Predicting Petrophysical Properties Using 3D Image Data

Fredrich, Joanne T.¹, Matthew M. Haney², Joshua A. White² (1) BP America, Houston, TX (2) Sandia National Laboratories, Albuquerque, NM

Geomaterials exhibit geometrically and topologically complex microstructures that control their macroscopic petrophysical properties. The availability of high-quality 3D image data, coupled with advances in numerical simulation methods, offers unprecedented opportunities for predicting physical and mechanical properties directly from microscopic image data. We consider 3D image data at 1-3 micron resolution obtained for natural and synthetic sandstones during synchrotron computed microtomography experiments performed at the Advanced Photon Source. We apply lattice Boltzmann (LB) methods to predict intrinsic permeability, and describe development of algorithms that improve the computational tractability of this memory-intensive numerical technique. The LB simulations also provide a rich description of the underlying pore-scale hydrodynamics. Next, we perform full-waveform, finite-difference modeling of wave propagation using a numerical approach analogous to the resonant bar technique. We initiate a broadband planar source of energy on one of two opposing boundaries of the digital rock model that are stress-free, while the other four boundaries are periodic and allow the wave to bounce between the two free surfaces for a long period of time. By Fourier-transforming the mean displacement field at the stress-free boundary opposite of the source, we study resonances in the amplitude spectrum in a way analogous to normal-mode seismology. We measure the frequency-dependent elastic properties, and also observe the transition in wave propagation behaviour from an effective medium regime to a regime of strong pore-scale scattering. Ultimately, this full-physics pore-scale modeling will improve estimates of reservoir quality and quantification of reserves, as well as guide development of improved recovery methods.