Real-Time Monitoring Model of PDC Bit Wear Based on GRU Neural Network

doi:10.12388/j.issn.1673-2677.2024.02.003

Abstract

Abstract:

Real-time monitoring of bit wear is crucial for accelerating drilling operations. However，it is challenging to measure on-site parameters that directly reflect levels of bit wear. Currently，there are few means of monitoring bit wear，and in most cases，determination of bit wear is empirically performed by technicians. Quantitatively evaluating the wear of PDC bit has always been a difficult task. The evaluation of bit wear is primarily based on rock breaking efficiency and mechanical specific energy. In this study，a model is proposed for real-time monitoring of PDC bit wear，based on a physical model to calculate mechanical specific energy. Moreover，the wavelet analysis and clustering algorithm are utilized to characterize the bit wear process. Finally，a monitoring model based on Gated Recurrent Unit （GRU） neural network is established，which maps drilling parameters to bit wear levels with 95% accuracy. The model is tested using data from Well A in Xinjiang Oilfield，which demonstrates the capability of the model to accurately estimate current bit wear levels. This model provides a solution for bit wear monitoring，aiding engineers in determining the optimal timing for bit replacement and thereby ensuring higher drilling efficiency.

Key words:

drilling, bit wear, clustering algorithm, wavelet analysis, GRU neural network, machine learning

摘要：

能够实时监测钻头磨损程度对于钻井提速是一个直观的参考目标。但钻井现场难以采集直接反映钻头磨损情况的参数，目前对钻头磨损程度的监测手段较少，主要依靠技术人员的经验判断。如何定量评估PDC钻头磨损程度一直是研究的难点。钻头磨损程度评价主要基于破岩效率和机械比能。通过物理模型计算机械比能，并通过小波分析、聚类算法表征钻头磨损过程，建立了基于门控循环单元（GRU）神经网络的PDC钻头磨损实时监测模型，形成了钻井参数与钻头磨损程度的映射关系，模型精度达95%。采用新疆油田A井数据对模型进行测试，结果表明该模型可以正确预测当前钻头磨损级别。该模型为钻头磨损监测提供了一种解决方案，可以辅助现场工程师判断起下钻时机，以保证更高的钻井效率。

关键词:

钻井, 钻头磨损, 聚类算法, 小波分析, GRU神经网络, 机器学习

CLC Number:

TE24 ','1');return false;" target="_blank">
TE24

ZHONG Yiming, KEDILIYA Palihati, BAI Jiashuai, WANG Chaochen, LI Qihao.

Real-Time Monitoring Model of PDC Bit Wear Based on GRU Neural Network

[J]. Xinjiang Oil & Gas, 2024, 21(2): 21-.

钟尹明, 柯迪丽娅·帕力哈提, 白佳帅, 王超尘, 李起豪.

基于GRU神经网络的PDC钻头磨损实时监测模型

[J]. 新疆石油天然气, 2024, 21(2): 21-.

References

［1］严博宇. 钻头磨损机理及改进技术研究［J］. 西部探矿工程，2023，35（11）：58-60.YAN B Y. Research on drill bit wear mechanism and improvement technology［J］. West-China Exploration Engineering，2023，35（11）：58-60.

［2］高德利，刘维，万绪新，等. PDC钻头钻井提速关键影响因素研究［J］. 石油钻探技术，2023，51（4）：20-34.GAO D L，LIU W，WAN X X，et al. Study on key factors influencing the ROP improvement of PDC bits［J］. Petroleum Drilling Techniques，2023，51（4）：20-34.

［3］刘少洋. 基于深度学习的多工况钻头磨损状态监测研究［D］. 黑龙江哈尔滨：哈尔滨理工大学，2022.LIU S Y. Research on multi-condition drill bit wear state monitoring based on deep learning［D］. Harbin，Heilongjiang：Harbin University of Science and Technology，2022.

［4］WILSON A. Real-time data analytics allows for bit-wear monitoring and prediction［J］. Journal of Petroleum Technology，2018，70（12）：67-68.

［5］WAUGHMAN R J，KENNER J V，MOORE R A. Real-time specific energy monitoring enhances the understanding of when to pull worn PDC bits［J］. SPE Drilling & Completion，2003，18（1）：59-67.

［6］宋先知，裴志君，王潘涛，等. 基于支持向量机回归的机械钻速智能预测［J］. 新疆石油天然气，2022，18（1）：14-20.SONG X Z，PEI Z J，WANG P T，et al. Intelligent prediction for rate of penetration based on support vector machine regression［J］. Xinjiang Oil & Gas，2022，18（1）：14-20.

［7］DUPRIEST F E，KOEDERITZ W L. Maximizing drill rates with real-time surveillance of mechanical specific energy［J］. Society of Petroleum Engineers，2005.

［8］ABBAS R K，HASSANPOUR A，HARE C，et al. Instantaneous monitoring of drill bit wear and specific energy as a criteria for the appropriate time for pulling out worn bits［C］. SPE Annual Caspian Technical Conference and Exhibition，Astana，Kazakhstan，November 2014.

［9］RASHIDI B，HARELAND G，FAZAELIZADEH M，et al. Comparative study using rock energy and drilling strength models［C］. 44th US Rock Mechanics Symposium and 5th US-Canada Rock Mechanics Symposium，Salt Lake City，USA，June 2010.

［10］李根生，穆总结，田守嶒，等. 冲击破岩钻井提速技术研究现状与发展建议［J］. 新疆石油天然气，2024，20（1）：1-12.LI G S，MU Z J，TIAN S C，et al. Research status and development proposal of ROP improvement technology with percussion rock-breaking method［J］. Xinjiang Oil & Gas，2024，20（1）：1-12.

［11］张佳伟，孟昭，纪国栋，等. PDC钻头破岩效率及稳定性室内试验研究［J］. 石油机械，2020，48（12）：35-43、51.ZHANG J W，MENG Z，JI G D，et al. Laboratory experimental study on rock breaking efficiency and stability of PDC bit［J］. China Petroleum Machinery，2020，48（12）：35-43，51.

［12］王斌. 双级PDC钻头破岩效率的实验研究［D］. 山东青岛：中国石油大学（华东），2020.WANG B. Experimental study on rock cutting efficiency of two stage PDC bit［D］. Qingdao，Shandong：China University of Petroleum （East China），2020.

［13］李宏波，罗平亚，白杨，等. 机器学习算法概述及其在钻井工程中的应用［J］. 新疆石油天然气，2022，18（1）：1-13.LI H B，LUO P Y，BAI Y，et al. Summary for machine learning algorithms and their applications in drilling engineering［J］. Xinjiang Oil & Gas，2022，18（1）：1-13.

［14］CHEN X Y，YANG J，GAO D. Drilling performance optimization based on mechanical specific energy technologies［J］. Drilling，2018，35：133-161.

［15］TEALE R. The concept of specific energy in rock drilling［J］. International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts，1965，2（1）：57-73.

［16］PESSIER R C，FEAR M J. Quantifying common drilling problems with mechanical specific energy and a bit-specific coefficient of sliding friction［C］. SPE Annual Technical Conference and Exhibition，Washington DC，USA，1992.

［17］ARMENTA M. Identifying inefficient drilling conditions using drilling-specific energy［C］. SPE Annual Technical Conference and Exhibition，Denver，Colorado，USA，September 2008.

［18］HAMERLY G，ELKAN C. Learning the k in k-means［J］. Advances in Neural Information Processing Systems，2003：16.

［19］CHO K，VAN MERRIËNBOER B，GULCEHRE C，et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation［C］. arXiv preprint arXiv：1406.1078，2014.

［20］王海涛，王建华，邱晨，等. 基于双向长短期记忆循环神经网络和条件随机场的钻井工况识别方法［J］. 石油钻采工艺，2023，45（5）：540-547、554.WANG H T，WANG J H，QIU C，et al. Recognition method of drilling conditions based on bi-directional long-short term memory recurrent neural network and conditional random field［J］. Oil Drilling & Production Technology，2023，45（5）：540-547，554.

［21］YU Y，SI X S，HU C H，et al. A review of recurrent neural networks：LSTM cells and network architectures［J］. Neural Computation，2019，31（7）：1235-1270.

[1]	SHENG Mao, JIANG Fumin, HU Shimeng, LYU Zhenhu, KONG Mingwei, SI Lijuan, . The In-Situ Stress Interpretation in Heterogeneous Glutenite Reservoirs and Fracturing Design Optimization [J]. Xinjiang Oil & Gas, 2024, 21(2): 1-.
[2]	TIAN Long, ZHU Zhihua, WANG Liwei, YU Jiawei, WANG Yifan. A Rock Drillability Characterization Method Based on Big Data and Unsupervised Clustering Algorithm [J]. Xinjiang Oil & Gas, 2024, 21(2): 29-.
[3]	LI Gensheng, MU Zongjie, TIAN Shouceng, HUANG Zhongwei, SUN Zhaowei. Research Status and Development Proposal of ROP Improvement Technology with Percussion Rock-Breaking Method [J]. Xinjiang Oil & Gas, 2024, 20(1): 1-12.
[4]	LIU Fan, LIU Yushuang, ZHANG Zhen, LI Yongjian, LIU Ce, MA Zhihu, . A Machine Learning-Based Bridging Particle Size Recommendation Method for Lost Circulation Control [J]. Xinjiang Oil & Gas, 2024, 20(1): 13-20.
[5]	SONG Xianzhi, GUO Yong, XIANG Dongmei, JU Pengfei, TAN Qiang, LIU Wei. Multi-Field Coupling Analysis of Wellbore Instability in Hutubi Anticline While Using Water-Based Drilling Fluid [J]. Xinjiang Oil & Gas, 2023, 19(4): 1-9.
[6]	LUO Ling , LI Yichang , WEI Liyao. Use of Metaverse Technology Clusters and Quantum Computing in CCUS：Research Status and Trends#br# [J]. Xinjiang Oil & Gas, 2023, 19(3): 86-94.
[7]	LI Hongbo, LUO Pingya, BAI Yang, LI Daoxiong, CHANG Shuang, LIU Xinguo . Summary for Machine Learning Algorithms and Their Applications in Drilling Engineering [J]. Xinjiang Oil & Gas, 2022, 18(1): 1-13.
[8]	SONG Xianzhi, PEI Zhijun, WANG Pantao, ZHANG Gonglingyan, YE Shanlin. Intelligent Prediction for Rate of Penetration Based on Support Vector Machine Regression [J]. Xinjiang Oil & Gas, 2022, 18(1): 14-20.