Abstract:The measurement of pore size distribution (PSD) in shale by low-field nuclear magnetic resonance (NMR) needs to be carried out at low echo time. However, pseudo-solid materials (i.e., kerogen and structural water) contained in shale can also be detected by NMR, resulting in a questionable PSD derived directly from the transverse relaxation (T2) spectra of oil-saturated shale (NMR-OS method). In this study, a new NMR-Deconvolution (ND) approach for the rapid and accurate characterization of PSD in shale was proposed. The Beta peak model is used to describe each T2 peak distribution of pseudo-solid materials and pore fluids. Estimation and constraint of the number of peaks and peak parameters in the T2 distribution are given by derivative analysis. Finally, the whole T2 spectra of pseudo-solid materials and pore fluids are fitted by the least square method. And verify it and carry out example application. Firstly, this method can fit symmetric and asymmetric peaks at the same time, and the attenuation rate of the peak is high, which can effectively avoid the situation that the baseline is far from the peak. Secondly, the number of peaks and the estimation and constraints of peak parameters can be obtained simply and quickly without complex sample pretreatment processes such as oil washing and drying, and then the solid like interference signals can be separated. Thirdly, for pores smaller than 10 nm, the PSD obtained by ND method compared well with the ones obtained by low-temperature nitrogen adsorption (LTNA) and NMR-O methods. Moreover, the absolute relative errors of pore volume measurement by ND method in the range of T2<1 ms is less than 15%, while the absolute relative error of PV obtained directly from the NMR T2 spectra of oil-saturated shales was up to 135%.