COMPONENT BASED APPROACH FOR SIMULATION
OF DYNAMIC HYBRID SYSTEMS
Gilles Hétreux, Jocelyne Perret, Jean-Marc Le Lann
Laboratoire de Génie Chimique (LGC - UMR 5503),
département PSE, groupe Génie Industriel
118 Route de Narbonne, 31077 Toulouse Cedex 04
E-mail: Gilles.Hetreux@ensiacet.fr, JeanMarc.LeLann@ensiacet.fr.
Abstract: PrODHyS is a general object-oriented environment which provides common and reusable
components designed for the development and the management of dynamic simulation of chemical
processes. The aim of this paper is to introduce a small part of the main reusable object components
available in this library and to illustrate its potentialities through the modelling and the simulation of a
reactive distillation column.
Keywords: Object oriented software component, Hybrid dynamic systems, object differential Petri nets
formalism, modelling and simulation of chemical process.
1. INTRODUCTION
Nowadays, dynamic simulation is an essential tool in
chemical process design and analysis. Such
applications have benefited by the growing power of
computers and the development of robust numerical
methods. In parallel, faced to the complexity of the
modelled processes, object oriented approach has
deeply changed the way of working in software
engineering. Largely influenced by the evolution of
software technology, computer aided process
engineering (CAPE) has been naturally infiltrated by
this paradigm. The unification of works in this domain
performed since several years in our laboratory has led
to the development of PrODHyS. This software allows
a object oriented description of processes using a
library of general components. Moreover, it is built on
a dynamic hybrid simulator based on a rigorous
formalism. This communication aims at presenting the
general architecture of this environment and at
emphasizing the fundamental concepts. These aspects
are illustrated through the modelling and simulation of
a reactive distillation process.
2. THE ENVIRONMENT : PrODHyS
2.1. Main objectives of PrODHyS
The philosophy of PrODHyS is to provide a collection
of components designed to the accurate modelling of
both the structure and the behaviour of chemical
processes, in an objective of dynamic simulation [3].
The elaboration of a dynamic model able to simulate a
complex system often induces its decomposition in
more simple subsystems. This systemic and
hierarchical vision of processes is well fitted to a object
oriented approach. From the beginning, we have
exploited this approach in sight to :
· improve the software quality,
· produce libraries of extensible and reusable
components,
· facilitate the creation of these components by
developers but also by modellers, thanks to a
vocabulary nearer of its trade competence,
· provide afterward, these components to the user
through a graphical interface (GUI) in which each
object will have a "active image".
In fact, the fundamental base of process modelling rests
on the identification of the elementary components.
The composition of these elementary components
makes possible the built of elementary models of
simulation, themselves being components of more
complex and specialized simulation models. So, an
important part of the contribution of this work lies in
the determination and the design of these initial
common building blocks.
Most of real processes have complex behaviour,
primarily due to the coupling of many physicochemical
phenomena. For these reasons, some phenomena are
sometimes simplified or neglected according to the
required level of accuracy. Thus, the resulting dynamic
model often becomes a hybrid model [8], i.e. a set of
consecutive continuous models coordinated by a
discrete model. In this framework, PrODHyS used the
Object Differential Petri Nets (ODPN) formalism to
describe the simulation model associated with each
component. This rigorous formalism makes collaborate
differential algebraic equations systems with Petri nets
within the same structure. In addition, the resolutely
object orientation of the platform is extended until the
formalism which thus offers a high level of abstraction.
.....