Pub Date : 2026-01-16DOI: 10.1016/j.cie.2026.111814
Faqun Qi , Jiahui Kong , Anming Zhang , Hongjie Lin , Mi Li
Importance measures have been widely used in maintenance management as an essential decision-support indicator. This study proposes a novel degradation-based importance measure (DIM) method for a system comprising a critical and an auxiliary subsystem. DIM is defined as the expected decrease in system performance due to the deterioration of a subsystem during each inspection interval. A two-phase inspection policy is proposed for the system. The first phase involves analyzing the system’s performance level and determining the preventive maintenance (PM) requirements for the system. In the second phase, the DIM of subsystems is evaluated, and opportunities for minor repair (MR) of the subsystems are identified. Semi-regenerative technology is used to model the evolution process of the system, and the long-run average cost is calculated. Two numerical studies of the boring tool system and the axial piston pump system are provided to demonstrate the proposed method. For those systems, the optimal inspection period, PM threshold, and MR threshold are determined by minimizing the long-run average cost. Finally, two comparison experiments are conducted to illustrate the effectiveness of the proposed strategy and the applicability of the proposed DIM.
{"title":"Importance measured-based maintenance strategy for systems with auxiliary subsystems subject to degradation dependence","authors":"Faqun Qi , Jiahui Kong , Anming Zhang , Hongjie Lin , Mi Li","doi":"10.1016/j.cie.2026.111814","DOIUrl":"10.1016/j.cie.2026.111814","url":null,"abstract":"<div><div>Importance measures have been widely used in maintenance management as an essential decision-support indicator. This study proposes a novel degradation-based importance measure (DIM) method for a system comprising a critical and an auxiliary subsystem. DIM is defined as the expected decrease in system performance due to the deterioration of a subsystem during each inspection interval. A two-phase inspection policy is proposed for the system. The first phase involves analyzing the system’s performance level and determining the preventive maintenance (PM) requirements for the system. In the second phase, the DIM of subsystems is evaluated, and opportunities for minor repair (MR) of the subsystems are identified. Semi-regenerative technology is used to model the evolution process of the system, and the long-run average cost is calculated. Two numerical studies of the boring tool system and the axial piston pump system are provided to demonstrate the proposed method. For those systems, the optimal inspection period, PM threshold, and MR threshold are determined by minimizing the long-run average cost. Finally, two comparison experiments are conducted to illustrate the effectiveness of the proposed strategy and the applicability of the proposed DIM.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111814"},"PeriodicalIF":6.5,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037807","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2026-01-16DOI: 10.1016/j.cie.2026.111827
Hazem J. Smadi, Nader A. Al Theeb, Razan A. Khatatbeh
Battery swapping has recently emerged as a practical innovation in the electric vehicle market, largely because it greatly reduces the time and effort needed to recharge batteries. In this study, we present a new optimization model and solution method for electric vehicles that must swap their batteries at the nearest station after completing deliveries from the company depot to customers. The model combines mixed-integer linear programming and set covering into a single integrated framework, allowing us to determine not only the optimal number and locations of swapping stations but also the number of batteries each station should hold to minimize total operating cost. To test the model’s performance, we used both CPLEX and a Greedy heuristic to solve 25 datasets of different sizes within three hours, comparing the quality of the results and the required computation time. The findings show that the heuristic approach provides solutions that are competitive with CPLEX but in much less time. To demonstrate the model’s practical value, the farm dairy company’s distribution department is selected as a case study. The results indicate that the proposed model improves distribution quantities by up to 27% compared with the previous model.
{"title":"Optimization of integrated distribution systems with electric vehicles equipped with battery swapping technology","authors":"Hazem J. Smadi, Nader A. Al Theeb, Razan A. Khatatbeh","doi":"10.1016/j.cie.2026.111827","DOIUrl":"10.1016/j.cie.2026.111827","url":null,"abstract":"<div><div>Battery swapping has recently emerged as a practical innovation in the electric vehicle market, largely because it greatly reduces the time and effort needed to recharge batteries. In this study, we present a new optimization model and solution method for electric vehicles that must swap their batteries at the nearest station after completing deliveries from the company depot to customers. The model combines mixed-integer linear programming and set covering into a single integrated framework, allowing us to determine not only the optimal number and locations of swapping stations but also the number of batteries each station should hold to minimize total operating cost. To test the model’s performance, we used both CPLEX and a Greedy heuristic to solve 25 datasets of different sizes within three hours, comparing the quality of the results and the required computation time. The findings show that the heuristic approach provides solutions that are competitive with CPLEX but in much less time. To demonstrate the model’s practical value, the farm dairy company’s distribution department is selected as a case study. The results indicate that the proposed model improves distribution quantities by up to 27% compared with the previous model.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"214 ","pages":"Article 111827"},"PeriodicalIF":6.5,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146081473","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2026-01-16DOI: 10.1016/j.cie.2026.111823
Yunshenghao Qiu , Hailong Tian , Chuanhai Chen , Zhifeng Liu , Yuzhi Sun , Haoyuan Li , Yufei Li
As a critical manufacturing asset, the reliability of heavy-duty Computer Numerical Control (CNC) lathes directly affects product quality, production efficiency, and operational safety. This study presents a dynamic, maintenance-informed reliability analysis framework. It integrates subjective expert evaluation with objective failure data to improve the accuracy and timeliness of risk prioritisation. Objective information is incorporated by constructing maintenance-effect-informed reliability models from failure data, with the estimated failure intensities dynamically mapped onto the risk factor Occurrence O. Subjective expert judgement is represented using Interval-Valued Spherical Fuzzy Sets (IVSFSs) to capture uncertainty. A bargaining game mechanism is used to revise inconsistent expert evaluations, enhancing consensus and reducing individual bias. To reflect the evolving reliability state, the evaluation credibility decay method adjusts the influence of past evaluations over time. Current and historical expert inputs are aggregated using the Dombi operator to compute updated risk priority numbers, enabling timely tracking of risk evolution after each maintenance event. By fusing expert knowledge with operational failure data, the method delivers more adaptive and rational reliability analysis that provides meaningful support for both reliability design and maintenance planning. A case study on a D-type heavy-duty horizontal lathe from Company W demonstrates the effectiveness of the proposed approach.
{"title":"An improved dynamic reliability analysis method for heavy-duty CNC lathes considering maintenance","authors":"Yunshenghao Qiu , Hailong Tian , Chuanhai Chen , Zhifeng Liu , Yuzhi Sun , Haoyuan Li , Yufei Li","doi":"10.1016/j.cie.2026.111823","DOIUrl":"10.1016/j.cie.2026.111823","url":null,"abstract":"<div><div>As a critical manufacturing asset, the reliability of heavy-duty Computer Numerical Control (CNC) lathes directly affects product quality, production efficiency, and operational safety. This study presents a dynamic, maintenance-informed reliability analysis framework. It integrates subjective expert evaluation with objective failure data to improve the accuracy and timeliness of risk prioritisation. Objective information is incorporated by constructing maintenance-effect-informed reliability models from failure data, with the estimated failure intensities dynamically mapped onto the risk factor Occurrence O. Subjective expert judgement is represented using Interval-Valued Spherical Fuzzy Sets (IVSFSs) to capture uncertainty. A bargaining game mechanism is used to revise inconsistent expert evaluations, enhancing consensus and reducing individual bias. To reflect the evolving reliability state, the evaluation credibility decay method adjusts the influence of past evaluations over time. Current and historical expert inputs are aggregated using the Dombi operator to compute updated risk priority numbers, enabling timely tracking of risk evolution after each maintenance event. By fusing expert knowledge with operational failure data, the method delivers more adaptive and rational reliability analysis that provides meaningful support for both reliability design and maintenance planning. A case study on a D-type heavy-duty horizontal lathe from Company W demonstrates the effectiveness of the proposed approach.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111823"},"PeriodicalIF":6.5,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037805","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2026-01-15DOI: 10.1016/j.cie.2026.111824
Yanjiao Wang , Naiqi Liu , Xuejie Bai
Supply chain (SC) disruptions often result in significant economic losses and eroded consumer trust. This paper investigates the design of resilient SCs that effectively respond to facility disruptions while considering customer loyalty. To address disruption impacts, we deploy multiple resilience strategies for network reconstruction during disruption events. Given the unpredictability of facility disruptions, we construct a novel moment-based ambiguity set via statistical methods to quantify capacity failure fractions. We develop a two-stage adaptive distributionally robust optimization (ADRO) model that minimizes SC costs under this ambiguity set. While traditional risk measures like Conditional Value at Risk (CVaR) focuses on extreme tail risks, we extend our risk-neutral model to a risk-averse formulation using the mean absolute deviation from the median (MADM) criterion, which provides a new perspective by minimizing cost variability around the median. In terms of solution method, tailored Benders decomposition (BD) algorithms with multi-subproblem are designed for our ADRO model reformulations. The effectiveness of our ADRO methods and BD algorithm is demonstrated via a practical case study. The results indicate that our model can accommodate the risk preferences of decision-makers while simultaneously offering resilient and economic SC design schemes for decision-makers.
{"title":"Robustifying resilient supply chain against ambiguous facility disruptions under new risk-averse criteria","authors":"Yanjiao Wang , Naiqi Liu , Xuejie Bai","doi":"10.1016/j.cie.2026.111824","DOIUrl":"10.1016/j.cie.2026.111824","url":null,"abstract":"<div><div>Supply chain (SC) disruptions often result in significant economic losses and eroded consumer trust. This paper investigates the design of resilient SCs that effectively respond to facility disruptions while considering customer loyalty. To address disruption impacts, we deploy multiple resilience strategies for network reconstruction during disruption events. Given the unpredictability of facility disruptions, we construct a novel moment-based ambiguity set via statistical methods to quantify capacity failure fractions. We develop a two-stage adaptive distributionally robust optimization (ADRO) model that minimizes SC costs under this ambiguity set. While traditional risk measures like Conditional Value at Risk (CVaR) focuses on extreme tail risks, we extend our risk-neutral model to a risk-averse formulation using the mean absolute deviation from the median (MADM) criterion, which provides a new perspective by minimizing cost variability around the median. In terms of solution method, tailored Benders decomposition (BD) algorithms with multi-subproblem are designed for our ADRO model reformulations. The effectiveness of our ADRO methods and BD algorithm is demonstrated via a practical case study. The results indicate that our model can accommodate the risk preferences of decision-makers while simultaneously offering resilient and economic SC design schemes for decision-makers.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111824"},"PeriodicalIF":6.5,"publicationDate":"2026-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977747","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2026-01-14DOI: 10.1016/j.cie.2026.111828
Jannis David, Thomas Volling
The prevalence of work-related musculoskeletal disorders (WMSD) poses a significant challenge for construction companies. These disorders cause severe physical distress for affected workers, as well as reduced productivity, increased absenteeism, and escalating healthcare costs. The situation is exacerbated by ongoing labour shortages and shifting workforce demographics. To address this issue, we present a novel multi-objective decision support framework designed to optimise construction projects both ergonomically and economically. The proposed Bi-MRCPSP expands upon the multi-mode resource-constrained project scheduling problem (MRCPSP) by additionally incorporating worker equipment modes. We consider three objectives: (1) project duration, (2) resource availability cost, and (3) workers’ energy expenditure. Within this framework, three ergonomic interventions are integrated: (1) additional workforce, (2) planned recovery breaks, and (3) the use of exoskeletons. Applying the model to the installation of photovoltaic (PV) systems in residential homes demonstrates its validity and ability to support decision-making for the selection and implementation of interventions. A key finding is that exoskeletons enable more time- and cost-efficient ergonomic workplace designs, encouraging both companies and researchers to explore this technology further.
{"title":"Multi-objective ergonomic–economic project scheduling in construction: The case of photovoltaic system installation","authors":"Jannis David, Thomas Volling","doi":"10.1016/j.cie.2026.111828","DOIUrl":"10.1016/j.cie.2026.111828","url":null,"abstract":"<div><div>The prevalence of work-related musculoskeletal disorders (WMSD) poses a significant challenge for construction companies. These disorders cause severe physical distress for affected workers, as well as reduced productivity, increased absenteeism, and escalating healthcare costs. The situation is exacerbated by ongoing labour shortages and shifting workforce demographics. To address this issue, we present a novel multi-objective decision support framework designed to optimise construction projects both ergonomically and economically. The proposed Bi-MRCPSP expands upon the multi-mode resource-constrained project scheduling problem (MRCPSP) by additionally incorporating worker equipment modes. We consider three objectives: (1) project duration, (2) resource availability cost, and (3) workers’ energy expenditure. Within this framework, three ergonomic interventions are integrated: (1) additional workforce, (2) planned recovery breaks, and (3) the use of exoskeletons. Applying the model to the installation of photovoltaic (PV) systems in residential homes demonstrates its validity and ability to support decision-making for the selection and implementation of interventions. A key finding is that exoskeletons enable more time- and cost-efficient ergonomic workplace designs, encouraging both companies and researchers to explore this technology further.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111828"},"PeriodicalIF":6.5,"publicationDate":"2026-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977744","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2026-01-13DOI: 10.1016/j.cie.2026.111825
Gwang Kim , Youngchul Shin , Yoonjea Jeong
In this study, we address the challenge of reliable monitoring using unmanned aerial vehicles (UAVs) to minimize the sum of travel costs associated with monitoring activities over a specified period. UAV systems are prone to failures caused by uncertainties and unforeseen factors. These disruptions can interfere with system operations, thereby affecting overall performance. The model considers uncertainties related to UAV failure and aims to minimize additional losses incurred due to these uncertainties. We formulate the problem as a two-stage programming model, consisting of here-and-now decisions in the first stage and recourse decision in the second stage. We utilize the sample average approximation (SAA) approach to address the reliable monitoring problem with UAV failure. A solution methodology based on the scenario decomposition technique is employed to enhance the computational efficiency of the SAA method. In addition, numerical experiments are conducted to evaluate statistical estimates of the model bounds using SAA problems and to assess the performance of the proposed algorithm.
{"title":"Scenario decomposition approach for mobile multi-agent monitoring under failure","authors":"Gwang Kim , Youngchul Shin , Yoonjea Jeong","doi":"10.1016/j.cie.2026.111825","DOIUrl":"10.1016/j.cie.2026.111825","url":null,"abstract":"<div><div>In this study, we address the challenge of reliable monitoring using unmanned aerial vehicles (UAVs) to minimize the sum of travel costs associated with monitoring activities over a specified period. UAV systems are prone to failures caused by uncertainties and unforeseen factors. These disruptions can interfere with system operations, thereby affecting overall performance. The model considers uncertainties related to UAV failure and aims to minimize additional losses incurred due to these uncertainties. We formulate the problem as a two-stage programming model, consisting of <em>here-and-now</em> decisions in the first stage and <em>recourse decision</em> in the second stage. We utilize the sample average approximation (SAA) approach to address the reliable monitoring problem with UAV failure. A solution methodology based on the scenario decomposition technique is employed to enhance the computational efficiency of the SAA method. In addition, numerical experiments are conducted to evaluate statistical estimates of the model bounds using SAA problems and to assess the performance of the proposed algorithm.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111825"},"PeriodicalIF":6.5,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977748","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2026-01-12DOI: 10.1016/j.cie.2025.111799
Martin Schönheit , Janis S. Neufeld , Rainer Lasch
As climate challenges intensify, ecological objectives are gaining importance alongside traditional objectives in distributed scheduling, giving rise to distributed green scheduling problems. However, current models and objectives fail to capture key characteristics of geographically distributed manufacturing systems, particularly the emission intensity of electricity generation and the distribution of goods. Since the environmental impact of electricity consumption varies with local emission factors, they are critical in distributed permutation flowshop scheduling problems. Further, the validity of ecological optimization can be compromised, as energy savings may be offset by increased transportation-related emissions. Based on an experimental analysis calibrated to real-world European production networks and including makespan as an economic objective, we find that optimizing total energy consumption results in an average hypervolume RPD of 42.12%, questioning its validity as an indicator of environmental performance in distributed scheduling. Moreover, focusing solely on production-related emissions still results in an average deviation of 26.13%, highlighting the bias caused by neglecting the distribution stage — an effect that becomes more pronounced with increasing product weight. To further enhance real-world applicability, we assess the impact of eligibility constraints — arising from limited redundancy in tools and raw materials — on the potential to minimize both makespan and carbon emissions, and propose distance- and emission-aware strategies for factory qualification. Finally, the problem is solved using a novel parameter-less iterated greedy algorithm that incorporates problem-specific knowledge into speed factor adjustment, removes the need for parameter tuning, and demonstrates strong solution quality in extensive computational experiments.
{"title":"Towards holistic environmental awareness in distributed permutation flowshop scheduling: Integrating production and transportation emissions","authors":"Martin Schönheit , Janis S. Neufeld , Rainer Lasch","doi":"10.1016/j.cie.2025.111799","DOIUrl":"10.1016/j.cie.2025.111799","url":null,"abstract":"<div><div>As climate challenges intensify, ecological objectives are gaining importance alongside traditional objectives in distributed scheduling, giving rise to distributed green scheduling problems. However, current models and objectives fail to capture key characteristics of geographically distributed manufacturing systems, particularly the emission intensity of electricity generation and the distribution of goods. Since the environmental impact of electricity consumption varies with local emission factors, they are critical in distributed permutation flowshop scheduling problems. Further, the validity of ecological optimization can be compromised, as energy savings may be offset by increased transportation-related emissions. Based on an experimental analysis calibrated to real-world European production networks and including makespan as an economic objective, we find that optimizing total energy consumption results in an average hypervolume RPD of 42.12%, questioning its validity as an indicator of environmental performance in distributed scheduling. Moreover, focusing solely on production-related emissions still results in an average deviation of 26.13%, highlighting the bias caused by neglecting the distribution stage — an effect that becomes more pronounced with increasing product weight. To further enhance real-world applicability, we assess the impact of eligibility constraints — arising from limited redundancy in tools and raw materials — on the potential to minimize both makespan and carbon emissions, and propose distance- and emission-aware strategies for factory qualification. Finally, the problem is solved using a novel parameter-less iterated greedy algorithm that incorporates problem-specific knowledge into speed factor adjustment, removes the need for parameter tuning, and demonstrates strong solution quality in extensive computational experiments.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111799"},"PeriodicalIF":6.5,"publicationDate":"2026-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977738","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2026-01-12DOI: 10.1016/j.cie.2026.111815
Yuxuan Zhang , Liang Chen , Xiangyu Bao , Jianguang Su , Lei Zhang , Yu Zheng
The integrated planning of berth allocation, quay crane assignment, and yard assignment (BQCYAP) is crucial for improving the service and efficiency of container terminals. Since three resource allocations for numerous vessels must be considered simultaneously, the decision space of BQCYAP is vast. Conventional branch-and-price (B&P) algorithms often produce useless subproblems or require extra runtime to test them, which makes exact solutions challenging. This paper proposes a deep reinforcement learning (DRL)-enhanced B&P algorithm that selects efficient branching variables without testing cost. We formulate the B&P procedure as a tree Markov decision process (MDP) and develop a DRL method to train the branching policy. To leverage information from the column generation procedure, a tripartite graph is proposed to represent the node states consisting of original variables, master problem constraints, and columns. Numerical experiments on various instance sizes demonstrate that the branching policy trained by the proposed DRL method significantly reduces the search tree size, enabling the B&P algorithm to outperform commercial solvers. Furthermore, comparative results verify the effectiveness of the tree MDP-based return function and the tripartite graph-based state representation in improving the generalizability and stability of the DRL method.
{"title":"Deep reinforcement learning-enhanced branch-and-price algorithm for integrated planning of berth allocation, quay crane assignment, and yard assignment","authors":"Yuxuan Zhang , Liang Chen , Xiangyu Bao , Jianguang Su , Lei Zhang , Yu Zheng","doi":"10.1016/j.cie.2026.111815","DOIUrl":"10.1016/j.cie.2026.111815","url":null,"abstract":"<div><div>The integrated planning of berth allocation, quay crane assignment, and yard assignment (BQCYAP) is crucial for improving the service and efficiency of container terminals. Since three resource allocations for numerous vessels must be considered simultaneously, the decision space of BQCYAP is vast. Conventional branch-and-price (B&P) algorithms often produce useless subproblems or require extra runtime to test them, which makes exact solutions challenging. This paper proposes a deep reinforcement learning (DRL)-enhanced B&P algorithm that selects efficient branching variables without testing cost. We formulate the B&P procedure as a tree Markov decision process (MDP) and develop a DRL method to train the branching policy. To leverage information from the column generation procedure, a tripartite graph is proposed to represent the node states consisting of original variables, master problem constraints, and columns. Numerical experiments on various instance sizes demonstrate that the branching policy trained by the proposed DRL method significantly reduces the search tree size, enabling the B&P algorithm to outperform commercial solvers. Furthermore, comparative results verify the effectiveness of the tree MDP-based return function and the tripartite graph-based state representation in improving the generalizability and stability of the DRL method.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111815"},"PeriodicalIF":6.5,"publicationDate":"2026-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977753","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Efficient preparation and smooth operation of rail freight trains are essential for improving rail freight services and customer satisfaction. This study examines how automation and digitalisation − specifically Digital Automatic Coupling (DAC) and Virtual Coupling (VC) − can enable seamless rail freight transport within marshalling yards and along railway lines. For the first time, a combined simulation- and optimisation-based modelling approach is proposed to assess the impact of these technologies.
A multi-agent simulation model of the Hallsberg marshalling yard was developed to analyse train handling and yard capacity. A 10-hour shunting operation was simulated under manual coupling and DAC technology, comparing standard train lengths and longer trains. The results indicate a substantial increase in processed trains when DAC was applied. Standard-length trains increased from 7 (manual) to 9 and 12 with DAC types 4 and 5, respectively, with similar gains observed for longer trains.
Trains from the simulation’s departure yard were subsequently integrated into an optimisation model to assess their scheduling on the main railway line. dispatchers face challenges in optimising freight train routing, VC was proposed as a capacity-enhancing measure. The optimisation results showed that, with conventional timetables, only 70 freight trains could be scheduled while prioritising passenger services, whereas VC enables up to 128 freight trains − − an 82.86% capacity increase.
Overall, these results demonstrate that integrating DAC and VC technologies can significantly enhance the efficiency and capacity of rail freight operations and systems, offering substantial benefits to stakeholders across the sector.
{"title":"An approach for seamless rail freight: integration of virtual coupling and digital automatic coupling","authors":"Weiting Yang , Yuguang Wei , Evelin Krmac , Boban Djordjevic","doi":"10.1016/j.cie.2026.111810","DOIUrl":"10.1016/j.cie.2026.111810","url":null,"abstract":"<div><div>Efficient preparation and smooth operation of rail freight trains are essential for improving rail freight services and customer satisfaction. This study examines how automation and digitalisation − specifically Digital Automatic Coupling (DAC) and Virtual Coupling (VC) − can enable seamless rail freight transport within marshalling yards and along railway lines. For the first time, a combined simulation- and optimisation-based modelling approach is proposed to assess the impact of these technologies.</div><div>A multi-agent simulation model of the Hallsberg marshalling yard was developed to analyse train handling and yard capacity. A 10-hour shunting operation was simulated under manual coupling and DAC technology, comparing standard train lengths and longer trains. The results indicate a substantial increase in processed trains when DAC was applied. Standard-length trains increased from 7 (manual) to 9 and 12 with DAC types 4 and 5, respectively, with similar gains observed for longer trains.</div><div>Trains from the simulation’s departure yard were subsequently integrated into an optimisation model to assess their scheduling on the main railway line. dispatchers face challenges in optimising freight train routing, VC was proposed as a capacity-enhancing measure. The optimisation results showed that, with conventional timetables, only 70 freight trains could be scheduled while prioritising passenger services, whereas VC enables up to 128 freight trains − − an 82.86% capacity increase.</div><div>Overall, these results demonstrate that integrating DAC and VC technologies can significantly enhance the efficiency and capacity of rail freight operations and systems, offering substantial benefits to stakeholders across the sector.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111810"},"PeriodicalIF":6.5,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977752","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2026-01-10DOI: 10.1016/j.cie.2026.111811
John Maleyeff , Jingran Xu , Ruthairut Wootisarn
Simheuristics is a simulation optimization method that combines simulation with heuristic approaches to solve complex or combinatorically challenging problems. Its performance is considered effective when it converges on a good solution while minimizing the number of simulation runs. Repair part inventory policy is an increasingly important component of inventory management due to the proliferation of equipment and products that need frequent updating, overhaul, or repair. A repair inventory problem, where the repair can start only after all parts needed for the repair are available, is addressed using a two-phase simheuristics algorithm. The approach is unique because in phase 1 it employs a designed experiment to create a metamodel of simheuristics results which, in phase 2, becomes the initial solution presented to the simheuristics algorithm. Results show faster convergence compared to the use of a deterministic model that typically initializes a simheuristics algorithm.
{"title":"Simheuristics with metamodel initialization for determining repair system inventory policies","authors":"John Maleyeff , Jingran Xu , Ruthairut Wootisarn","doi":"10.1016/j.cie.2026.111811","DOIUrl":"10.1016/j.cie.2026.111811","url":null,"abstract":"<div><div>Simheuristics is a simulation optimization method that combines simulation with heuristic approaches to solve complex or combinatorically challenging problems. Its performance is considered effective when it converges on a good solution while minimizing the number of simulation runs. Repair part inventory policy is an increasingly important component of inventory management due to the proliferation of equipment and products that need frequent updating, overhaul, or repair. A repair inventory problem, where the repair can start only after all parts needed for the repair are available, is addressed using a two-phase simheuristics algorithm. The approach is unique because in phase 1 it employs a designed experiment to create a metamodel of simheuristics results which, in phase 2, becomes the initial solution presented to the simheuristics algorithm. Results show faster convergence compared to the use of a deterministic model that typically initializes a simheuristics algorithm.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111811"},"PeriodicalIF":6.5,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977755","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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