VALIDATION OF METHODS FOR MULTI-SCALE PORE SPACE RECONSTRUCTION AND THEIR USE IN PREDICTION OF FLOW PROPERTIES OF CARBONATE

ABSTRACT The creation of a 3D pore-scale model of a porous medium is an important step in quantitatively characterising the medium and predicting its two-phase and three-phase relative permeabilities and capillary pressures. Recently, we have developed pore architecture models (PAMs) to reconstruct 3D reservoir rocks from 2D thin section images along with a set of pore analysis tools (PATs) to quantitatively analyse the reconstructed pore systems. The PATs reveal the pore topology and geometry, from which we can construct pore network models and compare the original and reconstructed 3D microstructures. Compared to other published reconstruction and network extraction methods, the PAMs and PATs algorithms are computationally significantly more efficient. In this paper we describe a series of tests involving rock PAM reconstructions and extracted networks, and compare the flow properties predicted using network flow simulations. Specifically, we validate the reconstruction method through "self-reconstruction"; i.e. we take numerical thin sections from a reconstructed rock or known 3D tomograph to create a new reconstruction, for which the properties are compared with the original. First, we consider a fairly homogeneous sandstone rock, for which good agreement is found between the original and the new reconstructions in terms of numbers of pores, pore size distribution and connectivity. The permeability of the sandstone example is also closely matched and, in addition, two-phase relative permeabilities for both drainage and imbibition agree well. Second, the reconstruction method is applied to more heterogeneous carbonate rocks. When such materials possess distinct pore systems with wide ranges of pore sizes, we overcome some of the difficulties of the multiple scales by integrating multiple reconstructions based on different resolution input images. Two-phase relative permeabilities are compared for the individual and the combined scale reconstructions. The results reported here indicate that the smaller elements of the pore system are important in governing aspects of the flow. The above suggests that the reconstruction process produces a good representation of homogeneous and some heterogeneous rocks, although prediction of multi-phase flow properties in multi-scale rocks will require larger reconstruction volumes, which seem to be within reach of our efficient methods