Application of Artificial Intelligence in Prediction of Wellbore Stability Using Well Logging and Drilling Data

支持向量机 人工神经网络 理论(学习稳定性) 钻井液 井筒 钻探 非线性系统 计算机科学 石油工程 工程类 人工智能 机器学习 机械工程 量子力学 物理
作者
Juntao Wu,Wei Liu,Hai Lin,Hailong Liu,Chengyong Peng
出处
期刊:50th U.S. Rock Mechanics/Geomechanics Symposium
标识
DOI:10.56952/arma-2023-0598
摘要

ABSTRACT Wellbore instability is one of the most critical challenges during drilling, often manifested as wellbore collapse, shrinkage, falling rocks, and formation fracturing, which may result in complex problems such as pipe sticking, high torque, mud loss, thus impeding the drilling progress and increasing the cost of the drilling operation. Conventional wellbore stability prediction relies on some deterministic physical models, involving some empirical coefficients which are difficult to determine and often dependent on field experience. In addition, some complex factors, such as natural fractures, cannot be explicitly and quantitatively characterized in existing wellbore stability prediction models. Artificial intelligence technique has shown unique advantages in nonlinear issues. The artificial intelligence technique is used to predict wellbore stability in this study, including artificial neural networks (ANNs) and support vector machine (SVM). The logging data and drilling data were collected from the field. According to the correlation analysis between influencing factors and wellbore enlargement rate, 16 parameters were extracted, such as mud density, formation density porosity, acoustic interval transit time, weight on bit as the input data of the models, and wellbore enlargement rate as output. Both SVM and ANNs models have exceptional performance in predicting wellbore stability. When the kernel of the SVM model is Linear, predictions perform optimally. In the ANNs model prediction results, the result performs optimally when the total number of neurons is 1024 in the hidden layer. Overall, ANNs model performs better than SVM model with a coefficient of determination (R2) of 0.991, therefore it is recommended to apply ANNs to predict wellbore stability. The present analysis supplies knowledge that can be used to predict wellbore stability problems before drilling, optimize drilling parameters, and reduce drilling accidents and costs. INTRODUCTION Wellbore instability refer to a series of responses resulting from mechanical, chemical and other effects of the rock around the wellbore. Wellbore instability is one of the most critical challenges during drilling, often manifested as wellbore collapse, shrinkage, falling rocks, and formation fracturing, which may result in complex problems such as pipe sticking, high torque, mud loss, thus impeding the drilling progress and increasing the cost of the drilling operation.

科研通智能强力驱动
Strongly Powered by AbleSci AI
科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
木子完成签到,获得积分10
1秒前
金元宝发布了新的文献求助10
2秒前
花无知发布了新的文献求助10
3秒前
xjc发布了新的文献求助10
4秒前
5秒前
8秒前
8秒前
脑洞疼应助科研通管家采纳,获得10
9秒前
9秒前
9秒前
9秒前
小蘑菇应助科研通管家采纳,获得10
9秒前
9秒前
10秒前
10秒前
10秒前
英俊的铭应助科研通管家采纳,获得10
10秒前
汉堡包应助科研通管家采纳,获得10
10秒前
10秒前
酷波er应助科研通管家采纳,获得10
10秒前
英姑应助科研通管家采纳,获得10
10秒前
赘婿应助科研通管家采纳,获得10
10秒前
Owen应助科研通管家采纳,获得10
10秒前
潇洒的以柳完成签到 ,获得积分10
10秒前
闻山发布了新的文献求助10
11秒前
赤侯发布了新的文献求助10
14秒前
JamesPei应助晓晓采纳,获得10
20秒前
illi完成签到,获得积分10
21秒前
研友_VZG7GZ应助myway采纳,获得10
21秒前
23秒前
赤侯完成签到,获得积分10
24秒前
朴素懿轩完成签到,获得积分10
26秒前
领导范儿应助ghhhn采纳,获得10
27秒前
shuo完成签到,获得积分10
28秒前
Akim应助极品小亮采纳,获得10
29秒前
心酸贝利完成签到,获得积分10
30秒前
情怀应助灰灰采纳,获得10
31秒前
31秒前
Leo完成签到,获得积分10
33秒前
35秒前
高分求助中
(应助此贴封号)【重要!!请各用户(尤其是新用户)详细阅读】【科研通的精品贴汇总】 10000
PowerCascade: A Synthetic Dataset for Cascading Failure Analysis in Power Systems 2000
Picture this! Including first nations fiction picture books in school library collections 1000
Signals, Systems, and Signal Processing 610
Unlocking Chemical Thinking: Reimagining Chemistry Teaching and Learning 555
Photodetectors: From Ultraviolet to Infrared 500
信任代码:AI 时代的传播重构 450
热门求助领域 (近24小时)
化学 材料科学 医学 生物 纳米技术 工程类 有机化学 化学工程 生物化学 计算机科学 物理 内科学 复合材料 催化作用 物理化学 光电子学 电极 细胞生物学 基因 无机化学
热门帖子
关注 科研通微信公众号,转发送积分 6357427
求助须知:如何正确求助?哪些是违规求助? 8172077
关于积分的说明 17206842
捐赠科研通 5413092
什么是DOI,文献DOI怎么找? 2864878
邀请新用户注册赠送积分活动 1842345
关于科研通互助平台的介绍 1690526