Towards Predicting Multi-Phase Flow in Porous Media Using Digital Rock Physics: Workflow to Test the Predictive Capability of Pore-Scale Modeling

Abstract Digital rock physics (DRP) has gained significant development in the last decade. At its current state, in general DRP cannot yet reliably a priori predict two-phase fluid flow properties without knowledge of wettability/contact angles, especially for non-water-wet rock. Prediction of two phase flow properties becomes even more challenging for carbonate rocks which are rarely water-wet and have a complex pore structure. The two main challenges are to model pore geometries and size distributions at a representative elementary volume (REV) and the representation of wettability. Advances in micro- and nano-CT imaging and computer capability may help solve the first challenge, although the translation from images to model input parameters and REV consideration remains a research topic. However, wettability distribution (as an input parameter) cannot be predicted and will remain to be the most significant problem in attempting to predict multi-phase flow properties. In this paper, we study the use of pore-scale imaging and modeling (DRP) to predict relative permeability curves. We recommend that we should shift focus from a priori prediction of fluid-flow properties, to instead investigate how much experimental special core analysis (SCAL) and imaging data is required as input to calibrate or constrain the model before computing two- or three-phase flow properties for field applications. In this paper we will focus on two main issues: 1- The role of DRP and how it can complement SCAL data, 2- How can we improve the predictive capability of DRP through the use of imaging data combined with benchmarking and tuning to match core-scale measurements. A case study and a recommended workflow to integrate DRP in the whole SCAL procedure is presented. In the case study, the measured primary drainage capillary pressure is used to modify the pore size distribution inferred from multi-scale imaging, while the waterflood capillary pressure is used to estimate contact angle. Then waterflood relative permeability is predicted and compared to experimental measurements: the water relative permeability lies within the range of measurements made on four core samples, while the oil relative permeability is over-predicted. The core samples used in the study are carbonate samples; however, the workflow described in the paper applies to both sandstone and carbonate rock. The workflow relies on using DRP as a complementary and integral technology to conventional SCAL that might help to reduce cost and fill some gaps instead of replacing SCAL.