ScienceDirect

Available online at www.sciencedirect.com

ScienceDirect
Transportation Research Procedia 00 (2018) 000–000
Available online at www.sciencedirect.com www.elsevier.com/locate/procedia
Transportation Research Procedia 00 (2018) 000–000

ScienceDirect www.elsevier.com/locate/procedia

Transportation Research Procedia 39 (2019) 92–103

www.elsevier.com/locate/procedia
Green Cities 2018
A systems analysis of empty Cities 2018logistics – a case study of
container
Green
A systems analysis ofMelbourne,
empty containerAustralialogistics – a case study of
Melbourne,
Lucille Gusaha, Robert Australia
Cameron-Rogers a
, & Russell G. Thompsona*
a The University ofa,Melbourne, Australia a*
Lucille Gusah , RobertEngineering,
Cameron-Rogers & Russell G.3010,
Thompson
a
Department of Infrastructure Parkville,

a
Department of Infrastructure Engineering, The University of Melbourne, Parkville, 3010, Australia
Abstract

Containerised
Abstract freight is a key component of global and urban trade. Empty containers are a necessary outcome of such trade and
have operational costs close to that of full containers yet are not a source of revenue as the latter is. A literature review revealed
regional repositioning
Containerised freight istoa be
keya component
complex problem
of globaldue andto urban
a multitude of interacting
trade. Empty parts
containers areand distributed
a necessary decision
outcome of making among
such trade and
stakeholders
have operationalwhich have
costs conflicting
close to that ofobjectives. The yet
full containers goalareisnot
to aoptimise
source oftherevenue
activities related
as the lattertois.urban goods movements
A literature and
review revealed
reduce the
regional total social cost
repositioning to beofathese
complexactivities.
problemTheduepurpose of this study
to a multitude is to develop
of interacting a detailed
parts description
and distributed and understanding
decision making among of
the empty container
stakeholders movements
which have in Melbourne
conflicting objectives.using
The agoalmodelling approach
is to optimise thedesigned
activitiesto related
represent complex
to urban systems.
goods This paper
movements and
explores
reduce thethe cause
total andcost
social effects of the
of these land-side
activities. Theempty
purposecontainer
of thisrepositioning. Then athe
study is to develop key characteristics
detailed description and of understanding
the system were of
captured
the emptyinto a conceptual
container movementsmodelinthat documents
Melbourne andaarticulates
using modellingthe logisticsdesigned
approach processes, interactions,
to represent power-relations,
complex systems. This problems
paper
and potential
explores solutions
the cause and of the freight
effects system. The
of the land-side emptyconceptual
container model was translated
repositioning. Then into a simulation
the key modelofinthe
characteristics thesystem
form of an
were
agent-based
captured intomodel. The agent-based
a conceptual model that model was aand
documents proof-of-concept which provided
articulates the logistics processes, an interactions,
opportunity power-relations,
to graphically imitate the
problems
logistics
and processes
potential and interactions
solutions of the freight of system.
the system
Theinconceptual
a natural, flexible
model wasandtranslated
dynamic way.
into aEmpty containers
simulation modelrepresent a sizeable
in the form of an
proportion of logistics activity, however previous studies on container freight operations in
agent-based model. The agent-based model was a proof-of-concept which provided an opportunity to graphically imitate Melbourne have tended to focus the
on
commercially
logistics valuable,
processes loaded containers
and interactions of the -system
not emptyin a containers. This paper
natural, flexible closes the
and dynamic way. gapEmpty
in understanding of emptyacontainer
containers represent sizeable
management
proportion ofinlogistics
Melbourne as represented
activity, in a concept
however previous model
studies on and agent-based
container freightmodel.
operations in Melbourne have tended to focus on
commercially valuable, loaded containers - not empty containers. This paper closes the gap in understanding of empty container
management in Melbourne as represented in a concept model and agent-based model.
© 2018 The Authors. Published by Elsevier B.V.
© 2019 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
Selection
© 2018 The
Selection andAuthors.
and peer-review underby
Published
peer-review under responsibility
Elsevier B.V.
responsibility of the
of the scientific
scientific committee
committee of ofGreen
Green Logistics
Logistics for for Greener
Greener Cities
Cities 2018.
2018.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
Keywords: Empty
Selection and Containers;under
peer-review Container Logistics; System
responsibility of the Dynamics; Agent-BasedofModel;
scientific committee GreenSimulation
LogisticsModel.
for Greener Cities 2018.

Keywords: Empty Containers; Container Logistics; System Dynamics; Agent-Based Model; Simulation Model.

*
Corresponding author Russell G. Thompsona
E-mail address: [email protected]
*
Corresponding author Russell G. Thompsona
2352-1465 © 2018 [email protected]
E-mail address: The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
Selection
2352-1465and peer-review
© 2018 under responsibility
The Authors. of the scientific
Published by Elsevier B.V. committee of Green Logistics for Greener Cities 2018.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
Selection and peer-review under responsibility of the scientific committee of Green Logistics for Greener Cities 2018.

2352-1465  2019 The Authors. Published by Elsevier B.V.

This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
Selection and peer-review under responsibility of the scientific committee of Green Logistics for Greener Cities 2018.
10.1016/j.trpro.2019.06.011
 Lucille Gusah et al. / Transportation Research Procedia 39 (2019) 92–103 93
2 Gusah, Cameron-Rogers, & Thompson / Transportation Research Procedia 99 (2018) 000–000

1 Introduction

The movement of goods, as part of the freight system, is a key component of the global economy and vital for the
functioning of cities such as Melbourne, Australia. The land-side movement of loaded container shipments occurs
between importers, exporters, inland depots, and ports. This necessitates the repositioning of empty containers, as
often containers are loaded for export at a different location to where they were initially unloaded. In the inner west
of Melbourne, nearly half the 10,000 truck trips occurring daily carry empty containers (Lucas, 2017).
Empty container repositioning results in higher operational costs due to increased travel distances for containers
and an increased number of container movements. Efficient transport of empty containers is important for the freight
industry as the associated operational costs are nearly equivalent to that of a full container, yet they result in no
direct profit as they do not contain goods to be traded.
Most studies concerning container freight operations in cities focus on commercially valuable, loaded containers
as opposed to empty containers. There is a need to increase understanding of the nature of empty container
movements in cities to aid the development of effective solutions for minimising their impacts. Minimising empty
container movements would considerably reduce operating costs and negative externalities such as congestion, air
quality and noise problems (Braekers, Janssens, & Caris, 2011).
An inherent tension of urban freight activity is that it is undertaken to satisfy the demand for goods by city
inhabitants, yet it also relies on resources and infrastructure that are shared with the local community with differing
interests. By framing the problem in this way, it becomes a city logistics problem with the overall goal to optimise
the activities related to urban goods movements and reduce the total social cost of these activities (Taniguchi,
Thompson, Yamada, & van Duin, 2001).
The empty container repositioning problem in Melbourne is the result of trade imbalances. Imported containers
generally contain lightweight manufactured consumer goods in 40ft long containers. Whereas export containers are
filled with dense agricultural goods, which tend to be heavier and therefore travel in smaller 20ft long containers due
to the capacity restrictions of equipment (Port of Melbourne Corporation, 2009; Victoria Univeristy Institute for
Supply Chain and Logistics, 2017). Melbourne has an import dominated port, the Port of Melbourne (PoM)
generates an abundance of empty containers (Mallon & Magaddino, 2001) since import containers generally cannot
also be used for export due to size and weight capacity restrictions. At the PoM, 25% of container throughput are
empty containers, 50% imported containers, and the final 25% are export containers (BITRE, 2016, p. 14).
This imbalance creates empty container movements on land, regional repositioning, and between ports through
the global repositioning of empty containers via maritime networks (Olivo, Di Francesco, & Zuddas, 2013). The
repatriation decisions surrounding the global repositioning of containers occurs on a lowest cost basis for shipping
lines, causing empty containers to be stockpiled in the hinterland of ports to hedge against shortages (Boile, 2006;
Pérez-Rodríguez & Holguín-Veras, 2014). As the dominant participant in the freight system, these decisions have
major downstream consequences on the operations of transport operators (TOs) and empty container parks (ECPs).
The research objective is to develop a detailed description and understanding of the empty container movements
in Melbourne in order to examine opportunities for improving the system. The study has been structured around the
following research questions:

1. How does the system of land-side empty container repositioning function in Melbourne?
2. Why does the system function as it does and what governs the behavior of players?
3. What are the outcomes and problems of the system?
4. How can the system be improved?

2 Literature review

There has been interest in city logistics and container transport problems since the 1970s for the former and
1980s for the latter. Common modelling approaches applicable for the problem are optimisation, simulation,
systems dynamics and heuristics (Taniguchi et al., 2001).
94 Lucille Gusah et al. / Transportation Research Procedia 39 (2019) 92–103
Gusah, Cameron-Rogers, & Thompson/ Transportation Research Procedia 99 (2018) 000–000 3

2.1 Optimisation models

Container transportation management has been addressed in the operational planning literature through two
focus areas: global repositioning on maritime networks and regional repositioning on land-side networks.
Furthermore, subproblems in regional repositioning relate to vehicle routing and container fleet management (Olivo
et al., 2013).
Early research in the field of empty container vehicle fleet management was reviewed by Dejax and Crainic
(1987) with Crainic and Kim (2007) incorporating more recent developments. Both papers highlight that insufficient
research had been undertaken in this field. Conversely, considerable work has been undertaken on the maritime
component of empty container repositioning (Cheung & Chen, 1998), along with work on rail-based land-side
movements (e.g., Powell and Carvalho (1998), Holmberg, Joborn, and Lundgren (1998)).
Crainic, Gendreau, and Dejax (1993) present a representative paper for addressing the problem of the land
container distribution system. Their model minimises the expected total operating cost, the formulations presented
are single and multicommodity and a two-stage restricted-recourse single commodity, stochastic model. Olivo et al.
(2013) developed their empty container repositioning model with adjustments to account for flexible leasing
arrangements and substitutions for containers, to more fully capture the range of options available to a shipping line.
Olivo et al. (2013) were able to provide a computational result, while other earlier models only providing a general
framework as is the case for Crainic et al. (1993). One major drawback of the Olivo et al. (2013) model is that it is
deterministic and does not explicitly consider the uncertainties of the system such as unexpected transportation
opportunities or delays.

2.2 Simulation models

Two categories of computer modelling for creating appropriate simulations are systems dynamics and agent-
based models. System dynamics models are well established in transport modelling; however, they have limited
application for freight transport and logistics at the urban level. The system dynamics model by Thaller, Niemannb,
Dahmenc, Clausend, and Leerkampe (2016) is one of the few examples of this. The model illustrated the internal
structures of system components, including freight demand generation and the wider effects on traffic and the
environment.
Agent-based models (ABM) have entered a period of popularity with many papers attempting to simulate the
freight supply chain (Li & Li, 2010; Pérez-Rodríguez & Holguín-Veras, 2014; Roorda, Cavalcante, McCabe, &
Kwan, 2010). Anand, Meijer, van Duin, Tavasszy, and Meijer (2016) developed an ABM for the city logistics
domain to demonstrate the usefulness of incorporating decision making of multiple stakeholders. The model shows
the potential of an agent-based approach to evaluate policy measures and reduce the negative effects of city logistics
processes.
ABMs have been used to assess mode choice variables in relation to intermodal and road freight with success
(Reis, 2014). By setting base assumptions, including loading requirements and decision makers, combined with
statistical information regarding freight quantities, an assessment can be made in comparing mode choice attributes.
For example, a micro-simulation model was developed for the Tokyo Metropolitan Area following the principles of
an adjusted four-step model to simulate truck and commodity flows across the urban system (Wisetjindawat, Sano,
Matsumoto, & Raothanachonkun, 2007). This model successfully reflected agent behaviour with rationalisation to
reduce costs, though it was limited by lacking vehicle routing and operation hours.

2.3 Literature synthesis

Recently, land-side container repositioning has garnered extra attention in operational planning literature.
Furthermore, empty containers represent a sizeable proportion of logistics activity at the PoM. Together, they
indicate that regional repositioning is a noteworthy problem, which should be further investigated. As mentioned,
previous studies on container freight operations in Melbourne have tended to focus on commercially valuable,
loaded containers not empty containers despite their growing importance. This presents a gap in understanding of
empty container management in the context of the Melbourne metropolitan region which could be addressed using
agent-based modelling.
 Lucille Gusah et al. / Transportation Research Procedia 39 (2019) 92–103 95
4 Gusah, Cameron-Rogers, & Thompson / Transportation Research Procedia 99 (2018) 000–000

The optimisation of regional empty container repositioning is an issue at the operational level. It has a highly
dynamic environment with inherent uncertainty, or stochasticity (Braekers et al., 2011). This presents a challenge
for sufficiently accurate optimisation modelling, which in-turn produces models that can be easily solved with the
resources available. Therefore, an ABM would be an appropriate and insightful approach suited to the complex
system that is empty container logistics. A detailed understanding is required of the stakeholders involved, their
inter-relations, and the degree to which they control their actions. The beliefs, desires and intentions of agents can be
modelled using an agent-based model, as agents can have objectives around which to optimise their behaviour. In
terms of resource requirements, simulation models are easier to build and understand, but they can be hard to debug
due to randomness and can also be costly to build due to their complexity.
3 Method
This study used and adapted the systems dynamics modelling procedure by Coyle (1996) as the approach was
designed to capture and represent complex systems. However, influence line diagrams, which are a key component
of system dynamics models, were not developed, as an agent-based modelling was the preferred approach.
Table 1. Modelling procedure.
Stage Outcomes
Stage 1: Problem The research problem detailed in Section 1 explored the causes and effects of the regional
recognition repositioning in Melbourne.
Stage 2: System An initial system description was generated based on a desktop survey which gathered
description information on the system regarding the stakeholders, interactions and other forms of
complexity that explain the systems behaviour. For this stage, the system description mainly
relied on existing literature of international cases and theory.
Stage 3: Qualitative The qualitative analysis refined the system description by having it reviewed by people with
analysis by semi- experience of the problem through semi-structured interviews. These interviews validated the
structured assumptions, simplifications, and definitions of the system description. They also provided
interviews and localised knowledge on the problems occurring in Melbourne. This stage also drew on the
concept modelling “pet theories” of the interview participants which are based on a wealth of experience in the
industry.
Stage 4: Simulation The conceptual model was translated into a simulation model in the form of an agent-based
modelling by agent- model. The model was created using the Java based simulation software AnyLogic. The
based modelling ABM represented the basic logistics chain for an import container transported by road which
enters the PoM full and returns empty after travelling to key delivery and storage nodes in the
city.

4 Results and analysis

4.1 Conceptual model

A conceptual model needed to be developed which could document the system to capture key operational details
and characteristics. The objectives of the model include the following. First, capturing system characteristics
including stakeholders, their interactions, with specific regards to timings and financial information. Second,
collation of information in a way that can be interpreted easily both for validation and potentially decision making.
Third, to serve as an input to the future simulation model.
The conceptual model was developed from information collected and confirmed during interviews. The findings
are displayed in three formats: stakeholder summary tables, logistics process maps, and a stakeholder interaction
diagram. The first set of findings describe how the system functions and the second set assesses stakeholder
motivations to explain why the system behaves as it does. Finally, the third set of data evaluates the issues produced
by the systems and documents potential solutions.

4.2 System description

Table 2 is a summary of the stakeholders of the freight system, their role, goals, and quantity within the system
in Melbourne. Industry stakeholders share a common goal as they are profit driven, however Resident Advocacy
96 Lucille Gusah et al. / Transportation Research Procedia 39 (2019) 92–103
Gusah, Cameron-Rogers, & Thompson/ Transportation Research Procedia 99 (2018) 000–000 5

Groups and the Government interest driven regarding the performance and externalities of the logistics system.
Table 2. Summary of system stakeholders.
Stakeholder Role Goal Quantity (approx..)
Shipping Lines Conducts the maritime stage of container movements Profit driven 30
Stevedores Operates port terminals and connects SLs with Profit driven 3
landside activities
Importers Destination for container in the import and purchaser Profit driven 1000+
of containerised goods
Transport Organises the landside transport between the ST and Profit driven 250
Operators the importer
Empty Container Receives, stores, inspects and services empty Profit driven or service 15
Parks (EPC) containers, and manages bulk runs driven
Empty Carriers Conducts bulk runs between EPCs and the port Profit driven
Resident Petition and protest about concerns for container Interest driven Several groups
Advocacy Groups movements
Government Administrative to enact policies to regulate operations Interest driven Several departments
Shipping Lines Conducts the maritime stage of container movements Profit driven 30

The process maps are broken into inbound and outbound container journeys and show the stages which
containers pass through and the activities performed by different stakeholders. The logistics chain diagrams are
shown in Figure 1 and Figure 2.

Figure 1. Process map - Inbound logistics chain

Figure 2. Process map - Outbound logistics chain

 Lucille Gusah et al. / Transportation Research Procedia 39 (2019) 92–103 97
6 Gusah, Cameron-Rogers, & Thompson / Transportation Research Procedia 99 (2018) 000–000

Table 2 highlights financial information of the charges and fines associated with logistics services, as well as
timing information regarding contractual restrictions and activity timings. For example, the operational of hours of
stevedores, being 24/7 operations, compared to importers who operate closer to 7am to 4pm during weekdays.
Table 3. Logistics interactions and exchanges.
Inbound Journey Outbound Journey
Time Shipping arrivals are 24/7, as are stevedore Empty Container Parks typically operate
operational hours. weekdays from 0730 to 1630.
Collecting containers from terminals takes Bulk runs take ~4 hours to complete.
around an hour.
Importers generally request delivery weekdays
from 0700 to 1600.
Financial Shipping Lines will charge ~$1000 per TEU Shipping Lines charge ~$100 per day for
to ship to Melbourne. demurrage.
Range of penalties for Transport Operators Shipping Lines pay Empty Container Parks
(late arrival is $55-$120). ~$50 per TEU for receipt, storage, and bulk
run to terminal.
Staging containers costs upwards of $50 due
to additional lifts. Empty carriers are paid ~$25 per container to
deliver to terminal.
Transport Operators charges between $175
and $540 / TEU for delivery.

4.3 System explanation

The interaction diagram in Figure 3 demonstrates the complexity of relationships between the stakeholders and
the operational and financial arrangements they have. As shown in the diagram, the stevedores and transport
operators are central to the flow of containers with several touch points of the flow of containers, both full and
empty, as opposed to empty carriers. Despite fulfilling the final stage of the system in returning empty containers to
the stevedores, empty carriers receive little communication and have a limited ability to prepare.

Figure 3. Stakeholder interaction diagram

98 Lucille Gusah et al. / Transportation Research Procedia 39 (2019) 92–103
Gusah, Cameron-Rogers, & Thompson/ Transportation Research Procedia 99 (2018) 000–000 7

Table 4. Frequency of different communication channels.

Key Communication type Frequency
1 Email, phone call Outset of deal, updates regarding timing.
2 Email, phone call Annual contracts.
3 Email, phone call, face-to-face 3 yearly deals. Individual ships as scheduled by SL.
4 1-Stop Slot bookings made daily with regular status updates.
5 Phone call Called when work is available.
6 Email, phone call When SL requires services.
In the stakeholder analysis in Figure 4 the shipping lines and stevedores hold the most power and interest in the
system which is reflected in their ability to dictate how other stakeholder’s operations run to suit their own.
Alternatively, the Empty Carriers have almost no power and instead are entirely reactive in the operations
leading to poor working conditions and concerns over driver well-being.

Figure 4. Stakeholder analysis matrix

4.4 System evaluation

Table 5 highlights the issues and potential solutions which were discussed during stakeholder interviews.
Returning to the previous example, the mismatch of hours between stevedores and importers requires transport
operators to stage containers, creating additional lift and journey costs for the transport operators to bear.
Additionally, it is highlighted that as Melbourne’s freight task continues, it will soon become impossible to deliver
containers within the requested time windows of importers and ECPs. A solution to this, presented is extended
operating hours for importers and ECPs until at least 8pm or 9pm. This solution works by allowing transport
operators to utilise off-peak container booking slots, so they are more likely to collect and deliver containers on their
first day of availability.
 Lucille Gusah et al. / Transportation Research Procedia 39 (2019) 92–103 99
8 Gusah, Cameron-Rogers, & Thompson / Transportation Research Procedia 99 (2018) 000–000

Table 5. System issues and solutions.

Inbound Journey Outbound Journey
Issues Mismatch of operating hours causing a need Importer misunderstanding of container flow
for staging intricacies
Unless Transport Operators can begin to Transport Operator may be unable to secure
deliver after hours, port will not be able to de-hire slot
move future demand Direct de-hire to terminal
Empty carriers work long hours for low pay
Solutions Extended operating hours Port Productivity Commissioner
Triangulation of containers Forecasting
New container movement study: BAHS –
Business Activity Harmonisation Study

4.5 Simulation model

The purpose of the simulation model was to demonstrate the potential of an agent-based modelling approach to
represent the container logistics chain in Melbourne with an appropriate level of complexity. The users of the model
are expected to be other researchers that will extend and refine the model. The objective is to develop an exploratory
model to understand empty container movement patterns in urban areas, and possibly test theories and solutions
which could improve freight operations and management.
The system boundaries have been set by considering three dimensions:

(i) The spatial scope has been limited to the Greater Melbourne region with focus on major freight activity
areas at the port and in the inner north-west.
(ii) The impact scope is restricted to truck movements of empty and full containers for the import leg of the
supply chain, until goods reach the importer’s distribution centre, and the export supply chain for empty
containers. The domestic export supply chain is excluded. Also excluded is the international supply chain
which would have involved an overseas manufacturer’s facility and the maritime transport of containers by
a shipping line.
(iii) The stakeholder scope will be limited to the following stakeholders that are considered highly influential
for empty container movements: empty container parks, labelled as a “container parks”; stevedores; road
transporter operator, labelled as a “carrier”; importer. Stakeholders that are excluded are: customers
brokers, freight forwarders, shipping lines and residents.

The logistics process that was modelled has been summarised in Figure 5.

Importer
Carrier Empty
places Importer
transports carrier
vehicle requests a
empty delivers to
order with vehicle
container Stevedore
Carrier

Carrier
Stevedore
picks up Importer ECP stores
requests
container receives empty
"bulk run"
from container container
delivery
Stevedore

Figure 5. Simplified logistics process modelled

100 Lucille Gusah et al. / Transportation Research Procedia 39 (2019) 92–103
Gusah, Cameron-Rogers, & Thompson/ Transportation Research Procedia 99 (2018) 000–000 9

4.6 Software development

A model of the container logistics chain in Melbourne was developed on the multi-method simulation software
AnyLogic 8.2.3. The model was coded in Java and offers graphical tools and a simple user-interface for model
construction and simulation.
The system is modelled with an agent-based approach which represents the components and interactions of the
logistics chain – importers, trucks, carriers, stevedore facilities and orders are modelled as separate agents with
custom behaviour. Then the agents are connected and interact within an environment that has its own dynamics,
which is a Geographic Information System (GIS) of Melbourne – so realistic travel times and routes on Melbourne’s
roads are modelled. Outputs from the model are dynamic animations of the simulation runs during which containers
are transported between stevedores to satisfy demand for a loaded or empty container (Figure 6). The locations and
population size of the importers, carrier and stevedore facilities are exogenous inputs in the form of database files
containing the facility latitude and longitude.
The model is built by identifying objects, or agents, and defining their behaviour. The importer’s behaviour is
defined by rules for special events which generate a demand for a container delivery. This causes the importer to
create an “order agent” at certain rate and send a message to the carrier agent to request one available truck.
Equipment limitations and uncertainty has been modelled for importers using a process flowchart that delays
delivery vehicles based on the number of forklifts available to unload or load the truck.

Figure 6. Overview of the AnyLogic simulation model

This behaviour was defined through a ‘statechart’ (shown in Figure 7), which defines a state with associated
actions and reactions. For example, if a carrier is waiting at the operator yard when they receive an order message,
then they can react by beginning to drive to the importer.
10 Lucille Gusah
Gusah, Cameron-Rogers, et al. / Transportation
& Thompson / TransportationResearch
ResearchProcedia
Procedia3999
(2019) 92–103
(2018) 000–000 101

Figure 7. Internal agent mechanisms for delivery trucks for the inbound logistics chain

4.7 Model output

The output simulation shows trucks moving along the roads in Melbourne to complete of loaded and empty
deliveries. At this stage in development the model imitates the logistics processes occurring in Melbourne for
imported containers to a simplified level. With a full-scale version of this model the system’s global behaviour can
emerge from many hundreds of concurrent individual behaviours. That is the benefit of an agent-based model, as it
is developed form the bottom up, the model maker can gain new and unexpected insights from the patterns that
emerge from the model.

4.8 Limitations

The conceptual models were developed accurately based on a combination of a desktop survey and the
interviews. However, due to time restrictions only a limited number of high-level interviews were conducted and so
missed the inputs of several key stakeholders of the system, particularly the stevedores and the shipping lines as
powerful influencers in the system. An inevitable result of this is that their views may not be reflected in the
conceptual model. Additionally, issues and solutions which have been presented in this study may have alternate and
even opposite positions for these key stakeholders.
The developed simulation model was sufficiently accurate to reflect a degree of the system’s complexities and its
variables. However, it was based on the conceptual model and so the above issues also carry into the ABM model.
Additionally, despite the model attempting to capturing a large degree of complexity it was simplified. However,
with more development time and training, this model could be further refined to allow solutions to be tested.
5 Conclusions
This study has found that the main cause of the regional repositioning in Melbourne is an intractable problem
due to an import dominated port resulting in inevitable significant empty container movements. The focus of the
study was on the sub-problems due to this abundance of empty containers. Therefore, the goal was to develop a
detailed description and understanding of the empty container movements in Melbourne to examine opportunities
for improving the system. Furthermore, in achieving this goal the study would fill a gap in literature regarding
empty container movements in Melbourne and a limited area of literary focus generally. Additionally, an ABM
model has not been used in the study of empty containers in Melbourne and providing an additional gap in literature.
An ABM was selected over optimization models, due to being the best means for capturing the complex, dynamic
system and providing a means for supporting decision making.
102 Lucille Gusah et al. / Transportation Research Procedia 39 (2019) 92–103
Gusah, Cameron-Rogers, & Thompson/ Transportation Research Procedia 99 (2018) 000–000 11

This research has addressed the objective through using a systems approach to city logistics and exploratory
modelling to capture the complexities of the system. The data underpinning the analysis was gathered initially
through a desktop survey of available literature, and second a series of high-level interviews with relevant and
experienced industry sources. The outcome of this was a conceptual model of the system which captured and
visually represented the data collected for logistics processes, timings, financials, issues, and possible solutions of
the system. This conceptual model was then used as a base for an agent-based model simulation of the system. The
simulation model was indicative of the benefits agent-based modelling and require further development to more
fully capture operational and behavioural complexities such as the mismatch of operational hours and staging at
carrier yards, however, in future such a model could be used to run experiments on the system.
The first set of research questions that this study attempted to address examined how and why the system of
land-side empty container repositioning functioned as it did in Melbourne. The freight system was found to be a
complex system with a range of interactions, variables, and participants. It operates generally linearly with
containers moving between participants conducting specific tasks from exporter to eventual de-hire. However, this
can be disrupted for a variety of reasons and result in inefficiencies and additional costs. For the second question,
the system functions as it does due to a mixture of market forces, power dynamics between participants, and
government policies though to a lesser degree. Generally, operations are run to achieve the lowest cost possible
whilst meeting contractual and client demands, as freight operations are generally a private sector endeavour. As a
multi-stakeholder system where participant objectives often conflict, a holistic approach is required to maximise
system efficiency and minimise externalities for the local community. The third research question examined the
outcomes and problems produced by the system. The problems and issues of the system are varied and include
mismatch of operating hours of participants and concerning working conditions for empty carriers. The final
research question explored how the system could be improved, and so a bank of solutions to address system issues
was compiled from desktop surveys and high-level interviews, including triangulation and an extension of importer
operating hours. Once an adequately complex model is developed, experiments can be conducted to estimate the
performance of schemes which hold promise for increasing the efficiency and sustainability of the freight system.
Future research directions include the application of the model. Once an adequately complex agent-based model
is developed, experiments can be conducted to estimate the performance of schemes which hold promise for
increasing the efficiency and sustainability of the freight system.
6 References
Anand, N., Meijer, D., van Duin, J., Tavasszy, L., & Meijer, S. (2016). Validation of an agent based model using a participatory simulation
gaming approach: The case of city logistics. Transportation Research Part C, 71, 489-499. doi:10.1016/j.trc.2016.08.002
BITRE. (2016). Waterline 58. Retrieved from Canberra,: https://bitre.gov.au/publications/2016/files/water_058.pdf
Boile, M. (2006). Empty Intermodal Container Management: Center for Advanced Infrastructure and Transportation.
Braekers, K., Janssens, G. K., & Caris, A. (2011). Challenges in Managing Empty Container Movements at Multiple Planning Levels. Transport
Reviews, 31(6), 681-708. doi:10.1080/01441647.2011.584979
Cheung, R. K., & Chen, C.-Y. (1998). A Two-Stage Stochastic Network Model and Solution Methods for the Dynamic Empty Container.
Transportation Science, 32(2), 142-162.
Coyle, R. G. (1996). System dynamics modelling: a practical approach. London: Chapman & Hall.
Crainic, T. G., Gendreau, M., & Dejax, P. J. (1993). Dynamic and Stochastic Models for the Allocation of Empty Containers. Operations
Research, 41(1), 102-302.
Crainic, T. G., & Kim, K. H. (2007). Intermodal Transportation, Chapter 8. Transportation, Handbooks in Operations Research and Management
Science, 14, 467-537.
Dejax, P. J., & Crainic, T. G. (1987). A Review of Empty Flows and Fleet Management Models in Freight Transportation. Transportation
Science, 21(4), 227-247.
Holmberg, K., Joborn, M., & Lundgren, J. T. (1998). Improved Empty Freight Car Distribution. Transportation Science, 32(2), 163-173.
Li, B., & Li, W.-f. (2010). Modeling and Simulation of Container Terminal Logistics Systems using Harvard Architecture and Agent-Based
Computing. Paper presented at the Winter Simulation Conference, Baltimore, USA.
Lucas, C. (2017, 7 March). Contain the rage: Western suburbs group rails against empty container 'scourge'. The Age. Retrieved from
http://www.theage.com.au/victoria/contain-the-rage-western-suburbs-group-rails-against-empty-container-scourge-20170703-
gx3t4b.html
Mallon, L., & Magaddino, J. (2001). An Integrated Approach to Managing Local Container Traffic Growth in the Long Beach - Los Angeles Port
Complex, Phase II: METRANS Transportation Center.
Olivo, A., Di Francesco, M., & Zuddas, P. (2013). An Optimization Model for the Inland Repositioning of Empty Containers. Maritime
Economics and Logistics, 15(3), 309-331. doi:http://www.palgrave-journals.com/mel/archive/index.html
Pérez-Rodríguez, N., & Holguín-Veras, J. (2014). The accumulation of empty containers in urban areas: Policy implications from a stochastic
formulation. Networks & Spatial Economics, 14(3/4), 379-408. doi:10.1007/s11067-014-9231-0
Port of Melbourne Corporation. (2009). Port of Melbourne and Dynon Rail Terminals - Container Logistics Chain Study: Full Report.
Melbourne: Port of Melbourne Corporation.
 Lucille Gusah et al. / Transportation Research Procedia 39 (2019) 92–103 103
12 Gusah, Cameron-Rogers, & Thompson / Transportation Research Procedia 99 (2018) 000–000

Powell, W. B., & Carvalho, T. A. (1998). Real-Time Optimization of Containers and Flatcars for Intermodal Operations. Transportation Science,
32(2), 110-126.
Reis, V. (2014). Analysis of mode choice variables in short-distance intermodal freight transport using an agent-based model. Transportation
Research Part A, 61, 100-120.
Roorda, M. J., Cavalcante, R., McCabe, S., & Kwan, H. (2010). A conceptual framework for agent-based modelling of logistics services.
Transportation Research Part E, 46, 18-31.
Taniguchi, E., Thompson, R., Yamada, T., & van Duin, R. (2001). City logistics: network modelling and intelligent transport systems.
Amsterdam, New York: Pergamon.
Thaller, C., Niemannb, F., Dahmenc, B., Clausend, U., & Leerkampe, B. (2016). Describing and explaining urban freight transport by System
Dynamics. Paper presented at the World Conference on Transport Research, Shanghai.
Victoria Univeristy Institute for Supply Chain and Logistics. (2017). Indicative estimates of commodities and truck volumes from container
destinations in Melbourne. Retrieved from Melbourne:
https://yoursay.infrastructurevictoria.com.au/application/files/1814/8843/1005/Victoria_University_Institute_for_Supply_Chain_and_
Logistics_Indicative_Estimates_of_Commodities_and_Truck_Volumes_from_Container_Destinations_in.PDF
Wisetjindawat, W., Sano, K., Matsumoto, S., & Raothanachonkun, P. (2007). Micro-Simulation Model for Modeling Freight Agents Interactions
in Urban Freight Movement. Paper presented at the 86th Annual Meeting of the Trasnportation Research Board, Washington DC.