Fracture Conductivity Prediction Based on Machine Learning

Xiaopeng Wang; Binqi Zhang; Jianbo Du; Dongdong Liu; Qilong Zhang; Xiaoqiang Liu

doi:10.1021/acsomega.4c00448

Fracture Conductivity Prediction Based on Machine Learning

ACS Omega. 2024 Mar 8;9(11):13469-13480. doi: 10.1021/acsomega.4c00448. eCollection 2024 Mar 19.

Authors

Xiaopeng Wang^{1

2}, Binqi Zhang^{1

2}, Jianbo Du^{1

2}, Dongdong Liu^{1

2}, Qilong Zhang^{1

2}, Xiaoqiang Liu^{3

4}

Affiliations

¹ State Key Laboratory of Offshore Oil Exploitation, Beijing 100028, China.
² Tianjin Branch of CNOOC Ltd., Tianjin 300450, China.
³ School of Energy Resources, China University of Geosciences Beijing, Beijing 100083, China.
⁴ School of Earth and Spacing Sciences, Peking University, Beijing 100871, China.

Abstract

Hydraulic fracturing technology is the main method to develop low-permeability reservoirs. Fracture conductivity is not only the basis of fracture optimization design but also one of the key parameters to determine the effect of hydraulic fracturing. However, current methods of calculating fracture conductivity require a lot of time and labor cost. This research proposes a fracture conductivity prediction model based on machine learning. The main controlling factors of fracture conductivity are determined using the Pearson coefficient method and gray correlation analysis. Example application shows that the R² values of the BP neural network model based on a genetic algorithm for predicting the fracture conductivity of block A and block B are 0.981 and 0.975, respectively, indicating that the machine learning model can accurately predict fracture conductivity.