A Software Factory for the Generation of CAPE-OPEN compliant Process Modelling Components
The increasingly growing complexity of chemical process modeling software has led to the emergence of interoperability needs. As a response, the CAPE-OPEN standard has been adopted by the process systems engineering community, and is widely used to integrate heterogeneous process modeling software. However, the component-based architecture imposed by this standard introduced additional complexity which is hard to master by process engineers. Indeed, developing and maintaining CAPE-OPEN compliant process modelling components requires additional knowledge about the standard specification, and the middleware.
Model-Driven Engineering is a software engineering paradigm that can provide methods and tools for reducing the complexity of producing such compliant components. This discipline aims at decoupling the business knowledge from the implementation details. The former is isolated as a platform-independent model; the latter is the refinement of the former for a specific targeted middleware, by the means of model transformations. Hence, the objective of this thesis is to address the design of a model-driven architecture for the generation of CAPE-OPEN compliant process modeling components. The proposed approach aims at decoupling the aspects involved in the development of such components, as separated expert views through which code generation is performed. Moreover, the approach allows to introduce a set of expected changes using domain-specific languages, and to automate change propagation, using a two-step model transformation. First, domain-specific models, that are platform independent, are automatically combined and refined in order to produce a platform-specific model. Next, code generation templates are used to generate part of the component’s code from the platform-specific model.
In order to realize such an approach, a state-of-the art study of the model-driven engineering principles and practices is carried out. Throughout this study, we focus on three key questions: how to represent knowledge, how to transform knowledge, and how to manage all the engineering activities in a single engineering process. Following this study, we build a model-driven prototype based on domain specific modeling and graph-based model transformations. The prototype allows generating parts of the component’s structure as well as part of its semantics. Finally, we validate the approach with respect to a set of criteria including formal verification of the intermediate models, as well as a set of expected changes impacting the considered expert views.
Defence : 03/15/2010 - 15h - Site Passy-kennedy - salle 550 Jury members : M. Mikal Ziane
M. Amar Ramdane- Chérif [Rapporteur]
Mme. Mireille Blay-Fornarino [Rapporteur]
Mme. Sylvie Cauvin
M. Tewfik Ziadi
M. Jean-François Perrot
M. Jean-Michel Bruel