Pore structure evaluation was of great importance in improving reservoirs characterization and validity prediction, especially in unconventional reservoirs. The nuclear magnetic resonance (NMR) log was considered to be advantageous in formation pore structure characterization, and many methods had been proposed to evaluate pore structure by using NMR log. However, NMR log can only be used to characterize rock pore structure under fully water saturation due to the domination of surface relaxation. In hydrocarbon-bearing formation, pore structure would be over or underestimated because of the effect of viscosity to the shape of NMR T2 spectrum.
In this study, to understand the effect of saturated hydrocarbon to NMR responses, 20 core samples, which were separately drilled from ultra-tight, low permeability and conventional sandstone reservoirs, were applied for laboratory NMR experimental measurements under four saturation conditions. These four saturation conditions included irreducible water saturation, fully water saturated, oil saturated and residual oil. To quantize the effect of crude oil viscosity to NMR T2 spectra, underground oil was collected and processed in laboratory to simulate the used oil with 3 different viscosities in the NMR experiment. NMR experiments under residual oil saturation was used to simulate field NMR log due to shallow invasion radius of field NMR tool. Experimental results illustrated that NMR T2 distributions were heavily affected by crude oil viscosity. For core samples drilled from low permeability sandstone reservoirs, and saturated with light oil, NMR T2 spectra all exhibited as bimodal distributions, even if they were unimodal under fully water saturation. The shapes and locations of the left spectra (reflecting micro to small pore size) were similar with those of irreducible water. However, the shapes and locations of the right spectra (reflecting free fluids) were predominantly determined by the property of used oil in the experiments, and they were similar with the bulk relaxation T2 distributions of oil. In this case, pore structure would be overestimated if field NMR log was directly used for pore structure characterization. Comparing with NMR T2 distributions acquired from core samples with high porosity and permeability, and fully water saturated, the morphology of T2 spectra hardly changes after the pore spaces saturated with light oil. However, once the viscosity of saturated oil increased, the location of the right peak moved to the left, and they still exhibited similar position with the bulkrelaxationT2 distributions of saturated oil. Under extreme conditions the NMR T2 spectra of free fluids overlapped with those of irreducible water once the viscosity of saturated oil increased to 110.0 mPa.s. Hence, formation pore structure should be underestimated by using NMR log in heavy oil-bearing reservoirs. In addition, for core samples drilled from ultra-tight sandstones, the NMR T2 distributions exhibited as similar morphological characteristics under four saturation conditions. The reason was that oil cannot be injected into the pore spaces due to high capillary pressure and poor pore connectivity in such type of rocks.
These experimental results could give us good indication that the shapes of field NMR T2 spectra should be first corrected to remove the effect of crude oil before they were used for formation pore structure characterization at a particular state of saturation.