Pore scale modeling of multiphase flow in heterogeneously wet media




Verma, Rahul (Ph. D. in petroleum engineering)

Journal Title

Journal ISSN

Volume Title



Pore scale simulation has recently become an important tool for understanding multiphase flow behavior in porous materials. It enables detailed mechanistic studies of upscaled flow parameters such as capillary-pressure saturation curves, residual saturation of each phase, and relative permeability. However, direct modeling of multiphase flow given the complex solid surfaces in a porous medium is a non-trivial problem. In this work, we develop a new quasi-static, variational level set formulation capable of handling trapped phases as well as wettability. We extend our previous work [1, 2] for simple geometries, and develop a new parallelized code enabling application of the method in larger geometries. We compare our model results against several experimental and semi-analytical datasets. The model is first applied to both homogeneous and heterogeneously wet rhomboidal pores, and compared against semi-analytical solutions derived by Mason and Morrow [3]. Subsequently, we focus on a quasi-2D micromodel study of fluid-fluid displacement for different wettabilities, which is quantified using the displacement efficiency and fractal dimension of the displacement patterns [4]. We then study classic experiments by Haines [5] and Leverett [6] for measuring the capillary pressure and relative permeability curves in sphere packs and sandpacks, respectively. We match trends in trapping in sandpacks during drainage/imbibition experiments by Pentland et al. [7], and also compare it against predictions by several other pore-scale models. Finally, we confirm the pore-scale hypothesis suggested by DiCarlo et al. [8] for explaining experimental observations of three-phase relative permeability of the intermediate-wet phase in sandpack experiments. For these three-phase experiments, we propose an approximation based on finding phases trapped between constant curvature surfaces, using two-phase simulations. We demonstrate the versatility of our methods by applying it to these disparate experimental datasets, and suggest future applications of our work.


LCSH Subject Headings