Scholarlink

<p>Experience from the oil and gas industry offers valuable insights that can be applied to renewable energy projects, particularly in offshore environments. This study explores the use of offshore drilling data to support the development of fixed offshore wind turbine foundations by improving predictions of the rate of penetration (ROP). Drilling data from seven wells in the X oilfield in the North Sea—focusing on the large-diameter 17-1/2″ interval—were analyzed using artificial neural network (ANN) regression. The model incorporated eight drilling parameters (depth, weight on bit, standpipe pressure, torque, maximum torque, surface round per minute, bit round per minute, and mud flow rate) across depths of 0–125&#xa0;m below seafloor. An initial ANN model trained on a single well achieved 0.946 of the coefficients of determination, outperforming previous methods such as the Bourgoyne and Young model approach. When data from all seven wells were combined, the model maintained a high coefficient of determination of 0.883, despite variations in lithology. Depth, standpipe pressure, surface round per minute, and torque were identified as key factors influencing ROP. This study highlights the potential of machine learning in offshore drilling applications.</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

Predicting rate of penetration in large-diameter offshore drilling using artificial neural networks

  • 期刊论文

  • 发表日期：

    2025年10月27日

查看全文
  • Muhammed Şuayip Akkuş,
  • Abbas Abbasov,
  • Şükrü Merey

摘要

Experience from the oil and gas industry offers valuable insights that can be applied to renewable energy projects, particularly in offshore environments. This study explores the use of offshore drilling data to support the development of fixed offshore wind turbine foundations by improving predictions of the rate of penetration (ROP). Drilling data from seven wells in the X oilfield in the North Sea—focusing on the large-diameter 17-1/2″ interval—were analyzed using artificial neural network (ANN) regression. The model incorporated eight drilling parameters (depth, weight on bit, standpipe pressure, torque, maximum torque, surface round per minute, bit round per minute, and mud flow rate) across depths of 0–125 m below seafloor. An initial ANN model trained on a single well achieved 0.946 of the coefficients of determination, outperforming previous methods such as the Bourgoyne and Young model approach. When data from all seven wells were combined, the model maintained a high coefficient of determination of 0.883, despite variations in lithology. Depth, standpipe pressure, surface round per minute, and torque were identified as key factors influencing ROP. This study highlights the potential of machine learning in offshore drilling applications.