Optimal siting of photovoltaic waste treatment facilities in China: A mixed-integer linear programming approach

Abstract

Facing the first wave of large-scale photovoltaic (PV) module decommissioning as the global leader in installed capacity, China urgently requires efficient nationwide PV waste recycling. This study projects regional PV waste volumes by 2050 using a Genetic Algorithm-optimized Backpropagation Neural Network (BPNN), revealing a distinct ring-shaped spatial distribution. Integrating multi-source data (transport, processing, operation, and other parameters), a Mixed-Integer Linear Programming (MILP) model is developed to minimize total costs and optimize the national siting of PV waste treatment facilities, comparing results with existing China-certified Waste Electrical and Electronic Equipment (WEEE) treatment enterprises. The analysis identifies 91 optimal locations for PV waste treatment facilities, concentrated in the Northwest and North China waste-intensive zone, utilizing rail for cost-effective transportation. Some large hubs achieve significant economies of scale, while smaller local plants face scale limitations despite cost benefits. Although 76 of the current 109 certified WEEE treatment enterprises can potentially upgrade for PV waste treatment, facility shortages may persist in some regions, necessitating mobile collection and temporary transfer solutions. Therefore, to address the research gap in sub-provincial PV waste generation and transport networks, this study establishes a decision-support framework aimed at achieving efficient, economical, and sustainable recycling by projecting China's future waste distribution pattern and optimizing the siting of treatment facilities for 2050.