Journal of Southwest Petroleum University(Science & Technology Edition) ›› 2024, Vol. 46 ›› Issue (3): 1-12.DOI: 10.11885/j.issn.1674-5086.2022.08.20.01

• GEOLOGY EXPLORATION •     Next Articles

Ground Stress Field Inversion and Fracture Prediction Based on MLR-ANN Algorithm

ZHANG Bohu1,2, HU Yao2, WANG Yan2, CHEN Wei2, LUO Chao3,4   

  1. 1. National Key Laboratory of Oil and Gas Reservoir Geology and Exploration, Southwest Petroleum University, Chengdu, Sichuan 610500, China;
    2. School of Geoscience and Technology, Southwest Petroleum University, Chengdu, Sichuan 610500, China;
    3. Provincial Key Laboratory of Shale Gas Evaluation and Exploitation of Sichuan, Chengdu, Sichuan 610056, China;
    4. Shale Gas Research Institute, Southwest Oil & Gas Field Company, PetroChina, Chengdu, Sichuan 610056, China
  • Received:2022-08-20 Published:2024-06-26

Abstract: Shale gas reservoirs are deeply buried in China, and the distribution law of ground stress is complex due to tectonic movement. It is difficult for traditional methods to reflect the magnitude and direction distribution of regional in-situ stress accurately. A coupling algorithm of multiple linear regression and artificial neural network is proposed to invert the shale gas reservoir and surrounding ground stress in Changning-Jianwu Block, southern Sichuan. Using the comprehensive fracture coefficient method, the reservoir fractures are predicted and the fracture development areas are divided. The in-situ stress in the study area is mainly compressive stress, and the direction is about NE115°. The stress around the fault caused by tectonic movement is relatively concentrated, and shear cracks are easy to develop. The cracks are mainly developed and medium developed. The study area has a high degree of fracture development in the upper part of the Wufeng Formation and the structural fault near the bottom of the Longmaxi Formation. The research results have important reference value for well pattern arrangement, fracturing optimization design and casing damage prevention of shale gas extraction.

Key words: multiple linear regression, artificial neural network, shale gas reservoir, ground stress field inversion, coupled algorithm, fracture prediction

CLC Number: