Predicting Relative Permeability from NMR Relaxation-Diffusion Responses Utilizing High Resolution Micro Xray-CT Images

AlGhamdi, Tariq M. (Saudi Aramco) | Arns, Christoph Hermann (The University of New South Wales)

OnePetro 

NMR relaxation measurements are routinely used in the petroleum industry to estimate permeability and to partition fluids to estimate irreducible water saturation. The shape of the relaxation time distribution is controlled by many mechanisms like pore-coupling in the presence of heterogeneity, internal gradient effects, and signal to noise ratio. However, given an anchoring of the relaxation time distribution, the logarithmic average of the NMR T2 distribution is a relatively robust measure and for rocks where a correlation between pore and throat size exist, a reliable estimate of permeability can often be made. In this work we utilize high resolution X-ray CT images Berea and Bentheimer sandstone and simulate the NMR relaxation-diffusion responses for the case of drainage by a non-wetting fluid at different magnetic field strength (2MHz, 12 MHz, and 400 MHz), calculating internal magnetic fields explicitely. The T2-D responses are projected onto the relaxation axis for each fluid and the SDR model used to predict absolute and relative permeabilities. The resulting correlations between NMR response and relative permeability are surprisingly strong. In particular, reasonable correlations exist between lattice Boltzmann derived relative permeability and NMR estimated relative permeability even for the effective permeability of the oil. This suggests that internal fields help in establishing a surface related/weighted relaxation mechanism for the non-wetting fluid. This methodology allows testing the applicability of SDR type relative permeability estimates for the purpose of log analysis. A variety of cross-correlations including resistivity information can be considered and correlations between relative permeability and NMR response are optimized by finding the best NMR acquisition sequence and interpretation (e.g. choosing optimal cut-offs).