Real-time and multi-objective optimization of rate-of-penetration using machine learning methods

Zhang, Chengkai; Song, Xianzhi; Liu, Zihao; Ma, Baodong; Lv, Zehao; Su, Yinao; Li, Gensheng; Zhu, Zhaopeng

doi:10.1016/j.geoen.2023.211568

Cited by 9 publications

(4 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Mendes 12 also presented a methodology based on a neural network model for ROP and a neuro-genetic adaptive controller to address the problem that relationships between operational variables affecting ROP are complex and not easily modeled. In addition, with the boom in ML algorithms approximately 2010, more and more ML methods are being used for ROP prediction, including Moran 13 , Arabjamaloei 14 , Esmaeili 15 , Ning 16 , Zare 17 , Bodaghi 18 , Hegde 19 , Mantha 20 , Hegde 21 , Anemangely 22 , Soares 7 , Sabah 23 , Felipe 2 , Korhan 24 , Li 25 , Mohammad 26 , Gan 27 , Hazbeh 28 , Salaheldin 29 , Zhang 30 , Ren 31 , Zhang 32 , Brenjkar 33 , Riazi 34 , Song 35 , Wang 36 , Mohammad 37 , Kaveh 38 and so on. Judging from the increasing number of articles published each year in recent years on the use of machine learning for ROP prediction, it can be amply demonstrated that ML methods are well suited for application in the field of ROP prediction.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Hybrid physics-machine learning models for predicting rate of penetration in the Halahatang oil field, Tarim Basin

Jiao,

Li,

et al. 2024

Sci Rep

Self Cite

View full text Add to dashboard Cite

Rate of penetration (ROP) is a key factor in drilling optimization, cost reduction and drilling cycle shortening. Due to the systematicity, complexity and uncertainty of drilling operations, however, it has always been a problem to establish a highly accurate and interpretable ROP prediction model to guide and optimize drilling operations. To solve this problem in the Tarim Basin, this study proposes four categories of hybrid physics-machine learning (ML) methods for modeling. One of which is residual modeling, in which an ML model learns to predict errors or residuals, via a physical model; the second is integrated coupling, in which the output of the physical model is used as an input to the ML model; the third is simple average, in which predictions from both the physical model and the ML model are combined; and the last is bootstrap aggregating (bagging), which follows the idea of ensemble learning to combine different physical models’ advantages. A total of 5655 real data points from the Halahatang oil field were used to test the performance of the various models. The results showed that the residual modeling model, with an R2 of 0.9936, had the best performance, followed by the simple average model and bagging with R2 values of 0.9394 and 0.5998, respectively. From the view of prediction accuracy, and model interpretability, the hybrid physics-ML model with residual modeling is the optimal method for ROP prediction.

show abstract

Section: Introductionmentioning

confidence: 99%

“…In 2023, Zhang used RF, ANN and SVM combined with real-time workflow to predict drilling speed in real time 35 and optimized drilling parameters through the NSG- III algorithm through an objective function for ROP and MSE to obtain a better real-time prediction effect during drilling.…”

Section: Introductionmentioning

confidence: 99%

Hybrid physics-machine learning models for predicting rate of penetration in the Halahatang oil field, Tarim Basin

Jiao,

Li,

et al. 2024

Sci Rep

Self Cite

View full text Add to dashboard Cite

show abstract

“…Hareland equation [6], Detournay equation [7], Motahhari equation [8], and the modified Yang equation [9]. Researchers [10][11][12] have determined the coefficients of drilling rate equations using field drilling data through multivariate regression to predict and optimize drilling rates. Physics-based drilling rate models align with drilling principles, exhibit good comprehensibility, have a simple form, and offer easy parameter determination.…”

Section: Introductionmentioning

confidence: 99%

“…However, the current literature on drilling rate prediction primarily relies on post-drilling analysis scenarios, where researchers use historical drilling data for modeling and prediction, making full use of all drilling data. Only a limited number of researchers [10,12,18,20,27] have worked on establishing drilling rate models based on real-time drilling scenarios, utilizing partial, real-time transmitted data. Few have studied the dynamic, uncertain, and real-time aspects of the drilling process.…”

Section: Introductionmentioning

confidence: 99%

Real-time prediction of oil and gas drilling rate based on physics-based model and particle filter method

Zhang,

Song,

2024

Preprint

View full text Add to dashboard Cite

Oil and gas drilling, essential for exploring and exploiting petroleum resources, involves significant time, labor, and costs, often exceeding $300,000 daily. Predicting the drilling rate (Rate of Penetration, ROP) accurately and promptly is crucial for improving efficiency and reducing expenses. In drilling, physicsbased and machine learning models are typically used for ROP forecasting. Physics-based models, while intuitive, often lack precision in complex conditions. Machine learning models, though precise, face challenges with data availability and training costs in real-time settings. This paper introduces a novel approach combining a physics-based model with a particle filter algorithm for real-time ROP prediction. It adapts the Bourgoyne-Young ROP model and Markov assumptions into a state space model, using the particle filter to estimate elusive coefficients through probability theory. This enables real-time data updates for more accurate ROP predictions. The proposed framework is evaluated against traditional models using open-source and field drilling datasets in post-drilling and real-time scenarios. Results show conventional physics-based models fall short in both scenarios, while machine learning and the new particle filter model show significant improvements. In post-drilling analysis, these models achieve under 5% mean relative error. For real-time predictions, machine learning models have over 20% error, but the particle filter model reduces this to approximately 15%. This highlights the particle filter model's superiority in accuracy and cost-effectiveness under dynamic and uncertain drilling conditions. This paper presents a robust, efficient solution for ROP prediction and optimization, marking a significant advancement in the drilling field.

show abstract

Modeling CO2 Loading Capacity of Diethanolamine (DEA) Aqueous Solutions Using Advanced Deep Learning and Machine Learning Algorithms: Application to Carbon Capture

Mahmoudzadeh,

Hadavimoghaddam,

Atashrouz

et al. 2024

Korean J. Chem. Eng.

View full text Add to dashboard Cite

Real-time and multi-objective optimization of rate-of-penetration using machine learning methods

Cited by 9 publications

References 35 publications

Hybrid physics-machine learning models for predicting rate of penetration in the Halahatang oil field, Tarim Basin

Hybrid physics-machine learning models for predicting rate of penetration in the Halahatang oil field, Tarim Basin

Real-time prediction of oil and gas drilling rate based on physics-based model and particle filter method

Modeling CO2 Loading Capacity of Diethanolamine (DEA) Aqueous Solutions Using Advanced Deep Learning and Machine Learning Algorithms: Application to Carbon Capture

Contact Info

Product

Resources

About