Spectral Peak Automatic Separation and Quantitative Evaluation Method of T 1- T 2 2D Nuclear Magnetic Resonance Logging

Summary

T 1-T2 2D nuclear magnetic resonance (2D NMR) logging is widely used in fluid identification in reservoir characterization. However, due to the influence of noise, the signals of different fluids would overlap, and the signals of the same fluid at different depths would have distinct distributions in 2D spectra, both of which would undermine the accuracy of fluid identification and volume calculation. To address these issues, we propose a spectral peak automatic separation (SPAS) method for quantitative fluid evaluation. According to the characteristics of fluid signals in 2D spectra, the novel amplitude descent clustering (ADC) algorithm was studied to automatically separate the multiple fluid peak signals in each 2D spectrum, and the Gaussian mixture model clustering (GMMC) algorithm was used to further divide the completely overlapping signals. Finally, the fluid types could be determined, and the volume of each component would be calculated based on the separated results. To validate its correctness, the proposed method, standalone GMMC algorithm, and blind source separation (BSS) method were used for fluid identification and volume calculation of the synthetic 2D NMR logging data. The accuracy of the processing results was compared, and the volume calculation error of the proposed SPAS method was the lowest. The proposed method was also applied to the logging interpretation of shale oil reservoirs and tight sandstone reservoirs, and the calculated oil volumes were consistent with the pyrolysis experimental results, and they were also consistent with the oil test results. The proposed method could automatically separate spectral peak signals based on 2D NMR characteristics, with high accuracy, stable, and reliable results, and could provide an effective method for 2D NMR logging interpretation and fluid evaluation.