This project is related to the mechanics of dense multiphase mixtures composed of solid grains and two (immiscible) fluids. Such systems are seldom found in geomechanical engineering (vadose zone in soils, seabed sediments, oil-water and gaz-water mixtures in natural reservoirs,…) as well as in many industrial processes (reactors, mixers, absorbant powders, …). While many research communities are actively working on transfers in such systems (unsaturated fluid flows, retention properties, transport of bubbles,...), the emphasis of this project is the coupling between the transfers and the mechanical behavior - in the sense of solid mechanics - i.e. the deformation and failure of granular materials partially saturated by two fluids. While deformation obvioulsy involve transfers, it makes the problem much more complex from fundamental and numerical points of view.
The objective is to develop a new numerical method for the direct simulation of multiphase granular systems at the microscale, enabling numerical experiments and the determination of effective properties through numerical homogenization. The proposed method can be summarized as a multiscale integration of a microcontinuum method (here Lattice-Boltzman, LBM) into an original pore-network approach of multiphase flow in deformable systems. The granular solid phase will be modelized using the discrete element method (DEM).
Left: decomposition of the pore space of a sphere packing by Regular Triangulation in the DEM-PFV method. Right: distribution of wetting and non-wetting phases resulting from a simulated drainage
The originality of the project is manyfold:
The microscale approach of complex systems is not original in itself, but applying this idea to deformable multiphase systems is relatively new, especially in the community of granular materials;
The pore-network vision of porous media has been very rarely applied to deformable materials, it enables the simulation of many thousands of solid particles in 3D, a bare minimum for reaching the REV size, yet beyond the reach of conventional methods due to computation times;
The classical pore-network models (including those developed by the PI until now) are based on analytical expressions of local quantities (volume and hydraulic conductivity of pores or pore throats, entry capillary pressure,...) which are hardly validated by any accurate measurement. In this proposal the local quantities will directly come from numerical results obtained at a smaller scale.
The project includes a direct comparison of the simulations with the data of tomography measurements during wetting-drying cycles on spherical beads. The experiments are being done independently of this project yet the setup has been designed in order to enable the comparison proposed here. Direct comparisons of this kind are extremely rare in the literature.
PI: Bruno Chareyre; Post-doc: Chao Yuan