Petroleum Research最新文献

{"title":"Full-field petroleum production forecasting: Is deep learning all you need?","authors":"Leonardo Kubota,&nbsp;Rafael dos Santos Gioria","doi":"10.1016/j.ptlrs.2026.01.007","DOIUrl":"10.1016/j.ptlrs.2026.01.007","url":null,"abstract":"<div><div>We investigate full-field petroleum production forecasting using machine learning. A data-driven proxy model that predicts field-scale production under various operational scenarios is proposed. It offers an alternative to time-consuming numerical simulators for assisting operational oilfield management decisions like designing waterflooding schedules and quantifying production loss caused by schedule delays. Its accuracy is validated with real-world datasets. Beyond its practical application, we leverage the proposed workflow to conduct a systematic comparison of several machine learning and deep learning algorithms using four datasets and three forecast horizons. Such critical and systematic analysis is long overdue and seems to be extremely rare in related works, which consistently favor increasingly intricate neural networks, despite the presence of relatively limited and sparse datasets, low-dimensional input variables and the existence of well-defined physical laws to guide input selection. Our comparative study suggests deep learning is an overkill in the context of full-field production forecasts, without any tangible benefits in terms of accuracy. Kernel-based models consistently outperform neural-based architectures in this multi-dimensional analysis. In addition, we find evidence that algorithmic generalization in the full-field context seems to be more influenced by the spatial density of production data than by dataset size. The data-driven proxy model and the investigative findings of this work can help engineers apply machine learning to petroleum production forecasting. On a broader scale, our results endorse the Green AI paradigm, which advocates for the use of compact architectures that deliver good accuracy with a low computational cost.</div></div>","PeriodicalId":19756,"journal":{"name":"Petroleum Research","volume":"11 3","pages":"Pages 743-760"},"PeriodicalIF":3.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148949219","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Drilling wellhead structure's fire resistance evaluation with gas well blowout fires","authors":"Yudan Peng,&nbsp;Guodong Ji,&nbsp;Weifeng Chen,&nbsp;Binhui Liu,&nbsp;Chen Hao,&nbsp;Wenqiang Lou,&nbsp;Guangming Fu,&nbsp;Jiying Chen","doi":"10.1016/j.ptlrs.2026.01.017","DOIUrl":"10.1016/j.ptlrs.2026.01.017","url":null,"abstract":"<div><div>To address the absence of fire resistance evaluation methods for drilling wellhead structures under gas well blowout fires, this study firstly conducted experiments to investigate the mechanical properties of wellhead materials under high-temperature conditions. Subsequently, a temperature model and a mechanical response model of the wellhead structure under fire were established and validated respectively based on thermo-mechanical coupling methods. Furthermore, a fire resistance limit evaluation method of wellhead structures in well blowout fires was developed. The results indicated that with clearance operations at the well site, a thermal radiation threshold existed; Only when the thermal radiation exceeded this threshold, the wellhead reached its fire resistance limit and failed. Without clearance operations at the well site, the wellhead structure was certain to reach its fire resistance limit and failed. The upper surface of the wellhead had the largest displacement and reached its fire resistance limit first. The convective heat transfer coefficient, pressure load on the upper surface of the metal sealing ring, and casing gravity load had minimal influence on the fire resistance limit. Without clearance operations at the well site, the wellhead surface temperature approached the flame temperature, and the rubber seal reaches its decomposition point, leading to seal failure. The research findings could provide valuable references for the safety assessment of wellheads during gas well blowout fires and contribute to the development of effective response strategies for blowout fire control.</div></div>","PeriodicalId":19756,"journal":{"name":"Petroleum Research","volume":"11 3","pages":"Pages 1008-1020"},"PeriodicalIF":3.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148949264","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Preparation and mechanistic analysis of high-selectivity water control systems","authors":"Zhizhen Li,&nbsp;Weiyu Chen,&nbsp;Yanhui Su,&nbsp;Xueli Geng,&nbsp;Bin Shi,&nbsp;Xiaobin Zheng,&nbsp;Qian Zhou,&nbsp;Gang Xie","doi":"10.1016/j.ptlrs.2025.11.008","DOIUrl":"10.1016/j.ptlrs.2025.11.008","url":null,"abstract":"<div><div>This paper presents a highly selective water control system designed to address declining oil production in offshore wells caused by rising water cuts while overcoming the poor selectivity of current chemical water control materials. The system was developed by copolymerizing octadecyl methacrylate and butyl methacrylate, followed by emulsion polymerization to prepare oil-soluble polymers. This resulted in a water-dispersible, high-selectivity water control system in the form of an emulsion dispersion. The feasibility of its field application was assessed via median particle size and reservoir stability tests, and the system's water control performance, visual performance, and underlying mechanism were evaluated. The test results showed that the median droplet diameter of the emulsion in the system was 0.431 μm, and it remained stable for at least 8 h under reservoir conditions. The system demonstrated excellent selective water-control capabilities, with a water-phase permeability blocking rate of 82.42% and an oil-phase permeability blocking rate of 5.02%. The mechanism of action was further explored via visualized flat plate tests and low-field nuclear magnetic resonance scanning displacement experiments on rock cores. The findings suggest that oil-soluble particles exhibit effective selective water-control performance and can be applied in water control operations in offshore oil fields.</div></div>","PeriodicalId":19756,"journal":{"name":"Petroleum Research","volume":"11 3","pages":"Pages 789-798"},"PeriodicalIF":3.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147331228","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Proppant transportation in vertical fractures with controlled roughness in different positions","authors":"Jun Li,&nbsp;Qingyuan Pu,&nbsp;Siyuan He,&nbsp;Mingyi Wu,&nbsp;Kewen Tang","doi":"10.1016/j.ptlrs.2025.12.003","DOIUrl":"10.1016/j.ptlrs.2025.12.003","url":null,"abstract":"<div><div>Effective proppant placement is essential for increasing the production of unconventional oil/gas reservoirs. It can substantially affect the fracture roughness. This paper presents a laboratory fracture model to quantify the effect of fracture roughness under well-controlled conditions. This model simulates fracture roughness by embedding particles into the walls. Via this model, the roughness can be well controlled and defined. The effect of fracture roughness location is quantified as an example to demonstrate the usefulness of the controllable fracture model. Using a fracture model with two-sided roughness, the relationships between roughness location, pump rate, and inlet location were established. The experimental results show that the rough surface in the middle resulted in a large void space in the fracture, which became more evident at a higher pump rate. This phenomenon has unexpected consequences for oil and gas production. Based on these experimental results, different sand injection modes were tested to reduce the void space caused by the surface roughness in the fractured middle. An injection mode that involves alternatively pumping proppants of different sizes was proposed to solve this problem effectively.</div></div>","PeriodicalId":19756,"journal":{"name":"Petroleum Research","volume":"11 3","pages":"Pages 815-826"},"PeriodicalIF":3.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148949155","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Effects of TPH contamination on soil and rice growth: Assessing the efficacy of sodium persulfate as a remediation agent","authors":"Farzad Rassaei","doi":"10.1016/j.ptlrs.2026.01.003","DOIUrl":"10.1016/j.ptlrs.2026.01.003","url":null,"abstract":"<div><div>Soil contamination by total petroleum hydrocarbons (TPH) poses a major risk to soil health and crop productivity. This study aims to investigate the effects of TPH on soil properties and rice growth factors, as well as the potential of sodium persulfate (SP) to mitigate negative impacts. The study includes four treatments: control (no SP or TPH), SP (0.0, 1.50 and 3.0%), TPH (TPH-contaminated soil, 4.0%, 6.0%, and 8.0%), and SP + TPH (SP and TPH levels). Results show that TPH contamination led to reductions in pH, organic matter, nutrient availability, root and shoot growth, chlorophyll content, plant height, grain and biomass yield. SP facilitates the restoration of soil physicochemical properties, mitigates toxicity, and enhances plant growth conditions by promoting both chemical oxidation and microbial degradation of TPH contaminants. Higher doses of SP are required to counteract higher TPH levels. The findings suggest that SP can be an effective amendment for remediating TPH-contaminated soils, but dosage and application should be tailored to site-specific conditions. This study highlights the potential of SP for sustainable management of TPH-contaminated agricultural land, but further work is needed to fully understand its impact on different plant species and to optimize remediation strategies for specific site conditions.</div></div>","PeriodicalId":19756,"journal":{"name":"Petroleum Research","volume":"11 3","pages":"Pages 1037-1051"},"PeriodicalIF":3.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148949262","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Spatially continuous regularization of the parameters of the capacitance–resistance model","authors":"Egor Illarionov ,&nbsp;Elizaveta Gladchenko ,&nbsp;Anton Voskresenskii ,&nbsp;Sergey Safonov ,&nbsp;Klemens Katterbauer","doi":"10.1016/j.ptlrs.2025.09.001","DOIUrl":"10.1016/j.ptlrs.2025.09.001","url":null,"abstract":"<div><div>The capacitance–resistance model (CRM) is a widely-used model for predicting well production rates because it requires the estimation of only a small number of parameters to describe production wells and the connectivity between injection and production wells. However, CRM parameters are not inherently spatially constrained, which can lead to unnatural results. In this research, we conducted spatially continuous regularization of CRM parameters and investigated how it affected model accuracy. A related goal was to investigate how this regularization improved the spatial interpolation of CRM coefficients. For regularization, we considered CRM parameters as values for a spatially continuous function, implemented this function using a neural network, and fitted it using production history data. We employed two benchmark datasets—the egg and Costa datasets—to compare two models: the unconstrained conventional CRM and the proposed spatially regularized CRM. We concluded that the spatially regularized CRM yielded accuracy close to that of the conventional CRM; however, it provided a more interpretable spatial distribution of the CRM parameters.</div></div>","PeriodicalId":19756,"journal":{"name":"Petroleum Research","volume":"11 2","pages":"Pages 431-440"},"PeriodicalIF":3.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148268460","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Alumina nanofluid enhances the heavy-oil upgrading process","authors":"Luay Ahmed Khamees ,&nbsp;Ghassan H. Abdul-Majeed ,&nbsp;Ayad A. Alhaleem","doi":"10.1016/j.ptlrs.2025.09.007","DOIUrl":"10.1016/j.ptlrs.2025.09.007","url":null,"abstract":"<div><div>Upgrading heavy crude oil remains one of the most critical and technically demanding processes in the petroleum industry, primarily due to the inherently poor quality of the various types of crude oil and the complexities associated with their transport and refining. This study investigates the impact of a nanofluid—composed of aluminum oxide nanoparticles dispersed in a kerosene solvent and stabilized with the surfactant sodium dodecylbenzenesulfonate—on enhancing the quality of crude oil extracted from an East Baghdad field. This nanofluid was applied using ultrasonic irradiation to ensure uniform dispersion and stability. The results demonstrated a substantial improvement in various crude oil properties. The viscosity was reduced significantly, from 58.15 cP to 5.58 cP, representing a 90.4% improvement. The API gravity increased markedly, from 19.63 to 30.49. Furthermore, the heavy metal content decreased considerably, with vanadium reduced from 109.67 ppm to 9.02 ppm (92% reduction) and the nickel content showing an 85% decrease. The sulfur content also declined sharply, from 4.422% to 0.77%, indicating an 83% improvement. These findings confirm the high efficacy of nanofluid-assisted ultrasonic treatment as a promising technique for upgrading heavy crude oil and enhancing its refining and transportation potential.</div></div>","PeriodicalId":19756,"journal":{"name":"Petroleum Research","volume":"11 2","pages":"Pages 451-470"},"PeriodicalIF":3.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148268462","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integrated electrochemical and machine learning framework for SiO2/CaCO3 under-deposits driven welded X65 carbon steel corrosion mitigation in sour service conditions","authors":"Adewale K. Ipadeola,&nbsp;Mostafa H. Sliem,&nbsp;Dana Abdeen,&nbsp;Nicholas Laycock,&nbsp;Ashwin RajKumar,&nbsp;Phaneendra K. Yalavarthy,&nbsp;Aboubakr M. Abdullah","doi":"10.1016/j.ptlrs.2025.09.006","DOIUrl":"10.1016/j.ptlrs.2025.09.006","url":null,"abstract":"<div><div>The integrity of welded X65 carbon steel (CS) pipelines in oil and gas systems is significantly compromised by preferential weldment corrosion (PWC) and pitting in sour conditions. This study clarifies that the effect of pre-corrosion conditioning and inorganic deposits (SiO₂ or CaCO₃) on welded CS pipelines, with SiO₂ deposits leading to the most substantial damage at weld metal (WM) and heat-affected zone (HAZ) relative to parent metal (PM). Due to localized acidification and differences in potential, corrosion rates (CRs) hierarchies are SiO₂-deposited &gt; CaCO₃-deposited &gt; non-deposited. Microstructural analysis reveals that deposited CS-WM has carbide redistribution and increase residual stress from the welding thermal process, with increased aggressive species and stabilized iron oxides, leading to intensified PWC. However, amine-based inhibitor CRW11 suppresses the CRs below 0.1 mmpy through chemisorbed films that disrupt microgalvanic coupling and protective corrosion product layers with inhibition efficiency (IE &gt; 85%). CS-PM-CaCO₃ has a high pit depth (68.3 ± 2.7 μm) and propagation rate (1.2 ± 0.2 mmpy), making it the most vulnerable region to pitting relative to CS-WM-CaCO₃ (52.5 ± 1.1 μm; 0.9 ± 0.1 mmpy) and CS-HAZ-CaCO₃ (29.3 ± 0.6 μm; 0.5 ± 0.1 mmpy). This proves that CaCO₃ deposit on welded CS creates a physical barrier for the surface nucleation effect and improves anodic dissolution. Machine learning models (Random Forest (RF), Decision Tree (DT), Gradient Boosting (GB), and Extreme Gradient Boosting (XGBoost)) accurately predict IE (<em>R</em>² = 0.99). However, RF performs best for data without deposits (RMSE = 0.9), while DT is most appropriate model for those with SiO₂ and CaCO₃ deposits (lowest RMSE = 0.5 and 0.2). These findings provide coupled electrochemical, microstructural, and machine learning frameworks for PWC mitigation, allowing for adaptive corrosion management methods in sour service pipes and material designs optimized for topology.</div></div>","PeriodicalId":19756,"journal":{"name":"Petroleum Research","volume":"11 2","pages":"Pages 613-632"},"PeriodicalIF":3.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148501869","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Research on segmented modelling and prediction of shale gas production based on time series analysis","authors":"Feifei Fang,&nbsp;Hongjian Chen,&nbsp;Sijie He,&nbsp;Jie Zhang,&nbsp;Wei Guo,&nbsp;Weixiang Jin,&nbsp;Yue Gong,&nbsp;ChuXiang Xia","doi":"10.1016/j.ptlrs.2025.09.009","DOIUrl":"10.1016/j.ptlrs.2025.09.009","url":null,"abstract":"<div><div>At present, the world is experiencing an unprecedented period of change, and the third wave of energy conversion is also emerging. Natural gas is widely regarded as an ideal bridge between traditional fossil fuels and new energy transformations due to its significant low-carbon advantages. Forecasting shale gas output each day is key for secure natural gas provision. However, the high dimension and nonlinear characteristics of shale gas production data present significant challenges for prediction. Therefore, this study proposes a segmented modeling method that divides the production cycle into unstable and stable periods and combines the IPSO–CNN-BiGRU-Attention hybrid model to model these two cycles respectively. Comparative results indicate segmented modeling offers superior predictive performance compared to whole production cycle modeling. Furthermore, when modeling different periods, the IPSO–CNN-BiGRU-Attention hybrid model demonstrates a superior predictive effect compared to traditional single time series models.</div></div>","PeriodicalId":19756,"journal":{"name":"Petroleum Research","volume":"11 2","pages":"Pages 399-414"},"PeriodicalIF":3.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148501950","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Research on intelligent kick early warning technology based on LSTM-AE unsupervised learning","authors":"Boyuan Li,&nbsp;Zhaoxue Guo,&nbsp;Xudong Wang,&nbsp;Gui Tang,&nbsp;Xing Zuo","doi":"10.1016/j.ptlrs.2025.10.002","DOIUrl":"10.1016/j.ptlrs.2025.10.002","url":null,"abstract":"<div><div>In oil and gas exploration, kick is a typical high-risk downhole incident. Currently, most intelligent kick detection methods belong to supervised learning, which depends on labeled samples and is prone to class imbalance issues in the training set, leading to reduced model accuracy and higher false alarm rates in practical applications. To address this, this paper focuses on the application of unsupervised learning in kick detection, proposing a kick detection method based on Long Short-Term Memory Autoencoder (LSTM-AE) combined with parameter trend change rules. The study integrates LSTM-AE with expert knowledge to develop an intelligent kick early warning system suitable for well sites, validated using field data from five wells. The results show that the average reconstruction mean squared error of the LSTM-AE is 0.017, with an average false alarm rate of 4.56% for kick detection, detecting kicks an average of 9.2 min earlier than manual detection. This outcome confirms that kick detection can utilize unsupervised learning methods, avoiding relying on labeled samples and class imbalance issues in the training set, thereby effectively improving model accuracy and generalization ability. Moreover, the proposed LSTM-AE demonstrates superior performance in kick detection, and the detection method combining the model and judgment rules has significant implications for monitoring and early warning of other types of downhole incidents.</div></div>","PeriodicalId":19756,"journal":{"name":"Petroleum Research","volume":"11 2","pages":"Pages 501-515"},"PeriodicalIF":3.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148501900","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}