Workflow for Simulators Based on Finite Element Method Felix C.G. Santos, Mardoqueu Vieira, and Maria Lencastre Federal University of Pernambuco - Department of Mechanical Engineering Rua Acadˆ emico H´ elio Ramos, s/n - Recife - PE 50740-530 - Brazil fcgs@demec.ufpe.br, {msv,mlpm}@cin.ufpe.br Abstract. Current workflow systems usually do not provide adequate support for workflow modeling. Real life work processes can be much richer in variations and more dynamic than a typical workflow model is capable of expressing; this means that the users need to be able to adjust workloads and modify workflow models on the fly [10]. Plexus, a system for the development of simulators [5,6], is a typical case where such a difficulty arises. Simulators provide an economical means of un- derstanding and evaluating the performance of abstract and real-world systems; their design and implementation is almost as complex as the systems being simulated, to be efficient they must be adaptable to an ever-increasing system complexity. The use of workflow technology helps thedevelopmentofmoreflexibleandversatilestrategies.Thispaperpro- posesaworkflowmanagementframework,calledGIG,forcontrollingthe simulator workflows in the Plexus context. 1 Introduction Due to tremendous ongoing activity in the fields of application of the Finite Ele- ment Method (FEM), there is a need for tools, which could help the development of simulators with a high reusability degree in both the academic and industrial worlds. Nowadays simulation systems supporting coupled multi-physics phenom- ena can be important predictive tools in many industrial activities. However, the need for suitable numerical tools, which could more appropriately simulate a large amount of coupled phenomena, and the need for computational environ- ments, which could help the building of those tools, are still a reality. The Finite Element Method is a way of implementing a mathematical theory of physical behavior. Simulations using FEM can become very complex, particularly when the designer wants to guarantee high levels of abstraction and reuse of the devel- oped solutions. Those requirements comprise the main strategies in saving the production costs of high quality simulation software. This work was done as a part of the activities of Plexus, a project for the development of a computational environment that helps the design and imple- mentation of simulation software for coupled phenomena, through flexible and friendly tools, based on the FEM [3,4,5,6]. By simulator we mean a computa- tional system aimed at obtaining approximate solutions to systems of coupled P.M.A. Sloot et al. (Eds.): ICCS 2003, LNCS 2658, pp. 555–564, 2003. c  Springer-Verlag Berlin Heidelberg 2003