Special Issue SIMULATION on Harbour and Maritime Simulation, 1998, to appear
Simulation and Planning of an Intermodal Container Terminal
Luca Maria Gambardella, Andrea E. Rizzoli, Marco Zaffalon
IDSIA - Istituto Dalle Molle di Studi sull’Intelligenza Artificiale
Corso Elvezia 36, 6900 Lugano, Switzerland
{luca,andrea,zaffalon}@idsia.ch
Abstract
A decision support system for the management of an intermodal container terminal is presented. Among the
problems to be solved, there are the spatial allocation of containers on the terminal yard, the allocation of
resources and the scheduling of operations in order to maximise a performance function based on some economic
indicators. These problems are solved using techniques from optimisation, like job-shop scheduling, genetic
algorithms or mixed-integer linear programming. At the terminal, the same problems are usually solved by the
terminal manager, only using his/her experience. The manager can trust computer generated solutions only by
validating them by means of a simulation model of the terminal. Thus, the simulation tool also becomes a means to
introduce new approaches into traditional settings.
In the present paper we focus on the resource allocation problem. We describe our modules for the optimisation
of the allocation process and for the simulation of the terminal. The former is based on integer linear
programming; the latter is a discrete event simulation tool, based on the process-oriented paradigm. The simulator
provides a test bed for checking the validity and the robustness of the policy computed by the optimisation module.
The case study of the Contship La Spezia Container Terminal, located in the Mediterranean Sea in Italy, is
examined.
1. Introduction
The management of an intermodal container
terminal is a complex process that involves a vast
number of decisions. Most of the world’s goods
which are traded daily are transported via intermodal
terminals. Goods arrive and leave on various
transportation means such as trucks, trains and
vessels. An intermodal container terminal plays a
fundamental role in routing goods to and from their
origins and destinations. It is a basic node in a
transportation network, where thousand of daily
decisions are taken to manage this sustained flow of
containers.
The advent of management information services and
data processing greatly improved the ability of
terminal managers to control the whole process, but
still raw data has to be analysed and treated to provide
some insight on the performance of terminal
operations. Simulation models have proven to be a
reliable and convenient tool to support the decision
makers in the daily operations in many cases (Hayuth
et al. 1994, Blümel, 1997, Bruzzone and Signorile,
1997). They provide a test-bed to assess the validity
of management policies and can be used to point out
problems such as conflicts in resource allocation and
terminal space management. These simulation tools
do not provide answers to question such as “how can I
minimise the time it takes to unload these two
incoming ships?” or “Should I unload the ship, or
wait for the train to arrive?”. In many cases, these
answers are yet to be provided by the terminal
managers, basing their decisions on experience in
solving these problems.
A substantial help to terminal managers can derive
from Decision Support Systems (DSSs) where
planning and management techniques, derived from
the Operations Research and Artificial Intelligence
fields, can be coupled with simulation models and
statistical data analysis tools. The role of simulation
becomes of paramount importance in such a setting:
human decision makers tend not to trust computer
generated management policies, unless they either
fully understand the way they were generated or are
provided with sufficient evidence of their validity.
This behaviour is often proven to be reasonable, since
very often computer generated policies are not
flexible enough in comparison to the complexity and
high stochasticity of real world operations.
A well designed simulation tool can be the middle
ground where decision makers compare their own
experience with the DSS generated management
policies and validate them. Under this point of view,
it is clear that the possible strength of mathematical
approaches to the optimisation of terminal processes
are highlighted in a proper way to terminal managers.

22. The problem
An intermodal container terminal is a place where
containers enter and leave by multiple means of
transport, as trucks, trains and vessels (I/O transport
means). We focus our attention on the case study of
La Spezia Container Terminal (LSCT), located in the
Tyrrhenian sea in Italy.
Containers arrive at LSCT by train, vessel or truck
and are stored in the terminal yard. Containers then
leave the terminal by the same 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 transhipped 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 amount of work processed by a container
terminal depends on the quantity of containers in
transit.
3. Decision support for terminal
management
Storing containers on the yard, allocating resources
in the terminal, and scheduling vessel loading and
unloading operations (L/U operations, for brevity) are
major problems in an intermodal container terminal.
To solve these problems we define an architecture
composed of three different but strictly connected
modules (Rizzoli et al., 1997) (see figure 3-1):
– a simulation model of the terminal, described in
terms of entities (work force, transport means,
storage areas, etc.) and processes (vessel
load/unload, shuttle truck movements, crane
operations, etc.);
– a set of forecasting models to analyse historical
data and to predict future events (Box et al., 1994;
Vemuri and Rogers, 1993), thus providing
estimates of the expected import and export flows;
– a planning system to optimise L/U operations,
resource allocation, and container locations on the
yard.
This architecture supports the terminal managers in
the evaluation of:
– vessels loading and unloading sequences in terms
of time and costs;
– resource allocations procedures;
– policies for container storage both in terms of
space and cost of operations.
This allows terminal managers to assess “what-if”
scenarios; for instance, what happens if the terminal
undergoes an increased input/output throughput, or
even if structural changes are made (e.g.: new berths
are built, new cranes are added).
As the forecasting module is described in previous
papers (Gambardella et al., 1996, Bontempi et al.,
1997), in the following sections we introduce the
other two modules of our architecture: the planner
and the terminal simulator. For each topic, we present
the major problems, the resolution methodologies and
the experimental results obtained at the current state
of the project.
3.1. The optimisation modules
In our study, we identified a series of problems,
placed at different representation levels, which can be
assisted by a computerised decision support system:
spatial allocation of container locations in the
terminal yard, allocation of terminal resources (yard
and quay cranes, work force, etc.), and scheduling of
terminal operations (e.g. container movements) in
order to maximise a performance function of
economical indicators. These problems also have
different planning horizons related to the speed of the
dynamics of the system they control: the spatial
allocation policy has a horizon of about one week,
while a few work shifts (about twenty-four hours) is
the horizon of the resource allocation policy. The
planning horizon of scheduling of terminal operations
can be as short as one hour. In this paper we will
Forecasting
Simulation
Planning
Figure 3-1 The modular system
architecture