Baresafe

BARESAFE (Numerical modeling of multi-scale fluid solid coupled systems : advanced hybrid methods for reduction of uncertainty on stability limits and optimization of nuclear safety barrier reliability) is a research project involved in the development of modeling and advanced numerical methods for simulation of large size systems involving multi-physics in the field of mechanics. It addresses the hard issue of stability analysis of dynamical systems submitted to external turbulent flows and aims to establish accurate stability maps applicable for heat exchanger design. The purpose is to provide stability limit modeling in a variety of configurations with a maximal accuracy in spite of the large scale of the systems to be considered. The challenge lies in predicting local effects possibly impacting global systems. Therefore the combination of several strategies convenient simultaneously for multi-physics, multi-scale and large size system computation is required. Based on empiric concepts, the heuristic models currently used in the framework of standard stability analysis suffer from a lack of predictive capabilities. On the other side, numerical approaches based on fully-coupled fluid solid dynamics system computation remain expensive and prohibitive due to the multi-physics patterns of physics and the large number of degrees of freedom to be involved. In this context, since experimentation is not reachable and numerical simulation alone is unavoidable, the project proposes an hybrid strategy in order to take advantages from both numerical local and empiric global solutions. The project is decomposed as follows : WP1 on modeling, WP2 on numerical scheme development for coupled computation, WP3 for improvement of High Performance Computing (HPC) environment, WP4 for numerical simulation of stability maps and WP5 for diffusion and dissemination of solvers and numerical solutions through Opensource libraries, softwares and data bases thanks to the common reference Opensource software coupling platform Salomé . The objective is devoted to numerical modeling of turbulence and its development in flows around moving obstacles and aims to enable accurate stability analysis of dynamical systems without degrading the models and without using penalizing CPU time consuming near-wall mesh refinement. From a numerical point of view advanced hybrid methods are addressed in order to go to real scale by controlling approximation error, uncertainty and accuracy of solutions. Optimized time and space discretization formulations for coupled computations are developed in order to face up with non-matching interfaces. Advanced optimization algorithms are proposed to enable fully-coupled parallel computations through iterative methods. The objective is to contribute to a scalability improvement factor of about 100 to make it possible to use thousands processors for reaching one billion of degrees of freedom models. All these developments are made possible by the combined contributions of all partners. Academic laboratories (IMFT, IMFS, LaMSID and IRIT) are involved in advanced computational methods for mechanics computation and HPC. Industrial partners of the nuclear domain (EDF, CEA et AREVA) share a real need in terms of increasing predictive capabilities of stability modeling when applied in a large scale context. Results of numerical simulations and large scale analysis will have a major impact in the framework of safety projects dealing with major issues addressed by nuclear safety authority and related to reliability of safety barriers equipping Pressurized Water Reactors (PWR). The program is proposed for 4 years in the framework of the 2011 NM program for a total amount of about 3500 k€ including 27% of ANR contribution. For its major contribution to Opensource software platform development the project formulates a Label requirement from the SystemAtic Competitiveness Pole (to which some project partners contribute).