Engineering factor analysis and intelligent prediction of CO2 storage parameters in shale gas reservoirs based on deep learning

CO2 injection has aroused great interest in developing shale gas reservoirs because it can achieve enhancing gas recovery (EGR). However, only some studies have studied the storage effect of the huff-n-puff scheme and fracture parameters during huff-n-puff. Looking at these engineering parameters is crucial for CO2 huff-n-puff to achieve CO2 storage. Besides, scholars have yet to use deep neural networks to intelligently predict the CO2 storage parameters of shale gas reservoirs. Thus, the CO2 huff-n-puff model considering various mechanisms is firstly introduced. Then, the effects of the huff-n-puff scheme and fracture parameters on the storage effect are studied systematically. Finally, the prediction model of CO2 storage capacity and storage factor is established using the Long Short-Term Memory (LSTM) network, and the model is applied to the field shale reservoir in New Albany Shale. Finally, the prediction model of CO2 storage capacity and storage factor is established using the LSTM network and applied to the field shale reservoir in New Albany Shale. This study provides understanding of CO2 huff-n-huff for effective carbon storage based on accurate simulation processes.