Scholarlink

<p>This study presents an integrated approach for quantitative fluid discrimination in the Ghar sandstone reservoir through the combined application of amplitude versus offset (AVO) analysis, Extended Elastic Impedance (EEI) inversion, and semi-supervised Probabilistic Neural Network (PNN) machine learning techniques. The methodology addresses critical challenges in hydrocarbon detection within complex reservoir environments where conventional seismic interpretation methods prove insufficient. The theoretical framework encompasses comprehensive AVO attribute generation utilizing Zoeppritz equation approximations, including intercept, gradient, AVO product, fluid factor, gas indicator, and Poisson’s Ratio Contrast attributes. Extended Elastic Impedance analysis was systematically applied across eight wells to determine optimal chi angles through cross-correlation analysis, identifying a consistent optimal chi angle of 32° for water saturation discrimination. The EEI methodology enables enhanced sensitivity to fluid variations through angle-dependent projections of elastic properties, addressing limitations in conventional elastic impedance approaches. Pre-stack simultaneous seismic inversion generated high-resolution P-impedance, S-impedance, and density volumes, providing foundation for comprehensive AVO attribute extraction and EEI analysis. Integration of multiple seismic attributes through EEI attribute volume generation enabled spatial extrapolation of reservoir property relationships from well locations to the complete survey area. The final stage employed semi-supervised PNN analysis utilizing EEI and AVO attribute maps as input parameters for training and prediction of oil, gas, and water distribution. Results demonstrate successful fluid discrimination with gas accumulations concentrated in structural crests, oil presence along southern flanks, and water distribution throughout broader reservoir intervals. Gas discrimination predictions were fully validated against production data and gas-oil contact surface, while oil distribution results can be verified using gas–oil and oil–water contact surfaces (GOC and OWC). However, water distribution probability maps remain unverified due to the absence of available well data for validation. The integrated approach provides a robust framework for reducing exploration risks and optimizing field development strategies in complex reservoir environments, demonstrating enhanced fluid discrimination capabilities through advanced seismic analysis techniques.</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

Quantitative fluid discrimination through AVO/EEI analysis and semi-supervised machine learning method: a case study from the Bahregansar field, Ghar sandstone reservoir

  • 期刊论文

  • 发表日期：

    2026年1月29日

查看全文
  • Hessam Mansouri Siahgoli,
  • Mohammad Ali Riahi,
  • Majid Nabi-Bidhendi,
  • Mohammad Emami Niri

摘要

This study presents an integrated approach for quantitative fluid discrimination in the Ghar sandstone reservoir through the combined application of amplitude versus offset (AVO) analysis, Extended Elastic Impedance (EEI) inversion, and semi-supervised Probabilistic Neural Network (PNN) machine learning techniques. The methodology addresses critical challenges in hydrocarbon detection within complex reservoir environments where conventional seismic interpretation methods prove insufficient. The theoretical framework encompasses comprehensive AVO attribute generation utilizing Zoeppritz equation approximations, including intercept, gradient, AVO product, fluid factor, gas indicator, and Poisson’s Ratio Contrast attributes. Extended Elastic Impedance analysis was systematically applied across eight wells to determine optimal chi angles through cross-correlation analysis, identifying a consistent optimal chi angle of 32° for water saturation discrimination. The EEI methodology enables enhanced sensitivity to fluid variations through angle-dependent projections of elastic properties, addressing limitations in conventional elastic impedance approaches. Pre-stack simultaneous seismic inversion generated high-resolution P-impedance, S-impedance, and density volumes, providing foundation for comprehensive AVO attribute extraction and EEI analysis. Integration of multiple seismic attributes through EEI attribute volume generation enabled spatial extrapolation of reservoir property relationships from well locations to the complete survey area. The final stage employed semi-supervised PNN analysis utilizing EEI and AVO attribute maps as input parameters for training and prediction of oil, gas, and water distribution. Results demonstrate successful fluid discrimination with gas accumulations concentrated in structural crests, oil presence along southern flanks, and water distribution throughout broader reservoir intervals. Gas discrimination predictions were fully validated against production data and gas-oil contact surface, while oil distribution results can be verified using gas–oil and oil–water contact surfaces (GOC and OWC). However, water distribution probability maps remain unverified due to the absence of available well data for validation. The integrated approach provides a robust framework for reducing exploration risks and optimizing field development strategies in complex reservoir environments, demonstrating enhanced fluid discrimination capabilities through advanced seismic analysis techniques.