Overflow-aware recycling bin location with random utilities: A case study of the commune of Estación Central, Chile

This study tackles the challenge of locating recycling bins to improve municipal waste collection. We develop a lexicographic mixed-integer optimization model with two sequential objectives: maximizing the amount of recycled glass and then minimizing bin overflow. The model considers how people decide where to recycle, which is captured through a multinomial logit (random-utility) choice model with an outside option. Demand is estimated at the census-block level using household counts and average glass-generation rates, and user access is represented through distance-based disutilities. We assume that users may leave glass outside the bins when these are full, reflecting the formation of micro-landfills. The model is applied to a real case in Estación Central, Santiago, Chile. In the case study, we evaluate the current configuration in terms of expected glass capture and overflow, solve the model to optimally relocate the existing bins without adding infrastructure, and conduct a sensitivity analysis by varying the number of bins. Results show that relocating existing bins can raise weekly recycling by about five metric tonnes and completely eliminate overflow without new infrastructure. Configurations with fewer bins also achieve high recycling while keeping expected overflow below 60 kg. These insights can guide local governments in designing more efficient and user-friendly recycling systems.