On the Implementation of Simulation-based on Representation by

Rules Methodology to Plan Port Logistics Operation

Anibal Tavares de Azevedo

Applied Science School, University of Campinas, Rua Pedro Zaccaria, 1300, Limeira, Brazil

Keywords: Simulation, Port Logistic Operation, Representation by Rules.

Abstract: This paper proposes a fast and agile methodology that enables easy incorporation of different business rules

for planning port logistics operations. The rules could be embedded in a simplified discrete-event and multi-

agent simulation scheme that clearly shows the impacts of different rules in each part of the port operation.

Furthermore, the developed approach enables an analysis of how operational decisions in one port could

affect subsequent ports.

https://orcid.org/0000-0003-1678-7795

1 INTRODUCTION

The introduction of containers into commerce

among countries enabled a higher degree of

efficiency in supply chains. Although, uncoordinated

port operations could lead to congestion or even

worst supply chain disruption, as noted by (Loh and

Thai, 2014): “The increased importance of ports

makes them a vulnerable node as a port-related

disruption can generate domino effect on a network

of supply chains. The vulnerability of ports thus

needs to be addressed to ensure the functionality of

ports and enhance supply chain resilience.”

Furthermore, unpredicted events like Evergreen

blocking Suez Canal could lead to a sudden increase

in demand for port operations. According to

(Leonard, 2021): “That is going to have a big impact

on the already stressed supply chain”.

1.1 Solution and Literature Review

A solution to deal with such kind of unpredicted

events, according to (Cholteeva, 2021), is: “Supply

chains will have to be agile, nimble and flexible to

counter problems like these. Overnight, batch-based

processing and planning simply won’t cut it. Real-

time, fully integrated and digitized supply chains are

needed to reduce the impact of events like these to a

minimum”.

The key concept to avoid uncoordinated port

operations is the ability to fast adapt and react to

unpredicted events considering all parts of the

system (Zavala-Alcívar et al., 2020). Although,

articles in literature integrate only some problems of

container ports: berth allocation and quay crane

scheduling (Bierwirth and Meisel, 2010; Yang et al,

2012); others integrate the allocation of berths and

yard operation planning (Hendriks et al., 2013); and

some integrate empty container allocation in the

yard with vehicle routing (Braekers et al., 2013).

From previous articles, it is possible to see that

integration is limited to propose specific models for

one stage or a combination for a few of them. This

could lead to planning without coordination among

stages and some could behave as bottlenecks for the

container flow through the port.

1.2 Contribution

The solution should encompass, as observed by

(Zeng and Yang, 2009), the following: “Many

complex systems such as manufacturing, supply

chain, and container terminals are too complex to be

modeled analytically. Discrete event simulation has

been a useful tool for evaluating the performance of

such systems. However, simulation can only

evaluate a given design, not provide more

optimization functions. Therefore, the integration of

simulation and optimization is need.".

438

Tavares de Azevedo, A.

On the Implementation of Simulation-based on Representation by Rules Methodology to Plan Port Logistics Operation.

DOI: 10.5220/0010602404380444

In Proceedings of the 11th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2021), pages 438-444

ISBN: 978-989-758-528-9

In this sense, we developed a methodology to

face random events and fast propose and evaluate a

new set of decisions for all stages of a complex

system like a container port is.

Based on simulation-optimization that employs

representation by Rules for port logistics (Azevedo

et al, 2018; Araújo et al., 2016, Azevedo et al.,

2014) for one or two stages in a port, we created a

general methodology, that could be easily adapted

and employed in all container port stages through

the application of four main steps:

Representation by Rules Methodology

I. Map agents, and their relations through the

system process. Select agents that will be

studied and the areas where they operate;

II. Describe all possible agent operations to

complete a process. For each operation

describe the rules that could be employed;

III. Combine rules into a hybrid simulation based

on discrete event simulation and agent

simulation;

IV. Test a different combination of rules and

identify the one that should be adopted

considering the best performance of the

overall system.

The next sections will describe in detail how to

employ this methodology for a specific part of a

container port operation.

2 THE PROBLEM

The step (I) to employ the Representation by Rules

methodology is to perform an adequate mapping of

the relevant agents and their relations that enable the

port processes.

In a container port, there are three main processes

related to container flow: import flow (IF), export

flow (EF), transshipment, or temporary flow (TF).

Figure 1 identifies the following port agents:

container ship, quay crane, vehicle, yard crane, and

yard storage blocks.

Figure 1: Container port agents and their relations.

Figure 1 also makes clear the relations among

agents through three main processes. One example is

that a quay crane will unload (IF or TF) or load (EF

or TF) a container into a container ship.

Two agents will be selected to illustrate the

methodology application: Container ship and quay

cranes which operate at the berth area and quay area.

2.1 Selection of Agents and Stages

Step (II) consists of the mapping agent’s cooperation

to perform a process. In the context of a container

port, ship unloading and loading operations could be

described as done in Figure 2.

Some process constraints should be observed:

1. The container ship stowage plan: this is a

ship map indicating where each container

will be or is positioned inside a container

ship;

2. Proper scheduling of two or more quay

cranes should consider the movement of

rail-mounted equipment, and keep a

minimum distance to avoid collisions.

Figure 2: Container ship and quay crane cooperation to

unloading and loading cargo process.

2.2 Creation of Agent’s Rules

From Figure 2 it is possible to determine which rules

could be related with agent:

1. Container ship:

1.1. Unloading rules (CUR): determine which

containers should be unloaded. Two

examples are: remove only the containers

whose destination is the current port, or

also remove more containers to reduce the

number of future blocking containers.

1.2. Loading rules (CLR): should determine the

position where a container will be stored.

On the Implementation of Simulation-based on Representation by Rules Methodology to Plan Port Logistics Operation

439

Since containers on a ship are organized in

stacks, depending on the position selected

this could result in more or less blocking

containers. A blocking container hampers a

container that is downwards on the same

stack and should be unloaded.

Figure 3 describes how the ship organizes containers

in stacks. Furthermore, stacks are organized in pairs

of odd bays (for containers of 20

feet) or even bays

(for containers of 40

feet).

Figure 3: Arrangement of containers in a ship.

Figure 4 details the organization in the 13

bay in

terms of rows and columns.

Figure 4: Arrangement of containers in 13

bay.

Each square with a number indicates that space is

occupied with a container. The number inside a

square specifies the destination port of a container.

Furthermore, Figure 4 illustrates the container

in position (row, column) = (4, 1), which destination

port is 5, is a blocking container. This occurs

because, once the ship arrives at port 2, this

container should be unloaded to allow the target

container on position (3, 1) to be unloaded.

Once the container ship rules determine which

containers will be moved in each bay, the total

workload per bay will be computed. Then, quay

cranes will employ rules to compute the total time

necessary to perform all operations.

2. Quay cranes:

2.1 Initial position rules (QIR): determine

which position is the more adequate to start

the quay cranes work. No matter is for

unloading, loading, or both operations on

the ship.

2.2 Movement rules (QMR): should observe

physical constraints like one quay crane

could not overpass another one since they

are rail-mounted, and should keep a secure

distance to avoid accidents.

Figure 5 illustrates in detail two quay cranes

allocation in terms of 20’ bays workload.

Figure 5: Initial position of quay cranes 1 and 2.

It is also important to stress that container ship rules

consider the number of containers moved and quay

cranes rules are related to the total time to perform

movements. One manner to merge rules is to employ

a simplified algorithm base on discrete-event

simulation and multi-agent simulation.

2.3 Combining Rules using Simulation

Instead of using a complete simulation framework

that encompasses discrete-event and multi-agent

with a great computational burden, we created an

algorithm that considers the main aspects of both

paradigms with minimal coding. It is also important

to stress that this scheme could be expanded and

generalized for a higher number of agents (ships and

quay cranes) or areas (Yard Area, for example).

Furthermore, the algorithm is a form of a

function in which parameters are the rules that will

be applied for an agent. The algorithm returns the

total time necessary to perform specified operations.

In this case, it means the necessary time to perform

operations to unload and load containers in a ship in

one port.

Another important aspect is that the container

ship is represented by a vector of matrices B whose

element values must be integer numbers

representing the final port destination if space is

occupied by a container, or zero if the space is

empty. This matrix representation encompasses the

ship organization of containers as described in

Figures 3 and 4. The scalar p indicates which port is

the current one. Vector TB indicates which blocking

containers will be unloaded temporarily that will be

SIMULTECH 2021 - 11th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

440

reloaded with containers from the current port from

vector TP. The vector VB is the total workload per

container ship bay that will be used to compute the

total time necessary to perform all unloading or

loading tasks using quay cranes. The vector VB

encompasses the scheme described in Figure 5.

Simulation(B, CUR, CLR, QIR, QMR)

Begin

# Unloading operations.

[B, VB, T] = unloading(B, CUR, p)

Tmov = TotalTime(VB, QIR, QMR)

# Loading operations.

[B, VB, T] = loading(B, CLR, p)

Tmov = Tmov + TotalTime(VB,QIR,QMR)

return Tmov

End

2.4 Testing Combination of Rules

Figure 2 describes each agent operation to ensure

that the unloading and loading cargo processes will

be done. These processes could be related to agent

rules as done in Table 1.

Table 1: Agents, their operations, and related possible

rules.

Agent Operation Rules

Ship Unload CUR1, CUR2

Load CLR1, CLR2

Quay Crane Initial position QR1, QR2

Move

Work

For ships, there are two sets of possible rules:

unloading and loading rules in the sense detailed in

subsection 2.2.

For the cranes, although there are three possible

operations it is possible to create just one set of

rules. This could be done for the following reasons:

A. Quay crane movement is restricted to be

one-directional without losing the

possibility to achieve an optimal solution

(Chen et al., 2014); Furthermore,

movement rules (QMR) described in

subsection 2.2 will be followed;

B. Quay cranes are assumed, without loss of

generality, to be equal and with a

deterministic processing time.

From (A) and (B), the only difference between quay

cranes will be the decision on the initial position of

each quay crane.

Since the set of rules for each operation is

defined, the decision problem could be simplified to

choose the best combination of rules that will

produce the minimal container ship stay time in port.

Tables 2 and 3 give the eight possible combinations

of rules for the set of rules described in Table 1.

Table 2: First four possible combinations of rules.

Combination Unloading Loading Crane

1 CUR1 CLR1 QR1

2 QR2

3 CLR2 QR1

4 QR2

Table 3: Last four possible combinations of rules.

Combination Unloading Loading Crane

5 CUR2 CLR1 QR1

6 QR2

7 CLR2 QR1

8 QR2

Tables 2 and 3 could be summarized in terms of

sequences of numbers as done in Table 4.

Table 4: Summarizing Table 2 and 3 in terms of numbers.

Combination [1, 2, 3, 4, 5, 6, 7, 8, 9]

Unloading [1, 1, 1, 1, 2, 2, 2, 2]

Loading [1, 1, 2, 2, 1, 1, 2, 2]

Crane [1, 2, 1, 2, 1, 2, 1, 2]

One important computational aspect is how to use

combination numbers to determine which rule

should be applied. For this purpose, Equation (1) is

useful.

(math.floor(i/N)) % M + 1 (1)

where: i is an integer number, N is the number of

repetitions of the digit that belongs to the set {1, …,

M}.

From Equation (1) is possible to convert the

numbers on the combination sequence into other

sequences:

I. Unloading sequence: N = 4 and M = 2

will produce: [1, 1, 1, 1, 2, 2, 2, 2];

II. Loading sequence: N = 2 and M = 2

will produce: [1, 1, 2, 2, 1, 1, 2, 2];

III. Crane sequence: N = 1 and M = 2 will

produce: [1, 2, 1, 2, 1, 2, 1, 2].

On the Implementation of Simulation-based on Representation by Rules Methodology to Plan Port Logistics Operation

441

This encoding helps to provide the following

function that translates an integer number (from

combination sequence) into other sequence numbers.

def translateNum2Rules(i,nrules):

nr = len(nrules)

x = [0]*nr

for t in range(0,nr):

N = np.prod(nrules[0:t])

I = (math.floor(s/N)) % nrules[t]+1

x[t] = i

return x

3 SIMULATION EXAMPLE

Consider a simplified version of a ship with an

initial cargo, for didactic purposes, are as shown in

Figure 6.

Figure 6: Simplified container ship.

The combination of rules that will be performed is 1

which means, according to Table 2, CUR1, CLR1,

QR1.

Since the container ship arrived at port 2, it is

necessary to unload all import containers with the

number 2 and ones that are blocking its movement

(CUR1).

Another unloading rule could consider removing

all containers that could be blocking containers as

the container 3 in bay 1 (CUR2).

Observe that the unloading rule will generate the

workload output for quay cranes operation.

To remove containers from the ship two quay

cranes will be employed using the QR1 rule which

initial position in bays will be the one presented in

Figure 7.

Figure 7: Initial position using QR1.

Considering that quay cranes could work in parallel

without a collision and the time necessary to move

one crane to another bay is one workload time, then

the total time to perform unloading operations will

be 5 units of time.

The next step is to plan the spaces on the ship

where export containers will be loaded. Suppose

that the number of containers for each port

destination is as detailed in Table 5.

Table 5: Export containers in port 2.

Destination 3 4 5

# 2 2 1

Additionally, the blocking container, which

destination is port 4, that was moved during

unloading of container 2 in bay 1, are already

counted in Table 5.

The loading rule chosen for the next step is

CLR1 which means start to search for a position in

ship containers with the lowest port destination (3).

The position should avoid producing blocking

containers and begin from the bay with the lowest

number to highest number filling as shown in Figure

Figure 8: Future ship arrangement after loading containers

to port 3.

Now the containers which port destination is 4 will

be distributed along bays, but they turn to be

blocking containers as shown in Figure 9.

Figure 9: Future ship arrangement after loading containers

to port 4.

Finally, the position for the container to port 5 is

also determined.

SIMULTECH 2021 - 11th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

442

Figure 10: Future ship arrangement after loading container

After determining all positions of containers that

will be loaded, the workload per bay is computed

and will be the same as shown in Figure 7. As a

result, the total time to perform loading operations

will be 5 units of time. Another combination of

rules, like one with LR2 which is the reverse of

LR1, could lead to another total stay time value.

The process to choose the best combination is to

evaluate all possible combinations and pick the one

with the lowest total stay time for the ship.

Although, this task could be a computational burden

task for a decision in multiple ports as shown in

Section 4.

4 EXTENDING FOR MULTIPLE

PORTS

Section 3 showed how to evaluate one possible

combination of rules using a simulation that

produced total time to perform operations in one

port.

Additionally, the methodology could be applied

for all ports that a ship will pass during its travel.

Figure 11 explains how to do an integrated

evaluation for more than one port.

Figure 11: Extending the methodology for more than one

port.

Employing the combination of rules 1, as described

in section 3, the container ship arrangement will

change after port 2 as shown in Figure 11.

The extension made in Figure 11 could be

replicated between ports 3 and 4, and ports 4 and 5.

By doing this, it is possible to measure the impact of

chose different rules through several ports.

Although this extension is interesting for a more

detailed evaluation it will bring more complexity to

simulate all combinations of rules as will be shown

in Section 5.

5 MULTI-PORT OPTIMIZATION

PERSPECTIVE

Section 4 showed how to evaluate a solution for

several ports. A general representation that enables

the test of several combinations of rules is shown in

Figure 12.

Figure 12: Using rules for more than one port.

In Figure 12, only the combination of rules 1 was

applied for every port. It is important to stress that

this is one possible combination in 5

= 390.625

alternatives of combinations.

Furthermore, as the number of agents and

operations increases, the search for an optimal

combination of rules could be a computational

burden task.

One possibility is to employ flexible methods to

deal with combinatorial problems like metaheuristics

as genetic algorithms, for example (Azevedo et al.,

2014; Azevedo et al. 2018).

6 CONCLUSIONS

We extended the representation by rules approach to

be a methodology to tackle any port process related

to any agent with its corresponding operations.

Furthermore, we illustrated how this approach

could be used to evaluate, for example, the total time

that a container ship will take to perform all

unloading and loading operations through ports.

Ideas of future works are:

• It could all be extended to consider more

port agents and their operations in more

stages like the Yard (Zhen et al., 2013) ;

On the Implementation of Simulation-based on Representation by Rules Methodology to Plan Port Logistics Operation

443

• The proposed simulation modeling could

be coupled with an optimization tool for a

combinatorial problem on a simulation-

optimization scheme;

• An automatic generator of rules could be

created for each operation;

• A Monte Carlo approach could be

incorporated to tackle uncertainty on

demand variation.

REFERENCES

Araújo, E. J., Chaves, A. A., Salles, L. L. N., Azevedo, A.

T. (2016). Pareto clustering search applied for 3D

container ship loading plan problem, Expert Systems

with Applications, 44, 50-57.

Azevedo, A. T., Ribeiro, C. M., Sena, G. J., Chaves, A.

A., Salles, L. L. N., Moretti, A. C. (2014). Solving the

3D Container ship loading planning problem by

representation by rules and meta-heuristics,

International Journal of Data Analysis Techniques

and Strategies, 6(3), 228-260.

Azevedo, A. T., Salles L. L. N., Chaves, A. A., Moretti, A.

C. (2018). Solving the 3D stowage planning problem

integrated with the quay crane scheduling problem by

representation by rules and genetic algorithm, Applied

Soft Computing, 65, 495-516.

Bierwirth, C., Meisel, F. (2010), A survey of berth

allocation and quay crane scheduling problems in

container terminals, European Journal of Operational

Research 202 (3), 615 – 627.

Braekers, K., Caris, A., Janssens, G. (2013) Integrated

planning of loaded and empty container movements,

OR Spectrum 35 (2), 457–478.

Chen, J., Leeb, D.-H., Gohc, M., (2014) An effective

mathematical formulation for the unidirectional

cluster-based quay crane scheduling problem,

European Journal of Operational Research, Vol. 232,

No. 1, 198-208.

Cholteeva, Y. (2021). Offshore explainer: The Evergreen

Marine blockage at the Suez Canal, https://www.

offshore-technology.com/features/offshore-explainer-

the-evergreen-marine-blockage-at-the-suez-canal/,

26th, March.

Hendriks, M. P. M., Lefeber, E., Udding, J. T. (2013)

Simultaneous berth allocation and yard planning at

tactical level, OR Spectrum 35 (2), 441–456.

Leonard, M. (2021). Ever Given and the Suez Canal: A

list of affected ships and what delays mean for

shippers, https://www.supplychaindive.com/news/ever-

given-suez-canal-evergreen-blocked-shippers-

cargo/597295/, 21th, March.

Loh, H. S., Thai V. V. (2014). Managing Port-Related

Supply Chain Disruptions: A Conceptual Paper, The

Asian Journal of Shipping and Logistics, 30(1), 97-

116.

Yang, C., Wang, X., Li, Z. (2012) An optimization

approach for coupling problem of berth allocation

problem and quay crane assignment in container

terminal, Computers & Industrial Engineering 63 (1),

243–253.

Zavala-Alcívar, A., Verdecho, M.-J., Alfaro-Saiz, J.-J.

(2020). A Conceptual Framework to Manage

Resilience and Increase Sustainability in the Supply

Chain, Sustainability, 12, 6300, 2071-1050.

Zeng, Q., Yang, Z. (2009) Integrating simulation and

optimization to schedule loading operations in

container terminals, Computers & Operations

Research 36 (6), 1935–1944.

Zhen, L., Jiang, X., Lee, L.H., Chew, E. P. (2013). A

Review on Yard Management in Container Terminals,

Vol.12 No.4, pp. 289-305, Industrial Engineering &

Management Systems, Korean Institute of Industrial

Engineers.

SIMULTECH 2021 - 11th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

444