Fast Approximation Heuristics for Multi-Objective Vehicle Routing Problems

The article describes an investigation of the use of fast approximation heuristics for multi-objective vehicle routing problems (MO-VRP). We first present a constructive heuristic based on the savings approach, which we generalize to fit the particular multi-objective nature of the problem. Then, an iterative phase based on local search improves the solutions towards the Pareto-front. Experimental investigations on benchmark instances taken from the literature show that the required computational effort for approximating such problems heavily depends on the underlying structures of the data sets. The insights gained in our study are particularly valuable when giving recommendations on how to solve a particular MO-VRP or even a particular MO-VRP instance, e. g. by means of a posteriori or interactive optimization approaches.