An optimization model for multi-factory remanufacturing planning considering fatigue risk

The Multi-factory Remanufacturing Process Optimization Problem (MRPOP) involves product allocation, disassembly scheduling, workstation activation, recovered-subassembly transportation, and human-related operational risks. This study proposes a human-centric chance-constrained optimization framework for MRPOP under a workstation–worker binding assumption. A mixed-integer programming model is developed to maximize expected system profit while controlling fatigue-threshold violation probabilities at active workstations. To solve the resulting stochastic discrete problem, a Risk-Aware Adaptive Large Neighborhood Search with Path Relinking (ALNS-PR) is designed with risk-aware operators, skill-compatible repair, progressive stochastic evaluation, and elite-guided path relinking. Experiments on 10 multi-scale cases show that ALNS-PR reproduces exact solutions on small instances and obtains high-quality feasible solutions for larger cases. It improves the CPLEX incumbent feasible solution by up to 19.6%. Monte Carlo validation reduces fatigue-threshold violation rate to 4.2% (95% CI: 3.8–4.6%), compared with 49.8%. Sensitivity analysis indicates that α=0.05 provides a reasonable profit-risk trade-off.