On the selection of primary variables in numerical formulation for modeling multiphase flow in porous media

Selecting the proper primary variables is a critical step in efficiently modeling the highly nonlinear problem of multiphase subsurface flow in a heterogeneous porous-fractured media. Current simulation and ground modeling techniques consist of (1) spatial discretization of mass and/or heat conservation equations using finite difference or finite element methods; (2) fully implicit time discretization; (3) solving the nonlinear, discrete algebraic equations using a Newton iterative scheme. Previous modeling efforts indicate that the choice of primary variables for a Newton iteration not only impacts computational performance of a numerical code, but may also determine the feasibility of a numerical modeling study in many field applications. This paper presents an analysis and general recommendations for selecting primary variables in simulating multiphase, subsurface flow for one-active phase (Richards' equation), two-phase (gas and liquid) and three-phase (gas, water and nonaqueous phase liquid or NAPL) conditions. In many cases, a dynamic variable switching or variable substitution scheme may have to be used in order to achieve optimal numerical performance and robustness. The selection of primary variables depends in general on the sensitivity of the system of equations to the variables selected at given phase and flow conditions. We will present a series of numerical tests and large-scale field simulation examples, including modeling one (active)-phase, two-phase and three-phase flow problems in multi-dimensional, porous-fractured subsurface systems.