Scholarlink

<p>Fractures play an essential role in the study of reservoir permeability and to improve reservoir modeling, and optimized hydraulic fracturing. This work focuses on a comprehensive study of the fractures identification and fracture properties estimation using well log data in the hydrocarbon reservoir and gas hydrates reservoir of the Krishna-Godavari basin. The natural fractures and breakouts are easily identified from advanced log Formation Micro Imager (FMI). But this tool is costly and not available in many wells and also, the low signatures of fractures in well-log data make it difficult to derive accurate information. Therefore, we adopted an integrated approach such FMI, conventional log responses, semi-automated machine learning (ML) and dual porosity model to identify the fractures. First, the prominent fracture zones have been identified from image of FMI and as well as conventional well log responses. Then, we evaluated the model’s performance of various classifier ML on training and test data, and found that the Random Forest is most suitable techniques for our data set. The model achieved an accuracy of 83.5% on test data and an accuracy of 87.81% and 91% on two validation well data. Lastly, the dual porosity model which&#xa0;is a&#xa0;very effective method was developed from raw resistivity data of FMI log. In this model, bimodal &amp; complex porosity distribution appears near fracture zones and unimodal represents near non-fractured zones. We estimated the various fracture properties fracture density, trace length, aperture, and secondary porosity respectively. The results show that natural fractures have dip angles ranging from 22.53° to 66.55°, with fracture aperture ranging from 2.2 to 11.06&#xa0;mm with mean value of 4.68&#xa0;mm and standard deviation of ± 3.46&#xa0;mm and the fracture porosity ranged from 0.5 to 9% with mean 2.92% and standard deviation of ± 2.82% respectively. This framework provides valuable insights into fractured zone identification, characterization and improves reservoir modeling especially for tight reservoir and gas hydrate formation.</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

A Comprehensive Study on Fracture Identification and Fractures Properties Estimation from Well Log Data in the Hydrocarbon Bearing Formation Using Semi-automated Machine Learning and Dual Porosity Model

  • 期刊论文

  • 发表日期：

    2025年7月18日

查看全文
  • Dip Kumar Singha,
  • Arnab Sarkar

摘要

Fractures play an essential role in the study of reservoir permeability and to improve reservoir modeling, and optimized hydraulic fracturing. This work focuses on a comprehensive study of the fractures identification and fracture properties estimation using well log data in the hydrocarbon reservoir and gas hydrates reservoir of the Krishna-Godavari basin. The natural fractures and breakouts are easily identified from advanced log Formation Micro Imager (FMI). But this tool is costly and not available in many wells and also, the low signatures of fractures in well-log data make it difficult to derive accurate information. Therefore, we adopted an integrated approach such FMI, conventional log responses, semi-automated machine learning (ML) and dual porosity model to identify the fractures. First, the prominent fracture zones have been identified from image of FMI and as well as conventional well log responses. Then, we evaluated the model’s performance of various classifier ML on training and test data, and found that the Random Forest is most suitable techniques for our data set. The model achieved an accuracy of 83.5% on test data and an accuracy of 87.81% and 91% on two validation well data. Lastly, the dual porosity model which is a very effective method was developed from raw resistivity data of FMI log. In this model, bimodal & complex porosity distribution appears near fracture zones and unimodal represents near non-fractured zones. We estimated the various fracture properties fracture density, trace length, aperture, and secondary porosity respectively. The results show that natural fractures have dip angles ranging from 22.53° to 66.55°, with fracture aperture ranging from 2.2 to 11.06 mm with mean value of 4.68 mm and standard deviation of ± 3.46 mm and the fracture porosity ranged from 0.5 to 9% with mean 2.92% and standard deviation of ± 2.82% respectively. This framework provides valuable insights into fractured zone identification, characterization and improves reservoir modeling especially for tight reservoir and gas hydrate formation.