SIMULATION FOR THE EVALUATION OF OPTIMISED OPERATIONS POLICIES IN A
CONTAINER TERMINAL
A. E. Rizzoli, L. M. Gambardella, M. Zaffalon, M. Mastrolilli
Istituto Dalle Molle di Studi sull’Intelligenza Artificiale
Corso Elvezia 36, Lugano, Switzerland
Email: { andrea , luca, zaffalon, monaldo}@idsia.ch
Keywords: intermodal terminal management,
simulation and optimisation.
ABSTRACT
A simulation model must be used to evaluate the impact
of new operations policies, not only to validate the policies,
but also as a tool to convince the decision-makers of the
potential advantages in adopting the proposed
enhancements in the management. Terminal resource
allocation policies and ship loading/unloading policies are
obtained by means of Operations Research techniques for
the case study of La Spezia Container Terminal (LSCT); a
simulation model of the terminal is designed, implemented
and validated. The simulation model is used to test the
policies and to assess their robustness in front of the
inherent stochasticity of the real world.
INTRODUCTION
The management of an intermodal container terminal is
a complex task, which involves a great number of decisions
to be taken at different levels, from strategic development
down to the single move of a container. Most terminals rely
on information management systems, often interfaced with
automated data gathering devices, in order to take informed
decisions. Almost as often, the sheer amount of the
information makes nearly impossible for the human
operators to see the "big picture", that is, the terminal in its
complexity, considering the multiple interactions of the
various concurrent processes, such as yard planning,
resource allocation, ship loading and unloading. It is clear
that Operations Research techniques could bring the
advantage of making a better use of the available
information and, consequently, to increase the overall
performance. This is not an easy task, the mathematical
models describing the terminal processes must be studied
and designed with great care, often with a conflicting
objective in mind: the model must include all the
significant characteristics and, at the same time, it must be
simple to be computationally solvable. During the model
design phase we are obliged to make simplifying
assumptions, which eventually may reveal themselves as
correct, but which are often unacceptable from the point of
view of the decision makers, since they might conflict with
their personal experience. It is pointless to design a model
in which the decision-makers have no trust, unless you
convince them to trust it. Simulation is the tool that can be
used to build this trust.
We have studied the intermodal container terminal of
La Spezia, operated by Contship SpA, and in four years we
have investigated different processes such as the container
flow to and from the terminal, by land and sea, the yard
storage allocation policies, the resource allocation policies,
the load and unload scheduling policies. With the aid of the
ISO 9001 documentation provided by LSCT we have
tracked down the current decisional processes and we have
noted that the terminal decision makers have adopted a
distributed decision making scheme, centered on the ship
loading and unloading process. This management choice
has enabled the port to reduce the size of the problem,
allocating decision making resources to each ship which
"competes" for the accomplishment of their task:
processing the whole ship in due time. While this approach
guarantees the best service for the customers (the shipping
companies) thus making La Spezia an "attractive"
destination, often it neglects the possibility of synergies in
the use of the yard resources. The historical data we have
processed supported by the terminal operators’ experience,
shows that often there are resource conflicts on the yard
cranes, which access the storage area of the terminal. It
may happen that, despite the yard planners tend to organise
the yard in order to de-couple the effect of two ships
accessing the same storage area, two yard cranes must
concurrently access the same area. For this purpose we
have designed resource allocation algorithms and loading
and unloading operations scheduling algorithms which will
be described in the section titled “Synthesis of management
policies”.

The models employed in the synthesis of the optimised
policies are deterministic and include a set of assumptions
which made possible the applicability of known algorithm
for their fast and efficient solution (even if we had to
devise an enhancement to classical job-shop scheduling
theory). Even if these models were calibrated and validated
before their employ, they do not intend to represent all the
terminal behaviour, but to limit their applicability to the
phenomena they want to describe. To assess the impact of
the optimal policies obtained from these models, we have
designed a micro simulation model of the container
terminal. It is based on stochastic discrete-event simulation
and it can be used to perform a trace-driven simulation,
using the historical data sets, which are continuously
collected during terminal operations. The terminal model is
able to operate to replicate the current "distributed"
operations control policy, but also to incorporate the
optimised policies, to compare their effect with the current
management practices. In section “Simulation” we describe
the design and the implementation of the discrete event
simulation model.
The model has been calibrated on one week of terminal
operations, while another week has been used to validate it.
A panel of terminal experts has positively judged the
validation results. The model has been used to show to the
panel the expected impact of the proposed policies and how
the terminal performance may increase while the operation
costs decrease, thanks to a more coordinated usage of the
terminal resources. The model has also been employed to
test the robustness of the policies, in front of unexpected
events and of variable service times. These results are
provided in subsection “calibration and validation” and
they are discussed in the final “Discussion” section.
THE PROBLEM AND THE CASE STUDY SITE
Containers arrive at La Spezia Container Terminal
(LSCT) by train, vessel or truck and are stored in the
terminal yard. Containers then leave the terminal by the
same transport means to reach their next destinations. The
flow of containers is composed of an import flow, i.e.
containers unloaded from ships, to be either transshipped or
directed to the final destinations by trucks and trains, and
an export flow, i.e. containers loaded on ships leaving the
terminal.
In the LSCT, containers are stacked up to the fifth level
on the yard by rail-mounted cranes (yard cranes) which
unload trucks and trains. This stack height is quite unusual
and is due to the lack of space on the yard. LSCT is a
terminal with a high traffic on a small yard and therefore
the management of space is a critical issue. Quay cranes
unload vessels and place containers on shuttle trucks,
which move them to storage locations in the yard. Loading
a vessel is a similar process, where the shuttle receives the
container from the yard cranes and moves it to the proper
quay.
The problems of the LSCT terminal are related to the
current workflow management procedures. All the work
processes relative to a ship are planned, supervised and
managed by a single ship planner, who does not know the
decisions made by other ship planners, which are working
on other ships present at the same time in the terminal.
After an analysis phase, we identified three areas where
mathematical models, optimisation and simulation could be
useful: dynamic allocation of storage areas on the yard,
allocation of the resources required to perform the ship
loading and unloading (L/U) operations, and finally the
scheduling L/U operations. It is easy to show that these
management actions are interdependent, a modification of
the policy used to decide where to place a container on the
yard has a direct impact on the resource (crane) which will
move it. In the same way, if the set of resources assigned to
unload and load a ship varies, the optimal sequence of L/U
operations will change.
For simplicity, we decided to analyse the three
problems separately, since their time scales are different:
decisions regarding yard space planning have an horizon of
some days, while resource allocation is made on an horizon
of 24 hours and L/U operations are scheduled before each
work shift, that is, every six hours.
In this paper we present the results relative to resource
allocation and L/U operations scheduling, since we haven’t
yet tested the effectiveness of the yard allocation policies.
SYNTHESIS OF MANAGEMENT POLICIES
In this section we summarize the work done in the
synthesis of the resource allocation and scheduling
policies. We also outline how the two policies can be used
together to take operational decisions in the terminal.
Resource Allocation
The role of the resource allocation (RA) module is to
determine the best allocation of resources for vessel
loading and unloading operations, with the objective of
maximizing the profit, given by the difference between
income and expenses (Zaffalon and Gambardella, 1998,
Zaffalon et al. 1998). The income is a term proportional to
the number of moved container, whereas the expenses are a