Additive manufacturing (AM) is a revolutionary technology gaining substantial interest from academia and industry. Facing supply chain (SC) disruption risks, AM helps SC partners restore capacities through in-house and on-demand production. Compared to common resilience strategies, e.g. using backup suppliers and outsourcing, AM can reduce structural SC redundancy and enhance responsiveness to market demands. However, AM's impacts on SCs under the ripple effect remain insufficiently explored. This work investigates a SC resilience improvement problem by combining a novel AM strategy with inventory redundancy. For the problem, a dynamic Bayesian network is applied to portray the ripple effect, and a problem-specific Markov decision process is proposed to quantify the impacts of resilience strategies. A new mixed-integer non-linear non-convex optimisation model is established to minimise the disruption risk, and a Q-learning-based genetic algorithm is designed for solving large-scale problems. Key managerial insights from the case study include: (i) the proposed approach assists SC managers in prioritising key partners and implementing differentiated strategies based on partners' positions within the SC under limited budgets; and (ii) the temporal factor is critical, necessitating AM machine rentals for immediate post-disruption response and early AM machine purchases to ensure long-term resilience.

Disassembly line balancing and sequencing (DLBS) problem has received considerable attention from both enterprises and researchers, driven by its significant impact on end-of-life (EOL) product processing efficiency. Most existing studies focus on parallel lines with shared workstations, handling different products simultaneously. However, the consecutive connected lines, implemented in practice for processing complex EOL products, remain unexplored in the existing literature. This study addresses the parallel DLBS problem considering line consecutive connectivity, where products undergo initial disassembly on the main line before being transferred to branch lines for further processing, aiming to minimize the overall system cost. We propose a mixed-integer linear programming (MILP) model and then develop a customized logic-based Benders decomposition (LBBD) approach to improve computational efficiency. The LBBD method divides the problem into a master task assignment problem, which is enhanced by some valid inequalities and solved in the branch-and-cut framework, and a task sequencing subproblem, tackled via the dynamic programming approach. Numerical results demonstrate the effectiveness and efficiency of our proposed LBBD algorithm.

The integrated scheduling of charging and electric bus (EB) trips has attracted significant attention and has been extensively studied in deterministic settings. However, in real-world operations, uncertainties in travel times and energy consumption can substantially influence the feasibility and effectiveness of charging plans. Moreover, vehicle-to-grid (V2G) technology is a promising solution with the potential to enhance the balance of the energy system. However, it also introduces several challenges, such as the potential reduction in battery lifespan. To date, no research has comprehensively addressed the integrated optimization of trip scheduling and charging, taking into account both V2G and battery degradation under stochastic conditions. This study proposes a two-stage stochastic optimization model to solve this joint problem, aiming to minimize the expected total operational cost. To solve the model, the sample average approximation method is employed. Numerical experiments demonstrate that the proposed model can offer a comprehensive and optimal EB system planning solution for decision-makers.

The electrification of urban transport has made battery electric buses (BEBs) an important option for reducing carbon emissions and improving urban air quality. However, the high investment cost of charging infrastructure and the uncertainty in effective usable battery capacity at the day-ahead scheduling stage—caused by accumulated degradation, heterogeneous operating conditions, and imperfect state estimation—create major challenges for charging infrastructure siting and daily bus operations. This study proposes a joint optimization model for infrastructure siting and BEB charging scheduling, in which effective capacity uncertainty is handled using a distributionally robust optimization (DRO) framework. To solve the resulting mixed-integer nonlinear program efficiently, we develop a matheuristic decomposition method that integrates Adaptive Large Neighborhood Search (ALNS) with small gaps relative to a relaxation-based lower bound. Computational experiments based on real-world bus route data indicate that the proposed framework obtains high-quality solutions with small gaps relative to a relaxation-based lower bound, performs better than representative benchmark heuristics, and scales well to large instances.

Pedestrian safety at urban intersections is influenced by complex interactions among environmental conditions, social cues, and boundedly rational behavioral responses, making intervention planning under uncertainty particularly challenging. Existing studies have largely emphasized descriptive analysis or predictive modeling, with limited attention to prescriptive decision support under data scarcity. This study proposes a Bayesian-network-based robust optimization framework for pedestrian safety planning, complemented by a large language model (LLM)-assisted parameter-elicitation process. We first construct a three-layer Bayesian network based on the stimulus–organism–response paradigm to represent the propagation of risk from environmental cues through latent psychological mechanisms to behavioral violations and accident risk. To support model initialization when site-specific behavioral data are limited, we introduce an LLM-assisted elicitation protocol that maps literature-based qualitative evidence to intervention mechanisms, effect directions, and qualitative strength classes. Numerical parameter ranges are subsequently assigned through explicit mapping rules rather than generated directly by the LLM. We then formulate a bi-objective robust optimization model that distinguishes between physical interventions, which alter root-node distributions, and cognitive interventions, which modify conditional probability tables. Using the ϵ-constraint method, the framework generates Pareto-optimal intervention portfolios and evaluates cost–risk trade-offs under behavioral uncertainty and adverse operating scenarios.

The increasing global demand for perishable agricultural products necessitates advancements in cold chain logistics. Cross-docking, known for its efficiency, is particularly well-suited for the transfer and distribution of such goods. However, truck scheduling at cold chain cross-dock terminals (CDTs) presents unique challenges, including product perishability, stringent time windows, and temperature-controlled environments. This work investigates a truck scheduling problem within a cold chain CDT, explicitly addressing uncertainties in refrigerated product damage (affecting supply) and repackaging times. A two-stage stochastic programming model is developed to capture these uncertainties. To solve this model, a scenario reduction approach employing K-means++ and K-medoids clustering is used, followed by Sample Average Approximation. Small-scale instances are solved optimally using CPLEX. For larger instances, a novel hybrid heuristic algorithm, combining the global search capabilities of Genetic Algorithms with the local search capabilities of Adaptive Large Neighborhood Search and Simulated Annealing, is proposed. Numerical experiments demonstrate the effectiveness of this algorithm, and sensitivity analysis provides valuable managerial insights.

Combining forecasts from diverse models via weighted sums enhances accuracy. Feature-based methods, which link data features to forecasting model performance, offer a promising approach to assigning weights. Currently, limited attention has been paid to the importance of reliable performance measurement. In this work, we propose an improved feature-based forecast combination method, particularly using the rolling origin evaluation to obtain reliable performance measurements of forecasting models. Experimental results based on the M4 competition data show that our method outperforms the state-of-the-art. The proposed method exhibits robustness to parameter variations.

This work investigates the permutation flowshop scheduling problem where each operation of any job is performed on a lot processing machine with uniform capacity. More than one job can be processed in the same lot, and the operations of all jobs in the lot are of the same completion time. Any job can be split and processed in consecutive lots, if necessary. The objective is to minimize the maximum completion time of the last operation of job, i.e., the makespan. We first examine the complexity of the considered problem, and provide a polynomial approximation algorithm when there are m = 2 operations in the flowshop. We further explore three special cases with m = 2 and present optimal solutions for each case, respectively. Moreover, we provide an m-approximation algorithm for the situation where there are m >= 2 operations. Finally, the efficiency of the approximation algorithm is demonstrated via numerical experiments.

Sustainable aviation supply chains (SCs) are increasingly exposed to risks arising from environmental regulations, social responsibility pressures, and economic uncertainties. These risks are associated with different SC members and may propagate through operational dependencies among suppliers, maintenance service providers, and airline operators. To support systematic risk assessment, this study proposes a hybrid Analytical Hierarchy Process-Bayesian network (AHP-BN) framework for sustainable aviation SC risk management. The intended contribution is a contextual and structural extension of existing AHP-BN logic to member-level sustainability risk propagation in aviation SCs, rather than a claim that AHP-BN integration itself is fundamentally new. The proposed framework first classifies sustainability risks into environmental, social, and economic dimensions and identifies the risk exposure relationship between SC members and risk factors. For the weighting component, Analytical Hierarchy Process (AHP) is used to derive relative importance weights from specified illustrative pairwise comparison matrices in the numerical experiment. Bayesian network (BN) is employed to model probabilistic dependencies among nodes defined by SC members and risk factors. The two methods are coupled through a weighted expected risk index, which integrates AHP-derived weights, member-specific exposure intensities, probabilities inferred by BN, and losses associated with different risk states. A numerical illustration based on a synthetic aviation SC with suppliers, maintenance service providers, and airline operators is conducted to demonstrate the computational procedure and diagnostic use of the proposed framework rather than to validate an empirical risk profile of the aviation industry. Within this illustrative setting, cost volatility, supplier reliability, emissions regulation, and sustainable aviation fuel availability emerge as the major contributors to the overall risk index under the assumed inputs. The analysis further indicates that the proposed framework can identify critical active pairs of SC members and risk factors, reveal vulnerabilities at the levels of SC members and sustainability dimensions, and provide a transparent decision-support tool for sustainable aviation SC risk assessment, while the resulting rankings should be interpreted as conditional outputs under the assumed input parameters.

The transition to Industry 5.0 is reshaping supply chain management by emphasizing resilience, sustainability, and human-centricity. However, studies on food supply chains have not simultaneously examined returnable transport items and human-centricity under disruptions. To fill this gap, this paper examines a novel resilience improvement problem for the food closed-loop supply chain, considering two resilience strategies under disruptions. The objective is to minimize the total expected operational cost under worst-case scenarios. For the problem, we develop a distributionally robust chance-constrained programming model and, by exploiting its structural properties, derive an approximate mixed-integer linear programming reformulation. A real-world case study and experiments on randomly generated instances validate the robustness of the proposed strategies and solution approach. Key results demonstrate that resilience requires the synergy of the proposed temporary worker recruitment and returnable transport item inventory redundancy strategies, as isolated approaches are ineffective.

In the context of Industry 5.0, human-robot collaboration (HRC) has emerged as a significant research area. For manufacturing companies, the prevalence of personnel absences due to disasters, viruses, and other factors is a global concern. Moreover, market and energy consumption uncertainties are increasingly prevalent. However, existing HRC research often neglects these uncertainties in the reconfigurable manufacturing systems (RMSs) that are not fully automated. To address these gaps, this study proposes an HRC-RMS framework, explicitly considering uncertain demand, energy consumption, and personnel absenteeism. The problem encompasses labour allocation, configuration optimisation, and production planning. The objective is to minimise the total cost, including reconfiguration, exploitation, energy consumption, inventory, and stockout costs. A two-stage stochastic programming model is introduced, where the first stage includes labour allocation, configuration optimisation, and planned production quantity decisions in all periods, while the second stage captures scenario-dependent recourse decisions through inventory and stockouts. To solve the model, we develop a column generation algorithm (CG) and a CG-GAT algorithm that integrates CG with a multi-head Graph Attention Network. We conduct numerical experiments to compare the performance of state-of-the-art algorithm and the proposed algorithms. Numerical results show that CG-GAT outperforms the others. This work further provides managerial insights.

Supply chain (SC) disruption risk assessment has attracted extensive attention, yet important gaps remain. Existing studies typically examine either forward or backward propagation, whereas in reality disruptions can spread simultaneously in both directions. Moreover, widely used methods such as Bayesian networks and simulations usually represent risks via discrete states, which oversimplifies their continuous and evolving nature. To bridge these gaps, this paper proposes a novel pressure wave-based approach inspired by fluid mechanics. We conceptualize disruptions as pressure signals that transmit forward and reflect backward between SC partners and explicitly quantifies disruption severity through wave intensity. Numerical experiments on a cluster supply chain demonstrate the model’s effectiveness in identifying critical nodes and evaluating mitigation strategies, such as inventory buffers. Our method provides a scalable and explainable alternative to conventional tools, offering managers a powerful lens for efficient risk assessment and evaluation of proactive resilience planning.

Supply chain (SC) disruption risk assessment and mitigation have attracted significant attention in both academia and practice. However, existing research predominantly focuses on unidirectional disruption propagation, either forward or backward, despite the reality that risks can propagate bi-directionally in complex supply chain networks. Furthermore, conventional assessment tools often concentrate on conceptualizing and quantifying risks, while risk mitigation requires mathematical optimization approaches. To bridge these gaps, this paper proposes a novel pressure wave-based approach inspired by fluid mechanics to assess bi-directional disruption propagation in cluster supply chain networks (CSCNs). The method conceptualizes disruptions as pressure signals that transmit between SC partners and explicitly quantifies disruption severity through wave intensity. By employing mathematical optimization, we develop a framework that assists managers in optimizing risk mitigation strategies, including inventory buffering and cross-chain cooperation. Numerical experiments demonstrate the effectiveness of the proposed method in explaining risk influencing factors, mitigating disruption risks, and achieving dynamic restructuring of SC structures. The results show that our approach reduces the Cluster Propagation Vulnerability Index (CPVI) by up to 40% compared to baseline models without optimization decisions.

In response to the challenges of uncertain market demand and production environment, this study addresses the risk-averse balancing and planning of a reconfigurable manufacturing system (RMS) under demand and processing time uncertainties. Amidst these uncertainties, the decision-maker’s risk attitude shapes the final decisions. Moreover, technological progress makes heterogeneous machines common in production to improve efficiency. However, the existing literature lacks models that address machine selection, uncertain processing times, and risk aversion comprehensively. To bridge this gap, we aim to maximize the profit by optimizing station utilization, machine rental cost, energy consumption, and product revenue. We introduce a risk-averse two-stage stochastic programming (TSSP) model that encompasses balancing and planning stages, incorporating time-of-use (TOU) electricity prices. Then, we prove that the problem is NP-hard. To solve it, we develop two algorithms: an improved genetic algorithm with stochastic variable neighborhood search (GASVNS) and a rule-based heuristic algorithm integrated with CPLEX (RBH). Using K-means clustering, the two algorithms are further enhanced, referred to as K-GASVNS and K-RBH, respectively. We conduct numerical experiments to compare the performance of the state-of-the-art algorithm and the proposed algorithms. Numerical results show that K-GASVNS outperforms the others in solution quality, while K-RBH has the best performance in terms of running time. This work further provides managerial insights.

Fueled by artificial intelligence advancements, the escalating demand for micro-products necessitates optimized production scheduling to manage diverse custom orders and foster sustainable industry growth efficiently. This study investigates the single-machine lot scheduling problem, characterized by a fixed lot capacity and constant delivery times, where order splitting is permitted. The objective is to minimize the maximum delivery completion time. For the offline problem (all information known a priori), we prove that an optimal rule yields polynomial-time solutions. For the online version (orders arrive dynamically), we propose a 2-competitive manual algorithm (DLDTF) and apply the Evolution of Heuristics (EoH) framework, integrating Large Language Models (LLMs) and Evolutionary Computation (EC), to automatically generate high-performance online heuristics. Computational experiments confirm the offline optimality rule and validate the online algorithms. Results indicate the manually designed online algorithm DLDTF performs robustly on small instances, whereas the automatically generated EoH-heuristics demonstrate superior performance for large-scale problems. Finally, managerial insights are derived from the analysis.

Dual-channel supply chain (DC-SC), which integrates both traditional offline and online channels, has significantly improved its performance. However, the increasing frequency of disruptions in DC-SCs, compounded by ripple effects, poses severe challenges to SC resilience. In response, one manufacturer has recently adopted exclusive stores in offline channels, which is operated solely by the manufacturer to sell its own products. There is no literature dedicated to DC-SC resilience that considers exclusive stores under ripple effects. Thus, we study a novel DC-SC resilience building problem with exclusive stores under ripple effects. The problem consists of (i) making pricing decisions, and (ii) planning on inventory redundancy, production and distribution. A conditional value-at-risk (CVaR)-oriented stochastic programming model with a dynamic Bayesian network (DBN)-based scenario set is established. Then, a novel scenario reduction (SR)-based method combined with probabilistic distance is designed. Numerical experiments indicate that the SR-based method guarantee solution quality relative to sample average approximation (SAA) benchmark while maintaining shorter running time, and the sensitivity analysis yields managerial insights.

With the frequency of sudden disasters, global supply chains are at risk of supply disruption. Backup production is an appropriate but costly method for supply chains to solve supply disruption problems. However, few studies have taken into account that suppliers may face capital-constrained problems in the event of supply disruption, which limits their abilities to adopt backup production during the supply disruption. This research examines a supply chain scenario where the capital-constrained supplier encounters the risks of supply disruptions and explores the effects of various backup production decisions and financing strategies. We demonstrate that with large potential market size, the capital-constrained supplier will adopt manufacturer financing with backup production to mitigate his financial pressure, and the manufacturer will offer financing to the capital-constrained supplier. With the small potential market size, when selecting bank financing without backup production, the supplier will achieve the highest profits, but the manufacturer will not offer financing to the capital-constrained supplier. Moreover, the manufacturer cannot obtain higher profits with backup production adoption and high unit production cost. These insights highlight the critical relationship between financing strategies and backup production in enhancing supply chain’s resilience.

Consensus mechanisms are algorithms that ensure the security and stability of blockchain networks by achieving agreement and verifying transaction integrity. Proof of Stake (PoS) stands as a widely acknowledged consensus algorithm, wherein the privilege to validate transactions is predicated upon participants’ stakes. However, long-term use of PoS may lead to wealth concentration among certain nodes, potentially undermining the network’s fairness and security. Therefore, we propose the Group-Polynomial-based Election Proof of Stake (GPE-PoS) consensus mechanism. GPE-PoS involves categorizing nodes, calculating polynomial values for each group, encrypting these values using Paillier encryption, and then allocating validation rights based on comparisons of polynomial values based on polynomial value comparisons to enhance system fairness. The fairness of the system is further fortified against Sybil attacks, which undermine its security and fairness, through the incorporation of digital certificates within GPE-PoS, thereby verifying participant identities. Simulation results confirm that GPE-PoS successfully maintains fairness and security in blockchain systems.

We investigate an electric bus overnight charging scheduling problem considering the uncertainty of bus charge-discharge efficiency. In line with scheduling practice, the vehicle-to-grid technology is considered in this problem. Given a set of appointed electric buses, our objective is to explore charging strategies that minimize the weighted sum of the total charging cost for the charging station, power grid load fluctuations, together with the negative value of rewards obtained from using V2G technology. Inspired by Abdelwahed et al., both discrete-time-optimization formulation and discrete-event-optimization formulation are established to model this problem under uncertain environments. We then employ K-means enhanced sample average approximation approach and K-means enhanced conditional value at risk approach to solve this problem. Numerical experiments are conducted to demonstrate the effectiveness of our approaches.

For a no-wait flow shop with continuous-flow characteristics, this study simultaneously considers machine setup times and rated processing speed constraints, aiming to minimize the sum of the maximum completion time and the maximum tardiness. First, lower bounds for the maximum completion time, the maximum tardiness, and the total objective function are developed. Second, a mixed-integer programming (MIP) model is formulated for the problem, and nonlinear elements are subsequently linearized via time discretization. Due to the computational complexity of the problem, two algorithms are proposed: a heuristic algorithm with fixed machine links and greedy rules (HAFG) and a genetic algorithm based on altering machine combinations (GAAM) for solving large-scale instances. The Earliest Due Date (EDD) rule is used as baselines for algorithmic comparison. To better understand the behaviors of the two algorithms, we observe the two components of the objective function separately. The results show that, compared with the EDD rule and GAAM, the HAFG algorithm tends to focus more on optimizing the maximum completion time. The performance of both algorithms is evaluated using their relative deviations from the developed lower bounds and is compared against the EDD rule. Numerical experiments demonstrate that both HAFG and GAAM significantly outperform the EDD rule. In large-scale instances, the HAFG algorithm achieves a gap of about 4%, while GAAM reaches a gap of about 3%, which is very close to the lower bound. In contrast, the EDD rule shows a deviation of about 10%. Combined with a sensitivity analysis on the number of machines, the proposed framework provides meaningful managerial insights for continuous-flow production environments.