Abstract
The design of a logistics system can be defined as a Location Routing Problem (LRP), which is to find the optimal number and locations of the distribution centers, simultaneously with the vehicle routing to minimize the total system cost. In this research, a three-phase hybrid heuristic approach (LH-SA), combining the Lagrangian heuristic (LH) and simulated annealing (SA), is developed to solve the multi-depot location routing problem (MDLRP). Firstly, the LH is applied to determine the facility location set and the allocation of customers. Then an independent vehicle routing problem (VRP) for each selected facility location is solved by SA. Finally, the SA is used to perform a global search for all routing moves. The performance of the proposed algorithm is tested on two different groups of benchmark instances and compared with other algorithms in the literature. The results show that the LH-SA is competitive and updates 20 best-known solutions.
