Rock-typing and permeability estimation of thin-bedded reservoir rock by NMR in the presence of diffusion coupling

Abstract

Conventional interpretation approaches to nuclear magnetic resonance (NMR) measurements of fluid-saturated reservoir rock rely on the assumption that the distributions of transverse relaxation time (T2) and pore size directly correlate. In practical scenarios this assumption is found not to apply in numerous multi-scale porosity structures as a result of what is known as “diffusion coupling” occurring between various pores. This problem has been analyzed in the context of individual pores, but less so for larger regions of interest. For gas reservoirs in particular it arises frequently in the case of thinly laminated reservoirs due to their characteristically small layer thickness and the subsequent shorter distances to be covered by mobile spins and the appreciably higher diffusion coefficients that characterize gas reservoirs. In such instances, rock-typing cannot directly be achieved from NMR measurements. This study employs NMR simulations on tomographic images for the interpretation of NMR measurements in the presence of interbed diffusion coupling. Knowledge about the magnetization decay of the coupling region is used together with prior knowledge of the individual rock types forming the layered rocks in a methodical treatment to establish the coupling strength (ξ_R). Following successful completion of rock-typing, the improvement in the estimation of vertical and horizontal permeabilities was evaluated, which relies on a proper definition of T2lm for each rock-type that corrected for diffusion coupling. The Lattice-Boltzmann (LB) method was also used to assess the enhancements in the NMR permeability estimations. Synthetic consolidated and unconsolidated laminated structures with two distinct grain sizes and various layer thicknesses are used to test the approach both numerically and experimentally. A relationship between strengthening pore coupling and reducing bed thickness was noted, together with the increase in the diffusion coefficient and the decrease in surface relaxivity. In instances of strong pore coupling, the T2 distribution was found to inaccurately represent the inherent bimodal distribution relative to various morphologies. Successful rock-typing was attained through the decoupling procedure by applying the value of (ξ_R) that consequently improve the NMR permeability estimation.