A hybrid metaheuristic for smart waste collection problems with workload concerns

Abstract

In this paper, a hybrid metaheuristic is developed to solve the smart waste collection problem with workload concerns. It is composed by: (i) a look-ahead heuristic aiming at deciding the days in which collection is necessary and which bins need to be collected (must-go) considering the present bin fill levels and future bin fill level predictions; and (ii) a simulated annealing/neighborhood search algorithm to choose the bins that are profitable to collect and the best route(s) to visit the bins. This algorithm was developed to find solutions within a relatively short amount of time (we considered two hours as reference), as required for practical operations. The proposed hybrid metaheuristic is applied to randomly-generated test instances of sizes up to 500 bins and to a real case study of recyclables collection, leading to results that demonstrate its effectiveness and usefulness in practice when dealing with large-size instances. For the real case study, involving a major waste management company in Portugal, the profit achieved by the hybrid metaheuristic is at least 45% higher than the profit obtained by the company, and, at the same time, compliance with maximum shift duration and route workload balance is clearly better when the metaheuristic is used than in current operations.