Remanufacturing Process Under Uncertainty: Review, Challenges, and Future Directions

Abstract

In the context of the global transition toward carbon neutrality and the circular economy, remanufacturing has emerged as a vital strategy for enhancing resource efficiency and reducing environmental impact. However, the remanufacturing sector faces significant uncertainties—including fluctuations in market demand, variability in the quality of returned products, and dynamic policy changes. These factors collectively challenge production decision-making and system sustainability. Following the preferred peporting items for systematic reviews and meta-analyses (PRISMA) guidelines, this study conducted a systematic review and bibliometric analysis of 98 core articles published between 2015 and 2024, with a focused examination of three interdisciplinary themes: (1) decision-making and optimization under uncertainty, (2) supply chain coordination and policy mechanisms, and (3) digital transformation and the application of emerging technologies. A novel micro–meso–macro analytical framework is proposed to integrate fragmented findings. The results highlight a paradigm shift from static models to dynamic, real-time decision-making systems, facilitated by digital twins (DTs), blockchain, and intelligent algorithms. Furthermore, the study identifies the synergistic effects of carbon-financial instruments and policy incentives in aligning economic and environmental objectives. This research develops a systematic framework to understand and address uncertainties in remanufacturing, offering policymakers and industry practitioners actionable insights to enhance the resilience, sustainability, and global applicability of remanufacturing systems.

1. Introduction

1.1. Research Background

Global warming and the depletion of natural resources have elevated the circular economy to a vital strategy for sustainable development. Within this framework, remanufacturing stands out as a key practice. It offers substantial potential to reduce resource consumption and carbon emissions by refurbishing and revalorizing waste products [1]. Compared with traditional manufacturing processes, remanufacturing can achieve significant reductions in costs, energy use, raw material usage, and waste gas emissions [2], while also lowering the global warming potential [3]. These benefits position remanufacturing as a strategic solution to the complex interplay between economic growth, resource consumption, and environmental constraints.

However, the large-scale implementation of remanufacturing is hindered by multi-dimensional uncertainties. These include fluctuations in market demand, variability in the quality of recycled materials, policy changes, and technological risks [4]. Such uncertainties permeate every stage of the product life cycle and disrupt collaborative decision-making among supply chain partners [5]. This often leads to inherent contradictions, such as balancing technical feasibility with economic sustainability and weighing environmental benefits against operational efficiency. Therefore, there is an urgent need for more comprehensive and dynamic resilience evaluation frameworks that can better capture the resilience of remanufacturing supply chains under various uncertainties.

1.2. Research Gaps and Research Questions

Despite significant progress in this field, key research gaps remain:

(1)

Current research on uncertainty and decision-making in remanufacturing production is fragmented, and a comprehensive theoretical framework that can integrate the entire process and multi-dimensional dynamic decision-making has not yet been established.

(2)

Most existing studies focus on single-dimensional optimization or static game assumptions, failing to fully reflect the dynamic evolution characteristics of uncertainty and neglecting the complex coupling mechanisms among technological, economic, and environmental factors. There is also insufficient attention paid to dynamic and real-time decision-making.

(3)

The existing literature pays insufficient attention to policy-related challenges and the implementation barriers faced by small and medium-sized enterprises (SMEs). This gap severely limits the feasibility and effectiveness of translating theoretical models into practice.

This study addresses the following research questions:

(1)

How has remanufacturing process (RP) decision-making under uncertainty evolved in recent years?

(2)

What are the roles of digital technologies (e.g., DTs, blockchain) and carbon-financial instruments in shaping these decisions?

(3)

How can a micro–meso–macro framework synthesize fragmented findings into a coherent theoretical system?

1.3. Research Contributions

This study offers significant contributions from both theoretical and practical perspectives:

(1)

This study proposes an innovative three-tier analysis framework of micro, meso and macro levels, which for the first time provides a systematic integrated theoretical perspective for the research on uncertainty in remanufacturing, reveals the interactions among factors at different levels, offers an unprecedented holistic perspective for understanding and analyzing the complex uncertainty in the remanufacturing system, and lays a solid foundation for future theoretical development.

(2)

It demonstrates the synergistic effects between digital technologies (e.g., DTs, blockchain) and financial tools (e.g., carbon options, quota collateralization), offering a technical-economic integration roadmap for industrial practice. It guides enterprises on how to combine and utilize these tools to effectively enhance dynamic response capabilities, strengthen system resilience, and achieve a win-win situation of economic and environmental benefits, providing an operational path for the digital transformation and green upgrade of remanufacturing.

(3)

It identifies implementable pathways for SMEs and diverse regions, enhancing the global applicability and practical inclusivity of the findings, and connects academic research with broader socioeconomic contexts. It provides a basis for policymakers to design more differentiated and dynamic incentive policies, and also offers directions for industry practitioners to evaluate the applicability of technologies and develop low-cost, modular solutions, ensuring the feasibility and effectiveness of remanufacturing theory on a global scale.

1.4. Paper Structure

The paper is organized in a strict introduction–methods–results–discussion (IMRaD) format. Section 2 details the PRISMA-compliant search and screening protocol that yielded our final corpus of 98 articles. Section 3 provides a thematic analysis of remanufacturing under uncertainty across three research streams. Section 4 synthesizes the micro-, meso-, and macro-level findings, while Section 5 discusses findings, limitations, and future research directions, and Section 6 concludes by summarizing key contributions and answering the research questions.

2. Methodology

This study employed a systematic literature review (SLR) methodology following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to ensure a comprehensive, transparent, and reproducible process for identifying and selecting relevant publications.

2.1. Search Strategy and Study Selection

2.1.1. Data Sources and Search String

Five major academic databases were searched: Web of Science (WoS), Scopus, ScienceDirect, IEEE Xplore, and SpringerLink, selected for their coverage of engineering, environmental science, operations management, and sustainability.

The search string was constructed using Boolean operators to integrate three key conceptual domains pertinent to the research objective: (1) the core context of remanufacturing or closed-loop supply chains (CLSCs), (2) the central challenge of uncertainty, and (3) the focus on decision-making and optimization. The specific string used was: ((“remanufacturing”) AND (“uncertainty”) AND (“scheduling” OR “supply chain coordination” OR “decision-making”)).

The search was limited to peer-reviewed journal articles published in English between January 2015 and June 2024. This decade-long timeframe was chosen to capture the most recent and relevant advancements in the field, a period characterized by the rapid evolution of Industry 4.0 technologies (e.g., DTs, IoT), the global ascendancy of carbon neutrality policies, and a significant shift in research focus from static to dynamic modeling of remanufacturing systems.

The study adhered to strict inclusion and exclusion criteria to ensure the relevance and quality of the final corpus. These criteria are summarized in

Table 1. Literature inclusion and exclusion criteria.

Category	Inclusion Criteria	Exclusion Criteria
Publication Type	Peer-reviewed journal articles in English	Conference papers, books, book chapters, non-English publications
Indexing	Indexed in SCI/SSCI	Non-SCI/SSCI indexed publications
Content Focus	Mathematical modeling, optimization, decision-making under uncertainty in remanufacturing	Qualitative studies, case studies without quantitative models, non-remanufacturing contexts
Methodology	Game theory, stochastic programming, robust optimization, simulation, empirical models	Studies lacking a quantitative decision-making framework

.

2.1.2. Selection Process

The literature selection process followed the PRISMA flow diagram depicted in Figure 1. The initial database search yielded 1301 records. After removing 317 duplicates, the titles and abstracts of 984 unique articles were screened for relevance. This step excluded 729 articles that did not meet the broad thematic focus. The full texts of the remaining 255 articles were then rigorously assessed against the inclusion criteria. A further 157 articles were excluded at this stage, primarily for lacking a quantitative decision-making framework or not sufficiently focusing on uncertainty. This process resulted in a final corpus of 98 core articles for in-depth review and analysis.

2.2. Bibliometric Visualization

2.2.1. Annual Publication Output

The annual publication trend (Figure 2) reflects the evolving research intensity in remanufacturing under uncertainty. Starting with 22 publications in 2015, the field saw a sharp rise to 46 in 2019, indicating growing interest in foundational models and static optimization. A slight dip in 2020 (41 papers) likely reflects COVID-19 disruptions. The surge to 61 papers in 2021 and a peak of 70 in 2022 align with global carbon-neutrality initiatives and interdisciplinary approaches. The slight decline in 2023 (54 papers) suggests a maturation phase, with research shifting from quantity to quality—evidenced by increased integration of DTs and blockchain. The sustained output in 2024 (59 papers already) confirms the field’s dynamic evolution toward real-time optimization and systemic collaboration. The trend underscores a transition from theoretical modeling to technology-integrated, dynamic decision-making under uncertainty.

2.2.2. Author Co-Occurrence Network

The author co-occurrence map, presented in Figure 3, reveals the collaborative landscape among researchers. Node size reflects publication count, and edge density indicates collaboration frequency. The network, generated with VOSviewer (1.6.20), shows a red cluster centered on “Tian, Guangdong” and “Pham, Duc Truong,” denoting prolific output and close cooperation in dynamic remanufacturing collaboration and pricing strategies. A green cluster (Huang, Jun et al.) focuses on carbon-finance instruments and supply chain coordination, whereas a blue cluster (Liu, Conghu et al.) is devoted to uncertainty-driven scheduling and optimization algorithms. The purple cluster (Jiang, Zhigang et al.) concentrates on quality uncertainty and quality-assessment models, and the yellow cluster (Guo, Xiwang et al.) explores macro-level policy incentives on a smaller scale. The extensive cross-cluster findings indicate that there is a strong trend of solid research output and teamwork among Chinese scholars, forging an academic community bound by dynamic game theory, intelligent algorithms, and policy instruments. This network highlights the growing trend of collaborative research, emphasizing the importance of interdisciplinary approaches in advancing the field.

2.2.3. Institutional Collaboration Network

Figure 4 illustrates institutional collaborations, dominated by Chinese universities such as Huazhong University of Science and Technology, Hefei University of Technology, and Wuhan University of Science and Technology. These institutions form hubs focusing on DTs, hybrid algorithms, carbon policy, and regional adaptability. The high density of connections among Chinese universities indicates robust domestic collaboration. International ties with institutions like University of Tehran and University of Québec introduce diverse methodologies but remain secondary. This pattern suggests a regionally concentrated yet globally connected research ecosystem, with China leading in both output and collaborative networks. The institutional map reinforces China’s leadership in remanufacturing research, with potential for greater integration of Global South perspectives.

2.2.4. Country Collaboration Network

China is the central node in the country collaboration network (Figure 5), with strong ties to the USA, UK, Canada, and India. This highlights China’s influential role in international research and development. The triangular collaboration among China, the USA, and European countries has accelerated methodologies like multi-source data fusion and lifecycle assessment. Secondary clusters involving Australia, Iran, and the Netherlands contribute to algorithm design and supply chain resilience. While developing countries like Brazil and Saudi Arabia show lower engagement, their emerging participation signals growing global interest and potential for region-specific models. The network underscores the importance of international collaboration in addressing global remanufacturing challenges, though diversity in regional representation remains limited.

2.2.5. Keywords Co-Occurrence Analysis

Keyword co-occurrence (Figure 6) reveals four thematic clusters:

(1)

Red (Macro-level Sustainability): Focus on “circular economy,” “sustainability,” and “management,” indicating a shift toward systemic and eco-strategic research.

(2)

Green (Scheduling and Inventory): Centers on “remanufacturing,” “inventory,” and “production planning,” emphasizing operational optimization under uncertainty.

(3)

Yellow (Reverse Logistics and Algorithms): Highlights “uncertainty,” “reverse logistics,” and “robust optimization,” reflecting algorithmic approaches to logistical challenges.

(4)

Blue (Supply Chain Contracts and Game Theory): Features “game theory,” “contracts,” and “competition,” exploring strategic interactions in CLSCs.

Early research (pre-2020) focused on green and blue clusters (operational and contractual optimization), while recent work (post-2020) increasingly integrates red and yellow themes (sustainability and algorithmic logistics), indicating a maturation toward multi-dimensional frameworks. The keyword evolution mirrors the field’s progression from isolated operational problems to integrated, sustainable, and technology-driven solutions.

3. Thematic Review of Remanufacturing Production Under Uncertainty

This section synthesizes findings from 98 core publications in remanufacturing research from the past decade, structured into three interdisciplinary themes: (1) decision-making and optimization under uncertainty, (2) supply chain and policy mechanisms, and (3) digital transformation and emerging technologies. As shown in Figure 7, this framework illustrates the synergistic interactions between three interdisciplinary themes and their collective role in mitigating system uncertainties.

3.1. Decision-Making and Optimization Under Uncertainty

The core of remanufacturing research lies in developing robust decision-making models to contend with pervasive uncertainties spanning demand, quality, and lead times [4,5,6,7]. This evolution, has progressed from static, single-period models to dynamic, adaptive frameworks that capture the temporal and interactive nature of real-world systems [8,9].

3.1.1. Pricing, Quality Evaluation, and Risk Hedging Strategies

Research has extensively addressed the impact of market demand fluctuations on optimal pricing strategies [10,11,12,13,14]. A critical parallel challenge is core quality uncertainty. Studies by Yanıkoğlu and Denizel (2021) [15] and Sun and Li (2023) [16] developed robust optimization models and accurate grading systems to mitigate value assessment attenuation. Yang et al. (2021) [17] further advanced this by considering discrete quality grades using the Phi-divergence method, while Liu et al. (2023) [18] introduced the Taguchi quality loss function for dynamic quality measurement [19,20]. The inherent uncertainty in consumer valuation of remanufactured products not only affects market pricing but also creates channel selection dilemmas and profit fluctuations for manufacturers [11,12,21,22]. A significant advancement has been the explicit incorporation of time-lag effects (e.g., delays in information feedback, quality processing, and policy implementation) into dynamic game models. Numerical analyses demonstrate that accounting for these lags leads to more conservative pricing strategies and a greater emphasis on quality control, underscoring the shift from static equilibria to time-conscious decision-making [10,16,17,23,24,25].

3.1.2. Integration of Carbon Policy and Financial Risk Management

Under carbon constraints, studies have explored production decisions under cap-and-trade policies [26,27], the hedging effect of carbon option contracts against demand risks [28,29,30], and the combination of trade credit insurance with carbon tax policies [31]. On the financing front, a significant body of work has emerged comparing traditional bank loans with innovative financing strategies like trade credit [32,33,34,35] and carbon quota pledge financing [36], analyzing their differential effects on easing capital constraints and optimizing product structure. The interplay between sales and leasing models under different quality scenarios has also been a point of economic comparison [37], revealing that sales models often hold advantages in high-quality recovery contexts.

3.1.3. Towards System Resilience and Adaptive Optimization

Research has moved beyond isolated problems to address production recovery scheduling under supply interruptions using branch-and-bound algorithms [9,38], cross-cycle flexible resource planning through two-stage stochastic programming [39,40,41], and the construction of stochastic dynamic models for collaborative quality-efficiency-service optimization [42,43]. To handle deep uncertainty without exact probability data, robust optimization frameworks like Φ-divergence models [17], scenario-based conditional value at risk (CVaR) approaches [44,45], and interval gray number models [46] have been employed. Zhu et al. (2022) [47] and Zhou et al. (2016) [48] formulated differential games to derive time-consistent production and emission-reduction strategies that co-evolve with market and policy changes, while Zhu and Goh (2024) [49] coupled consumer environmental awareness with evolving carbon policies in stochastic differential games. This body of work, which also includes robust lot-sizing [50] and Kalman filter-based estimation systems [51], collectively addresses a key research gap—the inadequate modeling of dynamic time-varying uncertainties—by embedding time dimensions, emergency responses, and dynamic resource allocation into a multi-level decision-making framework [4,5,24,43,51].

The characteristics, limitations, and future directions of this research stream are succinctly summarized in

Table 2. Summary of key research themes in remanufacturing under uncertainty.

Research Theme	Key Insights (Contributions and Findings)	Major Limitations and Future Gaps
Pricing, Quality and Risk Hedging	Contribution: Pioneered dynamic models with time-lag effects (e.g., Stackelberg Games). Finding: Integration of carbon finance (options) synergizes economic-environmental goals.	Relies on rational agent assumptions. Overlooks behavioral factors (e.g., consumer regret).
Supply Chain Coordination and Policy	Contribution: Quantified nonlinear impact of policies (subsidies, taxes). Finding: Option contracts effectively mitigate disruption risks.	Time lags in policy adjustment unmodeled. Under-explored consumer behavior dynamics.
Digital Transformation and Tech Integration	Contribution: Validated Blockchain-DT integration for trust (e.g., EtherTwin). Finding: AI/Heuristic algorithms (PSO-GA) improve success rates by 35–40%.	High computational complexity for large-scale problems. Techno-economic feasibility for SMEs unproven.

, which provides a comparative analysis across key sub-themes.

3.2. Supply Chain Coordination and Policy Mechanisms

Uncertainty permeates the entire remanufacturing supply chain, necessitating research into coordination mechanisms and the pivotal role of policy instruments in aligning the often-competing goals of economic efficiency and environmental performance [52,53,54].

3.2.1. The Central Role of Government Incentive Policies

A substantial number of studies have quantified the impact of subsidies, revealing their positive regulatory effect on corporate decision-making [55], identifying optimal levels under budget constraints [52], and demonstrating their ability to alleviate financial pressure while enhancing profits and social welfare in multi-entity systems [56,57]. The nonlinear influence of tax incentives on original equipment manufacturer (OEM) purchasing behavior [58] highlights the critical need for precise, dynamic policy design rather than one-size-fits-all approaches. Innovative mechanisms like trade credit insurance have been shown to synergize risk hedging with emission reduction incentives within carbon tax frameworks [31]. These studies collectively construct a vital transmission chain linking “policy tools, corporate responses, and system performance”.

3.2.2. The Intermediary Role of Consumer Behavior Heterogeneity

However, policy effectiveness is critically mediated by consumer behavior heterogeneity. Research has revealed how online reviews intensify penetration pricing strategies [59], how reference pricing and channel preferences interact in complex multi-shopping patterns [60], and how strategic waiting behavior and trade-in values adjust price elasticity [14,22]. Furthermore, consumer environmental awareness interacts with policy in complex ways; for instance, carbon taxes exert a double-edged effect on corporate decisions through consumer perception [61], and perceived value uncertainty is a key determinant in recycling channel selection [62]. This underscores that consumer behavior is not an external factor but a crucial intermediary variable that must be integrated into any holistic policy or supply chain model [22,59,60,63].

3.2.3. Collaborative Innovation for Supply Chain Resilience

The ultimate goal is supply chain collaborative innovation. Research has explored how pricing complexity affects closed-loop systems [13], expanded value creation through green investment models [64,65], and eased the impact of technology investment uncertainty via cash hedging strategies [35]. Theoretic foundations like closed-loop transformation theory [52] and comparisons of recycling modes based on consumer value perception [62] provide necessary guidance for supply chain reconstruction. Studies on option contracts [23,66,67] and risk-aversion models [68] have proven effective in coordinating decentralized supply chains and mitigating disruption risks. Inventory management strategies, such as those optimized for air conditioner remanufacturing and models considering stock-out control [69], further enhance operational stability. Ultimately, collaborative optimization mechanisms, such as carbon tax-subsidy synergy [53] or the joint optimization of carbon taxes and modular design [61,70], can achieve Pareto improvements in both economic and environmental performance, forging a robust “policy input--enterprise response--system output” closed-loop optimization mechanism [13,64,71].

3.3. Digital Transformation and Emerging Technologies

Digital technologies are fundamentally reshaping the remanufacturing paradigm by providing unprecedented capabilities to sense, model, and respond to uncertainties in real-time. Concurrently, the imperative for sustainability has driven research into holistic green strategies that span the entire product lifecycle, addressing the broad concept of green remanufacturing and its associated challenges [72,73,74].

3.3.1. DTs and Blockchain for Trust and Transparency

DTs and Blockchain are at the forefront of the digital transformation. Rather than isolated tools, they are increasingly treated as an integrated information-fusion layer. Putz et al. (2021) [75] demonstrated a blockchain-based DT system (EtherTwin) that automates data validation on a permissioned ledger, cutting manual reconciliation by 70%. Kapteyn et al. (2021) [76] embedded probabilistic graphical models into scalable DTs, reducing predictive-error variance by 18% in aerospace workflows, while Kantaros et al. (2021) [77] used lightweight DTs with IoT for adaptive scheduling, boosting overall equipment effectiveness by 12%. On the blockchain side, its power lies in creating decentralized trust. Smart contracts automate core return verification and payment, combating information asymmetry and enhancing consumer trust in remanufactured products [75,78,79], while tokenized carbon credit trading enables transparent ESG reporting and builds circular ecosystems [80]. However, as noted in the limitations, the techno-economic feasibility—especially regarding energy consumption, IoT deployment costs, legal enforceability, and scalability for SMEs—requires further scrutiny and constitutes a key future research direction [81,82,83].

3.3.2. AI and Heuristic Algorithms for Complex Optimization

Artificial intelligence (AI) and Heuristic Algorithms provide the computational muscle for optimization under uncertainty. Hybrid algorithms like PSO-GA have significantly improved the success rate of remanufactured parts matching [84]. The NGWO effectively solves multi-task scheduling problems in cloud-based environments with quality uncertainty [78], while improved discrete particle swarm optimization handles game-based scheduling with setup times [85,86]. Fuzzy robust programming [45], genetic-branch and bound algorithms [87], and multi-objective mixed-integer programming [45] are employed to tackle high mixed uncertainty and large-scale network design problems, often improving solution efficiency by 40% or more [87]. These algorithms, alongside constraint movement learning (CML) prediction for bottlenecks [88] and Benders decomposition for large-scale problems [89], are crucial for moving from theoretical models to practical, computable solutions for complex, NP-hard remanufacturing problems, such as disassembly line balancing [90,91] and reverse logistics network design [87,92].

3.3.3. System Integration and Technological Convergence

The convergence of digital technologies facilitates system-level integration and significantly enhances operational coherence and resilience across remanufacturing ecosystems. DTs serve as a core platform for real-time synchronization between physical and virtual systems, enabling dynamic carbon footprint tracking and life cycle assessment (LCA) that directly link operational decisions to environmental impacts [72,93]. For instance, the BDHDTPREMfg paradigm supports full lifecycle data penetration through DT technology, offering a novel pathway for uncertainty reduction and big-data-driven decision-making [81]. Integrated systems such as SLS-DED achieve micron-level monitoring accuracy, substantially shortening remanufacturing cycles and enhancing process reliability [94]. These technologies are increasingly deployed as an end-to-end information-fusion layer rather than isolated tools, providing a trustworthy and adaptive information backbone that continuously validates and refines production decisions [75,76,82].

Beyond individual applications, the interplay between technologies creates synergistic effects. Blockchain’s decentralized trust mechanism complements DTs by ensuring data immutability and auditability, which is critical for applications such as automated carbon credit trading and core quality verification [78,79,80]. AI algorithms leverage the high-fidelity data provided by IoT and DTs to drive predictive analytics and optimize complex scheduling and disassembly processes [84,95]. For example, adaptive genetic algorithms and fuzzy evaluation techniques have been shown to significantly improve assembly line balancing and parts-matching success rates [84,90]. Furthermore, the integration of edge computing with hybrid blockchain architectures (e.g., permissioned chains) helps reduce the computational and energy overhead associated with full-scale blockchain deployment, enhancing the feasibility of digital solutions for SMEs [83].

These technological advancements collectively enable macro-level circular economy integration in sectors such as tire remanufacturing [74] and photovoltaic module recycling [96], and optimize large-scale reverse logistics networks [92,97]. They also support the development of multi-level decision-support tools that integrate real-time data streams, predictive modeling, and collaborative platforms to address both operational and strategic challenges. By bridging the gap between physical processes and digital governance, emerging technologies serve as the technological backbone of the micro-meso-macro framework, ultimately enabling “precision perception, intelligent collaboration, and trustworthy governance” across remanufacturing ecosystems [81,83,97].

A comparative analysis of the key modeling and technological paradigms discussed across all sections is provided in

Table 3. Comparative analysis of key decision-making models and technologies in remanufacturing under uncertainty.

Model Type	Specific Model/Methodology	Motivation	Key Advantages	Major Limitations	Representative Studies
Dynamic Game Models	Stackelberg Game Pricing	Model strategic interactions under demand/cost uncertainty	Captures power asymmetry; quantifies equilibrium; integrates policy impacts	Assumes rational agents; often static/multi-period; struggles with behavioral factors	[31,59,98]
	Option Contracts (Put Options)	Hedge supply disruption/quality risks	Reduces over/understocking; flexible risk allocation; improves resilience	Requires precise cost estimation; contract enforcement challenges	[23,66,67]
Robust Optimization	Phi-Divergence Robust Model	Handle quality/distribution ambiguity	Performs well under severe uncertainty; avoids unrealistic assumptions	Computationally intensive; conservative solutions	[15,17]
	Scenario-Based CVaR Optimization	Minimize worst-case losses under uncertainty	Quantifies downside risk; balances cost/robustness	Curse of dimensionality; assumes fixed probabilities	[39,44]
LCA	Dynamic LCA + Carbon Footprint	Quantify real-time emissions; support eco-design	Links operations to environmental impact; identifies hotspots	Relies on coarse-grained data; lacks standardization; overlooks social dimensions	[72,93,99]

, highlighting their advantages, limitations, and applications.

4. Research Summary and Framework Synthesis

This study has systematically synthesized the literature on remanufacturing production under uncertainty through a novel three-tier analytical framework. This framework—spanning micro-level decision-making, meso-level coordination, and macro-level system perspectives—represents the main theoretical contribution of this paper. It provides a powerful, integrative lens to structure the vast body of existing research, trace its evolution, elucidate the interconnections between different research streams, and map out future trajectories. This review’s principal contribution is the development and systematic application of a tri-level micro-meso-macro framework, which offers a coherent structuring paradigm for the field of remanufacturing under uncertainty, as shown in Figure 8.

4.1. Micro-Level: Theoretical Breakthroughs in Adaptive Decision-Making

At the micro-level, a paradigm shift from static optimization to dynamic, adaptive decision-making is identified, driven by the need to respond to real-time uncertainties.

Pricing and quality control: Research has evolved from simple pricing models [10] to sophisticated frameworks incorporating time-lag effects and consumer behavior [16,22]. This evolution has been methodologically supported by interval gray number models [44,46] and robust markov decision processes [100], which have markedly improved the ability to handle market demand fluctuations. Quality uncertainty, a fundamental disturbance, is now addressed through robust optimization [15], Phi-divergence methods [17], and taguchi loss functions [18], significantly improving evaluation accuracy by up to 30% [91]. The application of DT monitoring [81,82] facilitates real-time quality mapping, further enhancing the efficiency of remanufacturing quality evaluation.

Risk hedging under policy: The integration of carbon finance (e.g., cap-and-trade [26], option contracts [28]) and innovative financing strategies (e.g., carbon quota pledges [36], trade credit [33]) into decision models exemplifies the synergy between economic and environmental objectives, helping to lower emission compliance costs by 20% [36].

Enabled by digitalization: Crucially, this shift is enabled by digital technologies. AI algorithms power dynamic scheduling and optimization [84,95], moving the field from experience-driven to data-driven paradigms. These breakthroughs in dynamic response mechanisms and policy synergy are applied in dynamic pricing under demand volatility and adaptive emission compliance strategies.

As synthesized in

Table 4. Synthesis of research findings across the three-tier framework.

Level	Core Theoretical Breakthroughs	Key Benefits	Major Limitations
Micro (Decision-Making)	Dynamic response models; real-time quality mapping via DTs; carbon-finance integration.	30% better quality evaluation; 20% lower emission costs; 15% quality improvement with time-lag models.	Static agent assumptions; behavioral factors unmodeled; SME barriers unaddressed.
Meso (Coordination)	Decentralized trust via blockchain; networked optimization; risk-hedging contracts.	40% faster disruption recovery; 35% higher parts-matching success; up to 14% profit rise.	Legal enforceability of smart contracts untested; high computational complexity.
Macro (System)	Policy-market coupling; lifecycle sustainability quantification; system-level circularity.	23% lower carbon emissions; 20% cost reduction; 15% higher resource efficiency.	Social dimensions omitted in LCA; cross-border policy conflicts; cultural gaps in developing economies.

, the micro-level is characterized by these advances, yielding quantifiable benefits. However, limitations such as static agent assumptions and overlooked SME financing barriers present key avenues for future research.

4.2. Meso-Level: System Reconfiguration for Enhanced Resilience

The meso-level analysis focuses on the coordination and resilience mechanisms that manage uncertainty across the supply chain network. The review reveals that resilience is achieved through technological integration and institutional innovation, with the core insights summarized in Table 4.

Building decentralized trust: Blockchain technology emerges as a key enabler, enhancing transparency and automating transactions to build consumer trust and combat information asymmetry [75,78,79].

Optimization and risk-hedging: The deep integration of intelligent algorithms (e.g., PSO-GA hybrids [84]) with traditional models optimizes complex operations like parts matching, improving success rates by 35%. Furthermore, risk is institutionalized through contract mechanisms like put options [66] and risk-aversion models [68], which effectively mitigate supply disruption and quality risks, leading to supply chain profit rises of up to 14% [48].

Designing for resilience: Architecturally, resilience is enhanced through loosely coupled system designs [101] that improve recoverability from disruptions by 40%, and through centralized configurations that benefit inventory management in short supply chains [25]. These studies, often employing Stackelberg game models [52], mark a clear paradigm shift from single-point optimization to networked collaboration, with practical applications in cross-factory scheduling in cloud remanufacturing and supply shortage risk hedging.

Table 4 highlights that the meso-level delivers significant improvements in disruption recovery and operational efficiency. Yet, it also faces challenges like the unproven legal enforceability of smart contracts and the computational complexity of multi-plant scheduling, which must be addressed to advance this field.

4.3. Macro-Level: Sustainable Pathways Through Policy-Technology Integration

At the macro-level, the synthesis concentrates on the policy and sustainability perspectives that guide the system-wide green transformation, with key findings summarized in Table 4.

Policy-market coupling: Research demonstrates the effectiveness of dynamic policy instruments, such as carbon tax-subsidy synergy models [53] and secondary subsidy mechanisms [80], in creating a positive cycle of “regulatory guidance-market response”. The nonlinear effects of policies like tax incentives [58] underscore the necessity for precise, dynamic design.

Lifecycle sustainability quantification: The adoption of lifecycle assessment (LCA) [93,99] and economy-energy-environment (3E) analysis [99] provides the methodological foundation for quantifying the sustainability of remanufacturing strategies, moving beyond qualitative claims to quantitative validation. For instance, AI-driven LCA [72] enables the linking of operational decisions to environmental impacts, facilitating regulatory dynamic adjustment mechanisms.

System-level circularity: Technological enablers like digital product passports [97] and modular design [70] support the practical implementation of circular economy models in specific industries, such as photovoltaic modules [96] and tire remanufacturing [74], achieving a 15% higher resource efficiency. This represents a transition from local technical enhancements to a broader system-ecological reconstruction.

Table 4 complements this conceptual model by synthesizing empirical findings and quantified performance outcomes across each level. By bridging operational research, sustainability analysis, and technological innovation, this framework establishes a foundational methodology for advancing both theoretical and practical endeavors in sustainable manufacturing systems.

5. Results and Discussion

5.1. Main Research Findings

Building upon the critical thematic analysis conducted in Section 3, which structured the literature into decision-making, supply chain policy, and digital transformation, four central findings emerge. These findings cut across the proposed micro-meso-macro framework, demonstrating the interconnected nature of modern remanufacturing research.

5.1.1. Transformation of Uncertain Decision-Making Paradigm

A substantial shift from static optimization to dynamic response is evident in recent research. In pricing strategy, the interval gray number model [44,56] and robust Markov decision processes [100] have been applied. These approaches have markedly improved the ability to handle market demand fluctuations. In quality control, the combination of the Taguchi quality loss function [18] and error propagation model [91] has effectively reduced the remanufacturing assembly defect rate. Through real-time data mapping, DT technology, has enhanced the efficiency of remanufacturing quality evaluation. This signifies a fundamental shift from experience-driven to data-driven decision-making.

5.1.2. Synergistic Innovation Effect of Policy Tools

Empirical evidence confirms the synergistic effect of policy tools. Specifically, the combination of carbon-quota pledge financing [36] and the secondary subsidy mechanism [80] has been tested to balance remanufacturing enterprises’ emission-reduction costs and profits. Blockchain technology [78] improves information transparency, significantly enhancing consumer trust in remanufactured products. This creates a positive cycle between policy incentives and market responses. However, the nonlinear effect of tax incentives [58] shows that policy design must establish a dynamic adjustment mechanism to optimize marginal benefits.

5.1.3. System Rebuilding Value of Technology Integration

The integration of intelligent algorithms with traditional models yields significant system-level benefits. For instance, the PSO-GA hybrid algorithm [84] improves the efficiency of remanufactured-parts matching, and the DEACA algorithm [102] reduces the risk of remanufacturing-project delays via buffer optimization. At the system level, the 3E analysis method [99] for the first time enables the collaborative optimization of the economic and environmental benefits of remanufacturing timing, delivering better decision-making accuracy than traditional methods.

5.1.4. Multi-Dimensional Improvement Paths for Supply Chain Resilience

Supply chain resilience is bolstered through multi-dimensional improvement paths. Research shows that a centralized configuration [25] is effective for inventory management in short supply chains. Through modular reconstruction, a loosely coupled system design [101] enhances the ability of CLSCs to recover from interruptions. The combined application of put option contracts [66] and a risk-aversion model [68] alongside ASP. NET provides an innovative solution for controlling the risks associated with remanufacturing supply shortages.

5.2. Discussion

5.2.1. Integrated Thematic Findings

This study presents an integrated review that systematically examines critical themes in remanufacturing, namely blockchain, intelligent algorithms, and dynamic modeling. Our analysis reveals the following:

(1)

Blockchain technology significantly enhances transparency and trust within remanufacturing supply chains. It ensures secure and tamper-resistant transactions for remanufactured products and provides a robust foundation for environmental compliance management. Furthermore, blockchain-supported smart contracts enable the automated execution of agreements, reducing transaction costs and improving operational efficiency.

(2)

Intelligent algorithms, including machine learning and optimization techniques, play a critical role in optimizing RPs. By leveraging real-time data and predictive analytics, these algorithms facilitate complex decision-making tasks such as production scheduling and quality control, thereby enhancing dynamic resource allocation and overall system resilience.

(3)

Dynamic modeling captures the time-varying nature of uncertainties in remanufacturing systems. The integration of real-time data streams supports the development of adaptive strategies to address fluctuations in market demand, supply chain disruptions, and changes in policy environments.

This study identifies a fundamental tension between the bounded rationality assumption of decision theory and the material closed-loop imperative of industrial ecology. Specifically, the former aims at individual profit maximization, while the latter focuses on overall system optimization. Notably, this research demonstrates that by incorporating environmental constraints (e.g., automated carbon quota trading) into distributed decision-making processes via blockchain-based smart contracts, synergy between these two paradigms can be achieved. This innovation aligns with the compatibility requirement in technology adoption theory, thereby offering a novel theoretical foundation and practical pathway for the diffusion of remanufacturing technologies.

5.2.2. Limitations of Current Research

Despite significant progress, research on RP decision-making under uncertainty faces several theoretical and practical bottlenecks that require urgent attention. These limitations are thematically categorized as follows:

(1)

Inadequate modeling of dynamic time-varying uncertainties and real-time response mechanisms

Existing research predominantly relies on static or multi-period models, which struggle to capture the dynamic evolution and nonlinear interactions of uncertainties such as fluctuating market demand, quality degradation of recycled materials, and policy adjustments. Time-dependent uncertainties arising from equipment aging, transportation losses, and technological obsolescence remain inadequately addressed. This limitation constrains the accuracy of long-term decision-making and fails to support the real-time adaptive optimization required in modern remanufacturing systems.

(2)

Lack of techno-economic assessments and industrial-scale validation for emerging technologies

While technologies such as blockchain and DTs show theoretical promise in enhancing transparency and efficiency, there is a critical lack of quantitative analysis regarding their economic feasibility (e.g., return on investment (ROI)) and operational effectiveness at an industrial scale. Key barriers include blockchain’s energy consumption, the unresolved cross-border legal validity of smart contracts, and the high cost of IoT deployment for DTs, which may exceed the affordability of low-margin remanufacturing businesses, particularly SMEs. The absence of a standardized cost–benefit assessment framework hinders their practical adoption.

(3)

Solation of technological, operational, and policy aspects, and weak cross-system integration

Current models struggle to reconcile decision-making conflicts and facilitate coordination among enterprises, supply chains, and policy systems. There is a significant gap in research on the synergistic integration of digital technologies (e.g., blockchain, DT) with policy instruments (e.g., carbon quotas, subsidies) to create automated, compliant, and resilient remanufacturing networks. Information silos and computational complexity further hamper the coordinated optimization of such cross-border and multi-level systems.

(4)

Insufficient focus on SME-specific and culturally adapted solutions

Most existing frameworks assume homogeneous market and policy environments, overlooking critical practical constraints faced by SMEs and developing economies. These include informal recycling sectors, varying levels of digital literacy, financing barriers, and regional disparities in carbon prices. The lack of low-cost, modular, and regionally adaptable solutions seriously limits the global applicability and practical implementation of proposed theoretical models.

5.3. Future Research Directions

Based on the systematic review and critical analysis presented in this study, we identify four overarching future research themes that address the most pressing gaps in remanufacturing under uncertainty.

First, there is a critical need to develop dynamic real-time decision-making systems that can adapt to time-varying uncertainties. Future research should focus on creating optimization models that integrate real-time data streams from IoT sensors with predictive analytics and DT technologies. Such systems would overcome the limitations of current static or multi-period models by capturing nonlinear and lagged effects in remanufacturing systems. Hybrid simulation-optimization frameworks, federated learning for cross-factory data collaboration, and model predictive control (MPC) integrated with DTs represent promising methodological approaches. These systems should be validated through pilot implementations in high-variability sectors such as automotive and electronics remanufacturing, with particular attention to SMEs.

Second, the techno-economic validation of emerging technologies remains substantially underexplored. While technologies like blockchain and DTs show theoretical promise, empirical studies are needed to evaluate their economic viability and operational effectiveness at industrial scale. Future research should conduct case-based cost–benefit analyses focusing on ROI, implementation latency, data integrity, and carbon footprint reduction. Lightweight DT deployments using low-cost sensors and cloud analytics, along with permissioned blockchain architectures incorporating edge computing, offer particularly valuable research directions for making these technologies accessible to resource-constrained SMEs.

Third, researchers should develop integrated digital-policy-supply chain frameworks that combine digital technologies with policy instruments to enable automated, compliant, and resilient remanufacturing networks. This research theme addresses the current isolation of technological, operational, and policy aspects in the literature by exploring their synergistic potential. Multi-agent simulation models incorporating regulatory constraints, smart contracts for automated carbon trading and subsidy disbursement, and system dynamics modeling to capture policy-technology adoption feedback represent innovative methodological approaches. The output of this research should include open-source decision-support toolkits adaptable to diverse regulatory and market contexts.

Finally, there is an urgent need for SME-specific and culturally adapted solutions that address the unique challenges faced by small manufacturers and developing economies. Future research should develop low-cost, modular, and regionally adaptable solutions that consider informal sectors, varying levels of digital literacy, and financing barriers. Ethnographic studies, participatory design workshops, and lightweight ERP and CRM integrations for remanufacturing would provide valuable methodological foundations. This research should deliver culturally adapted implementation frameworks, training modules, and policy briefs tailored for governments and industry associations operating in emerging markets.

This comprehensive research agenda not only addresses the identified theoretical and practical gaps but also provides a clear pathway for transitioning from isolated technological innovations to integrated, scalable, and sustainable remanufacturing ecosystems. By pursuing these research directions, scholars and practitioners can significantly advance both the theory and practice of remanufacturing under uncertainty.

6. Conclusions

This study addressed three core research questions through a systematic review of 98 publications on remanufacturing production decision-making under uncertainty, published between 2015 and 2024. By integrating a PRISMA-compliant systematic review with comprehensive bibliometric analysis and a novel micro–meso–macro analytical framework, our methodology provided both quantitative and qualitative insights into the field’s evolution, ensuring rigor and reproducibility while offering a structured synthesis of diverse research streams.

This analysis reveals that the field has evolved significantly from static, single-objective optimization models toward dynamic, adaptive paradigms that incorporate real-time data integration and intelligent algorithms. Digital technologies (DTs, blockchain) and carbon-financial instruments play enabling and synergistic roles—DTs enhance transparency and enable real-time simulation, while blockchain provides trust and traceability in reverse logistics. When combined with policy tools like carbon quotas and subsidies, they create powerful mechanisms for aligning economic incentives with environmental goals. The proposed three-tier framework effectively synthesizes these developments by demonstrating how micro-level advances (dynamic pricing, quality control), meso-level coordination (supply chain contracts, technology integration), and macro-level systems (policy-market synergy, lifecycle sustainability) are interdependent and collectively drive the field toward a more holistic understanding.

For researchers, this study offers four principal contributions: (1) a reproducible SLR protocol; (2) a comprehensive bibliometric mapping of research clusters; (3) a novel analytical framework to guide future research; and (4) a consolidated agenda that addresses critical gaps in dynamic modeling and technology integration. For remanufacturing professionals and policymakers, our findings provide actionable insights for the following: (1) diagnosing uncertainty challenges across both operational and strategic levels; (2) validating investments in digital technologies; and (3) designing dynamic policy instruments that can adapt to evolving market conditions. Furthermore, the emphasis on SME-specific solutions highlights the critical need for the following: (1) scalable and adaptable technologies, and (2) tailored policy support mechanisms to broaden industry adoption and enhance inclusivity.

While this review provides a comprehensive synthesis, its focus on English-language SCI/SSCI articles may overlook insights from other sources. Additionally, the rapid evolution of digital technologies necessitates ongoing research beyond this review’s scope. Future work should develop the dynamic, validated, and inclusive models needed to advance remanufacturing toward a sustainable ecosystem characterized by precision perception, intelligent collaboration, and trustworthy governance.