A simulated annealing-based approach for a real case study of vehicle routing problem with a heterogeneous fleet and time windows

We propose a meta-heuristic algorithm for addressing a real case of vehicle routing problem with a heterogeneous fleet and time windows. Our proposal consists of generating an initial feasible solution using the solution of a travelling salesman problem and, subsequently, refining it through simulated annealing. In this work, we evaluate the suitability of our development to planning the distribution of a regional distribution centre of a food company to customers within the urban perimeter of Cali, Colombia. Following the company's customer satisfaction goals, our primary aim is to route to reach a service level of 95% while reducing operational costs. Our results on our real case study show that our meta-heuristic algorithm finds better routes than current ones, in terms of costs and service levels, in a relatively short time.