Data-Driven Network Design and Optimization for Sustainable Semiconductor Reverse Supply Chains under Uncertainty
PhD candidate: Arash Amirteimoori
This PhD develops a data-driven framework for sustainable multi-echelon reverse logistics in semiconductor industry under uncertainty. It addresses AI demand ($1T by 2030), e-waste (62–82 Mt), scarcity, geopolitical risks in NL/EU CE context.
Core contribution: Closed-loop Multi-Echelon Reverse Logistics Network integrating forward/reverse chains. Features AI collection/classification (vision/ML for 10R sorting), recovery facilities (repair, remanufacture, repurpose, recycle, recovery), AV fleets with stochastic routing under weather, traffic, no-entry. Sources: data-centers (1–3 yrs), by-products.
Methodology: MILP for strategic design (locations, capacities, flows min cost/emissions) with RL/MARL and parallel computing for dynamic AV routing, dispatch, allocation. Hierarchical: exact solvers baselines; scalable RL for NP-hard stochastic. Validation: synthetic/empirical data, sensitivity, simulation, benchmarking.
Key contributions: Semiconductor-specific design with stochastic transport, forward-reverse synergies; PRL/MARL for disruptions; net-zero (solar facilities/AVs); simultaneous network-transport opt—absent in prior WEEE MILP/sim studies.
Expected outcomes: Optimized configs with higher recovery rates, reduced energy/kg-km, emission/cost cuts, resilience under disruptions, advancing scalable data-driven circular semiconductor chains.
