Development of visual prediction model for shale gas wells production based on screening main controlling factors

Abstract

For shale gas development, clarification of the main controlling factors of production and estimated ultimate recovery (EUR) with high accuracy is indispensable. The selection of 16 critical parameters directed toward the visual output of the objective function were the most influential factors determined through a sensitivity analysis. Based on the fundamental parameters, the distance correlation coefficient was used to clarify the main controlling factors affecting the EUR of shale gas wells in Weiyuan block. Then, visual forecasting models of EUR were established using Response Surface Method (RSM), Multi-layer Feedforward Neural Network (MLFNN) and Least Square Support Vector Machine (LSSVM). Furthermore, the models developed by the three methods are compared and analyzed. The field application results of the model indicated that the model based on the LSSVM has the best field application effect. The proposed model is a serviceable tool for EUR prediction. In addition, the use of the model is efficient and convenient, and only six main controlling factors can be used to achieve the prediction of EUR. The results of this study can be extended as the main controlling factors analysis and the development of EUR visual model of shale gas wells in other blocks.