Optimization of the Storage Sites for Export, Inbound and

Reorganize Containers by Timing Location

Mohammed Saleh

, Attariuas Hicham

, M. L. Ben Maâti

and Hatem Taha

Computer Science and System Engineering (CSSE), Abdelmalek Essaadi University, Av. Khenifra, Tetouan, Morocco

Research Operational and Statistic Applied(LAROSA), Abdelmalek Essaâdi University, Av. Khenifra, Tetouan, Morocco

Keywords: Container storage, container time, container terminal, artificial intelligence.

Abstract: Today, seaports subtend an increasing growth of containers stacking. Countries are striving to get the most

benefit from it and increase their share of this sector's resources, as well as optimizing their competitiveness.

Despite this increase, the ports suffer from many problems, including how to take the appropriate decision to

store and empty containers of various kinds. In this paper, we propose a method for storing containers at (El

Qasr El Saghir) terminal in Morocco, based on the hypothesis of time dynamics for choosing the optimal

location for the container in the yard. This hypothesis provides ideal storage locations for containers arranged

by time to avoid the accumulation of containers, reduce the forced movement of previously stored containers.

As well as facilitate the decision to relocate containers stored in the terminal to allow the provision of new

storage places, reducing time and operating cost. We propose to apply artificial intelligence (particularly

ANN) to this methodology (a case study on El Qasr El Saghir); for example, deciding for stacking containers

with different departure dates; because the parameters of our methodology are compatible with the ANN

algorithm.

1 INTRODUCTION

Without the ports, countries remain isolated from the

world. With it, communication horizons are opened,

the economy is strengthened and policies remain

independent, as it is one of the most important sectors

that support all industries, productive sectors, sustain

all sectors of the national economy which contributes

to promoting foreign trade and increasing

investments. Container terminals increase their

competitiveness and here it is clear that container

terminals face increasing pressure to maintain quality

of service to increase container productivity. The

increasing competition to improve efficiency at

container terminals has attracted the attention of

process analysts. According to (Stahlbock & Voß,

2007), (Murty et al., 2005) (Vis & de Koster, 2003),

overviews of operations and process problems at

container terminals and references on methods for

solving problems of these processes. In this paper, we

https://orcid.org/0000-0001-9394-1659

https://orcid.org/0000-0000-0000-0000

https://orcid.org/0000-0001-8025-5289

examine one of these important operational problems:

allocating storage locations for inbound containers,

export, and reshuffles.

According to (Salebeh, T. & Debo, A., 2020) the

storage yard of the container terminal is divided into

large storage areas called blocks. Figure 1 shows a

typical storage block in a station. Usually, these

containers are either outbound or inbound. Each type

has different characteristics (e.g., type, arrival

time/date, departure time/date, weight). Containers of

different sizes arrive through ships and are loaded

onto external trucks or trains called incoming

containers and on the other hand, export containers

are brought by trucks or other means of transport to

be loaded onto ships where the time of arrival of the

means of transport is unknown compared to the time

of arrival and departure of ships (Murty et al., 2005),

(Zhang et al., 2003).

The slot is the smallest storage unit in the location

for one container, where the container in this storage

Saleh, M., Hicham, A., Ben Maâti, M. and Taha, H.

Optimization of the Storage Sites for Export, Inbound, and Reorganize Containers by Timing Location.

DOI: 10.5220/0010733700003101

In Proceedings of the 2nd International Conference on Big Data, Modelling and Machine Learning (BML 2021), pages 333-338

ISBN: 978-989-758-559-3

333

position is naturally located with its longest side

either on the floor or precisely on another container

(of the same size). Containers are stacked one on top

of the other to form a column and the pillars are

placed together by the longest of them side by side to

form a bay.

The block is formed by placing many bays

together so that the container's doors are in one

direction at the front and the end in the other

direction. The locations of the same level in a block

form a tier. Columns that stack next to each other in

the block form a bay.

Joining two bays together in the block constitute

the Paired-Bay. In general, ternary (row, tire, bay)

specifies (r, t, b) coordinates of the (container)

location in the block. The length of the block is

determined based on the number of bays grouped by

the design of the storage yard and it is usually

approximately 20. The width (the number of rows)

and height (the number of tiers) of the block are

determined by the type of yard cranes used in the

block. The most common type of yard cranes, Rubber

Tire Gantry Cranes (RTGC), generally have a bridge

spanning six bays of containers stacked from four to

five levels high.

Figure 1: A container - Storage Paired-Bay in block

Railroad Structures Fixed Quay Cranes (RMGC),

which are larger than its predecessor, as it extends

over 13 rows of containers stacked at six heights and

there are other types of yard cranes, of different

widths and heights, for stacking (Salebeh, T. & Debo,

A., 2020).

Furthermore, we found in literary studies that

some container terminals adopt different storage

methods that meet their need. The large container

terminals have separate storage areas for export and

import containers and temporary areas designated for

containers that have just been emptied from or loaded

onto ships. Sometimes the export and import

containers are mixed into blocks, sometimes even in

bays. The adoption of these methods forces the station

to rearrange mixed containers based on species

distinction and the reorganization of future retrieval

operations. This increases the cost, time, effort of the

cranes; there is no doubt that this reduces the

productivity and competitiveness of the terminal.

This process (rearrangement) requires informed and

appropriate decision-making as it affects the

efficiency of the container yard's productivity (Kim

& Park, 2003). (Cobo, 2018) Present a general

dynamic pattern of container arrival and expected

handling effort for storage yard equipment by

proposing a methodology for allocating space for

outbound container storage and one place for each

outbound container.

According to (Voß et al., 2004), in the first part,

sites in the yard are reserved before the ship’s arrival

and then the containers are grouped, the weight

principle is adopted between the lighter and then

heavier containers to ensure the stability of the vessel.

In the second part, some terminals use online

procedures that do not require a pre-booking

reservation of space and that adopt scattered

planning. When a new container arrives, the berthing

location of the vessel is determined and then a good

location for the container is located in the specified

space in real-time, according to the category.

According to (Zhang et al., 2003) presented his

study on two levels; the first level worked to reduce

berthing by merging the work of the (Quayside

Crane) and storage yard cranes (RTGs) to scatter

containers between different blocks, whether

outgoing or incoming. The second level provides a

detailed solution to reduce the outbound container

transport distance between the block in the storage

space (block) and the ship's berthing place. In this

paper, we study in detail the problem of allocating

storage locations to determine the optimal location for

the container storage in the terminal yard.

According to known stowage schemes for ships

and container handling sequence, the problem of

optimized site allocation arises when import/export

containers arrive; these site allocation decisions are

made continuously in real-time as inbound/outbound

containers arrive and depart. This paper shows the

sequence of handling container storage according to

the characteristics referred to in the information of

arrival containers (Figure 3).

This method can deal with the problem of

container's locations in blocks, especially in bays,

because it enough to solve the problem of allocating

a site for the container in the block, allocating the site

in the bay and therefore the best slot in the bay is the

best slot for the entire block in the terminal. If the

activities in the storage yard are not properly

coordinated, ill-advised and costly rearrangements

BML 2021 - INTERNATIONAL CONFERENCE ON BIG DATA, MODELLING AND MACHINE LEARNING (BML’21)

334

are carried out and cause traffic congestion from the

carriers. Indeed this one-stack problem is very

complex (Avriel et al., 2000). Hence it is only natural

that studying the detailed site allocation problems of

a single stack is ineffective in the block.

In this approach, we first present a two-part

methodology of bay-timing. The first is the allocation

of double bays in the block and the second part

addresses the problem of allocating ideal sites for

storing containers. In general; the methodology

divides the blocks into Paired-Bays, in which the

incoming, outgoing, and refrigerated containers are

stored with taking into account the characteristics of

each type. The methodology is mainly based on the

exact time for each container location, which gives

the Paired-Bay the time mechanism. Then we

determine the ideal location for the container by the

previous content of the bays with the inclusion of the

variables related to the bay and the containers (Figure.

3). The methodology and its flexibility are discussed

in the section (Methodology).

Sharing, analyzing, and utilizing operational data

in real-time is essential, so we need the best ways to

use it; applying artificial intelligence is a perfect

solution for processing the big data generated by this

method or similar methods. Dealing with the huge

amount of data resulting from this problem is a

challenge in itself, in addition to making appropriate

and well-informed decisions. Undoubtedly, applying

artificial intelligence techniques to analyze this data

will be a much more intuitive and effective method

than any traditional method can do. The rest of the

paper is structured as follows: Section 2 presents the

methodology and scenarios. Section 3 presents

Artificial Intelligence. Finally, Section 4 concludes

the study.

2 THE METHODOLOGY OF BAY-

TIMING

When a ship arrives, the person responsible for

managing the terminal has information about the

containers (number, type, time/date of the arrival,

time/date of the departure, etc); he will allocate places

for these containers in a way that does not affect the

departure of the previous containers. This issue raised

an important question: How does one know what is

the date and time of receiving a container and how

can we reduce the time, cost, and operational effort?

Looking for an answer to this question ensures that

the container is not buried in the depths of the piles

when it is time to leave the yard, which reduces the

number of operations required to reach the container

in question. We propose the methodology of "Bay-

Timing"; It is a methodology for managing locations

within a block, which operates the location inside

bays according to time, specifically chronic two bays

in one block. This study demands that the container's

location in the storage yard be arranged according to

the departure time as if each slot of the container is

defined by date-time and therefore the processing of

containers storage begins according to the containers

with the farthest time range.

First, we formulate the following definitions:

Paired-Bay: is the even division of bays in a single

block.

Chronic-Bay: is a suggested method for linking the

bay with a specific date-time based on the time range

of the containers that are stored in the bay so that

maintains the arrangement of storage and disbursing

and also rearranging the containers.

Figure 2: Attributes used in the methodology

Reserved Bay: is a Paired-Bay that is reserved in

each block with the aim of flexibility to re-store the

containers in exceptional cases that do not comply

with the methodological standards for storage, such

as the one that expired the period allowed in the bay

designated for storage in the terminal yard and the

storage in it is processed on an individual basis.

The Paired-Bay reservation methodology

provides the ease and flexibility of storing a batch of

containers with different dates (known or unknown);

this is what we will try to clarify through this section;

the Paired-Bay methodology divides a block

consisting of six rows, four layers, eight bays (four

Paired-Bays (8/2 = 4). So, one block capacity = 6 * 4

* 8 = 192 locations. Figure 3. shows only one row

with four tiers in the length of four Paired-Bay.

Optimization of the Storage Sites for Export, Inbound, and Reorganize Containers by Timing Location

335

Figure 3: Paired-Bay in the block

In general, for each location, whether vacant or

full, there is a row number, tire number, and bay

number, and this location is denoted by the symbol

(P). We present the following notation that used in the

Paired-Bay methodology (Figure 3.):

N: the total number of slots. n: the rank of slots.

X: the Input(the container).

s: slot, r: row, t: tier, b: bay, pb: Paired-Bay,

Pe: Position_empty (vacancy slot)

fs: full_slot, tes: typical_empty_slot

ad: arrival_date, ld : left_date, t: container_type.

tg: time group of arrival container(probability

variable).

tc (Cz): the input(container) that located under the

empty_slot (the element of decision for stocking)

Pr: the probability of the time group.

Pr(ad)(tg): the probability of the time group of the

arrival container(tg) with arrival date(ad).

Pr(ld)(tc): the probability of the already storage

container(tc) with the left date(ld).

See Figure: 4. which represents the configuration

of the Paired-Bay method. We will process the

situation as in real-time, so we have 48 slots, 29 slots

are full, and 19 slots are empty (6 * 4 = 24 slots in

every bay, 48 in every Paired-Bay), we will illustrate

three scenarios based on Figure: 4.

The objective function which selects a location of

an arrival container and minimizes the total expected

number of relocation movements can be formulated

as follows:

𝑓

𝑥  

𝑃𝑒



𝑟1,𝑡,𝑏



,𝑡𝑔𝑙𝑑𝑡𝑐𝑙𝑑

𝑃𝑒



𝑟,𝑡,𝑏



, 𝑡𝑔𝑙𝑑  𝑡𝑐𝑙𝑑



(1)

..







..







𝑡𝑔 𝑛,𝑎𝑑,𝑙𝑑,𝑡



𝑃𝑒𝑟,𝑡,𝑏









(2)

2.1 Scenarios

For illustration, we assume one Paired-Bay of six

rows and four tires at most, as shown in Figure: 4.

There are 24 locations in the one bay and 48 locations

in the Paired-Bay, each location represented by a

square. In (Bay A), the reserved locations are 15 and

14 locations in (Bay B), vacant locations display

empty squares. The containers are stored in

chronological order according to the time dimension

rule (Farest date is First In) and in the export

according to the (Earliest Date First Out). The

numbers represent their distinct chronological order

in the squares. This method constitutes a flexible

construction for dynamic storage and container

retrieval; the presence of prior storage means that

there is a given configuration of the block, even if

partially, so storing containers based on time implies

that the retrieval sequence process will be well

defined and this, in turn, affects many aspects.

Figure 4: Empty and full slots in the Paired-Bay

Including reshuffling containers if required. So in

this Paired-Bay, we will have 19 empty spaces ready

to receive the new batch of containers. There are

several scenarios in which the method of the store,

departure, and rearrange/reshuffle will be represented

according to our methodology.

2.1.1 Scenario 1 (5 Containers Arrive with

the Unknown Departure Date)

Assuming the configuration shown in Figure 4.

specified on (25-Jan) and five new containers are

arrived on (27-Jan), but the date of departure is

unknown, so they are stored based on the permissible

period of stay of containers in the yard It is from (4 -

7 days) and also can be adapted according to the laws

in force at the station.

In (Bay a), we have container one on location (r2,

t1, Ba) being reshuffled on top of container three on

location Pe(r1, t4, Ba), in this case, this change

provides (4 locations) vacant to store containers (with

the same date), so that the permissible period (for

these containers whose departure date is unknown)

does not exceed the yard (max date) and do not store

the unknown departure date containers on top of

containers have a known departure date, after this

process the (Bay a) considered full, noting that the

remaining vacancies in the (Bay a) It cannot be stored

because the containers under these vacant places have

a known departure date and we cannot put containers

on top of them whose departure date is unknown.

In (Bay b), we will have the same condition that

container 2 in position (r2, t1, Bb) will be reshuffled

BML 2021 - INTERNATIONAL CONFERENCE ON BIG DATA, MODELLING AND MACHINE LEARNING (BML’21)

336

to location Pe(r1, t4, Bb) over the site (r1, t3, Bb)

Full_slot(fs), here also four vacancies were provided

ideal locations for storing the remaining containers

unknown departure date (r2, t, Bb), we will only

operate one location. (Scenario drawing see Figure

(5)). See Figure 5. Scenario 1.

2.1.2 Scenario 2 (5 Containers Arrive with

Known Departure Date)

We assume the arrival of five new containers on (27-

Jan) with known dates, grouped as two containers on

(3-Feb), one container on (1-Feb), and two containers

on (5-Feb). To store the two departing containers on

(5-Feb), either they are stored contrary to the

chronicle range (non-adherence to the methodology)

or in case of searching for a typical location based on

the methodology, rearrange the row(r2) in the bay

(Ba) or in (Bb) is the optimal solution for storing

these two containers. In this case, the

identification/selection of optimal locations for

storing the remaining two containers departing on (3-

Feb) will be clear and relatively fast, which will be on

top of the previous containers (r2, t3 & t4, Ba or Bb).

So the remaining container that has a departure

date (1-Feb) has four vacancies locations according to

the current configuration, which is respectively (r1, t,

Ba, or Bb), but according to the chronic method, the

ideal location for this container is (1-Feb).

However, according to the chronic method, the

ideal location for this container (1-Feb) is (r2, t1, Bb)

being empty. See Figure 5. Scenario 2.

2.1.3 Scenario 3 (Reshuffled-containers

(Reserved Bay))

The re-Reshuffled scenario illustrates how to provide

vacant locations in the event of new containers arrival

with reducing the operating cost. This process is

carried out in the spare time of the equipment.

According to the previous figure (Figure 4.), the

container (24) that existed at the location (r6, t1, Ba)

should move to the Reserved-Bay according to

methodology. Because the methodology cannot deal

with it due to exceeding the permissible period for

staying in the yard and also emptying location (r2, t1,

Ba) to allow storing new containers to which the

balancing methodology applies. The ideal location in

this case for the pre-existing container (r2, t1, Ba ) Is

the position (r1, t4, Ba) according to the time.

This process will provide eight vacant locations,

which are (r2 & r6, t, Ba); likewise, the column (r2 &

r6) in (Bay b) will be rearranged so that the container

(20) (r6, t1, Bb) will be moved into the Reserved-Bay

because it has exceeded the period; this process

provides four vacancies in the column (r6, t, Bb); the

column marked with row r2 is rearranged so the

container (r2, t1, Bb) takes the vacant position (r1, t4,

Bb) so we will get four vacant locations as well.

In general, the rearrangement according to the

Paired-Bay methodology provides perfect vacant

columns (r, t) within the Paired-Bay (4 * 4 = 16),

enabling us to deal with any new storage with any

period because it is empty.

Figure 4: An illustration of the Scenarios

According to the parid-pay methodology and

according to the data in the column (r5, t, Ba) and the

column (r3, t, Bb) with a sequential date 25-27/Jan,

the two columns can be combined according to their

time-range sequence and this provides an ideal vacant

column for storing new containers. See Figure 5.

Scenario 3, 4. Previous scenarios reinforce our

approach, merging two bays facilitates re-mixing and

rearrangement of containers during equipment's spare

time and provides good flexibility as well; as the two

bays will share vacant sites in this situation.

According to (Salebeh, T. & Debo, A., 2020), any

container that has not been decided upon (for

whatever reason), for example not being disbursed on

time, is either moved to another location outside the

yard, or a bay is allocated for containers that are not

disbursed on time, according to the internal laws of

container terminal management, a delay fine is

imposed on these cases.

3 ARTIFICIAL INTELLIGENCE -

ARTIFICIAL NEURAL

NETWORKS

Artificial intelligence brings about change in

management and operation processes. It can reduce

human errors and make operations faster. Therefore,

AI itself is part of a broader process of automation

and improvement of port operations. The vision of

artificial intelligence is represented by rational steps

and organizational boundaries that help to improve

key performance indicators; it serves to achieve

Optimization of the Storage Sites for Export, Inbound, and Reorganize Containers by Timing Location

337

common goals among the parties to artificial

intelligence and promotes beneficial opportunities for

all parties in real-time through prior knowledge of

container movement patterns. We are looking

forward to applying artificial intelligence, especially

neural networks, in addressing the risk and errors of

specifying vacant sites and choosing the optimal

location for the container and this is what we will try

to study in the next research by applying it to the data

of the station (El Qasr El Saghir - Morocco) as a case

study. In this paper, we focus mainly on the proposed

methodology, and the proposal of neural networks in

addressing this problem falls within the scope of

planned future work because of its good ability to

identify patterns and the diversity of methods of real-

time prediction in the ideal empty location. Artificial

Neural Networks (ANNs) are computer programs

whose main goal is to simulate how the human brain

processes information. ANN networks learn (or are

trained) through experience with appropriate learning

models and pool their knowledge by discovering

patterns and relationships in data reference

(Agatonovic-Kustrin & Beresford, 2000).

4 CONCLUSIONS

This study looked at the problem of storing containers

in real-time at the container terminal. The problem

was identified and a two-stage practical solution

approach was proposed. The first phase, dividing the

yard block into dual bays, involves the use of a

proposed methodology for bay timing, while the

time-bound container group approach is used in the

second phase, which specifies the optimal location of

the containers. The results of this study can be

practically implemented by the El Qasr El Saghir

station in Morocco. Through the simple scenarios, the

possibility of the methodology appears in helping

decision-makers store each container and track

storage conditions. When the proposed method is

applied to the reality in the station, it results in large,

repetitive, and diverse data that require collection,

purification, and processing to prove the effectiveness

of the proposed method. In the future paper, we will

process this big data using artificial intelligence to

verify the effectiveness of the method.

ACKNOWLEDGEMENTS

This work was supported by the Laboratory of

Computer Science and System Engineering (CSSE)

in the Faculty of Sciences - Abdelmalek Essaâdi

University.

REFERENCES

Agatonovic-Kustrin, S., & Beresford, R. (2000). Basic

concepts of artificial neural network (ANN) modeling

and its application in pharmaceutical research. Journal

of Pharmaceutical and Biomedical Analysis, 22(5),

717–727. https://doi.org/10.1016/S0731-

7085(99)00272-1

Avriel, M., Penn, M., & Shpirer, N. (2000). Container ship

stowage problem: Complexity and connection to the

coloring of circle graphs. Discrete Applied

Mathematics, 103(1–3), 271–279.

https://doi.org/10.1016/S0166-218X(99)00245-0

Cobo, P. T. (2018). Optimization of yard operations in

container terminals from an energy efficiency

approach. Undefined. /paper/Optimization-of-yard-

operations-in-container-from-

Cobo/f61c1de797c769f015849513e021bd6af5cadbf2

Kim, K. H., & Park, K. T. (2003). A note on a dynamic

space-allocation method for outbound containers.

European Journal of Operational Research, 148(1),

92–101. https://doi.org/10.1016/S0377-

2217(02)00333-8

Murty, K. G., Liu, J., Wan, Y., & Linn, R. (2005). A

decision support system for operations in a container

terminal. Decision Support Systems, 39(3), 309–332.

https://doi.org/10.1016/j.dss.2003.11.002

Salebeh, T. & Debo, A. (2020). Study of Service Levels in

Lattakia International Container Terminal LICT.

Stahlbock, R., & Voß, S. (2007). Operations research at

container terminals: A literature update. OR Spectrum,

30(1), 1–52. https://doi.org/10.1007/s00291-007-0100-

Vis, I. F. A., & de Koster, R. (2003). Transshipment of

containers at a container terminal: An overview.

European Journal of Operational Research, 147(1), 1–

16. https://doi.org/10.1016/S0377-2217(02)00293-X

Voß, S., Stahlbock, R., & Steenken, D. (2004). Container

terminal operation and operations research—A

classification and literature review. OR Spectrum,

26(1), 3–49. https://doi.org/10.1007/s00291-003-0157-

Zhang, C., Liu, J., Wan, Y., Murty, K. G., & Linn, R. J.

(2003). Storage space allocation in container terminals.

Transportation Research Part B: Methodological,

37(10), 883–903. https://doi.org/10.1016/S0191-

2615(02)00089-9

BML 2021 - INTERNATIONAL CONFERENCE ON BIG DATA, MODELLING AND MACHINE LEARNING (BML’21)

338