| Summary |
This study investigates the Optimal Districting Planning Integrated with Vehicle Routing Problem Considering Overlapping Service Regions, which is an extension of the districting planning problem. We are interested in solving optimal general overlapping service regions districting planning problem, which consider the general overlapping service regions, pre-determined depot and customer point coordinates, and stochastic demands in order to help logistics companies make the long-term decision. The concept of “general overlapping service regions” in the study is that multiple service areas can overlap each other to improve the flexibility of vehicle routing and fleet deployment, and reduce the total cost of vehicle routing operations for logistics companies.
This study uses Monte Carlo Simulation (MCS) to realize stochastic demands, and then uses the genetic algorithm proposed by Hsieh (2018) to solve the optimal vehicle routing planning in a single day, and get a total vehicle routing operating cost. Through MCS to simulate multi-day consumer demands and use above method to solve the optimal vehicle routing planning for each day. Then, compute mean as the long-term expected total operating cost, which is the evaluation benchmark of one service region planning. This study uses greedy algorithm, tabu search algorithm and tabu search with optimal computing budget allocation method to solve the general overlapping service region planning. The experimental results show that there is no significant difference between the mean of three algorithms in the statistical paired t test. However, in terms of mean, the average long-term expected total operating cost of the solution obtained by greedy algorithm is lower, and the total time of this algorithm does not exceed 5.55% of the two kind of tabu search algorithms. This study analyzes different number of customers based on randomly generated sample questions, and results show that the computation time of the greedy algorithm grows linearly with the number of customer. Therefore, even if it is a large-scale experimental problem, the greedy algorithm can obtain a solution of service region planning with excellently quality effectively in a reasonable time. According to the experimental results, when solving the problem of “Optimal districting planning integrated with vehicle routing problem considering overlapping service regions”, we recommend decision-makers to use the greedy algorithm as an effective decision-making method. |
