Using machine learning technology based on fuzzy neural network and geophysical data to predict the shale gas sweet spots: A case study of lower silurian longmaxi formation in Wy Block, Sichuan Basin

Compared with shale gas in north American, it is older, more mature, and significantly difference in China. Taking the Longmaxi Formation of WY block in Sichuan Basin as an example, the quantitative prediction method of sweet spots for high mature shale gas is formed. Firstly, the geophysical response characteristics of shale gas reservoir is studied, and the physical characteristics of the sweet spots are determined. Then, the distributions of reservoir evaluation parameters, such as total organic carbon, brittleness, and others, are obtained by pre-stack simultaneous inversion. Finally, fuzzy neural network method is carried out to predict the distribution of sweet spots and to deploy wells. Shale gas sweet spots in this block vertically concentrate within 30 m above the bottom of the Longmaxi Formation. And two grades of sweet spots are evaluated. The quantitative relationship between the production results of horizontal wells and the evaluation is established with Fuzzy neural network method.