A METHODOLOGY TO REDUCE THE EXPERIMENT DESIGN
PHASE WHEN IMPROVING LOGISTIC SYSTEMS THROUGH
SIMULATION
Roman Buil¤, Daniel Riera¤, Miquel A. Piera¤ and Antoni Guasch+
¤ Departament de Telecomunicaci´o i d’Enginyeria de Sistemes
Universitat Aut`onoma de Barcelona, Bellaterra, Barcelona, Catalonia
fRoman.Buil,Daniel.Riera,MiquelAngel.Pierag@uab.es
+ Institut de Rob`otica i Inform´atica Industrial
UPC/CSIC, Barcelona, Catalonia, Spain
Toni.Guasch@upc.es
ABSTRACT
Simulation models have proved to be useful to tackle the evaluation of alternative operating
procedures for some systems. It is widely acknowledged that simulation is a powerful
computer-based tool that enables decision-makers in business and industry to improve
operational and organizational e±ciency. However, when applying simulation techniques
to improve the performance of complex and logistic systems, several limitations arise due
to its inability to evaluate more than a fraction of the immense range of options available.
The state space of a simulation model represents all the possible ways in which the
process can be executed. The state space of an operational plan represents the possible
ways in which the operational plan can be executed, subject to resource, timing, and
synchronisation constraints.
The high number of decision variables in present logistic systems usually can lead to a
huge scenario design at the experimentation stage in a simulation project, which makes
practically impossible its computational handling.
A new methodology to reduce the possible alternatives to be evaluated is presented in this
paper. The main ideas is to model the logistic system using Coloured Petri-Nets (CPN),
from these, to generate a Constraint Satisfaction Problem (CSP) extracting constraints
from each structure detected in CPN. When the CSP is created, Constraint Programming
(CP) is used to reduce the domain of the decision variables, that is, to reduce the number
of alternatives to test in simulation.
Keywords: : Constraint Satisfaction Problems, Planning, Simulation, Production
systems, Coloured Petri-Nets.
1 Introduction
World-wide market competition, high product quality
requirements, together with random demands instead
of steady demands, are some key factors which have
forced industry to change both:
² Traditional rigid and/or non-automated production
architectures (such as Flow Shop, Job Shop)
towards Flexible Manufacturing Systems (FMS).
² Conventional production planning methodology
towards new heuristic based methodologies,
which could cope with a large amount of decision
variables inherent to present production architectures.
The exact optimal solution of a production, a distribution
or a transport system-planning problem is quite
complex and di±cult, even impossible to obtain.
Simulation limitations arise from the inability to evaluate
more than a fraction of the immense range of options
available. Most commercial discrete event oriented
simulation packages are designed to be used as
an analysis tool. That is, the system to be studied
is modelled, perturbed, parameterized and simulated
to predict which changes would cause in a real sys-
.....