In this paper, we deal with the physical distribution problem that comprises grouping and assigning deliveries to a heterogeneous fleet of vehicles aiming to minimize the total freight cost. This problem arises in several practical situations in which the customers to be served are geographically close enough to each other such that distances traveled between consecutive stops can be regarded as irrelevant when compared to the overall vehicle costs, and thus not considered when establishing the freight rate to be paid to third-party carriers. This work proposes a generalized bin-packing problem resolution method with additional constraint found in logistics and freight transportation practical situations. A VNDS (Variable Neighborhood Decomposition Search) meta-heuristic proposed to solve this problem. Computational experiments conducted using benchmark instances problems from the literature evidence that our heuristic produces improved results (optimal solution or near lower limit in all instances) in very short times (5.10% faster than the VNS).
variable neighborhood decomposition search; bi-dimensional variable sized bin-packing problem; meta-heuristic