In the traditional vehicle routing problems, customer demands are usually assumed that they can not be split.
That is to say, a customer can only be severed by a vehicle. In fact, split delivery requires fewer vehicles, which reduces
transportation costs. This paper analyzes the solution’s characteristics of split delivery vehicle routing problem, and also
proves the situations that customers’ demands can not be split. A clustering algorithm meeting the solution’s characteristics
is designed to solve this problem. Compared with ant colony algorithm and tabu search algorithm, the proposed algorithm is
demonstrated to obtain optimal solution more effectively through the simulation, and it is an effective method to solve split
delivery vehicle routing problem.