Pseudo-cost based branching is a popular branching strategy used by Mixed Integer Programming (MIP) solvers. This strategy relies on pseudo-cost updates from various parts of the search tree for making accurate branching decisions. Since such updates are not instantly available during a distributed computation, parallel MIP solver implementations that use pseudo-cost branching may not perform well when the underlying cluster is scaled horizontally. To address this issue, we propose integrating a repository of pre-calculated pseudo-costs into a parallel implementation of CPLEX. Although all the facilities needed for such an implementation are currently not available, experiments with hard-to-solve instances indicate that the proposed implementation can help limit the number of nodes explored during the distributed computation.
WEEE recycling is critical for global environmental governance and sustainable resource development, especially in developing countries where the coexistence of weak formal sectors and rampant informal recycling leads to resource waste and severe pollution. Existing research predominantly focuses on government subsidies for formal recyclers, yet it overlooks the integration of governance of informal sectors. This study constructs a dual-channel (formal/informal) price competition model incorporating consumer environmental awareness, systematically evaluating the effects of government subsidy policies and integration governance policies on recycling volumes, resource reuse efficiency, and environmental benefits. The findings indicate that government subsidy policies are constrained by a threshold: their effectiveness is significantly limited if subsidy intensity fails to bridge the processing cost gap between formal and informal sectors. Conversely, the implementation of integration governance policies requires that informal recyclers' profits under such policies are not lower than those under government subsidies. Policy selection depends on dynamic matching between government investment and consumer environmental awareness: with low investment, integration governance outperforms in regions with high environmental awareness; with sufficient investment, subsidies yield more comprehensive benefits. Furthermore, an extended "subsidy + integration" synergistic policy is proposed to leverage the advantages of both approaches, achieving full resource utilization and stable environmental benefits.
The rapid development of e-commerce has prompted firms with logistics advantages to establish self-run logistics. However, ambiguity remains regarding whether firms with logistics disadvantages should join shared logistics alliances. To address this issue, this study investigates the optimal logistics choices of two competing firms through a game theory model. The results indicate that TPL has an incentive to enhance green technology levels and reduce logistics capacity constraints. Furthermore, when the TPL's green technology level is low, the firm with logistics advantages is more likely to establish self-run logistics. Under certain conditions, the firm with logistics disadvantages joins the shared logistics alliance, leading to a win-win outcome for both parties. Notably, when the firm with logistics advantages has a low (high) unit environmental impact, the firm with logistics disadvantages choosing to join the shared logistics alliance will result in the smallest (largest) environmental impact, with social welfare reaching the highest (lowest) level. Therefore, the government must encourage firms with logistics advantages to invest in green technologies to reduce unit environmental impact and promote the establishment of shared logistics alliances to enhance environmental sustainability and social welfare. This research aims to provide theoretical guidance for firms' logistics choices and the government's policy guidelines.
Setting research and development (R&D) output targets is a critical managerial task. Data envelopment analysis (DEA), a classical tool for constructing production possibility sets and frontiers, can support target setting. However, conventional DEA is designed for ex-post analysis and lacks predictive capability, which limits its suitability for forward-looking R&D target setting. In addition, DEA typically follows an efficiency-maximization principle, often producing overly ambitious targets that are difficult to achieve in the short term. To address these limitations, this paper integrates artificial neural network (ANN) with DEA to enhance predictive capability and enable ex-ante analysis. It also introduces a gradual efficiency improvement principle, recognizing that efficiency gains in practice are usually incremental rather than immediate. This principle allows the degree of efficiency improvement to be flexibly controlled, making target setting more realistic and manageable. The proposed approach is applied to 3,660 Chinese listed companies. When 2022 data are used to predict the R&D output targets required to achieve different efficiency-improvement levels in the following year, the results demonstrate that the approach provides effective support for practical R&D target setting.
This study employs game-theoretic models to analyse a dual-channel supply chain system, where a manufacturer sells its product in two ways: (i) directly online to consumers and (ii) by supplying a retailer who sells the product offline. Now, Showrooming poses a significant threat to the retailer, as some customers visit the offline store to inspect the product but ultimately purchase it online, attracted by better prices. This paper addresses this challenge and proposes an in-person discount strategy that provides exclusive discounts only to showrooming customers. When a customer visits the store but searches for the product in the manufacturer's online channel, potential showrooming behaviour is detected through geolocation or browsing data, and a personalised discount is offered through the customer's mobile device to encourage in-store purchase. This mechanism is facilitated through collaboration between the retailer and the manufacturer's online channel. Findings reveal that the strategy reduces the impact of showrooming and improves retailer profitability, although it may reduce manufacturer profit under certain market conditions. When the manufacturer leads and both showroomer proportion and price competition are high, increased sales effort yields win-win outcomes and lowers selling prices. Under retailer leadership, win-win results occur, but often with a higher offline price.
Urban waste generation is increasing worldwide at a dramatic pace, making it crucial to develop efficient, cost-effective, and environmentally responsible waste management systems for fast-growing urban areas. This paper addresses a bi-objective facility location problem arising from the real-world waste industry, in which different classes of recyclable urban waste must be collected from sources and delivered to treatment or disposal facilities. The decisions are the number and location of the new intermediate transfer facilities to be opened and the optimal waste flow across the network. The goals are the minimization of the total costs and the CO2 emissions. We present a single-period mixed integer linear programming model and then extend it to a multi-period setting, which better reflects the dynamics of waste production with seasonal fluctuations and generalize to further applications. We apply an & varepsilon;-constraint algorithm to solve our models on two real-world case studies, obtaining approximated-but-well-structured Pareto sets of non-dominated solutions with 25% reduction of emissions with respect to the current state. The efficacy of the models is confirmed by further computational experiments on randomly created instances, showing that the models can be employed for analogous applications.
Developing a well-structured biomass supply chain (BSC) is important to ensure continuous access to bioenergy at a reasonable cost. Variations in feedstock location and type, along with seasonality, demand dispersion and logistical challenges, collectively undermine the viability of BSCs. Therefore, establishing a resilient network that preserves its operational efficiency under disruption is crucial. Quantifying the multifaceted nature of resilience is challenging, particularly in multi-echelon supply chains due to disruption propagation. As such, this study proposes a systematic analysis framework relying on agent-based modeling to quantitatively assess the resilience of a BSC network through the lens of the triple resilience capacities (i.e. absorptive, adaptive and restorative). This framework also models temporal behavior and interdependence among entities. Several recovery strategies are incorporated into the model to restore the functionality of critical nodes. A real case study of the remote, off-grid communities in Quebec is considered to validate the efficiency of the proposed framework. The outcomes underscore the necessity of implementing recovery strategies and the importance of inventory management at intermediary nodes to effectively harness disruption effects. More precisely, extra inventories at bio-hubs reduce the severity of disruption propagation, enhance responsiveness and lead to a higher share of bioenergy adoption in downstream echelons.
This paper introduces a hybrid algorithm (TSQL) that combines tabu search and Q-learning to address the challenges of container loading problems (CLPs). Recognized as a crucial logistics optimization task, CLP aims to maximize volume utilization within containers. However, large-scale CLPs are computationally intensive, limiting the applicability of exact solutions. The proposed TSQL algorithm leverages Q-learning to dynamically guide the neighborhood exploration process within tabu search, achieving an effective balance between exploration and exploitation. Comparative analysis on benchmark datasets reveals that TSQL consistently surpasses traditional tabu search in both convergence speed and solution quality, especially in scenarios with high problem complexity. To further validate its effectiveness, TSQL was benchmarked against several state-of-the-art algorithms across seven problem classes and achieved the second-best average performance overall, outperforming many hybrid and metaheuristic alternatives. This hybrid approach offers significant potential for logistics applications that involve complex loading and packing decisions and can support more sustainable operations by improving space utilization.
We study how to plan wind and solar expansion in a hydro-dominated power system when resources are intermittent and uncertain. We formulate a multistage stochastic generation expansion model that co-optimizes where and how much wind and solar to build together with detailed hydro-wind-solar operation. The planning horizon consists of one expansion stage followed by weekly operational stages with intra-day blocks, and three stochastic inputs: water inflows, wind power, and solar power. The model is solved with Stochastic Dual Dynamic Programming (SDDP), which co-optimizes expansion and short-term operation under uncertainty. We compare the resulting SDDP policy with two alternatives: a deterministic Model Predictive Control (MPC) policy and a Na & iuml;ve expansion policy whose operations are optimized by SDDP. The model is applied to the province of Qu & eacute;bec, with expansion budgets of up to 1.5 GW of wind and 3 GW of solar across 10 candidate zones. All approaches invest to these limits but allocate capacity differently. Out-of-sample simulations show that SDDP delivers higher effective renewable generation, much lower curtailment, and substantial cost savings - about 102 M$ per year relative to Na & iuml;ve and 1,614 M$ relative to MPC, while relying less on extreme demand response.
Timely discharge is a critical component of patient satisfaction and operational efficiency in hospitals. In Indian healthcare settings, patients availing cashless medical insurance often face significant delays during the discharge process - averaging 8 h compared to 4 h for those paying upfront. These delays not only reduce patient satisfaction but also lead to inefficient resource utilization and increased indirect costs for hospitals and insurers. This study investigates the discharge process for cashless patients in a multi-specialty hospital using a real case-based discrete-event simulation model. The model captures process inefficiencies and evaluates the impact of resource augmentation based on hospital policies. Three simulation scenarios are tested: baseline, delay reduction and resource enhancement. Results reveal that targeted interventions can reduce discharge time by up to 38.6%, significantly increasing the number of patients discharged on the same day. The study offers a quantitative framework for hospital administrators to optimize discharge operations, enhance patient experience and reduce indirect costs associated with prolonged bed occupancy. It also contributes to healthcare operations literature by modeling payer-provider interactions within third-party administrator insurance systems, an area where data-driven research remains limited in the context of emerging healthcare markets.
Encroachment and product diversification are two effective strategies that manufacturers can leverage to gain a competitive edge and increase revenue. This study focuses on a manufacturer producing two quality-differentiated products, exploring various strategies for market encroachment. In addition to conventional retail channel, the manufacturer can pursue encroachment by directly selling one or both products through its own channel or the agency channel provided by the platform retailer. We show that in the case of direct channel encroachment, only when the selling cost is very low, the manufacturer will choose to encroach through both products. As the selling cost rises, encroaching with one single product becomes the preferred strategy, and at a high selling cost, the manufacturer will abandon encroachment. There is a similar conclusion when the manufacturer encroaches with agency channel, and the choice of the manufacturer depends on the commission rate. Moreover, only the single product through the agency channel encroachment strategy can lead to a win-win outcome for the manufacturer and the retailer under certain conditions. Furthermore, we identify a critical selling cost threshold that dictates the manufacturer's approach to encroachment. These findings provide insights into how manufacturers can navigate encroachment dynamics in markets with differentiated products.
Technological innovation and a low-carbon economy are of great significance for the high-quality development of China's industrial sector. The article aims to assess the overall efficiency of Chinese technology innovation and carbon emissions. A comprehensive theoretical framework is developed to evaluate this relationship, employing a novel two-stage DEA approach that integrates non-cooperative and cooperative models, considering undesirable outputs and Variable Return to Scale (VRS) assumptions. The methodology is applied to assess 29 provinces in China. The results indicate a misalignment between technology innovation, energy consumption and carbon emission trajectories, revealing overall inefficiency and posing challenges to achieving net zero emissions. This research underscores China's commitment to prioritize technological and energy advancements for CO2 reduction, particularly in light of its renewed dedication at COP28, emphasizing the critical role of innovation in bridging the emission gap and meeting ambitious climate goals. Additionally, the outcome of this study provides clear policy implications for enhancing the comprehensive efficiencies of both the technology innovation process and carbon emissions process in China.
We use a bi-level framework to address how transmission planning may adapt to the cost of damage from emissions and imperfect competition in both greenfield and brownfield settings. Our upper level comprises a welfare-maximising transmission planner who internalises the damage cost in anticipation of generation operations at the lower level by profit-maximising firms. In a greenfield setting, i.e. without any existing capacity, complete CO2 taxation on firms is optimal under perfect competition, and the CO2 tax monotonically increases transmission capacity. By contrast, Cournot behaviour by industry chokes off consumption, thereby limiting the damage from CO2 emissions and rendering partial CO2 taxation optimal. In a brownfield Western European setting with perfect competition, only complete internalisation of the cost of damage from emissions provides the price signal to reduce consumption and to spur generation investment as in the greenfield setting. However, the brownfield setting with Cournot behaviour leads to fossil-fuel expansion by price-taking fringe firms, which exacerbates damage from CO2 emissions vis-& agrave;-vis the greenfield setting. Hence, it is optimal to fully internalise the cost of damage from emissions with complementary expansion of the transmission network in a brownfield setting even with market power in contrast to the result in a greenfield setting.
This study proposes a novel hybrid framework that combines data envelopment analysis (DEA) with multi-target machine learning (MTML) to evaluate and predict the performance of new decision-making units (DMUs) in large-scale datasets. Although prior research has integrated DEA with ML, most studies have been confined to predicting a single efficiency score and have not provided benchmarking information such as reference sets and improvement targets which is central to DEA's practical value. These limitations reduce the applicability of such models for realistic decision support, because inefficient DMUs require not only efficiency scores but also feasible and balanced improvement targets. Moreover, conventional approaches often necessitate rerunning the DEA model entirely whenever new DMUs appear, imposing substantial computational costs in the big-data era. To address these gaps, this study introduces a multi-target prediction approach capable of simultaneously estimating CRS and VRS efficiency scores, reference sets and return-to-scale (RTS) information for new DMUs. We examine three multi-target modeling strategies: (1) a DEA-coupled multi-target to single-target transformation, (2) direct multi-target modeling (DMM) and (3) multi-task learning (MTL). By expanding DEA-based prediction from a single indicator to multiple outputs, the study offers both theoretical contributions and practical value for benchmarking applications.
In present-day workplaces, open workspaces are becoming more and more popular. Inevitably, employees spend a substantial portion of their lives at work. In such an environment, the proper layout of employees with varying dispositions is a regular challenge. This study identifies the location's ranking based on employee preferences using the best-worst method (BWM) and the weighted aggregated sum product assessment (WASPAS) method. This article presents a novel mixed-integer non-linear programming (MINLP) model that incorporates employees' interaction rates and preferences to reach an optimal layout. Employees with task conflicts are included in the presented model as a principal assumption. The proposed methodology is examined using an open workspace as a real-life case study. Also, several sensitivity analyses are performed to provide beneficial managerial insights.
This paper examines the use of in-store customers as delivery couriers in a centralized crowd-shipping system, targeting the growing need for efficient last-mile delivery in urban areas. We consider a brick-and-mortar retail setting where shoppers are offered compensation to deliver time-sensitive online orders. To manage this process, we propose a Markov Decision Process (MDP) model that captures key uncertainties, including the stochastic arrival of orders and crowd-shippers, and the probabilistic acceptance of delivery offers. Our solution approach integrates Neural Approximate Dynamic Programming (NeurADP) for adaptive order-to-shopper assignment with a Deep Double Q-Network (DDQN) for dynamic pricing. This joint optimization strategy enables multi-drop routing and accounts for offer acceptance uncertainty, aligning more closely with real-world operations. Experimental results demonstrate that the integrated NeurADP + DDQN policy achieves notable improvements in delivery cost efficiency, with up to 6.7% savings over NeurADP with fixed pricing and approximately 18% over myopic baselines. We also show that allowing flexible delivery delays and enabling multi-destination routing further reduces operational costs by 8% and 17%, respectively. These findings underscore the advantages of dynamic, forward-looking policies in crowd-shipping systems and offer practical guidance for urban logistics operators.
An EV aggregator managing multiple EV garages is a distributed EV storage system that can play a significant role in the electricity market of the area where it sits. We analyze the ability of such an EV storage system to exercise market power. For this, we use two models. The first one represents the EV storage system as a price maker, and the second one allows the system operator to freely operate the EV storage system for the best of the power system as a whole. Considering analytical outcomes from a stylized model and numerical simulations from a realistic case study, we found that the EV storage system has limited ability to exercise market power. Thus, the outcomes of the price-maker model and those of the model in which the system operator is in control of the EV storage system are generally similar. This is so for a number of reasons, namely, (i) the EV storage system needs to charge prior to producing (discharging), (ii) its energy/power capacity is comparatively small, (iii) it is generally located at a single node of the power system and (iv) the step-wise supply curve of most power systems is increasingly flat.
