Proceedings of the TMCE 2004, April 13-17, 2004, Lausanne, Switzerland, Edited by Horváth and Xirouchakis  2004 Millpress, Rotterdam, ISBN 1 A NEW APPROACH OF INTEROPERABILITY BETWEEN CAD AND SIMULATION MODELS Okba Hamri *,+ Jean Claude Léon + + Laboratoire Sols, Solides, Structures UMR CNRS 5521, INPG, UJF FRANCE {Okba.Hamri, Jean-Claude.Leon}@hmg.inpg.fr Franca Giannini * *Istituto per la Matematica Applicata e Tecnologie Informatiche C.N.R ITALY Franca.Giannini@ imati.cnr.ge.it ABSTRACT During the product life-cycle modelling activity several simulations and analysis have to be performed. Multiple views of a product are then required to perform the involved development tasks. CAD models and simulation models are examples of such models that are required to perform a structural analysis. Within a co-operative design environment, to faster the development process, several activities are run in parallel. Thus, designers may need to perform additional modifications even after an analysis has been already conducted on a given version of the CAD model. Such operation involves a questioning of the validity of the analyses already carried out. Reappraisal does not necessarily mean that a new analysis is required, but more exactly it is necessary to evaluate or quantify its impact on the validity of the results previously obtained. This evaluation is not possible if there is no real interaction between both models. If CAD and simulation models are interoperable, some software mechanism could be provided to evaluate the impact of geometry changes onto the associated simulation model, thus enforcing consistency among these models. The approach proposed in this paper addresses such interoperability issue between CAD and simulation models using feature information from the CAD model.. KEYWORDS CAD models, polyhedral models, simulation models, interoperability, form features. 1. INTRODUCTION Currently, a CAD system contains only part of the information required for structural analysis, namely, geometrical data. In order to generate a Finite Element (FE) model, the CAD geometry needs to be adapted to suits the hypotheses of mechanical model needed. Furthermore, information about boundary conditions needs to be supplied. These tasks cannot be performed solely on the basis of a geometrical model (Dabke et al. 94, Rezayat 96, Belaziz et al. 00, Mobley et al. 98, Sheffer et al. 00), but require also engineering expertise (Fine et al. 00b). Therefore, a direct automatic transition from a CAD model to a finite element model is not feasible (Véron et al. 01, Fine 01). Generally, the difficulty in the analysis model preparation is the generation of a mesh for complex or detailed models that forms the basis for simulations. The mesh may also become too complex or difficult to adapt for more different simulations, using different categories of meshes (e.g. volume or surface) or hypotheses (shell, plate, beams, triangles, quadrangles, ...). In any case, the mesh obtained is not inter-operable, i.e. no relationship with the original CAD model is available though some elements of solution have been brought through FE remeshing approaches (François et al. 99). Moreover, no software or efficient data-processing approaches currently exist making possible to carry out adaptation directly, starting from a pre-existing mesh or polyhedral model to speed up the overall simulation process (Fine et al. 02). The waste of time and the expense caused by such a situation are not