Xinjiang Petroleum Geology ›› 2025, Vol. 46 ›› Issue (5): 622-629.doi: 10.7657/XJPG20250513

• APPLICATION OF TECHNOLOGY • Previous Articles     Next Articles

A Logging-Based Permeability Prediction Method Based on Dual-Driven Model for Low-Permeability Gas Reservoirs: A Case Study of Dongfang Gas Field in Yinggehai Basin

WU Bohan(), LI Fang, TANG Di, WU Yixiong, LUO Yuhu, XIAO Dazhi, ZHANG Shunchao   

  1. Research Institute, Hainan Branch, CNOOC (China) Co., Ltd., Haikou, Hainan 570311, China
  • Received:2024-12-24 Revised:2025-01-19 Online:2025-10-01 Published:2025-09-30

Abstract:

Offshore low-permeability reservoirs are characterized by fine lithology, poor physical property, and strong heterogeneity, making permeability prediction highly challenging. To address this issue, a regression committee learning machine (RCLM) driven by both data and physics was developed for predicting permeability based on logging data for low-permeability reservoirs. On this basis, sweet spot evaluation and dynamic permeability prediction were conducted. The results show that compared with a simple learning machine, the RCLM not only guarantees the prediction accuracy but also achieves higher prediction stability; in comparison with conventional porosity-permeability models, the RCLM obtains superior accuracy (up to 94% within half an order of magnitude). The comprehensive logging-based sweet spot index established using logging curves and petrophysical parameters can be used to effectively identify sweet spots in reservoirs. The newly drilled wells have verified the applicability of the dynamic-static permeability transformation model, which can be used to predict well test-derived permeability during regionally progressive exploration and development. The proposed method has been successfully applied in reservoir evaluation in the Dongfang gas field of the Yinggehai Basin, demonstrating its practical value. This method may provide robust support for efficiently designing exploration and development plan for offshore gas fields.

Key words: Yinggehai Basin, Dongfang gas field, low-permeability reservoir, logging-based permeability, dynamic permeability, sweet spot evaluation, regression committee learning machine

CLC Number: