SPE 115535 Predictive Pore-Scale Modeling: From Three-Dimensional Images to Multiphase Flow Simulations

We demonstrate and validate predictive pore-scale modeling: we start with three-dimensional images of small rock samples obtained using micro-CT scanning with a resolution of a few microns, extract networks from these images and then predict multiphase flow properties by simulating capillary-controlled displacement. We study two sand packs, a poorly consolidated sandstone, Berea sandstone and a carbonate. Single-phase flow properties can be computed on a binarized image directly: we calculate the absolute permeability, resistivity and NMR response. We also extract topologically equivalent networks of pores and throats using a maximal ball method. As a quality control we compare single-phase predictions on these networks with those obtained on the images and from experiment: the permeability and NMR response are similar although we tend to underestimate the resistivity. Networks representing consolidated media tend to over-estimate the magnetization decay in an NMR experiment. We then compute multiphase properties, including capillary pressure, relative permeability and NMR response as a function of wettability (the contact angle distribution assigned to pores and throats). Experimental data, where available, is used to validate our predictions; where we know the wettability and pore structure, we are able to predict multiphase flow properties accurately. We show how relative permeability and capillary pressure is affected by rock type – principally the coordination number of the pores and the pore size distribution – and wettability. We suggest that predictive pore-scale modeling combined with micro-CT imaging is a useful tool, complementary to special core analysis, for the determination of single and multiphase flow properties. Introduction Three-dimensional (3D) images of rock microstructure are the basis for the prediction of rock flow properties. These images are commonly generated using X-ray computed tomography (Hazlet, 1995; Arns et al., 2004), stochastic microstructural modeling, (Adler et al., 1990; Liang et al., 1998; Okabe and Blunt, 2004) or process-based simulation of rock forming processes (Bryant et al., 1993, Bakke and Øren, 1997; Øren and Bakke, 2003; Jin et al., 2003). These images are then used to compute macroscopic flow properties by solving numerically the continuum flow equations governing fluid transport. Several single-phase properties such as permeability, electrical resistivity and NMR response have been predicted from 3D images (Øren et al., 2002; Arns et al., 2004; Knackstedt et al., 2004, Sakellariou et al., 2007) and these have been shown to be in good agreement with conventional laboratory measurements (Arns et. al., 2001; 2002 and 2004). An alternative approach to the prediction of transport properties directly on 3D images is the use of pore-scale network models (Fatt, 1956; Chatzis and Dullien, 1977; Bryant et al., 1993; Blunt et al., 2002). To make accurate predictions they should be derived from 3D images (Lindquist et al., 1996; Bakke and Øren, 1997; Delerue et al., 2002) so as to make them topologically similar to the original samples. Capillary pressure, relative permeability and formation factor have been predicted from network models and have been shown to be in good agreement with experimental data (Bakke and Øren, 1997; Øren and Bakke, 2003, Valvatne and Blunt, 2004, Piri and Blunt 2005b). We generate 3D images of sand packs, a poorly consolidated sandstone, a sandstone and a carbonate using X-ray computed tomography. We then use a maximal ball algorithm (Silin et al., 2003; Silin and Patzek, 2006; Al-Kharusi and Blunt, 2007; Dong, 2007) to extract topologically equivalent networks. For the sand packs we measure permeability, formation factor and magnetization decay which were compared successfully with predictions on the 3D images and networks. Computed singlephase properties for the images and networks of the poorly consolidated sandstone, sandstone and carbonate are in good agreement. We also compute multiphase properties, including capillary pressure, relative permeability and NMR response, as