. [23] A.I. Malik, B. Sarkar, Disruption management in a constrained multi-product
imperfect production system, J. Manuf. Syst. 56 (2020) 227–240, https://doi.org/
Acknowledgments 10.1016/j.jmsy.2020.05.015.
[24] Y.W. Sang, J.P. Tan, W. Liu, A new many-objective green dynamic scheduling
disruption management approach for machining workshop based on green
This work is supported by the National Natural Science Foundation manufacturing, J. Clean Prod. 297 (2021), 126489, https://doi.org/10.1016/j.
of China (NO. 51779028), and the Dalian Science and Technology jclepro.2021.126489.
Innovation Fund (NO. 2022JJ12GX015). [25] F. Cheraghchi, I. Abualhaol, R. Falcon, R. Abielmona, B. Raahemi, E. Petriu,
Modeling the speed-based vessel schedule recovery problem using evolutionary
multiobjective optimization, Inf. Sci. (N Y) 448 (2018) 53–74, https://doi.org/
References 10.1016/j.ins.2018.03.013.
[26] X.L. Han, L.N. Qian, Z.Q. Lu, Dynamic disruption management for container
[1] Y.M. Chen, Review of dry bulk shipping market in 2021 and prospect in 2022, terminal resources allocation problem in periodic environment, J. Tongji
World Shipping 03 (2022) 1–8, https://doi.org/10.16176/j.cnki.21- University (Nat. Sci.) 46 (02) (2018) 264–272, https://doi.org/10.11908/j.
1284.2022.03.001. issn.0253-374x.2018.02.019.
[2] J.Y. Wang, X.Y. Zhang, B.D. Yang, W.Q. Guo, F.L. Liu, Optimization of Cargo [27] O.F. Abioye, M.A. Dulebenets, M. Kavoosi, J. Pasha, O. Theophilus, Vessel schedule
handling machinery allocation and ship traffic organization for bulk cargo ports, recovery in liner shipping: modeling alternative recovery options, IEEE Trans.
Navig. China 45 (04) (2022) 146–154, https://doi.org/10.3969/j.issn.1000- Intelligent Transp. Syst. 22 (2021) 6420–6434, https://doi.org/10.1109/
4653.2022.04.022. TITS.2020.2992120.
[3] Z. Cao, W.Y. Wang, Y. Jiang, X.L. Xu, Y.Z. Xu, Z.J. Guo, Joint berth allocation and [28] J.J. van der Steeg, M. Oudshoorn, N. Yorke-Smith, Berth planning and real-time
ship loader scheduling under the rotary loading mode in coal export terminals, disruption recovery: a simulation study for a tidal port, Flex. Serv. Manuf. J. 35
Transp. Res. Part B Methodol. 162 (2022) 229–260, https://doi.org/10.1016/j. (2023) 70–110, https://doi.org/10.1007/s10696-022-09473-8.
trb.2022.06.004. [29] A.R. De, J.W. Wang, M.K. Tiwari, Fuel bunker management strategies within
sustainable container shipping operation considering disruption and recovery

12
J. Wang et al. Advanced Engineering Informatics 56 (2023) 101989

policies, IEEE Trans. Eng. Manag. 68 (2021) 1089–1111, https://doi.org/10.1109/ agglomeration based on generalized cost, J. Adv. Transp. 2022 (2022) 5027133,
TEM.2019.2923342. https://doi.org/10.1155/2022/5027133.
[30] Z. Elmi, P. Singh, V.K. Meriga, K. Goniewicz, M. Borowska-Stefanska, [38] A. Rezaeipanah, S.S. Matoori, G. Ahmadi, A hybrid algorithm for the university
S. Wisniewski, M.A. Dulebenets, Uncertainties in liner shipping and ship schedule course timetabling problem using the improved parallel genetic algorithm and
recovery: a state-of-the-art review, J. Mar. Sci. Eng. 10 (5) (2022) 563, https://doi. local search, Appl. Intell. 51 (2021) 467–492, https://doi.org/10.1007/s10489-
org/10.3390/jmse10050563. 020-01833-x.
[31] J.H. Chen, C.L. Zhuang, C. Yang, Z. Wan, X. Zeng, J.Y. Yao, Fleet co-deployment for [39] H. Ansari, A. Aijaz, V. Kumar, M. Junejo, S. Shah, M. Saad, Z. Khaskheli, I.
liner shipping alliance: Vessel pool operation with uncertain demand, Ocean Coast Memon Qasim, Customizable algorithm for university timetabling and scheduling,
Manag. 214 (2021), 105923, https://doi.org/10.1016/j.ocecoaman.2021.105923. in: in: 8TH International Conference On Information And Communication
[32] J.H. Chen, J. Ye, A.T. Liu, Y.J. Fei, Z. Wan, X.T. Huang, Robust optimization of Technologies, 2019, pp. 113–117, https://doi.org/10.1109/
liner shipping alliance fleet scheduling with consideration of sulfur emission ICICT47744.2019.9001953.
restrictions and slot exchange, Ann. Oper. Res. (2022), https://doi.org/10.1007/ [40] T. Arbaoui, J.P. Boufflet, A. Moukrim, Lower bounds and compact mathematical
s10479-022-04590-x. formulations for spacing soft constraints for university examination timetabling
[33] H.G. Bo, X. Zhang, Y.T. Pan, Study on scheduling method based on disruption problems, Comput. Oper. Res. 106 (2019) 133–142, https://doi.org/10.1016/j.
management for hybrid N-wait flow shop, Operation Res. Manag. Sci. 25 (03) cor.2019.02.013.
(2016) 246–254, https://doi.org/10.12005/orms.2016.0109. [41] L.Q. Ding, Multimodal transport information sharing platform with mixed time
[34] G.G. Jiang, S.L. Wang, H.K. Lo, Z. Liang, Modeling cost variability in a bottleneck window constraints based on big data, J. Cloud Comput. 9 (1) (2020) 11, https://
model with degradable capacity, Transportmetrica B: Transport Dyn. 10 (2022) doi.org/10.1186/s13677-020-0153-8.
84–110, https://doi.org/10.1080/21680566.2021.1962430. [42] V.F. Rofatto, M.T. Matsuoka, I. Klein, M.R. Veronez, L.G. da Silveira, On the effects
[35] X. Li, Y. Luo, T.Q. Wang, P. Jia, H.B. Kuang, An integrated approach for optimizing of hard and soft equality constraints in the iterative outlier elimination procedure,
bi-modal transit networks fed by shared bikes, Transp. Res. Part E Logist. Transp. PLoS One 15 (8) (2020), e0238145, https://doi.org/10.1371/journal.
Rev. 141 (2020) 102016, https://doi.org/10.1016/j.tre.2020.102016. pone.0238145.
[36] F. Mueller, A. Aravazhi, A new generalized travel cost based connectivity metric [43] T.L. June, J.H. Obit, Y.B. Leau, J. Bolongkikit, Implementation of constraint
applied to Scandinavian airports, Transp. Res. Part D Transp. Environ. 81 (2020) programming and simulated annealing for examination timetabling problem, The
102280, https://doi.org/10.1016/j.trd.2020.102280. Computational Sci. Technol. (2019) 175–184, https://doi.org/10.1007/978-981-
[37] M. Yue, S.H. Ma, W. Zhou, X.F. Chen, Estimation Markov decision process of 13-2622-6_18.
multimodal trip chain between integrated transportation hubs in urban

13