A new technique for synthetizing capillary pressure (Pc) curves using NMR logs in tight gas sandstone reservoirs

Formation pore structure prediction is of great importance in tight gas sands evaluation, and NMR logs are considered to be an advantage in this aspect. The common method that is used to evaluate formation pore structure from NMR logs is transforming the NMR T-2 spectra as capillary pressure curves and then extracting the pore throat radius distribution from the synthetized capillary pressure curves and predicting pore structure evaluation parameters. However, field applications demonstrate that the existing methods are not always practicable; additionally, many existing methods involve input parameters that are difficult to obtain, especially in tight gas sandstone reservoirs. This study is based on the analysis of the morphology of lab mercury injection capillary pressure (MICP) and nuclear magnetic resonance (NMR). This experiment used 20 core samples that were drilled from tight gas sands of the central Sichuan basin. Our findings indicate that rock pore structure, the values of mercury injection saturation (S-Hg) under every capillary pressure (P-c), and pore component percentage compositions are interrelated. To quantitatively express this interrelationship, seven T2cutoffS, which are used to classify NMR T-2 spectra into eight parts, are defined to calculate eight pore component percentage compositions from the NMR logs. A novel technique, which connects the S(Hg)s and eight pore component percentage compositions, is proposed to construct pseudo P-c curves. Once this technique is extended to field applications, consecutively synthetic P-c curves can be predicted in the intervals with which field NMR logs were acquired. Additionally, pore throat size distribution and pore structure evaluation parameters can also be calculated. A field example of the Sichuan basin illustrates that this technique is reliable and that the predicted results from the field NMR logs match the core experimental measurements very well. The greatest advantage of this proposed technique is that no intermediate parameters need to be acquired from other methods. (C) 2016 Elsevier B.V. All rights reserved.