• Joint MSW optimisation under stochastic waste generation and covariance. • JRP-inspired coordination minimises expected costs and operational risks. • Standard metaheuristic reliably solves the complex stochastic MINLP. • Adaptive system reconfiguration outperforms static MSW setups. • Yields actionable policy insights tailored to local waste covariance. This paper addresses the joint optimisation of bin location, type selection, allocation, and collection frequency for urban waste management, explicitly considering stochastic waste generation. Waste originates from multiple nodes, also potential bin sites, with bins having specific capacities and installation/operating costs. Bin emptying is coordinated via a novel mechanism inspired by the periodic-review Joint Replenishment Problem (JRP). The comprehensive cost function covers bin placement, waste delivery, JRP-based collection ordering (major/minor), penalties for bin over/underutilisation and truck overcapacity, and disposal facility operations. The objective is to find optimal bin locations, types, allocations, and collection frequencies that minimise the long-run expected total system cost rate. Given the problem’s inherent complexity, a robust solution framework is employed, embedding an efficient heuristic algorithm (optimising collection frequency) within a standard metaheuristic to navigate the combinatorial search space. Numerical experiments evaluate the solution method’s performance and the model’s behaviour under diverse operating conditions. Key results demonstrate that the covariance structure of waste generation significantly dictates optimal system configuration and overall cost-effectiveness; high inter-node correlation, for instance, necessitates more flexible capacity provision and frequency planning to effectively mitigate amplified penalty risks. The study also reveals strong system sensitivity to variations in truck and bin capacities, as well as key cost parameters. Crucially, allowing the system to adapt its configuration consistently outperforms static approaches, highlighting the significant value of dynamic resource reallocation. These findings offer valuable, data-driven managerial insights for designing robust and cost-effective waste management systems, particularly when operating under inherent uncertainty.
Bertolini et al. (Wed,) studied this question.