Scholarlink

<p>Shale reservoirs exhibit complex geological characteristics including ultra-low permeability, nanoscale pores, and significant heterogeneity, making them challenging yet essential unconventional energy resources. Their successful development requires advanced horizontal well stimulation techniques that address reservoir heterogeneity and complex fracture network creation, necessitating a shift from conventional uniform fracturing to intelligent, geology-driven design strategies. This study presents an integrated framework that bridges comprehensive fracturing potential (CFP) evaluation, physics-constrained reservoir classification, and differential fracturing design optimization. First, we develop an adaptive weight determination method based on information entropy theory, incorporating mutual information correction and kernel density estimation to objectively quantify the contributions of mobility and fracability parameters in CFP assessment. Second, we propose a physics-constrained deep clustering algorithm that integrates CFP evaluation results into the reservoir classification process through specialized regularization terms and custom loss functions, categorizing reservoirs into three classes while preserving physical relationships between reservoir properties. Combined with quadratic discriminant analysis (QDA) and SHapley Additive exPlanations (SHAP) interpretation, we establish a quantitative evaluation framework that achieves 90.8% classification accuracy on an independent test set. Third, we establish a differential non-uniform fracturing design methodology based on spatial fracturing potential distribution, demonstrating significant production enhancement compared to conventional uniform designs in numerical simulations and providing a promising approach for field applications. This integrated framework systematically connects quantitative CFP evaluation, geologically-constrained reservoir classification, and targeted fracturing design into a complete workflow from reservoir characterization to stimulation optimization, offering significant potential for improving development economics in heterogeneous shale gas reservoirs.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Physics-constrained deep clustering for integrated reservoir classification and differential fracturing design in shale formations

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月20日

查看全文
  • Yuxuan Deng,
  • Wendong Wang,
  • Yuliang Su,
  • Xinyu Zhuang,
  • Mingwei Jia,
  • Feng Jia

摘要

Shale reservoirs exhibit complex geological characteristics including ultra-low permeability, nanoscale pores, and significant heterogeneity, making them challenging yet essential unconventional energy resources. Their successful development requires advanced horizontal well stimulation techniques that address reservoir heterogeneity and complex fracture network creation, necessitating a shift from conventional uniform fracturing to intelligent, geology-driven design strategies. This study presents an integrated framework that bridges comprehensive fracturing potential (CFP) evaluation, physics-constrained reservoir classification, and differential fracturing design optimization. First, we develop an adaptive weight determination method based on information entropy theory, incorporating mutual information correction and kernel density estimation to objectively quantify the contributions of mobility and fracability parameters in CFP assessment. Second, we propose a physics-constrained deep clustering algorithm that integrates CFP evaluation results into the reservoir classification process through specialized regularization terms and custom loss functions, categorizing reservoirs into three classes while preserving physical relationships between reservoir properties. Combined with quadratic discriminant analysis (QDA) and SHapley Additive exPlanations (SHAP) interpretation, we establish a quantitative evaluation framework that achieves 90.8% classification accuracy on an independent test set. Third, we establish a differential non-uniform fracturing design methodology based on spatial fracturing potential distribution, demonstrating significant production enhancement compared to conventional uniform designs in numerical simulations and providing a promising approach for field applications. This integrated framework systematically connects quantitative CFP evaluation, geologically-constrained reservoir classification, and targeted fracturing design into a complete workflow from reservoir characterization to stimulation optimization, offering significant potential for improving development economics in heterogeneous shale gas reservoirs.