Journal of Loss Prevention in The Process Industries最新文献

{"title":"Explainable AI-driven predictive maintenance for mitigating process safety risks in safety-critical industrial equipment","authors":"Zainab Ali Bu Sinnah","doi":"10.1016/j.jlp.2025.105907","DOIUrl":"10.1016/j.jlp.2025.105907","url":null,"abstract":"<div><div>Artificial Intelligence (AI) is increasingly transforming process safety by enabling early detection of equipment failures that could escalate into fires, explosions, or toxic releases. This study presents an interpretable hybrid machine learning framework that integrates ensemble tree classifiers with bio-inspired optimization algorithms for predictive maintenance in industrial settings. While demonstrated on CNC machinery, the framework is generalizable to safety-critical process equipment, enabling early detection of operational anomalies that could potentially escalate into process safety hazards. Using a twelve-month dataset of 2500 operating cycles from machinery representative of chemical and process plants, recursive feature elimination identified seven key process variables: hydraulic and coolant pressures, coolant and hydraulic-oil temperatures, spindle speed, torque, and cutting force that capture essential thermomechanical behavior associated with unsafe operating conditions. The hybrid models, validated through stratified 5-fold cross-validation, achieved test accuracies exceeding 0.98 and demonstrated robustness to industrial variability. Fourier Amplitude Sensitivity Test (FAST) analysis provided transparent, physically interpretable insights, highlighting torque and hydraulic pressure as dominant predictors of potential process hazards, while revealing synergistic effects of spindle speed and cutting force. By combining real-world sensor data, advanced optimization, and explainable AI, this framework enables proactive identification of safety-critical equipment degradation, supports inherently safer operations, and addresses key challenges of trustworthiness and interpretability in AI for process safety.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105907"},"PeriodicalIF":4.2,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938721","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Real-time prediction of gas leakage and diffusion for buried natural gas pipelines by deep learning and dimensionality reduction methods","authors":"Ziqi Han ,&nbsp;Jiansong Wu ,&nbsp;Jitao Cai ,&nbsp;Congze Wang ,&nbsp;Tong Xu ,&nbsp;Yuntao Li","doi":"10.1016/j.jlp.2025.105868","DOIUrl":"10.1016/j.jlp.2025.105868","url":null,"abstract":"<div><div>Buried natural gas pipelines are essential for urban energy supply, but frequent leakage incidents, especially small “pinhole” leaks in the soil, pose serious safety and environmental risks. Real-time prediction of gas concentration distribution in the soil is crucial for timely detection and prevention. Due to efficiency and accuracy concerns, traditional methods, such as statistical empirical model and numerical model are difficult to apply in on-site prediction. This paper proposes a machine learning model, Conv-<span><math><mrow><mi>β</mi></mrow></math></span> VAE-iTransformer, which integrates dimensionality reduction and multivariate time-series prediction techniques for buried gas pipeline leakage prediction. The model is trained on a dataset generated from a validated numerical model, covering various leakage conditions, including different pressures, locations, and aperture sizes. The evaluation results show the prediction model demonstrates strong generalization and robustness in predicting gas pipeline leakage, with a Mean Absolute Percentage Error (MAPE) of less than 3 % across various pressure scenarios, and a MAPE of 1.85 % at specific measurement points. Furthermore, comparative experiments demonstrate that our model outperforms others in terms of both prediction range and accuracy. Overall, this study provides an effective solution for the real-time prediction of natural gas diffusion dynamics in the soil, offering a valuable tool for risk assessment and emergency disposal of buried gas pipeline leakage.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105868"},"PeriodicalIF":4.2,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145621908","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Risk assessment and key factors analysis of confined space operation using knowledge graph, association rules mining and Bayesian network","authors":"Huaying Cui ,&nbsp;Jinlong Zhao ,&nbsp;Dina Zhang ,&nbsp;Xin Kong ,&nbsp;Jianping Zhang","doi":"10.1016/j.jlp.2025.105884","DOIUrl":"10.1016/j.jlp.2025.105884","url":null,"abstract":"<div><div>Confined space operation involves working in (semi-)enclosed spaces. While confined space is an important workspace in chemical industry and urban development, there is also an increased risk of injury or even death due to hazardous factors, such as limited entry and exit area or a lack of adequate ventilation. In this paper, an intelligent method was proposed combining knowledge graph (KG), association rules mining (ARM) and Bayesian network (BN) to assess the risk and determine the key factors for confined space operation. First, a causative indicator system was established using 601 previous accidents, for which KG was also constructed to allow automatic extraction of accident causes. Based on the association rules determined by ARM, a risk assessment method was developed using BN. The key factors were analyzed and countermeasures were proposed. The results show that the association between failure to conduct ventilation detection on the site and failure to wear safety protective equipment demonstrates significant correlation strength, while association rule between inadequate safety education and training and failure to wear safety protective equipment also high. By the analysis in BN, it can be seen that the probability of confined space operation accidents is significantly higher (59 %) with the baseline probability of nodes in BN. The other important factors include failure to wear safety protective equipment, blindly rescue, insufficient provision of protective equipment and operation without a license. This study can evaluate the risk and determine key factors in a data-driven manner to reduce the subjectivity, which provides a reference for the targeted safety management of confined space operation.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105884"},"PeriodicalIF":4.2,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145796869","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Vulnerability assessment of offshore platform emergency response system based on cloud model","authors":"Yue Liu,&nbsp;Guoming Chen","doi":"10.1016/j.jlp.2025.105842","DOIUrl":"10.1016/j.jlp.2025.105842","url":null,"abstract":"<div><div>The working environment of deep-sea platforms is complex and harsh, with frequent incidents of fire and explosion. In the event of an accident, there is a high probability of significant casualties, environmental pollution, and economic losses. The assessment of vulnerabilities in emergency systems is instrumental in facilitating a comprehensive understanding of the safety risks confronted by offshore platforms. This assessment enables the timely identification of potential vulnerabilities in emergency systems, thereby reducing the likelihood of accidents. The present paper puts forth a vulnerability assessment model for emergency systems of offshore platforms. This model is based on the theoretical framework of cloud model theory. Firstly, an assessment index system for the vulnerability of offshore drilling platform emergencies was constructed. This system was constructed by applying the principles of human factors engineering to the specific characteristics of the emergency management system of offshore drilling platforms. Expert experience is combined with subjective weighting of risk factors through expert scoring. The objective weighting method, CRITIC, is then used to calculate the correlations and conflicts between the indicators. Finally, game theory is used to calculate the composite weights for the evaluation factors, providing the optimal weightings. In consideration of the factors that engender vulnerability, the cloud model—which is characterized by its fuzzy and random properties—was adopted for the purpose of assessing the vulnerability of the emergency response system. A standard cloud diagram was established based on the quantitative grading standards. The digital characteristic values of each risk indicator were calculated using MATLAB software, and a cloud risk assessment diagram was generated based on this, which can visually display the vulnerability level of each indicator. The research content provides a theoretical basis and reference for the daily safety management and emergency response decision-making of offshore drilling platforms.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105842"},"PeriodicalIF":4.2,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145691438","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prediction of the exothermic onset temperature of aromatic nitro compounds by the SiRMS descriptors from molecular structures","authors":"Yinyan Zhang ,&nbsp;Xin Zhang ,&nbsp;Shengjie Niu ,&nbsp;Yong Pan","doi":"10.1016/j.jlp.2025.105876","DOIUrl":"10.1016/j.jlp.2025.105876","url":null,"abstract":"<div><div>The reactivity hazards of aromatic nitro compounds remain one of the most serious concern in the chemical industry in spite of continual research and attention devoted to them. The exothermic onset temperature (To) is one of the most important thermal stability parameters for risk assessment and safe management of aromatic nitro compounds. The quantitative structure–property relationship (QSPR) between the exothermic onset temperature and molecular structure of aromatic nitro compounds was investigated. 63 aromatic nitro compounds were collected. The simplex representation of molecular structure (SiRMS) descriptors was calculated. The six SiRMS descriptors were screened out by the genetic algorithm–multivariable linear regression (GA-MLR) method as input variables to establish an MLR model and two nonlinear models, support vector machines (SVM) and random forest (RF). The performance and parameters of the models were examined, and Williams plots were used to determine the range of applicability domain of the models. The results showed that the performance of the nonlinear model was better than that of the linear model, and the RF model performed the best. Compared with the existing models, the three developed models have significantly improved in model fitting ability, stability, and predictive ability. This paper provides a new method for predicting To of aromatic nitro compounds for engineering.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105876"},"PeriodicalIF":4.2,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145747529","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Consequence analysis on the critical role of transient pool formation in cold ammonia release and gas dispersion using CFD approach","authors":"Michelle Xin Yi Ng ,&nbsp;Kai Xiang Yu ,&nbsp;Joon Yoon Ten ,&nbsp;Khang Wei Tan ,&nbsp;Weng Hoong Lam ,&nbsp;Peng Chee Tan ,&nbsp;Thomas Shean Yaw Choong ,&nbsp;Parthiban Siwayanan ,&nbsp;Binghui Chen ,&nbsp;Kek Seong Kim ,&nbsp;Zhen Hong Ban","doi":"10.1016/j.jlp.2025.105870","DOIUrl":"10.1016/j.jlp.2025.105870","url":null,"abstract":"<div><div>Accurate consequence modelling and risk assessment is crucial to mitigate the toxicity risks of ammonia. However, available refrigerated liquefied ammonia studies have assumed key parameters including pool size, pool shape, and lack of flashing during initial release. This work proposed a dynamic pool area estimation method for more accurate evaporation rate calculation during continuous release, rather than assume initial pool characteristics. For releases into the atmosphere, Computational Fluid Dynamics (CFD) was used for improved simulation of ammonia flashing, evaporation, and gas dispersion. The impact of wind speed, release rate, and release pressure on pool formation and ammonia gas concentration were investigated. The increase in wind speeds extended pool length, without significantly affecting size, and increased evaporation rates, but decreased hazardous zone sizes as dilution effect was stronger under wind field influence. The increasing release rates through larger hole sizes increased pool sizes, followed by higher evaporation rates and larger hazardous zones. For larger release pressures, higher release velocities formed pools further from the source, impacting initial ammonia cloud location. The results offer insights into emergency response and evacuation plan development, and the model can provide quantitative data needed for advanced digital technologies such as Artificial Intelligence (AI)-based safety systems.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105870"},"PeriodicalIF":4.2,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145621872","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Uniform corrosion assessment in oil and gas pipelines using commercial corrosion prediction models – Part 2: uncertainty and risk considerations based on statistical analyses of gravimetric laboratorial tests, offshore pipelines inspections, and manufacturing tolerances","authors":"João Pedro Bachega Cruz ,&nbsp;Edilson Gabriel Veruz ,&nbsp;Idalina Vieira Aoki ,&nbsp;Adriana Miralles Schleder ,&nbsp;Gilberto Francisco Martha de Souza ,&nbsp;Gustavo Leitão Vaz ,&nbsp;Leonardo Oliveira de Barros ,&nbsp;Rene Thiago Capelari Orlowski ,&nbsp;Marcelo Ramos Martins","doi":"10.1016/j.jlp.2025.105836","DOIUrl":"10.1016/j.jlp.2025.105836","url":null,"abstract":"<div><div>Uniform Corrosion stands out as the main damage mechanism attacking the inner surface of oil and gas pipelines, especially in the offshore context, resulting in equipment losses and economic, safety, and environmental consequences. This challenge highlights the literature growing interest for uniform corrosion prediction/inference. Discussion about uniform corrosion assessment, related parameters, and modelling and data collection procedures and limitations were presented in this study's Part 1, resulting from applications of commercial models in operational cases. In counterpoint, over-criticism and oversimplification about commercial models' performance were noticed in a literature review cycle, with inappropriate/insufficient applications and/or superficial evaluations, suggesting imprecise and accurate results, respectively, both approaches recurrently neglecting statistical and uncertainty discussions. This research compares different scenarios through an uncertainty evaluation perspective, analysing information from laboratorial gravimetric tests rehearsing operation, offshore inspections in christmas trees pipelines, and industrial manufacturing tolerances. As results, inherent uncertainty is identified for uniform corrosion process, uncertainty-based adequate prediction range and production lines' hazardous position are established, and discussion about uncertainties is addressed. Procedures for comparing identical laboratorial tests and models' predictions against measurements allow a performance comparison between available models, characterizing their prediction domains in accordance with their predictive behaviour in terms of conservatism.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105836"},"PeriodicalIF":4.2,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145546584","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Inversion of the source of gas pipeline leaks using a Gaussian plume optimization approach","authors":"Peng Gao ,&nbsp;Jun-Cheng Jiang ,&nbsp;Qian Xu ,&nbsp;Jie Wu ,&nbsp;Yan Tang ,&nbsp;An-Chi Huang","doi":"10.1016/j.jlp.2025.105909","DOIUrl":"10.1016/j.jlp.2025.105909","url":null,"abstract":"<div><div>This investigation establishes a source-term inversion framework for hazardous gas escapes in chemical industrial parks by employing a hybrid Genetic Algorithm–Particle Swarm Optimization (GA-PSO) and an enhanced Gaussian plume model. The following are the enhancements: a ground-reflection coefficient to mitigate boundary effects, an effective source height corrected by the jet trajectory under temperature differences and by leakage inclination, and search robustness through adaptive crossover/mutation and Sobol quasi-random initialization. Forward simulations are combined with concentration measurements along mobile patrol routes to estimate the location, release rate, and inclination of the breach. The proposed approach significantly reduces relative errors in comparison to the conventional plume model. It provides the greatest performance at low wind speeds and denser sensor spacing, while accuracy is degraded by higher winds or sparse sensors. The sensitivity to wind speed and monitoring distance is quantified, providing actionable guidance for rapid localization and emergency response in chemical parks.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105909"},"PeriodicalIF":4.2,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938719","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Experimental investigation of time–frequency characteristics of acoustic signals during boilover in small-scale oil tanks","authors":"Yanshan Sha ,&nbsp;Dongliang Chen ,&nbsp;Feiyang Wu ,&nbsp;Qiang Cao ,&nbsp;Yucong Zhou ,&nbsp;Weihua Zhang ,&nbsp;Xin Huang ,&nbsp;Minghui Wang","doi":"10.1016/j.jlp.2025.105881","DOIUrl":"10.1016/j.jlp.2025.105881","url":null,"abstract":"<div><div>To systematically examine the relationship between the combustion acoustic signals and boilover in oil-storage-tank fires, and to accurately analyze the boilover severity index, small-scale tank experiments were conducted via time-frequency analysis of combustion acoustic signals. By performing small-scale tank fire tests with crude oil and diesel oil, and processing the acoustic data with wavelet denoising and MATLAB routines, the present investigation developed a boiling-state classification framework that relies on the signals’ time-domain waveform, probability density function (PDF), and power spectral density (PSD). In parallel, an integrated analysis was performed to couple these acoustic signatures with flame morphology and temporal evolution. The results demonstrate a responsive relationship between the evolution of the combustion acoustic signals and the boilover stage. Correlation analysis of boilover acoustic signatures with fire dynamics reveals two phenomena: overflow and splash, each displaying distinct acoustic stages and evolutionary trends.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105881"},"PeriodicalIF":4.2,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145836690","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessing intrinsic hazard characterization of conventional and eco-friendly firecracker powders based on dust combustion and explosion parameters","authors":"Chin-Lung Chiang ,&nbsp;Wei-Chun Chen ,&nbsp;Cheng-Li Chuang ,&nbsp;Yi-Ying Ho ,&nbsp;Jia-You Gao ,&nbsp;Paul Amyotte ,&nbsp;Yu-Chi Cheng","doi":"10.1016/j.jlp.2025.105874","DOIUrl":"10.1016/j.jlp.2025.105874","url":null,"abstract":"<div><div>Firecracker powders present inherent dust explosion hazards during manufacturing, handling, and storage. This study evaluated the explosion sensitivity and severity of two commercial firecracker formulations (conventional and eco-friendly) based on key combustibility parameters. The conventional powder exhibited a minimum ignition energy (<em>MIE</em>) of 9 mJ, a minimum ignition temperature (<em>MITC)</em> of 280 °C, a minimum explosible concentration (<em>MEC</em>) of 95 g/m³, and a deflagration index (<em>K</em>_st) of 555 bar m/s. In comparison, the eco-friendly formulation showed <em>MIE</em> &gt; 1000 mJ, <em>MITC</em> = 300 °C, <em>MEC</em> = 250 g/m³, and <em>K</em>_st = 75 bar m/s. The <em>K</em>_st results classified the conventional powder as St-3 and the eco-friendly type as St-1, indicating a substantial difference in explosion severity. Overall, the results suggest that newer formulations designed with environmental considerations may also provide enhanced intrinsic safety. This work provides practical insights into pyrotechnic hazard classification and supports the development of safer and more sustainable firecracker products.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105874"},"PeriodicalIF":4.2,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145747526","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}