Techno-economic integration evaluation in shale gas development based on ensemble learning

Abstract

For the development of shale gas, the accurate prediction of estimated ultimate recovery (EUR) has invariably been a hot and arduous issue that has attracted abundant attention from researchers. However, the intricate relationship between EUR and economic benefits of shale gas wells is frequently disregarded. Therefore, based on the basic geological and engineering parameters, this study carried out a joint multi-task modeling of investment cost and EUR evaluation, and creatively constituted a techno-economic integration evaluation framework for shale gas wells with internal rate of return (IRR) as the economic benefit evaluation target. Furthermore, the interaction graphs of investment cost and EUR on IRR are delineated to intuitively exemplify the relationship between EUR and economic benefits (IRR). The validity of the model is verified by the field data from 231 wells. The results show that the techno-economic integration evaluation framework of Blendstacking, which integrates multi-task joint modeling and integrated learning, can reliably evaluate investment costs and EUR. Concurrently, based on the evaluation results, the accurate prediction of IRR is realized. The mean prediction errors of investment cost and EUR are inferior to 50 ×104USD and 1400 ×104m3, respectively, and the mean error of IRR is regulated within 2.0%. This work can quickly and effectively predict the economic benefits of gas wells under complex geological and engineering factors, which facilitates expeditiously developing decision making. The research method can be extended to the economic benefit evaluation of other instance well datasets.