Bordeaux, Nouvelle-Aquitaine, France
Specialties: 3D meshes, 3D geometry, applied mathematics, 3D rendering, algorithmics
Enseignement de l'informatique MPII et MPSI option informatique
LEMMA is a high-level R&D company (15 employees) specialized in scientific computing, editing its own CFD/CSD software ANANAS. Coding new models and methods inside this home-made software. I was in charge of the Computational Structure Dynamics solver and of Fluid-Structure coupling. Detailed achievements: - Developping an industrial Computational Structure Dynamics solver from scratch - Large-displacement formalism; - Material laws (hyper-elasticity, finite strain plasticity); - Contact algorithms; - Crack propagation modelisation XFEM 3D from scratch, stress intensity factors computation - Thick shell models - Thermal effects and couplings. - Coding the Fluid-Structure interaction coupling scheme - Developing various meshing tools for industrial use - Occasional commercial involvment (prospection, industrial conferences) - Technical support for clients - Designing and delivering technical courses and trainings for our clients - Team work, software validation and industrialization - documentation writing and video tutorials recording
Meshing technologies, metric-based mesh adaptation, moving mesh simulations, compressible fluids, ALE schemes, Fluid-Structure interaction. International conferences in English. My PhD thesis deals with time-evolving simulations involving fixed or moving geometries. It attempts to partly fulfill the growing demand of industrials on this type of computations, notably by improving the accuracy as well as the efficiency (CPU time) of these simulations. Anisotropic metric-based mesh adaptation strategies, which have now reached a certain level of maturity on steady problems, offer good perspectives to enhance time-evolving simulations, but their extension in this context is far from straightforward. As for their application to moving mesh simulations, only few attempts can be listed so far and only a minority address complex three-dimensional real-life problems. This study proposes several novelties on these questions, notably the extension of multi-scale anisotropic metric-based mesh adaptation to unsteady problems, for both fixed and moving domains. Besides, mainly for CPU reduction purpose, a genuine strategy has been adopted to handle moving mesh simulations. It is notably demonstrated in practice that it is possible to move 3D complex objects undergoing large displacements using only connectivity changes and vertex movements, which comes to keep the number of vertices of the moving mesh constant throughout the simulation. Limiting the number of mesh operations allowed enables to considerably reduce CPU time as time is saved both on the meshing and on the solver parts. Finally, a new scheme extending the classical fixed-topology Arbitrary- Lagrangian-Eulerian framework to variable-topology moving meshes is proposed and its validity has been assessed on two dimensional test cases. All these methods have been applied to compressible CFD simulations around complex geometries in 2D and 3D.