Pub Date : 2025-11-24DOI: 10.1016/j.jlp.2025.105868
Ziqi Han , Jiansong Wu , Jitao Cai , Congze Wang , Tong Xu , Yuntao Li
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- 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.
埋地天然气管道是城市能源供应的关键，但泄漏事件频发，特别是土壤中细小的“针孔”泄漏，给城市安全与环境带来了严重风险。实时预测土壤中气体浓度分布对及时发现和预防至关重要。由于效率和准确性的问题，传统的统计经验模型和数值模型等方法难以应用于现场预测。本文提出了一种融合降维和多元时间序列预测技术的机器学习模型Conv-β vee - itransformer，用于埋地输气管道泄漏预测。该模型是在经过验证的数值模型生成的数据集上进行训练的，该数据集涵盖了各种泄漏条件，包括不同的压力、位置和孔径大小。评价结果表明，该预测模型具有较强的通用性和鲁棒性，在各种压力情景下的平均绝对百分比误差（MAPE）小于3%，在特定测点的MAPE为1.85%。此外，对比实验表明，我们的模型在预测范围和精度方面都优于其他模型。总体而言，本研究为天然气在土壤中的扩散动态实时预测提供了有效的解决方案，为埋地输气管道泄漏风险评估和应急处置提供了有价值的工具。
{"title":"Real-time prediction of gas leakage and diffusion for buried natural gas pipelines by deep learning and dimensionality reduction methods","authors":"Ziqi Han , Jiansong Wu , Jitao Cai , Congze Wang , Tong Xu , 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":"2025-11-24","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}
In chemical devices, irrational layouts accelerate the propagation of synergistic and domino effects across devices, thereby increasing the severity of accidents. In this paper, the failure probability of each chemical device is estimated by dynamic target device time to failure (ttf) and escalation threshold when synergistic effects of multiple fires are considered. The failure probabilities are further converted into accident damage costs. Then, the optimization model considering the synergistic effects is constructed by combining the other costs. The layout model is solved by combining the simulated annealing (SA) algorithm and the particle swarm optimization (PSO) (PSO-SA) algorithm. Finally, as a case study, ethane, ethanol, and acetic acid tank farm layouts are derived considering synergistic effects and without synergistic effects. When synergistic effects are not considered, the tank layout is more centralized. At the same time, other related costs are reduced while pipeline costs are increased. One of the most hazardous tanks was then selected as the initial tank for accident risk analysis. This work may provide some support for designers in terms of chemical device layout.
{"title":"Layout optimization and risk analysis of chemical devices under the synergistic effects of multiple fires","authors":"Di Xiao, Hui-hui Lu, Shu-Yu Chen, Xiang Liu, Jia-Jia Jiang, Jun-Cheng Jiang","doi":"10.1016/j.jlp.2025.105866","DOIUrl":"10.1016/j.jlp.2025.105866","url":null,"abstract":"<div><div>In chemical devices, irrational layouts accelerate the propagation of synergistic and domino effects across devices, thereby increasing the severity of accidents. In this paper, the failure probability of each chemical device is estimated by dynamic target device time to failure (<em>ttf</em>) and escalation threshold when synergistic effects of multiple fires are considered. The failure probabilities are further converted into accident damage costs. Then, the optimization model considering the synergistic effects is constructed by combining the other costs. The layout model is solved by combining the simulated annealing (SA) algorithm and the particle swarm optimization (PSO) (PSO-SA) algorithm. Finally, as a case study, ethane, ethanol, and acetic acid tank farm layouts are derived considering synergistic effects and without synergistic effects. When synergistic effects are not considered, the tank layout is more centralized. At the same time, other related costs are reduced while pipeline costs are increased. One of the most hazardous tanks was then selected as the initial tank for accident risk analysis. This work may provide some support for designers in terms of chemical device layout.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105866"},"PeriodicalIF":4.2,"publicationDate":"2025-11-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145621871","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}
A novel dry water (DW) fire extinguishing material was developed to enhance the extinguishing efficiency of early-stage oil spill fires. The material is composed of a hydrophobic fumed silica shell and an aqueous core that has been modified with four potassium salts: potassium carbonate (K2CO3), potassium bicarbonate (KHCO3), potassium acetate (CH3COOK), and potassium oxalate (K2C2O4). Bulk density, water retention, fluidity, particle size distribution, and thermogravimetric behaviour were the physical and thermal properties of the modified DW samples that were systematically assessed. Potassium salt-modified DW outperformed unmodified DW in fire suppression experiments conducted on n-heptane pool fires, achieving superior cooling performance and faster flame extinction speed. It is important to note that the shortest extinguishing times were obtained by DW modified with potassium oxalate and potassium carbonate, which were 3 and 4 s, respectively. Within 150 s, all formulations achieved a reduction in core flame temperatures below 200 °C, surpassing the performance of commercial ABC dry powder agents. These results offer an optimistic foundation for the creation of high-efficiency, environmentally friendly fire extinguishing materials that are suitable for oil-related fire situations.
{"title":"Performance and fire suppression efficiency of potassium salt-modified dry water agents","authors":"Weiyi Ding , Feihao Zhu , Jiaping Zhao , Jun-Cheng Jiang , An-Chi Huang","doi":"10.1016/j.jlp.2025.105869","DOIUrl":"10.1016/j.jlp.2025.105869","url":null,"abstract":"<div><div>A novel dry water (DW) fire extinguishing material was developed to enhance the extinguishing efficiency of early-stage oil spill fires. The material is composed of a hydrophobic fumed silica shell and an aqueous core that has been modified with four potassium salts: potassium carbonate (K<sub>2</sub>CO<sub>3</sub>), potassium bicarbonate (KHCO<sub>3</sub>), potassium acetate (CH<sub>3</sub>COOK), and potassium oxalate (K<sub>2</sub>C<sub>2</sub>O<sub>4</sub>). Bulk density, water retention, fluidity, particle size distribution, and thermogravimetric behaviour were the physical and thermal properties of the modified DW samples that were systematically assessed. Potassium salt-modified DW outperformed unmodified DW in fire suppression experiments conducted on n-heptane pool fires, achieving superior cooling performance and faster flame extinction speed. It is important to note that the shortest extinguishing times were obtained by DW modified with potassium oxalate and potassium carbonate, which were 3 and 4 s, respectively. Within 150 s, all formulations achieved a reduction in core flame temperatures below 200 °C, surpassing the performance of commercial ABC dry powder agents. These results offer an optimistic foundation for the creation of high-efficiency, environmentally friendly fire extinguishing materials that are suitable for oil-related fire situations.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105869"},"PeriodicalIF":4.2,"publicationDate":"2025-11-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145621870","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}
Pub Date : 2025-11-23DOI: 10.1016/j.jlp.2025.105864
Zhanzhong Wang, Tingting Li, Meng Yang, Zhihao Wu
—As industrialization continues to advance, the volume of hazardous materials transportation continues to increase. Given the inherent risks associated with hazardous materials, such as flammability and explosiveness, accidents involving hazardous materials vehicles during transportation can have catastrophic consequences. To mitigate the risk of accidents during hazardous materials transportation caused by factors such as road congestion or sudden incidents, this paper proposes a real-time risk-based dynamic optimization model for dangerous materials transportation routes, guiding vehicles to avoid congested or incident-affected sections of the road. A dual-objective initial path planning model for transportation risk and cost is constructed to obtain the optimal driving path for hazardous materials vehicles under static road network conditions. Based on the initial path, vulnerability indicators are applied to evaluate real-time road segment risks, and the optimal path is selected to minimize dynamic risks, thereby achieving dynamic guidance for hazardous materials vehicles. Taking the road network of Changchun City, Jilin Province, China, as an example, this paper verifies that the proposed model can effectively reduce potential risks during transportation and enhance the safety of the road transportation network. This paper provides a dynamic path optimization algorithm to assess risk levels in different regions at different times, achieving dynamic optimization of hazardous materials vehicle routes.
{"title":"Dynamic optimization of hazardous materials vehicle transportation routes based on real-time risk","authors":"Zhanzhong Wang, Tingting Li, Meng Yang, Zhihao Wu","doi":"10.1016/j.jlp.2025.105864","DOIUrl":"10.1016/j.jlp.2025.105864","url":null,"abstract":"<div><div>—As industrialization continues to advance, the volume of hazardous materials transportation continues to increase. Given the inherent risks associated with hazardous materials, such as flammability and explosiveness, accidents involving hazardous materials vehicles during transportation can have catastrophic consequences. To mitigate the risk of accidents during hazardous materials transportation caused by factors such as road congestion or sudden incidents, this paper proposes a real-time risk-based dynamic optimization model for dangerous materials transportation routes, guiding vehicles to avoid congested or incident-affected sections of the road. A dual-objective initial path planning model for transportation risk and cost is constructed to obtain the optimal driving path for hazardous materials vehicles under static road network conditions. Based on the initial path, vulnerability indicators are applied to evaluate real-time road segment risks, and the optimal path is selected to minimize dynamic risks, thereby achieving dynamic guidance for hazardous materials vehicles. Taking the road network of Changchun City, Jilin Province, China, as an example, this paper verifies that the proposed model can effectively reduce potential risks during transportation and enhance the safety of the road transportation network. This paper provides a dynamic path optimization algorithm to assess risk levels in different regions at different times, achieving dynamic optimization of hazardous materials vehicle routes.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105864"},"PeriodicalIF":4.2,"publicationDate":"2025-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145621907","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}
Pub Date : 2025-11-20DOI: 10.1016/j.jlp.2025.105852
Peng Zhang , Chuan Wang , Wei Liu , Haoyu Su
In order to address the challenges posed by complex feature correlations, high uncertainty, and insufficient model generalization in predicting the corrosion depth of natural gas pipelines under small sample conditions, this paper proposes a hybrid deep learning framework that integrates physical mechanisms with data-driven approaches. The framework utilizes a Bayesian Network (BN) to identify seven critical features and constructs six interactive features based on physical-electrochemical corrosion mechanisms to enhance physical consistency. The model employs a three-stage architecture: XGBoost serves as the baseline model to learn global nonlinear trends and generate initial predictions. The Kolmogorov-Arnold Network (KAN) is first embedded to perform high-order feature modeling on the residuals of corrosion predictions, enhancing stable representation capabilities. The Gaussian Process (GP) performs residual smoothing correction in the embedded space and outputs a 95 % confidence interval. Validation based on 242 sets of sample data collected from excavation sites of buried pipelines in southern Mexico that have been in service for over 50 years.The findings indicate that by employing Bayesian methods for joint hyperparameter adjustment, the model attains a prediction performance of R2 = 0.9613 and a root mean square error (RMSE) of 0.2809 on a dataset comprising 242 groups. This enhancement in prediction accuracy is accompanied by a reduction in RMSE of over 50 % when compared to a solitary XGB model. A high R2 value indicates that the model possesses exceptional explanatory power and predictive accuracy, while the 95 % confidence interval provides reliable uncertainty boundaries for corrosion risk assessment and safety margin determination in engineering practice. The interpretability of the model was enhanced through the implementation of Shapley Additive Explanations (SHAP) and KAN weight analysis, which facilitated the visualization of both global and local feature contributions. The findings suggest that the water content (wc), dissolved chloride ions (cc), pH, and the interaction feature wc_rp exert a substantial influence on pipeline corrosion. This model achieves a balance between predictive accuracy, interpretability, and uncertainty quantification capabilities, thereby providing a reliable foundation for decision-making processes regarding pipeline corrosion monitoring and maintenance in scenarios involving small sample sizes.
{"title":"A hybrid deep learning model driven by physical mechanisms and data for predicting corrosion in natural gas pipelines","authors":"Peng Zhang , Chuan Wang , Wei Liu , Haoyu Su","doi":"10.1016/j.jlp.2025.105852","DOIUrl":"10.1016/j.jlp.2025.105852","url":null,"abstract":"<div><div>In order to address the challenges posed by complex feature correlations, high uncertainty, and insufficient model generalization in predicting the corrosion depth of natural gas pipelines under small sample conditions, this paper proposes a hybrid deep learning framework that integrates physical mechanisms with data-driven approaches. The framework utilizes a Bayesian Network (BN) to identify seven critical features and constructs six interactive features based on physical-electrochemical corrosion mechanisms to enhance physical consistency. The model employs a three-stage architecture: XGBoost serves as the baseline model to learn global nonlinear trends and generate initial predictions. The Kolmogorov-Arnold Network (KAN) is first embedded to perform high-order feature modeling on the residuals of corrosion predictions, enhancing stable representation capabilities. The Gaussian Process (GP) performs residual smoothing correction in the embedded space and outputs a 95 % confidence interval. Validation based on 242 sets of sample data collected from excavation sites of buried pipelines in southern Mexico that have been in service for over 50 years.The findings indicate that by employing Bayesian methods for joint hyperparameter adjustment, the model attains a prediction performance of R<sup>2</sup> = 0.9613 and a root mean square error (RMSE) of 0.2809 on a dataset comprising 242 groups. This enhancement in prediction accuracy is accompanied by a reduction in RMSE of over 50 % when compared to a solitary XGB model. A high R<sup>2</sup> value indicates that the model possesses exceptional explanatory power and predictive accuracy, while the 95 % confidence interval provides reliable uncertainty boundaries for corrosion risk assessment and safety margin determination in engineering practice. The interpretability of the model was enhanced through the implementation of Shapley Additive Explanations (SHAP) and KAN weight analysis, which facilitated the visualization of both global and local feature contributions. The findings suggest that the water content (wc), dissolved chloride ions (cc), pH, and the interaction feature wc_rp exert a substantial influence on pipeline corrosion. This model achieves a balance between predictive accuracy, interpretability, and uncertainty quantification capabilities, thereby providing a reliable foundation for decision-making processes regarding pipeline corrosion monitoring and maintenance in scenarios involving small sample sizes.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105852"},"PeriodicalIF":4.2,"publicationDate":"2025-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145691439","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}
Pub Date : 2025-11-19DOI: 10.1016/j.jlp.2025.105851
Tianya Zhang, Hui Hu, Bin Zhang
1-Chloro-2,4-dinitrobenzene (CDNB) is extensively used as a crucial chemical intermediate in the pharmaceutical, dye, and pesticide industries. The presence of two nitro functional groups in its molecular structure results in substantial heat release during thermal decomposition. An insufficient understanding of its decomposition behavior can lead to severe thermal runaway incidents in industrial production. To investigate the thermal hazard characteristics of CDNB under various acidic environments, this study was designed based on a representative industrial accident scenario. A systematic analysis was carried out using differential scanning calorimetry (DSC) and accelerating rate calorimetry (ARC), combined with kinetic modeling, to thoroughly investigate the thermal decomposition behavior and runaway potential of CDNB. Experimental data revealed that the decomposition enthalpy of CDNB exceeds 4000 J/g, demonstrating intense exothermicity and a clear tendency for thermal runaway. Moreover, the time to maximum rate under adiabatic conditions (TMRad) was found to be less than 1 h, with a corrected adiabatic temperature rise (ΔTad,f) of 1680.4 K. Based on these thermal safety parameters and risk matrix assessment, the thermal runaway risk level of CDNB was evaluated as Level 3, corresponding to an “unacceptable risk” category. Notably, in the ARC experiment, the addition of sulfuric and nitric acids significantly lowered the initial decomposition temperature of CDNB. Their catalytic effects became more pronounced with increasing acid concentrations, with the temperatures being reduced by up to 40.2 °C and 50.0 °C, respectively. Under different acidic conditions, the activation energy of CDNB decreased by 14.3–99.0 kJ/mol, significantly increasing the likelihood of thermal hazard events. This study provides essential theoretical support for risk assessment and control in the safe industrial handling of CDNB.
{"title":"Effect of acidic conditions on the thermal hazard of 1-chloro-2,4-dinitrobenzene","authors":"Tianya Zhang, Hui Hu, Bin Zhang","doi":"10.1016/j.jlp.2025.105851","DOIUrl":"10.1016/j.jlp.2025.105851","url":null,"abstract":"<div><div>1-Chloro-2,4-dinitrobenzene (CDNB) is extensively used as a crucial chemical intermediate in the pharmaceutical, dye, and pesticide industries. The presence of two nitro functional groups in its molecular structure results in substantial heat release during thermal decomposition. An insufficient understanding of its decomposition behavior can lead to severe thermal runaway incidents in industrial production. To investigate the thermal hazard characteristics of CDNB under various acidic environments, this study was designed based on a representative industrial accident scenario. A systematic analysis was carried out using differential scanning calorimetry (DSC) and accelerating rate calorimetry (ARC), combined with kinetic modeling, to thoroughly investigate the thermal decomposition behavior and runaway potential of CDNB. Experimental data revealed that the decomposition enthalpy of CDNB exceeds 4000 J/g, demonstrating intense exothermicity and a clear tendency for thermal runaway. Moreover, the time to maximum rate under adiabatic conditions (TMR<sub>ad</sub>) was found to be less than 1 h, with a corrected adiabatic temperature rise (ΔT<sub>ad</sub>,f) of 1680.4 K. Based on these thermal safety parameters and risk matrix assessment, the thermal runaway risk level of CDNB was evaluated as Level 3, corresponding to an “unacceptable risk” category. Notably, in the ARC experiment, the addition of sulfuric and nitric acids significantly lowered the initial decomposition temperature of CDNB. Their catalytic effects became more pronounced with increasing acid concentrations, with the temperatures being reduced by up to 40.2 °C and 50.0 °C, respectively. Under different acidic conditions, the activation energy of CDNB decreased by 14.3–99.0 kJ/mol, significantly increasing the likelihood of thermal hazard events. This study provides essential theoretical support for risk assessment and control in the safe industrial handling of CDNB.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105851"},"PeriodicalIF":4.2,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145621944","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}
Pub Date : 2025-11-19DOI: 10.1016/j.jlp.2025.105848
Saizhe Ding , Tong Lu , Xin Lv , Yuxin Zhang , Rong Deng , Xinyan Huang
Unsafe behavior is one of the main causes of on-site safety accidents, while safety training is critical for mitigating such workplace hazards and ensuring operational reliability. Therefore, to improve the effectiveness of safety training, this paper proposes a novel On-site AR-based Training System (OATS) to enhance training experience. The developed video see-through AR eliminates the heavy requirement of virtual environment modeling by superimposing training content onto the real world. Moreover, enhanced interaction enables users to engage with virtual elements beyond passive animation or Q&A sessions; meanwhile, the isometric locomotion method reduces motion discomfort by tracking real body movements. For the demonstration, laboratory safety training is conducted by comparing the proposed AR approaches with traditional video-based training involving 36 participants. Results showed that OATS outperformed traditional video-based training in knowledge acquisition, self-efficacy, and intrinsic motivation after training. Meanwhile, it demonstrated high usability (p = 0.005) and presence (p < 0.001) while maintaining low simulator sickness and task load. These findings confirm OATS's potential to improve educational experience and deliver reliable safety training.
{"title":"Augmented reality for enhancing educational experience in laboratory safety training","authors":"Saizhe Ding , Tong Lu , Xin Lv , Yuxin Zhang , Rong Deng , Xinyan Huang","doi":"10.1016/j.jlp.2025.105848","DOIUrl":"10.1016/j.jlp.2025.105848","url":null,"abstract":"<div><div>Unsafe behavior is one of the main causes of on-site safety accidents, while safety training is critical for mitigating such workplace hazards and ensuring operational reliability. Therefore, to improve the effectiveness of safety training, this paper proposes a novel On-site AR-based Training System (OATS) to enhance training experience. The developed video see-through AR eliminates the heavy requirement of virtual environment modeling by superimposing training content onto the real world. Moreover, enhanced interaction enables users to engage with virtual elements beyond passive animation or Q&A sessions; meanwhile, the isometric locomotion method reduces motion discomfort by tracking real body movements. For the demonstration, laboratory safety training is conducted by comparing the proposed AR approaches with traditional video-based training involving 36 participants. Results showed that OATS outperformed traditional video-based training in knowledge acquisition, self-efficacy, and intrinsic motivation after training. Meanwhile, it demonstrated high usability (p = 0.005) and presence (p < 0.001) while maintaining low simulator sickness and task load. These findings confirm OATS's potential to improve educational experience and deliver reliable safety training.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105848"},"PeriodicalIF":4.2,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145621943","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}
Pub Date : 2025-11-19DOI: 10.1016/j.jlp.2025.105849
Guojin Qin , Zijin Zhang , Xu Wang , Yihuan Wang
Climate change is reshaping the risk landscape for natural gas pipelines, with landslides emerging as a major driver of technological accidents triggered by natural hazards (Natech events). Conventional Natech risk models rarely incorporate climate-sensitive parameters such as groundwater levels and soil moisture, limiting their capacity to capture evolving threats. This study develops a probabilistic model that explicitly links climate-driven landslide susceptibility to pipeline vulnerability, providing a quantitative basis for assessing pipeline failure probability under different emission projection scenarios. Using Monte Carlo simulations across five regions in China, the results show that under high-emission pathways (SSP5-8.5), pipeline failure probability in summer increases dramatically. For example, from 0.320 to 0.943 in Xinjiang, 0.112 to 0.220 in Sichuan, and 0.087 to 0.188 in Hainan. In cold regions, winter failure probability more than doubles, rising from 0.206 to 0.501 in Heilongjiang and from 0.235 to 0.488 in Beijing. These shifts reveal an overall increase in risk, intensification of seasonal contrasts, and, in some areas, a reconfiguration of high-risk periods. Sensitivity analysis highlights groundwater levels and soil moisture as the dominant drivers, with regional differences shaped by precipitation regimes, permafrost thaw, and typhoon impacts. Building on these insights, this study proposes an AI-based condition-monitoring framework that integrates real-time climate and geotechnical data to support adaptive early warning and safety management.
{"title":"A probabilistic model for natural gas pipeline failure under climate-induced Natech hazards: Toward AI-based safety management","authors":"Guojin Qin , Zijin Zhang , Xu Wang , Yihuan Wang","doi":"10.1016/j.jlp.2025.105849","DOIUrl":"10.1016/j.jlp.2025.105849","url":null,"abstract":"<div><div>Climate change is reshaping the risk landscape for natural gas pipelines, with landslides emerging as a major driver of technological accidents triggered by natural hazards (Natech events). Conventional Natech risk models rarely incorporate climate-sensitive parameters such as groundwater levels and soil moisture, limiting their capacity to capture evolving threats. This study develops a probabilistic model that explicitly links climate-driven landslide susceptibility to pipeline vulnerability, providing a quantitative basis for assessing pipeline failure probability under different emission projection scenarios. Using Monte Carlo simulations across five regions in China, the results show that under high-emission pathways (SSP5-8.5), pipeline failure probability in summer increases dramatically. For example, from 0.320 to 0.943 in Xinjiang, 0.112 to 0.220 in Sichuan, and 0.087 to 0.188 in Hainan. In cold regions, winter failure probability more than doubles, rising from 0.206 to 0.501 in Heilongjiang and from 0.235 to 0.488 in Beijing. These shifts reveal an overall increase in risk, intensification of seasonal contrasts, and, in some areas, a reconfiguration of high-risk periods. Sensitivity analysis highlights groundwater levels and soil moisture as the dominant drivers, with regional differences shaped by precipitation regimes, permafrost thaw, and typhoon impacts. Building on these insights, this study proposes an AI-based condition-monitoring framework that integrates real-time climate and geotechnical data to support adaptive early warning and safety management.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105849"},"PeriodicalIF":4.2,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145621997","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}
Oil gathering and transportation pipelines are the crucial component in oilfield production systems, however, leaks can cause significant economic losses and environmental pollution. Distributed Vibration Sensing (DVS) technology has been effectively utilized for leak detection; nevertheless, minor leaks often generate weak signals that are difficult to accurately capture and analyze. Given the temperature difference between the oil inside the pipeline and the surrounding environment, even small leaks can lead to detectable changes in the ambient temperature near the leak point. Based on this insight, this study proposes an intelligent pipeline micro-leakage monitoring technique integrating distributed fiber-optic temperature and vibration signals to achieve accurate leakage identification and localization. First, utilizing a self-built distributed optical fiber test platform, vibration and temperature signals were collected under various conditions, including normal operation, leakage scenarios, and environmental interference. Subsequently, a systematic model selection process was implemented through the comparative evaluation of five deep learning architectures (ResNet, 2DCNN, CNN-LSTM, CNN-attention and CNN-LSTM-attention). The fusion of vibration and temperature signals at the decision level was performed to enhance recognition accuracy and improve localization performance. The CNN-LSTM-attention model emerged as the most suitable, demonstrating an accuracy rate of 99.52 % and achieving precise leak location within ±1 m. During model training, the Adam optimizer and L2 regularization were utilized to adjust learning rates and prevent overfitting, improving the model's generalization ability. Furthermore, SHAP interpretability analysis was applied to visualize feature contributions and validate the model's decision logic. Finally, a leakage detection and early warning software system was developed, facilitating immediate observation of leak locations and execution of responsive actions.
{"title":"Minor pipeline leak detection and localization using explainable deep learning with fusion of distributed fiber-optic vibration and temperature signals","authors":"Ruijiao Ma, Jiawei Liu, Wei Wu, Yang Yang, Xiaowei Liu, Shuai Zhang, Meng Zou, Yixin Zhang","doi":"10.1016/j.jlp.2025.105844","DOIUrl":"10.1016/j.jlp.2025.105844","url":null,"abstract":"<div><div>Oil gathering and transportation pipelines are the crucial component in oilfield production systems, however, leaks can cause significant economic losses and environmental pollution. Distributed Vibration Sensing (DVS) technology has been effectively utilized for leak detection; nevertheless, minor leaks often generate weak signals that are difficult to accurately capture and analyze. Given the temperature difference between the oil inside the pipeline and the surrounding environment, even small leaks can lead to detectable changes in the ambient temperature near the leak point. Based on this insight, this study proposes an intelligent pipeline micro-leakage monitoring technique integrating distributed fiber-optic temperature and vibration signals to achieve accurate leakage identification and localization. First, utilizing a self-built distributed optical fiber test platform, vibration and temperature signals were collected under various conditions, including normal operation, leakage scenarios, and environmental interference. Subsequently, a systematic model selection process was implemented through the comparative evaluation of five deep learning architectures (ResNet, 2DCNN, CNN-LSTM, CNN-attention and CNN-LSTM-attention). The fusion of vibration and temperature signals at the decision level was performed to enhance recognition accuracy and improve localization performance. The CNN-LSTM-attention model emerged as the most suitable, demonstrating an accuracy rate of 99.52 % and achieving precise leak location within ±1 m. During model training, the Adam optimizer and L2 regularization were utilized to adjust learning rates and prevent overfitting, improving the model's generalization ability. Furthermore, SHAP interpretability analysis was applied to visualize feature contributions and validate the model's decision logic. Finally, a leakage detection and early warning software system was developed, facilitating immediate observation of leak locations and execution of responsive actions.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105844"},"PeriodicalIF":4.2,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145578330","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}
Azo diisobutyronitrile (AIBN) is frequently employed as an initiator in rocket propellants, however it possesses intrinsic thermal risks. This study methodically examines the influence of sodium halides (NaCl, NaBr, NaI, and NaF) on the thermal stability and decomposition kinetics of AIBN by thermogravimetric-Infrared, differential scanning calorimetry, and accelerating rate calorimetry studies. Thermokinetic modelling employing the Kissinger–Akahira–Sunose, Flynn–Wall–Ozawa, and Starink methodologies demonstrated that sodium halides augment the apparent activation energy (Ea) of AIBN breakdown, particularly with NaF (Ea increased from roughly 145 ± 1.00 to 169 ± 1.00 kJ/mol). ARC studies demonstrated that NaBr considerably lowers the maximum heating rate of AIBN from 11.25 ± 0.30 °C/min to 10.11 ± 0.30 °C/min, hence reducing thermal risk. The simulation results using the multiple linear regression method show that when NaBr is present, the decomposition energy levels of AIBN and the reaction heat released are significantly reduced. Gaussian computations verified a negative Gibbs free energy (−56.93 kJ/mol), signifying spontaneous decomposition. These quantitative results offer significant insights for improving the safe storage and management of AIBN in practical applications.
{"title":"Effect of sodium halides on the thermal stability and thermokinetic of azo diisobutyl nitrile","authors":"Xin-Hao Wang , Yan-Long Guo , Yen-Chun Cheng , Jun-Cheng Jiang , An-Chi Huang","doi":"10.1016/j.jlp.2025.105850","DOIUrl":"10.1016/j.jlp.2025.105850","url":null,"abstract":"<div><div>Azo diisobutyronitrile (AIBN) is frequently employed as an initiator in rocket propellants, however it possesses intrinsic thermal risks. This study methodically examines the influence of sodium halides (NaCl, NaBr, NaI, and NaF) on the thermal stability and decomposition kinetics of AIBN by thermogravimetric-Infrared, differential scanning calorimetry, and accelerating rate calorimetry studies. Thermokinetic modelling employing the Kissinger–Akahira–Sunose, Flynn–Wall–Ozawa, and Starink methodologies demonstrated that sodium halides augment the apparent activation energy (<em>E</em><sub>a</sub>) of AIBN breakdown, particularly with NaF (<em>E</em><sub>a</sub> increased from roughly 145 ± 1.00 to 169 ± 1.00 kJ/mol). ARC studies demonstrated that NaBr considerably lowers the maximum heating rate of AIBN from 11.25 ± 0.30 °C/min to 10.11 ± 0.30 °C/min, hence reducing thermal risk. The simulation results using the multiple linear regression method show that when NaBr is present, the decomposition energy levels of AIBN and the reaction heat released are significantly reduced. Gaussian computations verified a negative Gibbs free energy (−56.93 kJ/mol), signifying spontaneous decomposition. These quantitative results offer significant insights for improving the safe storage and management of AIBN in practical applications.</div></div>","PeriodicalId":16291,"journal":{"name":"Journal of Loss Prevention in The Process Industries","volume":"100 ","pages":"Article 105850"},"PeriodicalIF":4.2,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145578331","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}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号