8.0 mm 8.0 mm 25.4 mm 25.4 mm (a) (b) 103 Number of Turns: Permeability (mD) 2 3 4 5 102 Calculation Increasing Tortuosity Simulation Measurement 101 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Calculated Porosity (%) (c) Figure 10. Validation of the permeability-tortuosity relationship. (a) A model of a 3-D-printed cylindrical sample with a three-turn helix. (b) A similar model using five turns. (c) Permeability values plotted as a function of the number of turns. The number of turns varied from two to five. Permeability values were computed using (14), simulated using LBM, or determined experimentally. Shaded regions illustrate the 5% error margin found in the dimensions of the 3-D-printed model. (Figure used with permission from [35].) Image registration was performed to align the wet and dry images. Powell's direction set method was used for maximizing the correlation coefficient between the two images. The search space was over the seven parameters of a similarity transform (three translations, three rotations, and one scaling). After registration, the cleat network was visible from the difference image which was then used for porosity and permeability computations. Permeability was found to be highly affected by the existence of minerals. Using demonetization techniques may enhance the recovery of CBM. Accurate chemical characterization can be obtained by using scanning electron microscope with energy dispersive spectrometry (SEM-EDS) and X-ray diffraction methods. Image registration was used successfully in aligning 2-D SEM images with micro-CT slices [39]. Other innovations Most DRP experiments rely on CT scans acquired under ambient temperature and pressure conditions. Rock properties strongly depend on the temperature and pressure conditions of the reservoir. An experimental study of the effect of confining pressure on CT scans was conducted in [40]. A high-pressurehigh-temperature (HPHT) X-ray cell has been manufactured to conduct such experiments. The cell was made from a 21-mmdiameter, 300-mm-long polyetheretherketone rod. A hole 8 mm in diameter was drilled through the rod to accommodate the rock sample. Lab-based experimental permeability values were measured on a Bentheim sandstone rock sample. The experiment was conducted under different confining pressures and the results were compared with numerical simulations conducted using the HPHT cells under comparable conditions. Results demonstrate good agreements between the numerical DRP model and experimental permeability values. However, elastic properties varied considerably between numerical simulations and laboratory results. This mismatch is likely caused by the influence of high pressure on pore structures. Figure 11 shows two segmented CT images acquired using confining pressures of 1 MPa and 20 MPa. These images were segmented using the watershed segmentation algorithm. The white IEEE Signal Processing Magazine | March 2018 | 129