The EU imports 16,000 tonnes of rare-earth materials annually from China, which accounts for 98% of its requirements. These materials are essential for the production of electric vehicles, but their supply is in jeopardy. Production has increased from 4,070 to 10,261 metric tons, resulting in an increase in prices of 152 percent. Worldwide, manufacturers' operational viability is in jeopardy as a result of this volatility, particularly as the electric vehicle revolution advances. A data-driven path forward is provided by ground-breaking research by Pérez, Lieder, Jeong, and Asif (2025). Their simulation-based decision support tool, which was published in the International Journal of Production Research, demonstrates that an additional 10.6% design investment can reduce the average cost of an electric machine by 18.6%, material demand by 14.7%, and cradle-to-gate impact by 38.7%.
This research helps manufacturers navigate the transition to circular economy models by bridging the gap between early product design decisions and end-of-life outcomes that won't happen for over a decade. More than just environmental rhetoric, the imperative of the circular economy Principles of the circular economy have evolved from aspirational targets for sustainability to competitive necessities. The Circularity Gap Report (2021) states that the circular economy has the potential to cut greenhouse gas emissions by 39% and reduce virgin resource use by 28% globally. However, the lack of implementation is frustrating. Despite estimates that remanufacturing can cost 40–65 percent less than new production, remanufacturing only makes up 1.9% of production in the EU. The study reveals a fundamental disconnect: it is impossible to fully link the effects of today's design choices to the vehicle's end-of-life (EOL) situation ten years later. As executives struggle to justify investments whose returns are uncertain and distant, this temporal gap paralyzes strategic planning.
With significant after-market potential estimated at 0.8 to 0.9 billion Euros in 2040, electric motors (also known as electric machines) have been identified as a "rising star" component that will experience increased demand alongside traction batteries. These components are particularly strategic due to the embedded rare-earth magnets. By 2050, according to a study, circular practices like recycling could meet up to 70% of the neodymium demand for electric vehicles. The need for action is clear: putting in place recovery infrastructure now will put manufacturers in a position to benefit from the wave of end-of-life electric vehicles that will arrive around 2035-2040 as the first generation of EVs reach the end of their useful lives. Supply Chain Management's Evolution of Advanced Analytics: From Static Models to AI-Enhanced Simulation We must follow the development of supply chain modeling approaches—a path that increasingly intersects with artificial intelligence—to comprehend the innovation represented by this research. Deterministic Optimization, First Generation (1990s-2000s) Production planning with known parameters was the primary focus of the initial remanufacturing models. Case studies focused on established remanufacturing sites with known capacity, emphasizing planning and inventory policies, and numerical models simulated hybrid manufacturing systems to manage both returned products and raw materials in order to find the best policies.
These strategies relied on presumptions about predictable processing times, predictable return flows, and stable demand, all of which are uncommon in actual circular supply chains. Probabilistic Simulation, Second Generation (2000-2010) The adoption of stochastic modeling was driven by recognizing the uncertainties in the supply chain. Based on cost, revenue, and profit, probabilistic models evaluated system performance through Monte Carlo simulations. However, their inability to capture the dynamic interactions between product lifecycles, component degradation on its own, and cascading material flows over multiple recovery cycles remained a limitation. Multi-Method Simulation: Third Generation (2010s-Now) The research conducted by Pérez et al. exemplifies the sophistication of this generation. Agent-Based (AB), Discrete Event (DE), and System Dynamics (SD) simulation modeling produces outcomes that are more realistic. Through manufacturing, utilization, and multiple recovery cycles, their model tracks individual parts as agents carrying variables like material type, cost, mass, and degradation state.
Given the above hierarchy, the ability to merge (i.e., assemble) individual part agents into a complete product is the main advantage of using an AB approach. In a similar vein, the product can be disassembled, or unmerged, back into its component parts, preserving the weight, cost, and identification numbers of each part while remaining trackable throughout the simulation runs. With previous modeling methods, this capability—tracking individual components through assembly, 15-year usage phases, disassembly, recovery, and reassembly into new products—was impossible. AI-enhanced adaptive systems are the upcoming fourth generation (current and future). Although it does not explicitly employ machine learning, the current study lays the groundwork for AI integration. Individual component histories, degradation patterns, recovery success rates under various conditions, cost-benefit relationships across thousands of scenarios, and extensive data structures that are ideal for training AI models are generated by the simulation. Layering AI capabilities onto this simulation infrastructure is the natural evolution: Predictive Analytics: The linear degradation function (D(t) = a t) used in the research was not as accurate as machine learning models trained on operational sensor data at predicting component degradation. Core collection can now be done proactively before critical damage thresholds are exceeded. Computer vision: Convolutional neural networks could replace manual core assessment with automated visual inspection, lowering labor costs, which make up a lot of the cost model's recovery expenses.
Reinforcement Learning: As return flows fluctuate unpredictably, adaptive algorithms could optimize inventory policies in real time to balance the availability of new parts and remanufactured parts. Digital Twins: The digital twin idea that is becoming more and more used in smart manufacturing is reflected in the agent-based architecture in which each component carries lifecycle data. These digital representations could be fed by real-time data from IoT-enabled products in the field, allowing for dynamic decision-making. Detailed Circular Manufacturing Use Cases for AI in Supply Chain Management Although the research does not directly employ AI, its findings highlight the areas in circular supply chains where AI applications would generate the greatest value: Use Case 1: Core Quality Predictive Assessment Remanufacturing success rate, followed by product-related factors like lifetime minimum, critical damage threshold, and degradation factor, emerged as the most crucial parameter for ensuring economically viable circular supply chains over the long term, according to the sensitivity analysis.
Through the following, AI can significantly boost success rate prediction: Embedded sensors monitor vibration patterns, temperature profiles, electrical performance degradation, and mechanical stress indicators during the utilization phase of Sensor Fusion Analytics. These multi-modal data streams are processed by machine learning algorithms to determine which parts can be reused rather than recycled and how long they will last. Anomaly Detection: Unsupervised learning finds unusual patterns of degradation that point to manufacturing flaws, abuse, or environmental factors that affect the health of a product. This information immediately flags cores unsuitable for recovery and contributes to design enhancements. Optimization Algorithms: AI finds the best return timing by weighing the value of extended utilization against the risk of degradation beyond recovery thresholds using predicted core quality and current inventory levels. According to the findings, 60 percent of the mass of magnets and 56 percent of the mass of regular steel could be used for remanufacturing or reuse. These percentages could be increased by 5-10% by improving quality prediction, resulting in material cost savings of millions at scale.
Automated Disassembly Planning is Use Case 2 The research uses sophisticated cascading logic: it is assumed that the disassembly sequence is the same as the assembly sequence, with the product breaking down into modules, subassemblies, and individual parts first. Throughout each of the aforementioned disassembly levels, the cascading logic for core evaluation is used repeatedly. This can be improved by robotic disassembly systems powered by AI: Vision-Guided Robotics: Without reprogramming, vision-guided robots are able to identify parts, locate fasteners, and adapt to product variations thanks to deep learning models trained on component images. Using actual processing times, part accessibility, and tooling requirements, reinforcement learning discovers more efficient sequences rather than assuming disassembly mirrors assembly. Quality-Dependent Routing: As components are taken out, computer vision evaluates damage and wear and dynamically routes them to recycling, remanufacturing, or reuse streams.
Workers = demand/1000 + 1 in the research's staffing equation assumes manual disassembly. In addition to enhancing consistency and speed, AI-driven automation could cut labor requirements by 40-60%. Use Case 3: Optimizing Inventory and Dynamic Demand Forecasting Over the course of 45 years, 10,000 electric machines are assumed to be in constant demand in the simulation. The effects of technology disruption, economic cyclicality, and seasonality are all present in real markets. Applications of AI include: Multi-Horizon Forecasting: LSTM neural networks incorporate external variables such as vehicle production schedules, economic indicators, commodity prices, and policy shifts (e.g., EV subsidies, emission regulations) into their forecasts of demand at daily, monthly, and annual horizons. Probabilistic Forecasting: AI generates probability distributions over demand scenarios rather than point estimates, allowing for robust optimization in the face of uncertainty. Multi-Echelon Inventory Optimization: AI accounts for the unique uncertainties of reverse logistics, where return timing and quality are stochastic, by simultaneously optimizing stock levels across new part suppliers, production facilities, remanufacturing centers, and distribution networks.
The quality and timing of returns, demand for remanufactured and reused products, requirements for materials that match, and uncertainties regarding disassembly and process time are among the challenges and uncertainties that closed-loop supply chains face, according to the research. AI is particularly well-suited to addressing these complexities due to its ability to deal with uncertainty in high dimensions. Use Case No. 4: Enhancing the Network of the Supply Chain The study's models are simplified networks with a single center for manufacturing and remanufacturing. Real-world implementations confront challenging network design issues: Considering transportation costs, labor availability, regulatory environments, and customer proximity, where should remanufacturing centers be situated? This old problem is solved by using circular-specific constraints and mixed-integer programming enhanced with machine learning for cost prediction. AI determines optimal material flows across network nodes in real time in response to transportation disruptions, capacity constraints, and shifting material prices as cores become available and production needs change.
Assessment of Supplier Risk AI models can process geopolitical news, trade data, and commodity markets to predict disruption probabilities, triggering proactive sourcing adjustments, given the research's focus on rare-earth supply chain risks. Use Case 5: Improving the Process Parameters Labor time for inspection, cleaning, disassembly, and specialized remanufacturing activities are among the many parameters included in the cost model. To figure out how much it would cost, experts gave time estimates for each remanufacturing and reuse step. As a result, costs associated with recovery, EOL labor and energy, and internal logistics can all be factored into the cost of a single restored component. These processes can be improved by AI through: Process Mining: Using execution log analysis, bottlenecks, rework loops, and efficiency differences between operators or equipment can be discovered. Prescriptive Analytics: Using machine learning, process parameters like the length of time it takes to clean, the temperature at which it heats, and the sequences of inspections are suggested to keep quality consistent while reducing costs. Adaptive Control: Instead of using predetermined procedures, AI adjusts processing parameters in real time as part conditions change.
Modern AI Services for Supply Chain Management The market for AI-powered supply chain solutions is changing quickly, but comprehensive platforms for circular manufacturing are still lacking: Platforms for businesses with the potential to go around Blue Yonder (JDA Software): The demand forecasting and inventory optimization features of their Luminate platform are made possible by machine learning. Neural network demand sensing and autonomous supply chain orchestration are two core capabilities. However, in order to handle reverse logistics, quality-dependent routing, and material tracking across multiple lifecycles, adaptation for circular flows would necessitate substantial customization. SAP Integrated Business Planning: S/4HANA's circular economy features support recycling and remanufacturing operations, and machine learning modules address demand volatility and supply response.
Between forward production planning and reverse material flows, integration challenges exist. Oracle Supply Chain Management Cloud: Blockchain-enabled traceability and predictive analytics effectively support forward supply chains. Circular applications necessitate custom development, particularly for the research's cascading EOL evaluation logic. Kinaxis RapidResponse: If extended with reverse logistics modules and quality assessment capabilities, their concurrent planning approach and machine learning-enhanced scenario analysis could support circular manufacturing. Solutions for the Circular Economy Maersk TradeLens: This blockchain platform offers container-level tracking and documentation, which is pertinent to the research's requirements for product traceability. Before EMs are taken off the market, the study emphasizes the need for product health indicators and product tracking capabilities to estimate their residual useful life.
The IBM Sterling Supply Chain Intelligence Suite includes IoT-integrated AI-powered visibility and risk management. Although circular-specific workflows necessitate customization, asset tracking capabilities could support the individual part agent tracking implemented in the simulation model. Manufacturing execution system with AI analytics from Rockwell Automation called Plex Systems is useful for shopfloor remanufacturing operations. According to the findings, in order to improve the stability of the process and decrease the amount of effort required for operational handling, it is necessary to have knowledge of the remanufacturing process, which can be supported by digital tools and automation. The Deficit in Circular Manufacturing An important point to make is that no major vendor provides complete out-of-the-box circular manufacturing solutions with integrated forward/reverse flows, quality-dependent cascading routing, multi-lifecycle material accounting, and linkages from design to EOL. Manufacturers face a challenge and technology providers stand to benefit from this gap. Part-level agents, hierarchical assembly/disassembly, degradation modeling, and integrated cost/environmental accounting are all part of the research's simulation architecture, which basically outlines the features needed for such a platform.
Deep Dive: A Case Study of the Scania Electric Machine Through a comprehensive examination of a Scania electric machine, the research demonstrates how simulation-based decision support can be applied to real-world manufacturing decisions. Material Structure and Design of Products There are twelve separate components that make up the EM. This kind of motor typically has a stator with fixed, symmetrical three-phase windings. A rotor housed within the stator houses a collection of specially shaped magnets. The distribution of the materials has strategic significance: Electrical Steel: The component with the highest mass (stator/rotor laminations) Rare-Earth Magnets (NdFeB): Critical supply chain risk and highest cost component Copper windings have a high value, but technically they can't be used again. Housing made of aluminum is a structural component that can be recycled. Insulation materials with varying recyclability include polymers and composites. Electromechanical transformation is made possible by magnets that are shaped specifically for the rotor. Typically, these magnets account for the majority of the cost of the materials. The torque, power density, speed, high-speed operation, and flux-weakening capability of magnets are affected by their material and configuration. The optimization experiments identified magnet-containing components as priority investment targets for circular design because of this concentration of costs.
Methodology for Design Parameterization The methodical quantification of design effort is a significant methodological advance. In AB simulations, a technique is used to parameterize design decisions for decision support. As a starting point, the design is based on the current Scania EM concept. Activities were listed to estimate additional timely efforts, prototyping, and testing based on the need to modify the design to facilitate reuse, remanufacturing, or recycling. The following three design methods were quantified: Modular architecture, standardized fasteners, accessible wear components, documented disassembly procedures, and improved durability for repeated processing are all part of the design for remanufacturing (an additional 15% effort). Similar to remanufacturing, Design for Reuse (12%) is optimized for minimal reconditioning and features robust connectors, protective coatings, and simplified critical junctions. Material purity, identification marking, simplified material separation, and avoiding hazardous substances are all part of the Design for Recycling (7% additional effort) strategy.
"Design approaches for reuse, remanufacturing, and recycling can be combined by allocating these to different parts of the EM," states the innovation, "recognizing that different parts within the same product can follow different EOL strategies." First Experiment: Cost-Effective Circular Design By varying EOL strategies across the 12 parts, the initial optimization experiment attempted to reduce total operational costs to a minimum. Parts P1, P8, P9, P11, and P12 should be considered for reuse, while part P10 should be remanufactured, as shown by the best combination. Additionally, it is suggested that the entire assembly of parts P8 through P12 be reused in module M2 and the subassemblies S3 and S4. The general design-for-circularity guidelines, which treat all components equally, are starkly opposed to this targeted approach. The strategic insight is to focus design investment on subassemblies of high value that use expensive materials. The outcomes are compelling: the total cost decreased by 17.7%, the carbon footprint decreased by 38.7% (from cradle to gate), material demand decreased by 14.7%, the average cost of EM decreased by 18.6%, and design investment was increased by 10.6%.
Financial validation suggests viability: To break even, approximately 31,000 units must be recovered using a 16 percent cost savings per EM and a high CAPEX estimate of 6 percent design investment—achievable given the scope of the target market. Second Experiment: A Circular Investment of Minimum Viability The objective of the second optimization was reversed: maintain economic viability in comparison to the linear baseline while minimizing design investment. The conclusion suggests that S3, as well as all of its components P8, P9, and P10, ought to be made to be simple to reuse. Recycling-friendly design should be used for all remaining components and assemblies. Implementations with limited resources or regulatory compliance issues are addressed in this scenario. Despite a design investment of +8.4 percent, total costs decreased by 6.6%, the carbon footprint decreased by 8.0%, and material demand decreased by 10.1%. Strategic Implication: Businesses can gradually implement circularity, concentrating first on subassembly S3 (the magnet-containing rotor assembly) while ensuring that the remaining components can be recycled. Risk is reduced, early wins are achieved, and organizational support for a broader circular transformation is developed with this phased approach.
The Pattern of Three Phases in Temporal Dynamics Across three 15-year phases, the longitudinal analysis reveals distinct performance patterns: Phase 1: The Investment Period (from 0 to 14). Only new (virgin) materials are used to make and sell EMs in the first EM group. Since the simulation is expanding, no benefits are anticipated at this stage. This is the commitment period that requires ongoing investment despite the lack of immediate returns, which executives must take into account when evaluating circular business cases. Phase 2: Peak Returns (14-30 years) When cores can be recovered after the utilisation phase, which occurs in year 14, the first benefits are realized. After another 14–15 years (in year 30), when these EMs exit the use phase, they go through the remanufacturing and reuse procedure once more. The second category of EMs consists of EMs that have been in use for 14–15 years and have been recovered. These EMs contain a mixture of new and remanufactured components. Cradle-to-gate impact reduction and cost savings reached their highest levels after year 30. Phase 3: Effects of quality degradation (ages 30 to 45) The third EM group, like the second, has a mix of new and remanufactured or reused parts. However, after two complete lifecycles, some of the remanufactured and reused components must be recycled.
Core quality overall suffers as a result of this. After year 30, the average cost savings fall to 13.2% from 23.6%. Similarly, the average cradle-to-gate impact decreases from 46.0% to 33.0% after year 30. The crucial finding is that infinite circularity is not possible. Recovery is limited by material fatigue to approximately two full lifecycles before recycling is required. Total material demand reduction plateaus at the observed 14.7% level due to this physical constraint, as opposed to more optimistic scenarios that assume indefinite remanufacturing. Analysis of Material-Specific Recovery Eight material categories with wildly varying recovery potential were tracked in the simulation: Eighty percent of the electrical steel mass collected during the simulation runs could be used for remanufacturing or reuse, and the remaining twenty percent could be recycled. Magnets and regular steel follow a similar pattern, with roughly 60% and 56% going to remanufacturing and reuse, respectively, and the remaining 40% and 44% going to recycling. Copper, for example, cannot be remanufactured or reused due to technical constraints; however, at the conclusion of the simulation period, it still possesses a high recycling potential of more than 1,500 tons.
Copper windings cannot be remade, which presents a significant design opportunity. Copper has a significant material cost as well as embedded energy. Modular winding cassettes or advanced joining methods like ultrasonic welding and laser bonding, both of which allow for non-destructive removal and reinstallation, have the potential to shift copper's value from recycling to reuse. Sensitivity Analysis: Important Factors for Success By varying seven different parameters in increments of 25%, Experiment 3 systematically investigated uncertainties from 100% to +100%. The most Important Parameters: Cost benefits are most affected by changes in the success rate of remanufacturing and reuse, with economic losses occurring when the success rate falls below 75%. This parameter reduces footprint savings, but the circular scenario still had lower emissions than the linear one. On the other hand, cost savings are greatest when lifetime minimum, lifetime maximum, and degradation factor are reduced. This is because the utilisation phase is shortened, allowing cores to be recovered earlier and with less damage.
Obstacles to Circulation: Some variations (marked in white) have not reduced costs or the carbon footprint, indicating that both outcomes are the same. This is the situation, for instance, when the return rate is reduced to zero, preventing cores from being utilized in remanufacturing and reuse processes. This highlights a fundamental requirement: proactive core acquisition strategies are necessary for circular manufacturing. Since the business model must guarantee product returns through deposits, buy-backs, leasing, or regulatory mandates, zero return rates eliminate all circular benefits. Priorities for the Manager: The allocation of investments is directly influenced by the sensitivity analysis: Process Excellence: To maximize success rates, put a lot of money into remanufacturing process technology, quality control, and operator training. Implement monitoring systems that optimize return timing for predictive maintenance to capture cores before excessive degradation occurs. Build products with modular, resistant to degradation architectures to increase the number of possible recovery cycles.
User Insights: Making Circular Manufacturing Systems Work Better The research provides operational insights that are crucial for implementation in addition to its quantitative findings: Planning for Infrastructure and Capacity The study assumes that the OEM performs remanufacturing and reuse in separate facilities from primary production. Depending on whether greenfield construction or adaptation of existing facilities are used, CAPEX investments typically range from 2 to 6 times the design investment. Initial planning guidance is provided by the staffing equation (workers = demand/1000 + 1), which assumes stable efficiency. In actual operations, there will be learning curves, equipment downtime, and throughput-affecting variability in core quality. Practical Advice: Begin by adapting existing facilities with a focus on high-volume, high-value components (lower CAPEX multiplier). Greenfield, specialized facilities are becoming more and more cost-effective as volumes grow and learning curves become more efficient.
Needs for Data Infrastructure The research emphasizes that operational data availability is a prerequisite for circular success. Parameter estimation is needed for the declining success rate function and degradation model. These must rely on conservative estimates during use because they lack embedded sensors and data connectivity, which reduces their economic viability. Requirements for the Technology Stack: Temperature, vibration, and electrical performance monitoring during use are examples of IoT sensors. Edge Computing: Real-time data processing and product-level anomaly detection Infrastructure for connectivity: cellular and satellite connections for remote asset monitoring Cloud Analytics: centralized platforms that collect performance data for the entire fleet RFID, QR codes, or blockchain traceability systems allow for individual part tracking across multiple lifecycles. ERP Integration: Smooth communication between planning systems and operational data Identification and database management skills are required to keep track of individual components over multiple assembly and disassembly cycles.
This requirement is reflected in the agent-based architecture of the simulation, where each component has a lifecycle history. Capabilities of the Organization and the Management of Change Implementing the circular economy successfully necessitates skills not found in conventional manufacturing. Variability is higher in remanufacturing than in new production. The study notes that different damage levels require different processing times, creating scheduling complexity. Instead of highly specialized assembly line operations, businesses require workforces that are adaptable and skilled. Priorities for Capability Development: Cross-training technicians in diagnosis, repair, and quality assessment, as opposed to narrowly specialized tasks in Process Engineering Coordination of the supply chain involves removing functional barriers that divide procurement, which traditionally focuses on suppliers, and reverse logistics, which focuses on customer returns. Customer Relationship Management: In order to achieve the assumed 90 percent return rate, proactive engagement—perhaps through service contracts, buy-back programs, or deposit systems—is required. The OEM-customer relationship becomes more long-term as a result. Market Development and the Regulatory Setting Findings are positioned within European regulatory frameworks by the research. The End-of-Life Vehicles Directive 2000/53/EC establishes legal requirements for recovery.
This regulatory reality is reflected in the mandatory recycling cost for non-recoverable parts in the simulation. The recent EU Critical Raw Materials Act may impose additional recovery requirements or incentives for rare earths. However, regulations on their own are not enough because the study assumes that remanufactured parts are "as good as new," even though they compete in spare parts markets that demand lower prices. The market barrier of customer acceptance of remanufactured goods is still unsolved by technical models. Strategies for Developing the Market: Ensure quality assurance by providing identical warranties for remanufactured goods. Transparency: Make it clear to customers that they will save money and help the environment. Performance Guarantees: Supply performance data demonstrating functionality that is as good as new Premium Positioning: Rather than compromising, portray remanufacturing as cutting-edge technology. Strategic Advice and Implications for Managers Manufacturing executives who are navigating circular transformation can benefit from the research's practical recommendations: Options for Strategic Design Implement component-level circular investment prioritization as the first recommendation. Create a priority matrix that takes into account the following, as opposed to a circular design that is common to all products: cost per component of the material Price fluctuation and criticality of the supply chain Possibilities of technical remanufacturing Projections for volume and return rates The Scania case demonstrates that subassembly S3 (rotor with rare-earth magnets) generates disproportionate returns for incremental investments of 10.6%.
Using simulation-based scenario testing, apply similar analysis to your product portfolio. Include Circular Considerations in Stage-Gate Processes as a Second Recommendation Circular considerations must be incorporated into initial development due to the temporal gap between design decisions and EOL outcomes (10-15+ years). Change the criteria for design review to include: Time estimates for disassembly and equipment requirements Recycling is made simpler by material purity and separation. Accessibility of wear components for remanufacturing Integration points for sensors for operational monitoring Cost models that cover the entire lifecycle (not just production and use) The third recommendation is to establish product health monitoring as a fundamental capability. Predictive core quality assessment is crucial, as demonstrated by the success rate sensitivity analysis. Analytics capabilities that translate operational data into remaining useful life estimates, customer-facing interfaces that promote timely returns before critical damage, and embedded sensors for performance monitoring during utilization are among the investment priorities.
Early in product development, collaborate with IoT platform providers and data analytics specialists. By 3-5 percent, retrofitting connectivity costs more than designing it in from the beginning. Implementation of Operations Step 4: Implement a Phased Rollout Strategy In light of significant CAPEX requirements and operational ambiguities: Phase 1 (Years 1-3): Pilot program to develop process expertise and validate economic assumptions with real data. Limited scope (one high-value subassembly). Phase 2 (Years 4-7): Scaled implementation broadens recovery scope as return volumes rise and learning curves lower costs. Phase 3: Full-scale operations that achieve steady-state economics as full utilization of first-generation products Financial risk is managed through this strategy, which also gradually builds organizational capabilities.
Recommendation 5: Start working with partners in reverse logistics early The study assumes return rates of 90%, which can only be achieved through proactive collection strategies. For core collection, think about partnerships with dealers and service networks, incentive structures that encourage returns (deposit-refunds, trade-in credits), regional consolidation centers that reduce transportation costs, and information systems that provide return visibility and forecasting. Deal with dealer agreements, which should include things like core collection goals and pay structures. Test deposit-refund programs in areas where customers are likely to accept them.
Sixth Recommendation: Invest in Inspection and Disassembly Automation In recovery operations, labor intensity is indicated by the staffing equation. Automation investments should focus on flexible reassembly lines that can accommodate mixed new and remanufactured components, disassembly robots with adaptive capabilities for handling product variations, automated inspection systems that use computer vision for quality assessment, cleaning systems that reduce manual handling and improve consistency, and disassembly robots. Prioritize the automation of high-volume, variable, and repetitive tasks (such as inspection and cleaning) over complex decisions that still require human judgment. Designing a New Business Model
7th Recommendation: Look into Other Types of Ownership The study is based on conventional sales with an optional buy-back option. Consider:
Product-as-a-Service: Lease/subscription models that remain under OEM ownership offer continuous performance data and guarantee 100 percent return rates. This eliminates uncertainty regarding return rates, but it necessitates distinct financial structures and customer value propositions. Performance-Based Contracting: Selling "operational hours" or "kilometers driven" rather than actual products ensures that OEM incentives are aligned with product longevity and effective maintenance, naturally supporting circular practices. Second-Life Marketplaces: Expand addressable markets by creating channels for recovered products in various applications (such as remanufactured electric machines for industrial equipment after automotive use).
Eighth Recommendation: Work with the industry to develop volume and standardization Economy of scale is restricted by volume constraints, which limit single-OEM circularity. Economy is improved by industry collaboration through: shared facilities for remanufacturing multiple OEM products standardized interfaces that permit component reuse across brands collaborations with recyclers to guarantee infrastructure for material recovery Common data standards for product lifecycle and health monitoring Participate in or lead industry consortia developing standards for the circular economy. Standard-setting pioneers have a disproportionate impact on ecosystem development. Evaluation of performance and Constant Improvement
Recommendation 9: Set Circular-Specific Key Performance Indicators Circular performance is not adequately captured by conventional financial and operational metrics. Establish:
Economic Indicators: Circular cost difference (new production versus remanufactured unit cost) Return on circular investment, or ROCI, compares operational savings to investments in design and infrastructure. Cost of core acquisition in relation to product value Environmental indicators: Material circularity index (percentage of mass remade or reused versus virgin) Carbon intensity per unit from cradle to gate across all lifecycles Supply chain risks are weighted by material criticality scores. Metrics for Operations: Success rates of remanufacturing according to component and damage level Time distributions and core return rates Processing times for recovery and productivity of workers Comparison of remanufactured and new products' quality conformance Retry Continue Edit Report dashboard with leading and lagging indicators that are examined quarterly with executive leadership and annually with board sustainability committees Tenth Recommendation: Make Simulation Capability an Important Strategic Asset The study shows that strategic decision-making benefits from simulation.
Utilizing simulation for training managers on circular supply chain dynamics, developing internal simulation capabilities or partnering with specialists, regularly updating models with operational data to improve accuracy, testing strategic scenarios (such as variations in demand, changes in technology, and shifts in regulatory requirements) prior to committing resources are all necessary. Create a dedicated circular economy modeling team that reports to operations or strategy functions. Through improved decision quality regarding multimillion-dollar infrastructure investments, the investment (two to three FTEs and software licenses) yields a return on investment (ROI). Case Study Insights and AI Capabilities Connected The Scania case study identifies specific areas in which the application of AI could enhance circular manufacturing success: Optimization of Magnet Recovery in AI Application 1 Parts P8, P9, and P10 in subassembly S3 containing rare-earth magnets were identified as the highest priority for circular investment by the optimization.
However, quality variability is brought about by magnet demagnetization as a result of mechanical and thermal stress. AI Solution: To predict magnet flux density degradation, use machine learning models that process operational data like temperature cycling, mechanical vibrations, and electrical load patterns. This allows: Selective recovery: Collect only EMs whose magnets exceed minimum performance thresholds Dynamic pricing: Give cores with predicted high-quality magnets a higher buy-back price. Optimize the process by sending high-quality magnets directly to reuse, medium-quality magnets to remagnetization, and low-quality magnets to recycling. Increase magnet recovery rate from 60% (the current simulation result) to 75-80%, resulting in a 20-25% increase in material savings on the component with the highest cost.
Adaptive Disassembly Sequencing is AI Application 2 EOL paths are evaluated by the cascading logic at the product, module, subassembly, and part levels. However, the research is based on the assumption that assembly mirrors fixed disassembly sequences. AI Solution: Agents with reinforcement learning learn the best disassembly sequences taking into account: Actual state of the component (some subassemblies may be worse off than others) Current requirements for inventory (prioritize recovering parts that are scarce) Tooling and operator proficiency levels Time constraints and goals for throughput Compared to fixed sequences, the expected impact is a 15-20% decrease in average disassembly time and a 25-30% reduction in labor costs (C_CD in the cost model).
Third AI Application: Boosting Return Rates Through Predictive Engagement All circular benefits are eliminated by zero return rates, according to the sensitivity analysis. The presumed return rate of 90% necessitates proactive customer involvement. AI Solution: Customers most likely to return cores are identified by predictive models based on: Patterns of vehicle use and wear indicators Behavior in the past (previous returns, service use) Aspects of geography and demographics Response to economic incentives Use personalized incentives and targeted communication campaigns to launch outreach at a time when predicted core quality is at its best for recovery. Expected Impact: Optimized timing will directly address the two parameters from the sensitivity analysis that are the most sensitive. This will result in an increase in return rate from 90% to more than 95% and an improvement in average core quality of 10-15%. Multi-Lifecycle Material Tracking is AI Application 4 As parts undergo second remanufacturing, the temporal analysis revealed effects of quality degradation in the third phase (years 30-45).
However, not all components degrade equally; some may continue to function after the third cycle. AI Solution: Create digital replicas of each component for tracking: Batch and date of manufacture Processes and materials utilized in manufacturing and remanufacturing complete history of operations across all lifecycles Evaluations of quality at each recovery point Instead of making generalizations, machine learning uses individual parts to forecast remaining lifecycle potential. Extending the viable recovery window for parts of the highest quality, possibly allowing selective third lifecycles for 10 to 15 percent of components, and further lowering the material demand beyond the plateau of 14.7% found in the research are the expected effects. Risks, Limitations, and Future Directions for Research While the research provides useful insights, there are a few drawbacks that should be taken into consideration: Simplifications and Model Assumptions Steady-State Economics: Observed volatility is not reflected in the 45-year constant demand and fixed material costs. Business case economics would be significantly impacted by the 152 percent rise in rare earth price that was mentioned in the introduction.
With Monte Carlo simulation, stochastic price modeling would provide confidence intervals for ROI projections. Technology Stasis: The study assumes that the design of electric machines will remain the same for 45 years. In point of fact, advancements in technology, such as advanced materials and magnet-free motors, may render recovery infrastructure obsolete. Technology disruption pathways should be included in scenario planning. Single OEM View: The model does not take into account competitive dynamics, the entry of third-party remanufacturers into the market, or cannibalization, in which remanufactured products replace new sales. Competitive circular economy strategies could be investigated using game-theoretic approaches. Linear Degradation: The assumption that D(t) = a t simplifies wear patterns too much. Stable operation, accelerated wear-out phases, and early infant mortality are all signs of nonlinear degradation in electric machines. Accuracy would be enhanced by utilizing physics-based degradation models such as Arrhenius equations for thermal aging and Paris Law for crack propagation. Risks of Implementation Organizational Opposition: Circular manufacturing necessitates a shift in culture.
Variability brought on by a mix of new and remanufactured components may be able to withstand production teams that have been optimized for efficiency with the same new parts. In technical models, change management and incentive alignment are important but not addressed. Variability in Quality: In practice, the "as good as new" assumption might not be true. Remanufactured parts may have higher failure rates despite rigorous testing, resulting in warranty costs and reputational risks. This risk can be managed with conservative initial implementations (remanufactured parts used in non-critical applications). Evolution of Regulation: The assumptions that are currently in place regarding recycling obligations and recovery incentives might change. Either way, carbon pricing, expanded producer responsibility, or right-to-repair legislation could significantly alter economics. Acceptance in the Market: The research does not model the elasticity of demand for products with remanufactured parts. Pricing strategies and positioning could be better understood through the results of consumer surveys and behavioral economics experiments.
Future Opportunities for Research Multi-Product Portfolio Optimization: Use shared platform analysis to cover all vehicle lineups. Reusing components from different products could significantly boost recovery economics. Model multi-facility configurations with regional variations in costs, regulations, and markets for network design optimization. Determine the ideal locations for distributed recovery capabilities versus specialized remanufacturing centers. Real-Time Adaptive Control: Instead of static strategic planning, integrate simulation models with live operational data streams to enable dynamic decision-making as conditions change. Innovation in the Circular Business Model: Formally model the effects of alternative ownership structures like leasing, performance contracting, and shared mobility on return rates, utilization intensities, and the economics of the system as a whole. Cross-Industry Collaboration: Look into opportunities for shared infrastructure in which a number of OEMs or industries (such as aerospace, industrial, and automotive) jointly operate recovery facilities to benefit from economies of scale. Artificial Intelligence Integration: Quantify the added value of machine learning-enhanced decision-making by explicitly implementing the AI use cases discussed in this essay within the simulation framework.
Conclusion:
AI-Enhanced Circular Manufacturing's Strategic Need Pérez, Lieder, Jeong, and Asif's study is a turning point in manufacturing strategy. They provide executives with the evidence required to justify circular investments by quantifying the benefits of circular practices to the economy as well as the environment through sophisticated simulation. Through 10.6 percent design investment, a cost reduction of 18.6 percent and an improvement in carbon footprint of 38.7 percent were achieved. This demonstrates that circularity provides both economic and environmental value simultaneously—the elusive "win-win" that sustainability advocates promise but rarely rigorously quantify. New infrastructure, distinct organizational capabilities, altered business models, and integrated data systems covering decades-long product lifecycles are all necessary for realizing these benefits. The sensitivity analysis reveals that operational excellence in remanufacturing processes, proactive core acquisition strategies, and predictive product health monitoring are necessary for success. At this point, artificial intelligence becomes not only useful but also necessary. With stochastic return flows, quality-dependent routing, multi-lifecycle material tracking, and cascading decision hierarchies, circular supply chains are more complicated than human cognitive capacity for real-time optimization. Computational intelligence is provided by AI to: Optimized return timing is made possible by accurately predicting individual component degradation. Improve consistency while lowering labor costs with automated quality assessment. Under uncertainty, optimize inventory policies by balancing new and remanufactured parts. Find effective sequences of disassembly that adapt to actual conditions. Maintain the integrity of the data by tracking materials through multiple lifecycles. Utilizing external economic and policy variables to forecast demand The architectural foundation for AI integration is provided by this study's simulation framework. Digital twin concepts are mirrored by agent-based modeling in which individual parts have lifecycle histories. The cost and environmental accounting structures offer optimization algorithms training data.
The capability of scenario testing enables the secure validation of AI-driven decisions prior to their actual implementation. The strategic imperative for manufacturing executives is crystal clear: material scarcity, regulatory pressure, and customer expectations are driving circular practices from voluntary sustainability initiatives to competitive necessities. The question is not whether circular manufacturing should be implemented, but rather how to strategically do so by starting with high-value components, gradually building capabilities, and utilizing AI to manage complexity. This study's winning formula is as follows: Strategic Design Investment: Focus on high-value components made of expensive, limited-supply materials in circular design (10-15 percent additional investment). Phased Implementation: Start with pilot programs on individual subassemblies and scale up as capabilities improve and volumes increase. Integrate sensors and connectivity into the data infrastructure to enable predictive core quality assessment and optimized return timing. Process Excellence: Maximize success rates (the most sensitive parameter) by investing in remanufacturing automation and quality control.
Innovation in the business model: Use proactive core acquisition strategies (buy-back, leasing, and deposits) to guarantee return rates above 85 percent. AI Integration: Implement adaptive optimization by incorporating machine learning capabilities into simulation foundations. Industry Collaboration: Take part in the development of standards and think about sharing infrastructure to take advantage of scale economies. Performance Management: Set circular-specific KPIs to monitor material circularity, recovery success rates, and the return on investment made in circularity. As material realities force manufacturing sector transformation, businesses that follow this formula—which combines strategic design, operational excellence, data infrastructure, and AI-enhanced decision-making—will be in a position to thrive. The road map is provided by the simulation-based decision support demonstrated in this study. The capability for execution is provided by the integration with artificial intelligence.
The design choices that manufacturers make today are the starting point for the future of the circular economy, which is not some far-off theory. In 2040, products that are currently in development will return for recovery. In order to realize their value, the infrastructure, capabilities, and data systems that are required must be built between the years 2025 and 2030. The quantitative basis for those investment decisions is provided by this research. Circularity must encompass all high-value, material-intensive components in order for the electric vehicle revolution to fulfill its environmental potential. A good place to start is with electric machines because of their strategic importance and rare-earth magnets. The scaled-up cost reduction of 18.6% and the material demand reduction of 14.7% result in the avoidance of millions of tonnes of virgin material extraction and billions of dollars in economic value. The question for leaders in manufacturing is whether or not your company will lead the way with circular manufacturing, gaining a competitive advantage, and contributing to genuine sustainability. Or will regulatory mandates and material scarcity necessitate costly, reactive compliance? There are currently simulation tools, AI technologies, and strategic frameworks. Now, vision, dedication, and action are required. Manufacturing's future is circular. Now is the time to act.
More Reading and References M. Pérez, M. Lieder, Y. Jeong, and F. Asif M. A. (2025). a remanufacturing case study for a simulation-based decision support tool for automotive industry circular manufacturing systems employing electric machines. https://doi.org/10.1080/00207543.2025.2464912, International Journal of Production Research. Economy of Circles (2021). https://www.circularity-gap.world/2021. Circularity Gap Report Foundation Ellen MacArthur (2013). Vol. Toward the Circular Economy 1. The Economic and Business Justification for a Quicker Transition. Commission of Europe. (2023). Act on Critical Raw Materials in Europe M. Lieder and A. Rashid (2016). A comprehensive examination of the manufacturing sector in the context of the implementation of the circular economy. 115, 36-51, Journal of Cleaner Production.
No comments:
Post a Comment