Laboratory of Mathematics and their Applications of Pau (LMAP), UMR CNRS 5142
Research Federation IPRA (LMAP-LFCR-DMEX-LaTEP-SIAME-CHLOE-OPERA)
Collège Sciences & Technology for Energy and Environment (STEE)
2018 : Vice-Président in charge of Digital Environnement at UPPA
2013 : Professor of Applied Mathematics at UPPA, France
2010 : Coordinator of BioFiReaDy ANR Project
2003 : MIT Fellowship Award
2002 : Permanent position at INSA Toulouse, France
2001 : PhD in Applied Mathematics, Univ. Grenoble I
1998 : MD in Applied Mathematics
1973 : Born in Strasbourg, France
My work focuses on scientific computing related to complex three-dimensional flows. My goal is to build mixed Eulerian-Lagrangian numerical methods, that is to say hybrid grid-particle schemes, aiming at computing flows in complex geometry and/or non-linear constitutive law.
Applications and fields of research are dedicated to complex fluid transport at the microscale:
in the context of porous media and geophysics at their pore scale, with a focus on CO2 mineral storage and environmental safety,
in the context of biological flows for the study of lung disease such as cystic fibrosis or ABPA.
In both these contexts, the transport phenomena, the fluid rheology, the domain complexity, multi-scale features and reactivity bring common concepts and methods.
The projects MucoReaDy (ANR 2021-2025) and formerly BioFiReaDy (ANR 2010-2014) aim at investigating mechanical dysfunction and airway clearance efficiency (or lack thereof) of the respiratory system, by a quantitative analysis of mucus motion.
A large part of the project is dedicated to healthy configuration, cystic fibrosis, ABPA and epithelium dysfunction consequent to consumption of cilia inhibiting drugs (eg nicotine) or recurrent virus and/or bacteria attack. Nevertheless, a large class of diseases is concerned by alteration of mucus mobility.
The flow within a porous object at the sub-millimeter scale, i.e. at the scale of its pores (pore-scale) is an essential link in the study of transfers within natural geological structures or artificial porous materials. These flows have an impact in particular on flows in deep geothermal energy, the securing of geological reservoirs, but above all from an ecological point of view the mineral capture of long-term CO2.
This project consists of developing a high-performance numerical method to simulate these flows by managing the different time scales involved in these configurations. In the case simplified from slow isothermal dissolution, this type of formalism has been shown to be among the most efficient and accurate on a global level. We will explore, in this project, its generalization to transport heat, crystallization and rapid dissolution. We will thus be able to explore the reconstruction of heterogeneities of a rock sample by dynamic imaging, by techniques of optimal control and of artificial intelligence, which we will confront.
Numerical homogenization, or upscaling, allows to describe an equivalent medium at a different scale with different constitutive laws but same features and behavior. This science heavily uses the averaged properties at the microscale, for transport laws depending on local porosity, or tortuosity index quantifying the upscaled diffusion from the molecular diffusion. It is then expectable to build the link between nanometer, micrometer and millimeter.
Another challenge is to build robust numerical methods in order to catch the effect of non-Newtonian features of fluids (with several orders of magnitude difference on viscosity due to high shear at pore-scale) and then effects of non-observable roughness at walls that can induced large discrepancy on main petrophysics values.
The Lagrangian methods I consider belong to the Vortex-in-Cell methods, whose main features are to transport cells of vorticity, and to compute their trajectory using grid-particle coupling.The motivation of this approach is to generalize the VIC methods sufficiently to perform large three-dimensional flow computation, as close as possible to real world configuration, and perform their optimal control. Consequently, the numerical methods presented have been developed with the concern of high computational efficiency.
