Abstract
This paper proposes a new two-phase hybrid metaheuristic for vehicle routing problem with time windows (VRPTW). The first phase is to minimize the number of routes by means of variable neighborhood search algorithm, while the second phase is mainly aimed at minimizing the total travel distance using tabu search algorithm. Three neighborhood search operators (All-exchange, All-2-opt, All-crossexchange) and two local serarch operators (All-relocate and ejection chain) are designed. To further lower the number of vehicles, the sum-of-squares route sizes is maximized in the first phase. A comparative test is implemented by our algorithm on the basis of 56 benchmark problems proposed by Solomon (1987), the mean number of vehicles and running time of this algorithm is very competitive comparing to previous metaheuristics, showing that the new two-phase hybrid metaheuristic is effective and fast.
