Introduction
The manufacturing industry is at a crossroads. The conventional "take-make-waste" model is proving to be unsustainable as global demand for electric vehicles grows at an unprecedented rate and material scarcity puts unprecedented pressure on supply chains. A simulation-based decision support tool that quantifies the economic and environmental benefits of circular manufacturing practices in the automotive industry, specifically focusing on electric machines (EMs) as a case study, is presented as a compelling solution in research that was published by Pérez et al. (2025) in the "International Journal of Production Research." This study comes at a crucial time. The European Union imports approximately 16,000 tonnes of rare-earth materials annually—98% of its needs—primarily from China. In recent years, prices for these essential materials increased by 152 percent, posing grave operational risks for manufacturers. Companies can reduce the average cost of electric machines by 18.6%, material demand by 14.7%, and the cradle-to-gate carbon impact by 38.7% with strategic design investments that add only 10.6%. The Circular Economy Imperative in Manufacturing
Principles of the circular economy are becoming more than just environmental stewardship; they are also becoming a competitive necessity. The research frames circularity as a strategic response to resource scarcity, demonstrating that remanufacturing and reuse practices can meet up to 70% of neodymium demand for electric vehicles by 2050. However, implementation faces significant hurdles: cost-effectiveness uncertainties, volume constraints, regulatory ambiguity, and most critically, the disconnect between early-stage design decisions and end-of-life (EOL) outcomes that won't materialize for over a decade.
The study addresses this temporal disconnect through a sophisticated simulation approach that compresses multiple product lifecycles into quantifiable scenarios. Previously, product designers and strategic planners lacked the ability to compare design decisions made today to recovery operations that will take place more than 15 years in the future. This capability is now available to manufacturers. Evolution of AI and Advanced Analytics in Supply Chain Management
While the Pérez et al. study doesn't explicitly employ artificial intelligence, it represents a crucial evolution in computational decision support for supply chain management that creates the foundation for AI integration. The research employs multi-method simulation modeling combining Agent-Based (AB) and Discrete-Event (DE) approaches—methodologies that are increasingly being enhanced with machine learning and AI capabilities in contemporary applications.
From Static Models to Dynamic Simulation
Static spreadsheets and linear programming models were used in traditional supply chain analysis, which failed to take into account the complexity of circular systems. There were several phases to the evolution: First Generation (1990s-2000s): Numerical optimization models focused on production planning with known capacities and deterministic demand. While these models were capable of determining the most effective policies for inventory, they failed to take into account the dynamic uncertainties that are a part of reverse logistics. Second Generation (2000-2010): System performance was first evaluated using stochastic inputs by probabilistic models and Monte Carlo simulations. However, these remained overly simplistic for capturing the cascading effects of circular material flows across multiple product lifecycles.
Third Generation (2010s-Present): Multi-method simulation combining AB, DE, and System Dynamics (SD) approaches emerged, enabling more realistic modeling of complex interactions. This generation is represented by the Pérez et al. research, which keeps detailed cost and environmental accounting while tracking individual parts through manufacturing, utilization, and multiple recovery cycles. Emerging Fourth Generation (Present-Future): Integration of AI and machine learning with simulation models enables real-time optimization, predictive maintenance scheduling, and adaptive decision-making based on operational data streams. This is a natural progression from the foundation laid by studies like Pérez et al. The Role of Simulation in Enabling AI-Ready Supply Chains
In a number of ways, the simulation framework developed in this study creates the infrastructure for AI integration: Digital Twin Development: By modeling products at the component level with individual agents representing parts, the system creates a digital twin architecture. Each part carries variables including material type, cost, mass, degradation state, and lifecycle history—precisely the data structure needed for machine learning algorithms to identify patterns and optimize decisions.
Data Generation for Training: Running 1,000+ iterations with varying parameters generates extensive synthetic data that can train AI models to predict optimal EOL strategies, forecast return rates, and estimate remanufacturing success probabilities under different conditions.
Scenario Testing Infrastructure: The simulation provides a safe environment to test AI-driven decisions before implementation, addressing one of the key barriers to AI adoption in manufacturing: the high cost of real-world experimentation with physical systems.
AI in Supply Chain Management: Detailed Use Cases
The transition from simulation-based decision support to AI-enhanced supply chain management opens several specific applications relevant to the circular manufacturing context:
Predictive Core Quality Assessment
The research identifies remanufacturing success rate as the most critical parameter affecting economic viability. This can be significantly improved by AI through: - Computer Vision Systems: Automating visual inspection of returned cores using convolutional neural networks trained on defect patterns, reducing the manual labor costs identified in the study's cost model
- Sensor Fusion Analytics: Processing real-time operational data from embedded IoT sensors during the utilization phase to predict component degradation with greater accuracy than the linear degradation model (D(t) = a × t) used in the research
- Predictive Maintenance Scheduling: Using machine learning to identify optimal return timing before critical damage thresholds are exceeded, maximizing reuse potential
The study notes that 80% of electrical steel mass could be channeled toward remanufacturing or reuse in optimal scenarios. AI-enhanced quality prediction could increase this percentage by identifying cores for early return before exceeding damage thresholds.
Dynamic Demand Forecasting and Inventory Optimization
The research assumes constant demand of 10,000 EMs annually over 45 years. In reality, demand fluctuates based on market conditions, technological changes, and economic cycles. AI applications include:
-Time-Series Forecasting: LSTM (Long Short-Term Memory) neural networks can predict demand patterns incorporating seasonality, market trends, and external economic indicators
- Reinforcement Learning for Inventory Policy: Adaptive algorithms that learn optimal stock levels for both new and remanufactured parts, balancing the availability uncertainties inherent in reverse logistics
- Multi-Echelon Optimization: AI systems that optimize inventory across the forward production flow, remanufacturing centers, and customer locations simultaneously
Automated Disassembly Sequencing
The cascading logic developed in the research (product → module → subassembly → part) evaluates EOL paths at each hierarchy level. AI can enhance this through:
- Robotic Process Automation: Vision-guided robots using deep learning for part recognition and adaptive disassembly in the face of product variations
- "Sequence Optimization": Algorithms that use actual processing times rather than estimates to learn effective disassembly paths - Anomaly Detection: Identifying unexpected component configurations or damage patterns that require specialist intervention
The study estimates staffing requirements using a linear function (workers = demand/1000 + 1). AI-driven automation could significantly reduce these labor requirements while improving consistency.
Supply Chain Network Optimization
The research models a simplified network with single production units and remanufacturing centers. AI enables:
- Location Intelligence: Optimization algorithms determining optimal placement of recovery centers considering transportation costs, regional regulations, and material flows
- Dynamic Routing: Real-time logistics optimization responding to changing material availability and transportation costs
- Supplier Risk Assessment: Machine learning models predicting supply disruptions based on geopolitical factors, particularly relevant given EU dependence on Chinese rare-earth materials
The Provider Landscape of Cutting-Edge AI in Supply Chain Management There are a number of technology providers working on AI solutions that can be applied to the problems with circular manufacturing that were found in the research: Enterprise Platforms
Blue Yonder, formerly JDA Software, provides demand forecasting and inventory optimization AI-powered supply chain solutions. Their Luminate platform could enhance the demand prediction aspects of the circular supply chain model, moving beyond the constant demand assumption.
For demand sensing and supply response management, machine learning is incorporated into SAP Integrated Business Planning. Integration with circular logistics would require extensions to handle reverse material flows and quality-dependent routing decisions.
Oracle Supply Chain Management Cloud provides AI-driven predictive analytics but primarily focuses on forward supply chains. Modules for core acquisition and quality assessment would be required for circular system adaptation. Specialized Circular Economy Solutions
Maersk's Trade Lens (blockchain-based supply chain platform) demonstrates how distributed ledger technology combined with IoT can provide the product tracking capabilities identified as critical in the research. The study emphasizes that product health indicators and tracking through multiple lifecycles are prerequisites for successful circular operations.
AI is used for supply chain visibility and risk management in the IBM Sterling Supply Chain Intelligence Suite. Their asset tracking capabilities could support the individual part agent tracking implemented in the simulation model.
New AI-Native Service Providers The integration of AI analytics into manufacturing execution systems by Plex Systems, which is now part of Rockwell Automation, is important for the shopfloor remanufacturing operations that the study identified as requiring additional research. Clear Metal, which was purchased by project 44, developed machine learning techniques for predicting port congestion and container tracking, which can be used to optimize reverse logistics in circular systems. However, there is a significant gap: none of these providers provide comprehensive circular manufacturing solutions with integrated forward and reverse flows, quality-dependent routing, and material tracking across multiple lifecycles. This represents a significant commercial opportunity.
Case Study Analysis: Scania Electric Machine Circularity
The research uses Scania's electric machine as the primary case study, providing concrete data to validate the simulation approach. The EM consists of 12 parts assembled into modules, with a designed utilization phase of 15 years and triangular distribution (13.5, 15, 16.5 years) to model variability.
Experiment 1: Cost Minimization Through Design Investment
The optimization identified specific parts for targeted design investment:
- P1, P8, P9, P11, and P12 are intended for reuse. - Part P₁₀: Designed for remanufacturing
- Subassemblies S₃ and S₄, Module M₂: Designed for integrated reuse
Instead of a sweeping investment in all components, this is a strategic allocation. The 10.6% additional design investment concentrated on high-value components containing expensive materials (particularly rare-earth magnets) yielded the 18.6% average cost reduction.
Key Insight for Practitioners: Not all components are worthy of investment in circular design. The simulation demonstrated that concentrating on subassembly S3, which includes components P8, P9, and P10, yields the greatest return on design investment. This targeted approach contrasts with generic design-for-circularity guidelines that treat all components equally.
Experiment 2: Circular Investment Minimum Viable The following was the model's recommendation for maximizing profitability while minimizing design investment: - Subassembly S3: Designed to be used again - All other parts: Designed for recycling
This creates a hybrid approach where the OEM captures value from high-potential components (S₃) while ensuring end-of-life material recovery through external recyclers for remaining parts. The 8.4% design investment still generated positive economic returns compared to the linear baseline, demonstrating that partial circularity can be commercially viable even with constrained budgets.
Practical Implication: Companies facing resource constraints can implement circularity incrementally, focusing first on high-value subassemblies rather than attempting complete system transformation. Risk is reduced and early successes are achieved with this phased approach, gaining organizational support for more extensive implementation. Lifecycle Phase Analysis
The temporal analysis revealed distinct performance patterns across three 15-year phases:
Phase 1 (Years 0-14): Pure ramp-up using virgin materials. Since the products haven't been used again, there are no benefits from the circular economy. This is the time in an investment where long-term commitment is required despite the lack of immediate returns. Phase 2 (Years 14-30): First recovery cycle. Cost savings peaked at 23.6% and cradle-to-gate impact reduction at 46.0% as cores returned in relatively good condition. Remanufacturing success rates remained high due to products operating within designed lifetime parameters.
Phase 3 (Years 30-45): Second recovery cycle. Benefits declined to 13.2% cost savings and 33.0% impact reduction as parts underwent second remanufacturing. The quality degradation across two lifecycles increased recycling activities, demonstrating the practical limits of multiple remanufacturing cycles.
Critical Finding: The declining benefits in Phase 3 highlight that infinite circularity is unrealistic. The study's assumption of two full lifecycle recoveries before recycling appears aligned with material fatigue realities. This constrains total material demand reduction to the 14.7% observed rather than more optimistic scenarios assuming indefinite circularity.
Material-Specific Recovery Rates
Eight material categories with varying recovery potential were tracked in the simulation: - Electrical Steel: 80% remanufactured/reused, 20% recycled (1,500+ tons total volume)
- Magnets (Rare-Earth NdFeB): ~60% remanufactured/reused, 40% recycled (critical supply chain risk mitigation)
- Standard Steel: 56% remanufactured/reused, 44% recycled
- Copper: 0% remanufactured or reused, 100% recycled (despite its high value and technical limitations) The inability to remanufacture copper windings due to technical constraints represents a design opportunity. Advanced joining techniques or modular winding designs could potentially shift some copper recovery from recycling to reuse, capturing higher value.
Sensitivity Analysis Insights
Experiment 3 systematically varied seven parameters to identify critical factors:
"Most Sensitive Parameters": 1. Remanufacturing Success Rate: Benefits to the environment persisted despite economic benefits being eliminated by deviations below -75%. 2. Lifetime Minimum/Maximum: Shorter lifetimes increased savings by returning less-degraded cores
3. Degradation Factor: Slower degradation improved recovery potential
Less Sensitive Parameters:
4. Cost of core acquisition 5. Critical damage threshold
6. Return rate (though zero return rate eliminated all circular benefits)
Managerial Takeaway: Investment priorities should focus on (1) developing remanufacturing process expertise and quality control, (2) implementing predictive maintenance to optimize return timing, and (3) product design for durability with modular degradation-resistant architectures.
Additionally, the sensitivity analysis revealed that reducing return rates to zero completely eliminates circular benefits (representing complete customer retention of used products). This underscores the criticality of the business model—circular manufacturing requires proactive core acquisition strategies, potentially through:
- Deposit-refund systems
- Buy-back guarantees incorporated into initial sales
- Product-as-service models retaining OEM ownership
- Regulatory mandates for return (Extended Producer Responsibility)
User Insights: Operationalizing Circular Manufacturing Systems
Beyond the quantitative findings, the research provides critical operational insights for manufacturing organizations implementing circular practices:
Needs for the Infrastructure The study assumes remanufacturing and reuse conducted by the OEM at dedicated facilities separate from primary production. This organizational design decision carries significant implications:
Capital Investment: The research acknowledges CAPEX investments are typically 2-6× the design investment depending on whether existing facilities are adapted versus greenfield construction. Using the 10.6% design investment and 6× multiplier, with 16% cost savings per EM, approximately 31,000 units must be recovered to achieve break-even (assuming 10% material price inflation over 15 years).
Capacity Planning: The staffing equation (workers = demand/1000 + 1) provides initial guidance for planning, but it makes the assumption that efficiency will remain constant. Real operations will face learning curves, equipment downtime, and variability in core quality affecting throughput.
Location Strategy: The model includes 3,000 km forward/reverse logistics and 500 km internal logistics assumptions. Optimal network design would minimize total transport while maintaining sufficient volume at each facility—a classical facility location problem complicated by uncertain return flows.
SCANIA MILITARY TRUCK
Data and Technology Dependencies
The availability of operational data is emphasized as a prerequisite for circular success in the research: Product Health Monitoring: Parameter estimation is required for the degradation model (D(t) = a t) and the declining success rate function (R(t) = -s t + b). Without embedded sensors and data connectivity during utilization, these must rely on conservative estimates, reducing economic viability.
Traceability Systems Requires sophisticated database management and identification (RFID, QR codes, or blockchain-based systems) to follow individual components through multiple assembly and disassembly cycles. The simulation's agent-based architecture where each part carries lifecycle history reflects this requirement.
Quality Assessment Technology: Visual inspection and technical evaluation labor costs constitute significant recovery expenses (C_eval in the cost model). Computer vision and automated testing investments could reduce these costs while improving consistency.
Organizational Capabilities
Successful circular implementation requires capabilities distinct from traditional manufacturing:
Process Engineering Expertise: Remanufacturing involves greater variability than new production. The study notes that different damage levels require different processing times, creating scheduling complexity. Organizations need flexible, skilled workforces rather than highly specialized assembly line operations.
Supply Chain Coordination: Managing both forward (supplier → production → customer) and reverse (customer → recovery → production) flows simultaneously requires integrated planning systems and coordinated incentives across functions.
Customer Relationship Management: Achieving the assumed 90% return rate requires proactive engagement, possibly through service contracts, buy-back programs, or deposit systems. This shifts the OEM-customer relationship from transactional to longitudinal.
Regulatory and Market Context
The research situates findings within European regulatory frameworks:
Extended Producer Responsibility: Directive 2000/53/EC on End-of-Life Vehicles creates legal obligations for recovery. The simulation's mandatory recycling cost for non-recoverable parts reflects this regulatory reality.
Critical Raw Materials Act: Recent EU legislation designating rare-earths as critical materials may introduce additional recovery mandates or incentives, improving the business case beyond the modeled scenarios.
Remanufactured Product Acceptance: The study assumes remanufactured parts are "as good as new" but compete in spare parts markets requiring lower pricing. Customer acceptance of remanufactured products remains a market barrier not captured in technical models.
Managerial Implications and Strategic Recommendations
The research generates several actionable recommendations for manufacturing executives:
Strategic Design Decisions
Recommendation 1: Implement Targeted Circular Design Investments
Rather than attempting comprehensive circular design across all products, focus initial investments on:
- High-value components with expensive materials (rare-earth magnets, precious metals)
- Parts with established quality control procedures and remanufacturing processes - Components with high failure independence (where single-part replacement is feasible)
The Scania case demonstrates that 10.6% incremental design investment concentrated on specific subassemblies generates substantial returns. Utilizing material costs, criticality, and technical feasibility, spreadsheet-based prioritization can identify ideal targets.
Recommendation 2: Establish Design-for-Remanufacturing Principles Early
Due to the temporal gap between design decisions and EOL outcomes (10-15+ years), circular considerations must be incorporated into initial development: - Modular architectures enabling non-destructive disassembly
- Standardized fasteners and documented disassembly sequences
- Material selection favoring durability and repeated processing
- Wear components that can be accessed for replacement and inspection
Recommendation 3: Develop Product Health Monitoring Capabilities
The success rate sensitivity analysis demonstrates that predictive core quality assessment is critical. Priorities for investments include: - Embedded sensors for performance monitoring during utilization
- Data connectivity infrastructure for real-time health tracking
- Capabilities for analytics that convert operational data into estimates of remaining useful life - Customer-facing interfaces promoting timely returns before critical damage
Operational Implementation
Recommendation 4: Pilot Before Scaling
Implement phased rollouts due to the substantial CAPEX requirements (two to six design investments) and operational uncertainties: Phase 1 (Years 1-3): Pilot program to develop process expertise and validate economic assumptions with a limited product scope (such as Subassembly S3 only). Phase 2 (Years 4-7): Scaled implementation broadening recovery scope as return volumes rise and learning curves lower costs. Phase 3 (Years 8+): Full-scale operations achieving steady-state economics as first-generation products complete utilization phases
This phased approach manages financial risk while building organizational capabilities.
Recommendation 5: Form Partnerships for Reverse Logistics
The research assumes 90% return rates, achievable only through proactive collection strategies. Consider:
- Partnerships with dealers and service networks for core collection
- Incentive structures (deposit-refunds, trade-in credits) encouraging returns
- Regional consolidation centers minimizing transport costs
- Information systems providing return visibility and forecasting
Recommendation 6: Invest in Remanufacturing Process Technology
The staffing equation indicates labor intensity in recovery operations. Automation investments should target:
- Disassembly robots with adaptive capabilities handling product variations
- Automated inspection systems using computer vision for quality assessment
- Cleaning systems that make cleaning less laborious and improve consistency - Flexible reassembly lines accommodating mixed new/remanufactured components
Business Model Innovation
Recommendation 7: Explore Product-as-Service Models
The research assumes traditional sales with optional buy-back. Alternative models could improve circular viability:
Subscription/Leasing: Retaining product ownership ensures 100% return rates while providing continuous performance data. This eliminates return rate uncertainty but requires different financial structures.
Performance-Based Contracting: Selling "operational hours" rather than products aligns OEM incentives with product longevity and efficient maintenance, naturally supporting circular practices.
Marketplace Platforms: Creating channels for second-life products (e.g., remanufactured EMs for lower-specification applications) expands addressable markets beyond spare parts.
Recommendation 8: Develop Circular Supply Chain Partnerships
Circularity with a single OEM is constrained by volume. Industry collaboration could improve economics:
- Shared remanufacturing facilities processing multiple OEM products
- Standardized component designs enabling cross-brand reuse (particularly for commodity parts)
- Joint ventures with recyclers ensuring material recovery infrastructure
- Industry-wide data standards for product health monitoring
Performance Measurement and Continuous Improvement
Recommendation 9: Establish Circular Economy Metrics
Circular performance cannot be adequately captured by conventional operational and financial KPIs. Implement:
Economic Metrics:
- Total cost of ownership (TCO) including acquisition, operation, and EOL handling
- Circular cost differential (cost savings per unit from remanufacturing vs. new production)
- Return on circular investment (ROCI) tracking design and infrastructure investments against operational savings
Environmental Metrics:
- Material circularity index (% mass remanufactured/reused vs. virgin)
- Cradle-to-gate carbon intensity per unit (including all lifecycles)
- Strategic supply chain risks are weighted by material criticality scores. **Operational Metrics:**
- Remanufacturing success rates by component and damage level
- Core return rates and timing distributions
- Recovery processing times and labor productivity
- Quality conformance of remanufactured vs. new products
Recommendation 10: Implement Simulation-Based Scenario Planning
The research demonstrates simulation value for strategic decision-making. Organizations should:
- Develop internal simulation capabilities or partner with specialists
- Regularly update models with operational data improving accuracy
- Before investing resources, test strategic scenarios (such as shifts in demand, technological advancement, and regulatory changes). - Use simulation for training managers on circular supply chain dynamics
Limitations and Future Research Directions
The study acknowledges a few limitations that require further investigation: Technology Evolution: The 45-year steady-state assumption doesn't account for EM design updates, new materials, or process innovations. Long-term projections would be more realistic if models included product generations and technological advancements. Market Dynamics: The market realities are oversimplified by constant demand of 10,000 units annually. Applicability would be enhanced by integration with econometric forecasting and market diffusion models. Competitive Dynamics: Single-OEM perspective ignores competitive responses, third-party remanufacturers, and cannibalization effects where remanufactured products displace new sales.
Material Price Volatility: The observed price fluctuations of rare earths are not reflected in fixed material costs (the 152% increase noted in the introduction). Supply chain risks would be better captured by stochastic price modeling. Shopfloor Operations: High-level process modeling needs extension to detailed production scheduling, quality control procedures, and failure mode analysis for remanufacturing operations.
Network Optimization: Single production site and recovery center should be extended to multi-facility network optimization considering regional variations in costs, regulations, and market conditions.
AI Integration: While the simulation creates infrastructure for AI applications, actual implementation of machine learning for predictive quality assessment, dynamic routing, and adaptive control requires further development.
Conclusion: Toward AI-Enhanced Circular Manufacturing
The research by Pérez et al. represents a significant advance in quantifying circular economy benefits in manufacturing contexts. The study gives manufacturing leaders the tools they need to make informed investments in circular practices by bridging product design decisions with long-term supply chain outcomes through sophisticated simulation. The 18.6% cost reduction, 38.7% carbon footprint improvement, and 14.7% material demand reduction achieved through 10.6% design investment demonstrate that circularity can deliver simultaneous economic and environmental benefits—crucial for justifying investments in resource-constrained organizations.
The implementation of these advantages, on the other hand, necessitates significant organizational transformation in the form of new infrastructure, distinct capabilities, altered business models, and integrated data systems. Remanufacturing success rates, product lifetimes, and return rates are important parameters that require operational excellence, according to the sensitivity analysis. Looking forward, the integration of AI and machine learning with the simulation-based framework offers promising opportunities. Automated disassembly systems, dynamic logistics optimization, adaptive inventory management, and predictive core quality assessment all have the potential to further enhance the business case while simultaneously lowering the risk of implementation. For manufacturing executives, the strategic imperative is clear: circular practices are transitioning from environmental compliance activities to competitive necessities. Companies that are able to close material loops profitably are increasingly favored by material scarcity, regulatory pressure, and customer expectations. 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 advanced analytics to manage complexity. The simulation-based decision support demonstrated in this research provides a roadmap. Organizations that invest in these capabilities now—combining strategic design, operational excellence, data infrastructure, and emerging AI technologies—will be positioned to thrive in the circular economy that material realities are forcing upon the manufacturing sector.
This essay synthesizes findings from: Pérez, M., Lieder, M., Jeong, Y., & Asif, F. M. A. (2025). A simulation-based decision support tool for circular manufacturing systems in the automotive industry using electric machines as a remanufacturing case study. International Journal of Production Research. https://doi.org/10.1080/00207543.2025.2464912
