We have shown that by estimating and minimizing several sources of numerical error, very accurate predictions of permeability can be derived directly from digitized microtomographic images at a small scale. The combination of appropriate choice of window sizes on the imaged core and the natural heterogeneity of the rock allowed us to derive permeability- porosity relationships across a range of porosity from four subsamples [≈ (2.7 mm)3]. The utility of small sample sizes highlights a potential for predicting properties from core material not suited for laboratory testing (e.g., drill cuttings, sidewall core and damaged core plugs).
Tomographic imaging facilities now have the ability to image samples in 3D with unprecedented speed and imaging resolution. The combination of high−brilliance X-ray sources, fast high-resolution X-ray detector systems, high-speed data networks, and large-scale computation gives one the ability to virtually reconstruct 3D images in real time. Computational techniques and hardware have similarly progressed to the point where one can calculate properties on 3D voxelated images over short time scales. One can envision the routine calculation of the porosity−permeability relationship via numerical techniques on core material for a specific reservoir. The total computational time required to calculate the data points in Fig. 2 was = 150 processor days, with moderate computational resources (e.g., a 64-node pc cluster) this permeability−porosity correlation could be generated in 60 h. Optimising the LB permeability solver or use of a finite difference code may further reduce the computational loads. The parallel development of these experimental and computational methods demonstrates the potential to develop a virtual core laboratory, a facility for the imaging and calculation of petrophysical properties on core material. This virtual lab has enhanced capabilities as one may obtain data from core material which to date have been considered unsuitable for testing. Moreover, numerical simulation offers the ability to measure local information (e.g.., flow paths, tortuosity) on complex samples, information which to date is not experimentally realisable.