ABSTRACT

Fluid injection into subsurface porous media may initiate fluid-driven fractures. Numerical modeling of immiscible fluid injection and fracture initiation in granular media is challenging since continuum analytical and numerical mechanical models are not able to capture fracture width, length and morphology in granular media. In this study, the discrete element method (DEM) is coupled with computational fluid dynamics (CFD) for solving fluid flow of two immiscible fluids at the pore/grain scale. The interface between the two fluids is modeled by the volume of fluid (VOF) method. The numerical model is verified against the analytical solution of a benchmark problem of a settling particle in a biphasic system. We perform numerical simulations of the injection of a fluid into a granular medium that is saturated with a second, immiscible fluid. Initially, the granular medium is under applied stress and saturated with a second fluid. With the increase of injection velocity, there is a transition from a radial flow regime without grain displacement to particle-displacement dominated regime. Results show that the injection of an immiscible fluid can initiate a fracture in the direction orthogonal to the minimum applied stress, as expected from energy considerations. However, the resulting displacement field around the fracture is much more complex than what is expected in a linear elastic solid, which leads to a more complex fracture morphology.

1. INTRODUCTION

Fluid injection into porous media has many applications in reservoir development, such as water flooding, hydraulic fracturing, subsurface CO2 storage, acidizing and chemical enhanced oil recovery (Espinoza and Santamarina, 2010; Lake et al., 2014; Sun et al., 2016; Zheng et al., 2019). The injection of fluid increases the local pore pressure and changes the stress state, which may result in fracture initiation (Hubbert and Willis, 1972). The invasion of an immiscible fluid into a porous medium is a three-phase, fluid-particle system, which is challenging to model numerically (Pozzetti and Peters, 2018).

This content is only available via PDF.
You can access this article if you purchase or spend a download.