A Production Prediction Method for Shale Gas Wells Based on Multiple Regression

The estimated ultimate recovery (EUR) of a single shale gas well is one the important evaluation indicators for scale and benefit development gas, which affected by many factors such as geological engineering, so its accurate prediction difficult. In order to realize recovery, this study considered 172 wells in Weiyuan block samples selected 19 engineering that affect wells. Furthermore, eight key controlling were means Pearson correlation coefficient maximum mutual information comprehensive method. data divided into training testing samples. Different numbers seven schemes designed. Based on factors, model was established through multiple regression methods. effectiveness verified analyzing result shows with increase size samples, error predicted gradually decreases gradually. When predicting well, average absolute less than 20% if number more 80. potential similar blocks without drilling, sum 10% reaches 60.