Scholarlink

<p>Reliable reservoir characterization in the offshore Nile Delta is often hindered by intricate channelized depositional architectures, pronounced lateral facies variability, and inconsistent seismic data quality. This study introduces an integrated workflow that combines post-stack acoustic impedance (AI) inversion with probabilistic neural network (PNN) modeling to improve reservoir assessment in the Denise Field, offshore the Nile Delta. The model-based post-stack inversion produced a high-resolution AI volume that exhibits strong agreement (&gt; 90%) with impedance derived from well logs and effectively distinguishes the upper and lower channel systems as low-impedance anomalies indicative of gas-bearing sands. To extend reservoir property prediction beyond well control, a PNN was trained using stepwise-optimized seismic attributes to predict clay volume (Vcl), effective porosity (ϕ), and water saturation (Sw). The PNN predictions exhibit excellent agreement with well-log data, with average correlation coefficients of ~ 0.98 for V<sub>cl</sub> and porosity and ~ 0.87 for water saturation. The resulting 3D property volumes reveal pronounced reservoir heterogeneity and identify laterally continuous channel cores with low shale content, high porosity, and low water saturation. High-quality reservoir zones are concentrated primarily in the western segment of the upper channel, where porosity locally exceeds 25%, and water saturation is generally below 30%. Integrated interpretation of AI, porosity, and saturation volumes confirms strong consistency between elastic and petrophysical indicators, providing independent validation of gas-charged sands and improved confidence in reservoir connectivity. The combined inversion–PNN workflow significantly reduces interpretation uncertainty, enhances predictions of reservoir properties in areas with sparse well control, and provides a robust basis for identifying producible hydrocarbon zones. These findings highlight the value of combining seismic inversion with machine-learning approaches to improve reservoir characterization and support development planning in the offshore Nile Delta and in comparable clastic gas systems worldwide.</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

Overcoming subsurface challenges in the Denise gas field (offshore Nile Delta): an integrated acoustic impedance and machine learning approach for reservoir characterization

  • 期刊论文

  • 发表日期：

    2026年7月23日

查看全文
  • Adel Mahmoud Negm,
  • Mohamed I. Abdel-Fattah,
  • Dhyaa H. Haddad,
  • Mohamed Reda,
  • Ahmed I. Albrkawy,
  • Mansour H. Al‑Hashim

摘要

Reliable reservoir characterization in the offshore Nile Delta is often hindered by intricate channelized depositional architectures, pronounced lateral facies variability, and inconsistent seismic data quality. This study introduces an integrated workflow that combines post-stack acoustic impedance (AI) inversion with probabilistic neural network (PNN) modeling to improve reservoir assessment in the Denise Field, offshore the Nile Delta. The model-based post-stack inversion produced a high-resolution AI volume that exhibits strong agreement (> 90%) with impedance derived from well logs and effectively distinguishes the upper and lower channel systems as low-impedance anomalies indicative of gas-bearing sands. To extend reservoir property prediction beyond well control, a PNN was trained using stepwise-optimized seismic attributes to predict clay volume (Vcl), effective porosity (ϕ), and water saturation (Sw). The PNN predictions exhibit excellent agreement with well-log data, with average correlation coefficients of ~ 0.98 for Vcl and porosity and ~ 0.87 for water saturation. The resulting 3D property volumes reveal pronounced reservoir heterogeneity and identify laterally continuous channel cores with low shale content, high porosity, and low water saturation. High-quality reservoir zones are concentrated primarily in the western segment of the upper channel, where porosity locally exceeds 25%, and water saturation is generally below 30%. Integrated interpretation of AI, porosity, and saturation volumes confirms strong consistency between elastic and petrophysical indicators, providing independent validation of gas-charged sands and improved confidence in reservoir connectivity. The combined inversion–PNN workflow significantly reduces interpretation uncertainty, enhances predictions of reservoir properties in areas with sparse well control, and provides a robust basis for identifying producible hydrocarbon zones. These findings highlight the value of combining seismic inversion with machine-learning approaches to improve reservoir characterization and support development planning in the offshore Nile Delta and in comparable clastic gas systems worldwide.