Predicting Photoelectric Logs in Challenging Conditions Using Machine Learning and Statistical Analysis DOI

Eassa Abdullah,

Reem AlYami

Published: Nov. 26, 2024

Abstract The photoelectric (PEF) log measures the absorption factor, pivotal for determining rock matrix properties. High factor values are typical in limestones, dolomites, clay, iron-bearing minerals, and heavy whereas sandstones exhibit lower values. In this study, actual logs were gathered from field alongside various other such as gallons per minute (GPM), standpipe pressure (SPP), rate of penetration (ROP), bulk density (RHOB). Utilizing a suite machine learning regression techniques—ridge regression, linear support vector machines (SVM), polynomial random forest, decision tree—this research aimed to predict using porosity data inputs. effectiveness these models was confirmed through their strong predictive accuracy relative ensemble demonstrated significant correlation coefficients low root mean square errors, illustrating robust capability at depths based on available drilling data.

Language: Английский

Reservoir quality drivers in the Oligo-Miocene Asmari Formation, Dezful Embayment, Iran: facies, diagenesis, and tectonic controls DOI
Roghayeh Fallah-Bagtash, Aram Bayet‐Goll,

Armin Omidpour

et al.

Marine and Petroleum Geology, Journal Year: 2025, Volume and Issue: unknown, P. 107279 - 107279

Published: Jan. 1, 2025

Language: Английский

Citations

1
Machine Learning-Based Prediction of Tribological Properties of Epoxy Composite Coating DOI Open Access
Han Yan,

Tan Junling,

Hui Chen

et al.

Polymers, Journal Year: 2025, Volume and Issue: 17(3), P. 282 - 282

Published: Jan. 22, 2025

Machine learning, being convenient and nondestructive, is beneficial for evaluating the tribological properties of coatings. Here, six machine learning algorithms, using a sericite/epoxy composite coating (SEC) as an example, were employed to assess impact filler content (10, 15, 20, 25, 30 wt%) mesh size on epoxy coatings under different loads. The results showed that gradient boosting regression model had superior accuracy stability compared other models, achieving friction coefficient wear rate prediction accuracies 93.7% 85.7%, respectively. This outperformed others, including decision trees, extreme boosting, Gaussian process regression. Feature importance sericite most significant influence properties. work provides valuable guidance engineering application this material.

Language: Английский

Citations

1
Development of a deep learning-based model for predicting of dominant seepage channels in oil reservoirs DOI Creative Commons
Chen Liu, Zenghua Zhang, Wensheng Zhou

et al.

Journal of Petroleum Exploration and Production Technology, Journal Year: 2025, Volume and Issue: 15(5)

Published: April 25, 2025

Language: Английский

Citations

0
Stacked machine learning models for accurate estimation of shear and Stoneley wave transit times in DSI log DOI Creative Commons

Donya Amerian,

Mohammadkazem Amiri,

Ali Safaei

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: March 14, 2025

Accurate estimates of the shear and Stoneley wave transit times are important for seismic analysis, rock mechanics, reservoir characterization. These parameters typically obtained from dipole sonic imager (DSI) logs instrumental in determining mechanical properties formations. However, DSI log may contain inconsistent missing data caused by various factors, such as salt layers spike phenomenon, which can cause difficulties analyzing interpreting data. This study addresses these challenges Log using machine learning methods common logs, including computed gamma ray (CGR), bulk density (RHOB), compressional time (DTC), well depth-based lithology different layers. Data two wells a field southern Iran were used. Outliers noise carefully removed to improve quality, normalization implemented ensure integrity. Then, invalid DTC values corrected used predict DTS DTST. Finally, predicted final models. Eight distinct models, Random Forest (RF), Gradient Boosting (GB), Support Vector Regression (SVR), Multiple Linear (MLR), Multivariate Polynomial (MPR), CatBoost, LightGBM, Artificial Neural Networks (ANN), independently trained evaluated. The results show that best among all approach facilitates subsurface interpretation evaluation provides strong foundation improving management future decision-making.

Language: Английский

Citations

0
Predicting Photoelectric Logs in Challenging Conditions Using Machine Learning and Statistical Analysis DOI

Eassa Abdullah,

Reem AlYami

Published: Nov. 26, 2024

Abstract The photoelectric (PEF) log measures the absorption factor, pivotal for determining rock matrix properties. High factor values are typical in limestones, dolomites, clay, iron-bearing minerals, and heavy whereas sandstones exhibit lower values. In this study, actual logs were gathered from field alongside various other such as gallons per minute (GPM), standpipe pressure (SPP), rate of penetration (ROP), bulk density (RHOB). Utilizing a suite machine learning regression techniques—ridge regression, linear support vector machines (SVM), polynomial random forest, decision tree—this research aimed to predict using porosity data inputs. effectiveness these models was confirmed through their strong predictive accuracy relative ensemble demonstrated significant correlation coefficients low root mean square errors, illustrating robust capability at depths based on available drilling data.

Language: Английский

Citations

0