The burgeoning discourse surrounding digital twins (DTs) has reached a critical point that requires the careful examination of the disparity between their conceptual idealisation and empirical reality. While the literature frequently presents the DT as a transformative force for Industry 4.0 and 5.0, analysis of current implementations reveals that a mere 16% of analysed systems achieve the closed-loop, bidirectional capabilities required of a true digital twin, with the majority remaining confined to the status of unidirectional 'digital shadows'. Consequently, future research must shift from definitional debates toward the development of rigorous empirical assessment frameworks that utilise measurable indicators—such as feedback closure, real-time reciprocity, and post-adjustment monitoring—to differentiate functionally autonomous systems from their descriptive counterparts.
Production of multiple products sequentially is widely practiced in industries. Their unpredictable demands and lead times generate inefficient supply chain flow in meeting demands. Besides, production-based business activities emit carbon. However, researchers developed conventional inventory models with deterministic/stochastic demand, ignoring environmental impact. Here we present a model assuming stochastic demands of products processed sequentially by incorporating the costs of operations and multi-source carbon emissions. A lot of a product is assumed to be multiple of its demand plus approximated safety stock. Usually, larger production in lesser setups minimizes setup and processing costs imposing higher holding and carbon emission costs (due to warehousing and temperature control), and vice versa. Consolidated shipments with fuel-efficient slower transportation decrease transportation cost and carbon emission, requiring higher lot sizes and safety stocks, increasing inventory-related carbon emission. This operations-emission cost trade-off encompasses balancing these conflicting objectives. We minimize this total cost via calculus method of differentiation, synchronizing the production flow of satisfying demand of a product during production of other products. The model is illustrated with an example problem and applied to a local Coca-Cola production plant of bottling five types of this beverage. Minimum total carbon emission cost is found to be about 30-32% of the minimum total cost for the former, and 12-13% for the latter. When standard deviation of demands and lead times are jointly increased by 30%, the minimum total cost increases by 5.44%. Finally, results of sensitivity analyses on various parameters and managerial implication of the optimal solution policy are highlighted.
In recent years, fine-tuned Large Language Models (LLMs) have become increasingly important in manufacturers’ improvement of production processes. Because of the heavy fine-tuning cost burden, firms within a supply chain have been witnessed to engage in fine-tuning collaboration, sharing efforts and results to reduce costs. Such collaboration, however, raises concerns about free-riding problems, which can be very typical in a co-opetitive supply chain. Therefore, we examine a manufacturing system including a supplier with self-brand and a manufacturer who both deploy their LLMs in production and make efforts on fine-tuning. We reveal that a higher intensity of fine-tuning collaboration creates a “suppression effect” which hampers stakeholders’ fine-tuning efforts out of the free-riding problems, and simultaneously a “synergy effect” which indicates that the overall market potential can be expanded by fine-tuning effort spillover, motivating them to invest more in fine-tuning. Driven by the dueling effects, the benefits of both the supplier and the manufacturer from fine-tuning collaboration exhibit non-monotonic relationships with respect to the collaboration intensity and the base product quality of the supplier. We further show that fine-tuning collaboration benefits the supplier more than the manufacturer in the co-opetitive relationship. Their incentives for collaboration are aligned if both the base product quality of the supplier and the collaboration intensity are relatively high.
Sustainable purchasing planning in inbond logistics integrates ESG (environmental, social, and governance) criteria, establishing it as a distinct function within the supply chain. This study addresses four interrelated research questions: how purchasing quantities can be linked to suppliers’ and transportation modes’ social performance in tactical planning; how order allocation can prioritize low-emission transportation to advance the environmental and governance pillars of ESG; what integrated decision-making approach can simultaneously maximize sustainability performance, minimize cost, and mitigate sustainability risk; and through what mechanisms higher supplier social performance reduces total carbon emissions, and how the social-environmental-economic dimensions can be optimized under increasing carbon costs. To answer these questions, we develop a multi-period interval multi-objective programming model that accounts for demand uncertainty and non-linearity. The model explicitly links order allocation to suppliers’ social performance scores, encouraging social sustainability competition, and incorporates carbon trading and caps to manage carbon risks. Key findings show that higher carbon prices are internalized into procurement budgets; social sourcing and emissions reduction are interdependent-emphasizing social criteria can lower both cost and emissions by restricting volumes; and the emphasis on social performance is bounded by shortage constraints, providing clear thresholds for setting sustainability policies without compromising service levels. The model’s applicability is demonstrated through a numerical example. These findings highlight the critical role of suppliers’ ESG performance in managing demand and sustainability under uncertainty.
Conventional Supply Chain Master Planning (SCMP) approaches are poorly suited to environments characterized by simultaneous supply, demand, and sales-channel disruptions particularly in dual-channel supply chains where asymmetric shifts between online and offline demand can create inventory misallocation, service deterioration, and cost escalation. Although Supply Chain Digital Twins (SCDTs) are increasingly discussed as enablers of disruption-aware planning, limited evidence exists on how SCDT frameworks support adaptive SCMP in multi-echelon dual-channel settings. This study develops a SCDT framework for adaptive SCMP that integrates demand forecasting, machine learning-based disruption detection, discrete-event simulation, and multi-objective optimization to dynamically revise sourcing, production, and inventory decisions in response to evolving system states. The framework is instantiated in a television manufacturing supply chain and evaluated against static and rolling-horizon planning approaches under supplier, demand, and channel-shift disruptions. Results show that the proposed SCDT improves service-cost performance, supports more balanced channel-level fulfillment, and enhances adaptive recovery behavior under disruption relative to benchmark approaches. The findings further show that channel-aware SCMP is especially important when demand shifts asymmetrically across sales channels.
Amid the growing emphasis on corporate environmental responsibility, some e-commerce platforms provide financial support to help manufacturers invest in green product development. This study examines cooperative green innovation through cost-sharing within a platform supply chain considering both selling mode choices and asymmetric market information. A game-theoretic framework is employed to analyze mechanisms influencing manufacturers’ green investment decisions, the platform’s optimal cost-sharing strategy, and choices of selling modes and information-sharing strategies. First, our analysis demonstrates that platform-enabled information sharing facilitates manufacturers in aligning greenness with realized demand, thereby amplifying the value of information disclosure. Contrary to conventional wisdom, by strategically augmenting its green cost-sharing proportion, the platform may have an incentive for voluntary information disclosure under the reselling mode. This incentive intensifies as consumer green awareness rises. Second, with low greenness competition intensity, the platform’s willingness to undertake green cost-sharing increases with price competition intensity, and agency mode is more advantageous for sharing green costs than reselling when the commission rate is high. Third, higher green cost-sharing proportions and greater consumer green awareness enhance the platform’s incentive to adopt the agency mode. Specifically, when consumer green awareness is high and the platform bears a substantial share of green investment costs, the agency mode with information sharing leads to a win-win outcome under weak price competition; in contrast, the reselling mode with information sharing is mutually beneficial under intense price competition. This study provides significant managerial insights into collaborative green research and development operations within platform supply chains.
The increasing prevalence of supply chain digitalization has widened the digital divide between firms and their partners, posing a significant collaboration challenge. Previous studies predominantly emphasize the negative effect of digitalization asymmetry on focal firms, with limited attention to its effect on dyadic performance. This study investigates how digitalization asymmetry affects buyer-supplier relationship (BSR) performance, with particular attention to the mediating role of joint learning and the moderating effects of trust and monitoring as two distinct governance mechanisms. Using a sample of 170 matched buyer-supplier dyads from China, this study tests the proposed model through regression analyses. The results reveal the double-edged role of digitalization asymmetry in BSRs, demonstrating an inverted U-shaped relationship between digitalization asymmetry and dyadic performance. The instantaneous indirect-effect analysis indicates a significant negative indirect effect through joint learning at a relatively high level of digitalization asymmetry. Furthermore, trust strengthens, whereas monitoring weakens, the effect of digitalization asymmetry. These findings provide theoretical insights for supply chain literature by advancing our understanding of the nonlinear dynamics of digitalization asymmetry in BSRs. For practitioners, this study provides practical insights on how to engage in effective collaboration with partners at varying levels of digitalization and implement governance strategies to address the digitalization asymmetry dilemma.
Delays in relief food delivery critically threaten post-disaster survival, yet existing studies lack a systematic approach to identifying and assessing these risk factors across different disaster contexts. Focusing on floods and storms, this study develops a two-stage risk analysis framework tailored to relief food delivery delay risks across disaster contexts. Using multi-source real-world data and grounded theory, it identifies 24 risk factors, including previously overlooked institutional political risks such as geopolitical conflicts, political considerations, and bureaucratic approval redundancy. Methodologically, this study integrates failure mode and effect analysis (FMEA), a belief-rule-based Bayesian network, and evidential reasoning to assess and prioritise risks, while employing spherical fuzzy set-based social network analysis and the best-worst method to determine expert and risk factor weights. The results show that the key risk factors differ between disaster types: bureaucratic approval redundancy, traffic paralysis, and concentrated warehouse distribution dominate in floods; severe weather, bureaucratic approval redundancy, and communication infrastructure breakdown prevail in storms. This study contributes to delay risk assessment by applying an integrated FMEA–BBN–ER framework to the underexplored context of relief food delivery under floods and storms, while refining its weighting architecture through combined background–SFS–SNA expert weighting and BWM-based risk factor weighting to account for expert and risk factor heterogeneity. It further provides disaster-specific insights by cautiously interpreting how contextual factors, including governance capacity, institutional settings, and cultural value orientations, may help explain these differences and inform targeted mitigation strategies.
As manufacturers bundle products, services, and software into customer solutions, they need to rethink both customer and dealer relationships. Existing literature mainly explores changes in manufacturer-customer ties. Less is known about how manufacturers that use dealers as intermediaries for customer solutions must adapt their relationships with dealers. This article draws on an abductive qualitative study comprising 47 interviews with manufacturers, dealers, and customers across six industries, complemented by an embedded process case. The research examines how greater solution orientation reshapes manufacturer–dealer relationships. The findings reveal three relationship modes: (1) manufacturer-led enablement, (2) joint coordination, and (3) strategic co-creation. These modes differ in governance, coordination, decision-making, and power. The study extends the relational-process view of customer solutions to dealer-mediated channels. We develop the relationship mode portfolio concept and show that advanced servitization does not necessarily eliminate dealers; instead, dealers assume different roles depending on solution requirements.
Textile SMEs are under pressure to adopt technologies that improve their operational efficiency, production visibility, resource allocation, and competitiveness, while also meeting demands for sustainability and responsible transformation. However, their financial constraints and lack of organizational readiness make this process difficult. This study proposes a conceptual framework to guide responsible technological transformation in textile SMEs, complemented with a decision-support maturity model (MM-RET3-SME) that links technology adoption to production efficiency, cost-related constraints, and sustainability-oriented performance. The framework details 20 internal and external driving forces classified into five categories, relevant Industry 4.0 technologies and defines 15 critical requirements in six categories. These elements are associated with operational readiness, cost-related constraints, and the conditions required for economically viable and responsible technological adoption. In addition, the proposed maturity model (Level 0–5) works as a decision-support tool for prioritizing investment actions according to maturity gaps. The research combines literature reviews, interviews with experts and a fuzzy analysis that validates the model and weights the criteria. The results show that: (i) the overall maturity level of the sector is at Level 2; (ii) enterprise size is a key factor, as the results place companies at different levels depending on their size: micro-enterprises (Level 1), small (Level 2) and medium-sized enterprises (Level 3); and (iii) the main gaps are in infrastructure and technology, data management, security and regulations, and costs and viability. From a production economics perspective, the model supports SMEs in linking technological readiness with investment prioritization, cost-related constraints, resource allocation, and operational performance under sustainability requirements.
With the rapid diffusion of electric vehicles, battery-swapping has emerged as a replenishment alternative that converts time savings into potential market adoption. Battery-swapping represents a product-service system, whose diffusion hinges on complementary innovations between product-side compatibility and service-side capacity. As battery-swapping networks attract investment from battery suppliers, vehicle manufacturers, and third-party firms, differences in capital commitment and control over service assets bring service leadership structures to the forefront of strategic analysis. Accordingly, we investigate product-service innovation and pricing decisions under three leadership regimes, and examine equity-based coordination through cross-shareholding. To capture demand heterogeneity arising from time-based competition between charging and swapping, we develop a circular Hotelling model of a swapping supply chain comprising a battery supplier, a manufacturer, and a service provider. Our results highlight three key findings. First, higher users time value generally expands swapping adoption under decentralized leadership, but may reduce adoption when the manufacturer self-operates service, unless coordination efficiency is sufficiently high. Second, leadership advantages are stage contingent. When time advantages and differentiation are strong, regimes that better internalize cross-stage gains achieve higher innovation levels, lower effective service prices, and greater penetration; as fast charging technologies converge and substitution intensifies, performance becomes increasingly sensitive to cost efficiency and asset turnover. Third, cross-shareholding can mitigate innovation spillovers and improve both supply-chain performance and consumer welfare, but its incentive effects depend on leadership structure: under third-party leadership, innovation is inverted-U in equity intensity, whereas under supplier leadership, stronger equity alignment more consistently promotes innovation, particularly in time-sensitive markets.
Accurate assessment of human worker skills is essential for human-centric automation in modern manufacturing systems. However, modelling these skills remains challenging due to variability in human performance and limited high-quality open-access datasets. This study proposes an approach for evaluating individual worker skill levels in assembly systems and assessing the reliability of internal company metrics. It introduces worker-specific learning curves that capture individual skill progression through an exponential model that incorporates three parameters: , representing worker-specific learning rate, , quantifying previous experience and, , correcting inaccuracies in the NMV (Nominal Measured Value) calculations or pre-training effects. An algorithm based on the trust-region method is used to fit the model using task execution data, effectively capturing individualized learning trajectories and distinguishing skill progression across workers.The proposed approach was validated using a real dataset collected from an assembly line over two weeks, where four workers performed tasks at both manual and collaborative workstations. As this dataset can be seen insufficient, a Moving Block Bootstrap (MBB) stability analysis was conducted to assess the robustness of the fitted learning curve parameters under resampling. The MBB results show partial stability, with some parameters remaining consistent under resampling while others vary more substantially. Comparison with company-reported skill levels reveals notable discrepancies, highlighting the need for improved assessment methodologies. Furthermore, the proposed approach was benchmarked against classical learning curve models. The dataset is released as open-access, enabling transparent and reproducible research and fostering advancements in skill modelling.
Digital subscription platforms such as Netflix and iQiyi frequently employ dynamic pricing, and users naturally perceive fairness concerns in response to such price changes. These fairness concerns significantly influence the equilibrium pricing structures of two-sided subscription platforms. This paper develops a two-period dynamic pricing game model to study how buyers' fairness concerns alter the platforms' pricing structure across buyers and sellers. We show that fairness concerns demonstrate flexibility in adjusting pricing strategies in the intertemporal game of two-sided markets. Different from previous studies, the side with stronger network effects will be subsidized rather than subject to more aggressive pricing decisions when fairness concerns are low. Furthermore, the platform's pricing strategy is jointly determined by retention rates and fairness concerns. Specifically, for buyers, the platforms adopt the ‘penetration-harvesting’ strategy when both the retention rate and fairness concerns are weak. However, if the retention rate is high, the ‘lock-in effect’ will be utilized to increase the price. For sellers, the platforms adopt an incremental pricing mechanism. We also find that the buyers' fairness concerns can lead to asymmetric pricing adoption, which alters the traditional seesaw effect (where price increases on one side correspond to decreases on the other). Additionally, the unit transportation cost of two-sided users and fairness concerns jointly determine the strategy selection of the platforms. Essentially, in two-sided markets with strong network externalities, moderate fairness concerns can enhance social welfare.
As consumer demand diversifies, consumer-to-manufacturer (C2M) customization has emerged as a new production paradigm. It leverages data collection and analysis of consumer preferences to inform the tailored product design and manufacturing. This paper investigates a platform-based C2M customization mode in which the platform can choose whether to share information to empower the manufacturer’s C2M customization production. By introducing a two-dimensional product matching framework that decomposes consumers’ matching preferences into functional matching and personalized matching, we examine the interplay between the manufacturer’s C2M customization and the platform’s information sharing strategy and investigate how the interaction impacts the supply chain members’ decisions and profits. We obtain several main results. Firstly, the platform’s information sharing does not consistently incentivize the manufacturer to invest in C2M customization, and may not always yield mutual benefits for both parties. Secondly, when the manufacturer’s C2M investment efficiency is low, the platform is more inclined to share consumer preference information with the manufacturer, thereby facilitating the manufacturer’s C2M customization investments at equilibrium. Conversely, the equilibrium strategy compels the manufacturer to collect consumer information to promote C2M customization, as the platform withholds information. Lastly, a win–win outcome occurs when the manufacturer’s C2M customization investment efficiency is not excessively high. We extend the model in several ways to verify robustness. Our findings provide managerial insights for manufacturers and platforms considering engagement in C2M customization.
Supply chain complexity reshapes firms’ innovation choices, yet its implications for innovation ambidexterity remain unclear. This study develops a biform game model to examine how complexity influences the balance between exploration and exploitation within and across firms. We show that complexity has a dual effect by expanding innovation opportunities while increasing coordination frictions. This trade-off generates threshold conditions under which within-firm ambidexterity becomes less sustainable and specialization becomes more attractive. Even so, the weakening of internal ambidexterity does not necessarily eliminate ambidexterity at the system level. Under heterogeneous participation conditions, role differentiation across firms can preserve an ambidextrous supply chain structure. We further show that coordination regimes, including standards, platforms, and integration investments, can mitigate frictions and shift these thresholds. Numerical analysis illustrates these patterns over a broader parameter range. The study extends research on innovation ambidexterity to complex interorganizational settings and offers implications for collaborative supply chain design.
Increasing stakeholder pressure to hold firms accountable for the sustainability performance of their supply networks creates a need to select sustainable suppliers and minimize associated risks. Firms increasingly use third-party sustainability performance information to screen and assess potential and existing suppliers. However, such third-party assessments are structurally limited and often unavailable for some suppliers, leaving firms without external sustainability information for them. Supply network-level characteristics are an overlooked factor for predicting supplier sustainability performance. Using a large-scale dataset of 792 firms embedded in 291 supply networks up to tier three, we develop a predictive model to estimate the sustainability performance based on firm- and supply network-level characteristics. Our analysis indicates that including a set of supply network-level characteristics significantly improves predictive performance, reducing the prediction error by 8.83%. We further show that considering a firm’s embeddedness within the supply network significantly reduces the prediction error, while surprisingly, its control within the supply network has no significant influence. Our study adds to the literature on supply networks, sustainability, and sustainability performance prediction by offering new insights regarding the predictive power of supply network-level characteristics and the relative importance of firm-level characteristics. We also provide managerial implications for supplier selection, monitoring, and risk management in supply networks.
Transparency is frequently advocated in supply chains because better information can reduce coordination failures and mitigate bargaining inefficiencies. Yet in power-imbalanced buyer–supplier relationships, disclosure can change what becomes salient in negotiation, shifting attention toward rents. We study whether profitability transparency can harm relationship quality and continuation when the upstream party has a strong outside option. In a pre-registered, incentivized online retailer–supplier bargaining experiment with stochastic demand and inventory risk (N = 200), retailers negotiate a wholesale price and order quantity against a supplier with an advantaged disagreement payoff. Retailers either observe the supplier's net unit cost (complete information) or receive only incomplete cost information that prevents precise inference and yields higher expected cost estimates (i.e., lower expected supplier profitability). As predicted, the economic and relational effects diverge: relative to complete information, incomplete information lowers the retailer's expected payoff implied by the negotiated agreement by 23.5%, yet significantly increases perceived fairness, satisfaction, and an incentivized measure of willingness to negotiate again. Conditioning on the retailer's payoff further amplifies the incomplete-information advantage in all three relational outcomes, consistent with a rent-salience channel in which evaluations depend on both own outcomes and inferred counterpart rents. The results tighten a boundary condition for transparency: when leverage is asymmetric, making upstream profitability transparent can carry relational costs that are operationally consequential for continuation. To preserve relationship quality, transparency initiatives may therefore require complements, such as gain-sharing rules or limits on how disclosed cost information is used.
Modern manufacturing systems face increasing pressure to remain cost-efficient, responsive, and resilient amid operational uncertainty and disruptive events. While traditional scheduling methods offer limited adaptability, existing AI-based solutions often overlook robustness and risk sensitivity in complex production settings. This study addresses this gap by developing a multimodal deep reinforcement learning (MDRL) framework that learns data-driven scheduling policies under routine stochastic variability and evaluates them through distributionally robust and tail-risk-based disruption stress tests. The proposed approach leverages a transformer-based algorithm enhanced with transformer encoders and cross-modal attention to processing diverse data sources. Robustness is assessed through scenario-based disruption stress testing and CVaR/DRO-based robustness scoring for post-training checkpoint selection. Experimental results reveal that the MDRL agent consistently favors mid-sized lot sizes (peaking around 58 units) and reduces setup activations by approximately 30%, promoting stable and efficient operations. Analyses for robustness show that while increasing the CVaR threshold from 0.5 to 0.99 raises worst-case cost by over 50%, it leads to a 26% drop in service level, highlighting critical trade-offs between risk aversion and responsiveness. These findings highlight the potential of MDRL to optimize operational performance while supporting systematic robustness evaluation in real-time production environments. The framework offers a scalable and intelligent decision-making tool for next-generation smart factories, with implications for operational efficiency, sustainability, and supply chain resilience.
