Statistically Generated Medium

• http://www.sciencedirect.com/science/article/pii/S0309170816301385
• 10.1016/j.advwatres.2016.05.009

Abstract — A quasi-static scheme based on pore space spatial statistics is presented to simulate pore-scale two-phase capillary-dominant displacement processes. The algorithm is coupled with computational fluid dynamics in order to evaluate saturation functions. Wettability heterogeneity in partial and fractional/mixed-wet media is implemented using a contact angle map. The simulation process is pixel-wised and performed directly on binary images. Bypassing and snap-off are tackled as non-wetting phase trapping mechanisms. Post-processing results include residual saturations, effective permeability and capillary pressure curves for drainage and imbibition scenarios. The primary advantages of the proposed workflow are eliminating pore space skeletisation/ discretization, superior time efficiency and minimal numerical drawbacks when compared to other direct or network-based simulation techniques.

• http://www.sciencedirect.com/science/article/pii/S0920410516302698
• 10.1016/j.petrol.2016.07.010

Abstract — Numerical experiments are performed to evaluate the capability and examine the computational costs and tradeoffs of numerical and analytical approaches for apparent gas permeability calculation. Considering the dominance of capillary force in pore-level microstructures, two-phase distribution maps are constructed within micro- and nano-scale media using a quasi-static pore morphology-based technique. At each equilibrium step, the gas saturation profile is extracted to conduct single phase flow simulations and predict the gas effective permeability. To investigate the effect of gas slippage on the accuracy of the results, both Computational Fluid Dynamics (CFD) and Direct Simulation Monte Carlo (DSMC) techniques are applied in a wide range of gas pressures. The results reveal that the continuum theory is valid for a specific pore size and gas pressure range and its accuracy is strongly dependent on the Knudsen number value. DSMC simulation case studies demonstrate that in case of proper medium characterization, a tuned Klinkenberg correlation is capable of providing satisfying predictions within both slippage and transition flow regimes.