Résumé
Light Non-Aqueous Phase Liquid (LNAPL) flow on the water table is highly mobile and is sensitive to the fluctuation of groundwater. This process is highly complex and involves the migration of three immiscible phases (i.e. water, LNAPL and air) which need the explicit definition of multiple parameters. A coupled experimental and numerical simulation methodology is performed by using Time Domain Reflectrometer (TDR) and multiphase simulation of a controlled environment to mimic the water table fluctuation and its effect on the LNAPL residual saturation. TDR probes are installed in different locations of a 2D tank (i.e. a cuboid box with relatively low off-plane thickness) and the bulk permittivity of the phases are measured through artificially imposed boundary conditions. The bulk permittivity is then translated into saturation of the three different phases. The translated residual saturations along with the previously measured porous media properties (e.g. porosity and saturated permeability) are then inserted into the numerical simulator (i.e. COMSOL Multiphysics®) and the migration of the three phase in porous media is simulated. The numerical exponents and entry pressures needed for the simulation of the multiphase flow are estimated using the temporal experimental values. The exponents of water LNAPL relative permeability were estimated to be around 2 while the exponents gas LNAPL relative permeability were estimated to be closer to 3. The results, simulated with the optimized parameters, are then evaluated with pictures taken from the transparent face of the 2D tank different stages of the experiment. The temporal evolution of different phase saturation has been compared and validated between the experimental results obtained and interpreted by the TDR probe measurements and the simulations. The relative error stays in the 5 % confidence level for most reported points and only in the highly dynamic flow time steps the error reaches around 12% which are discussed in the text and is accepted due to the highly nonlinear nature of the problem.