Accurate modeling of crude oil and brine interfacial tension via robust machine learning approaches DOI Creative Commons
Chunyan Liu, Jing Wang, Jinshu Wang

et al.

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Nov. 20, 2024

Interfacial tension (IFT) between water and crude oil is a crucial variable that enhanced recovery (EOR) techniques can adjust to increase extraction from depleted fields. Most of the developed intelligent models in literature are based on synthetic samples rather than real or brine total salinity each salt type. Hence, this study applies various machine learning approaches, such as Convolutional Neural Networks (CNN), Adaptive Boosting (AdaBoost), Decision Trees (DT), Random Forest (RF), K-Nearest Neighbors (KNN), Ensemble Learning, Support Vector Machines (SVM), Multi-Layer Perceptron Artificial (MLP-ANN) develop advanced for predicting IFT considering taking account type prevalent within phase, which represent realistic circumstances encountered reservoirs. These predictions factors like concentration salt, API oil, properties system (pressure temperature) using previously published experimental data. A sensitivity analysis, incorporating relevancy factor, performed highlight influence input parameters IFT. Among these models, Tree highlighted its high accuracy low training cost compared ANN-based evidenced by emerged evaluation metrics (R-squared 0.9796 mean square error 5e-4). It noted AdaBoost model least accurate with an R2 0.6696. Furthermore, analysis indicates molecular weight has smallest impact IFT, whereas temperature most significant effect. The smart may be used accurately estimate oil/brine without needing tedious, time-consuming expensive workflows.

Language: Английский

A Comprehensive Study on Optimizing Reservoir Potential: Advanced Geophysical Log Analysis of Zamzama Gas Field, Southern Indus Basin, Pakistan DOI
Saddam Hussain, Asad Atta, Chaohua Guo

et al.

Physics and Chemistry of the Earth Parts A/B/C, Journal Year: 2024, Volume and Issue: 135, P. 103640 - 103640

Published: May 20, 2024

Language: Английский

Citations

9
Developing flood mapping procedure through optimized machine learning techniques. Case study: Prahova river basin, Romania DOI Creative Commons
Daniel Constantin Diaconu, Romulus Costache, Abu Reza Md. Towfiqul Islam

et al.

Journal of Hydrology Regional Studies, Journal Year: 2024, Volume and Issue: 54, P. 101892 - 101892

Published: July 13, 2024

Prahova river basin located in the central-southern region of Romania. This study aims to assess susceptibility flooding by using state-of-the-art machine learning and optimization procedures. To achieve this goal, we employed ten flood-related variables as independent our models. These include slope angle, convergence index, distance from river, elevation, plan curvature, hydrological soil group, lithology, topographic wetness rainfall, land use. We used 158 flood locations dependent training four hybrid models: Deep Learning Neural Network-Statistical Index (DLNN-SI), Particle Swarm Optimization-Deep (PSO-DLNN-SI), Support Vector Machine-Statistical (SVM-SI), Optimization-Support (PSO-SVM-SI). Utilizing Statistical method, calculated coefficients for each predictor class or category. The PSO-DLNN-SI model demonstrated best performance, achieving an AUC-ROC curve 0.952. It's worth noting that application PSO algorithm significantly enhanced model's performance. Additionally, it's crucial highlight approximately 25 % exhibits a high very events. Taking into account precise results models applied present study, can state point view, current research contributes better understanding intensity with which floods affect different areas basin.

Language: Английский

Citations

5
Application of Deep Learning for Reservoir Porosity Prediction and self Organizing Map for Lithofacies Prediction DOI
Mazahir Hussain, Shuang Liu, Wakeel Hussain

et al.

Journal of Applied Geophysics, Journal Year: 2024, Volume and Issue: 230, P. 105502 - 105502

Published: Aug. 31, 2024

Language: Английский

Citations

5
Computational approach towards shear strength prediction of squat RC walls implementing ensemble and hybrid SVR paradigms DOI
Mudassir Iqbal, Babatunde Abiodun Salami, Mohsin Ali Khan

et al.

Materials Today Communications, Journal Year: 2024, Volume and Issue: 40, P. 109921 - 109921

Published: July 22, 2024

Language: Английский

Citations

4
Comparative Analysis of the Gas Extraction Effects With Different Numbers and Spacings of Boreholes Drilled Along a Coal Seam DOI Creative Commons

Yuexia Chen,

Tingxiang Chu, Xuexi Chen

et al.

Energy Science & Engineering, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 5, 2025

ABSTRACT The number and spacing of boreholes drilled along a coal seam are important parameters the borehole layout. Despite previous extensive research by many experts, quantitative visual comparisons gas extraction effects with different spacings numbers rare. Here, effective range, delimited 0.74 MPa isobaric surface lines, was simulated. around multiple sets at 60 days is wavy. At 50 days, lines single set approximately circular, whereas those elliptical. When five used, short semiaxis elliptical contour middle 59% greater than radius borehole. Among investigated, volume V5 peaks 5 m 90 days; but, top bottom working face concave inwards, that is, certain pressures exceed MPa, thereby increasing likelihood emissions. Thus, efficiency in this studied higher as 4 m. This approach, which takes into account superimposed effect seam, extraction, isolines, volume, offers theoretical guidance for arrangement boreholes.

Language: Английский

Citations

0
Real-Time Inversion of Formation Drillability and Concurrent Speedup Strategies for Microdrilling Time Optimization DOI

Huohai Yang,

Zhirong Li, Lin Gao

et al.

SPE Journal, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 16

Published: Feb. 1, 2025

Summary As the complexity of oil drilling engineering grows, real-time optimization parameters to improve efficiency and lower costs becomes an important task. In this research, we propose a novel combination categorical boosting (CatBoost) genetic algorithm (GA) for synchronous with intelligent inversion formation drillability. The intricate causal relationship between time is made clear by introducing Peter-Clark (PC) discovery algorithm. A prediction model then built using information, comparing performance five supervised learning models across metrics. Subsequently, was designed utilizing GA accurately anticipate drillability dynamically alter parameters. field experiments on two wells, approach greatly increased efficiency. CatBoost performed well through 10-fold cross-validation, determination coefficients (R²) 0.986 0.990, effectively inverted that cannot be directly obtained in real (usually calculated from logging data after well) reduced about 5% 8%, respectively, optimization. Furthermore, Shapley additive explanation (SHAP) methodology fully quantified impact each parameter enhanced interpretability model. This method breaks traditional limitation relying engineers’ experience, realizes during process, provides scientific decision support improving

Language: Английский

Citations

0
System for Real-Time Rate of Penetration Optimization Using Machine Learning with Integrated Preventive Safeguards Against Hole Cleaning Issues and Stick-Slip DOI
T. Robinson,

Pauziah Mohd Arshad,

Olav Revheim

et al.

SPE/IADC International Drilling Conference and Exhibition, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 25, 2025

Abstract Drilling related costs can contribute 30-70% of operators’ capital expenditures for well construction. To reduce costs, operators bit-on-bottom time and flat time. This work describes a drilling optimization advisory system utilizing machine learning (ML) with integrated safeguards preventing issues that might occur following parameter changes intended to increase rate penetration (ROP), such as hole cleaning (HC) which lead stuck pipe, or stick-slip reduces efficiency. builds on the authors’ previous publications ROP (OTC-31680-MS, SPE-214521-MS), incorporating modules targeted at prompt detection timely mitigation, ensuring advised do not potentially cause HC pack-offs. The safeguard utilized downhole Equivalent Circulating Density (ECD) estimation ML model (SPE-208675-MS), queried by optimizer estimate effects proposed changes, corresponding ROP, ECD. A configurable tolerance (expected) ECD from baseline parameters ensured any increases were acceptable. detector monitored frequency spectra surface rotary speed torque measurements, classifier probability symptoms’ presence. has been field-deployed in SE Asia since Q4 2023, no pipe incidents relating pack-offs occurring this version software use. further enhanced was deployed field operations Q2 2024; analysis historical data torsional vibration demonstrates identifies high performance, achieving precision 0.92 holdout (unseen) intervals five wells, all symptoms present identified. With identified based estimated probabilities, human monitoring staff are notified, automatically alters its behavior allow vibrations be mitigated order maintain efficiencies. Literature contains many works separate topics prevention, however these have previously incorporated into holistic balancing different, sometimes competing, objectives. effective integration optimization, pack-off risks vibrations, combined enabling increased efficiency while reducing leading non-productive time, contributing overall reduced construction

Language: Английский

Citations

0
Integrated net pay cut-off evaluation workflow for tight sandstone reservoirs: a case study of the Linxing gas field, Ordos Basin DOI Creative Commons
Li Li, Jie Yu, Tao Huang

et al.

Frontiers in Earth Science, Journal Year: 2025, Volume and Issue: 13

Published: Feb. 25, 2025

Net pay detection is a crucial stage in reservoir characterization, serving various purposes such as reserve estimation, modeling, simulation, and production planning. was quantified through the use of petrophysical cut-offs. However, these cut-offs varied according to core dynamic data, introducing uncertainty into evolution process. This challenge particularly pronounced tight sandstone reservoirs, characterized by low porosity. In Linxing gas field Ordos Basin, reservoirs Shiqianfeng, upper Shihezi, lower Shanxi, Taiyuan formations exhibited ultra-low porosity permeability, thereby complicating determination net study utilized extensive data from field, including 50 wells, testing 217 comprehensive well logging data. An analysis area’s gas-bearing characteristics presented, accompanied straightforward cut-off evaluation workflow. The shale volume evaluated identify sand, while permeability evaluations were conducted reservoir. Hydrocarbon saturation employed establish pay. Eight methods determine These include particle size for cut-off, statistical accumulation frequency, minimum pore throat radius, mercury injection capillary pressure, per meter index, cross-plot methods—based on fracturing test data—for bound water relative hydrocarbon Subsequently, divided two vertical sections; section (including fifth layer Shiqianfeng Shihezi formations) target this study, with determined follows: 20% volume, 6% porosity, 0.15 mD 40% saturation. validated against actual provides reliable basis calculation offering technical support future development production.

Language: Английский

Citations

0
A real-time prediction method for rate of penetration sequence in offshore deep wells drilling based on attention mechanism-enhanced BiLSTM model DOI
Qi Yuan, Miao He, Zhichao Chen

et al.

Ocean Engineering, Journal Year: 2025, Volume and Issue: 325, P. 120820 - 120820

Published: March 2, 2025

Language: Английский

Citations

0
Web-Based Baseflow Estimation in SWAT Considering Spatiotemporal Recession Characteristics Using Machine Learning DOI Open Access
Jimin Lee, Jeongho Han, Bernard A. Engel

et al.

Environments, Journal Year: 2025, Volume and Issue: 12(3), P. 94 - 94

Published: March 17, 2025

The increasing frequency and severity of hydrological extremes due to climate change necessitate accurate baseflow estimation effective watershed management for sustainable water resource use. Soil Water Assessment Tool (SWAT) is widely utilized modeling but shows limitations in simulation its uniform application the alpha factor across Hydrologic Response Units (HRUs), neglecting spatial temporal variability. To address these challenges, this study integrated SWAT with Tree-Based Pipeline Optimization (TPOT), an automated machine learning (AutoML) framework, predict HRU-specific factors. Furthermore, a user-friendly web-based program was developed improve accessibility practical optimized factors, supporting more predictions, even ungauged watersheds. proposed approach area significantly enhanced recession predictions compared traditional method. This improvement supported by key performance metrics, including Nash–Sutcliffe Efficiency (NSE), coefficient determination (R2), percent bias (PBIAS), mean absolute percentage error (MAPE). framework effectively improves accuracy practicality modeling, offering scalable innovative solutions face stress.

Language: Английский

Citations

0