| Summary: | 碩士 === 國立東華大學 === 運籌管理研究所 === 103 === In order to improve service quality and satisfy specific delivery requests from different kinds of customers, recently wholesalers are tending to provide more efficient and convenient distribution services rather than follow traditional approaches. Customers may have different preferred hours, and logistics service providers must deliver goods in different time windows. According to the pre-determined delivery routes and schedules, how should city logistics service providers react, respond, and mitigate the delays occurred at the upstream inbound intercity delivery vehicles?
This research is motivated by Ghiani et al. (2004) and Rabah and Mahmassani (2002). The former study mentioned that logistics and distribution systems have been recognized as the key contributors in global, national, and local economy. The later study analyzed the opportunities offered by information and communication technologies (ICT) to operate and control a logistics system in real-time with progressively reduced costs.
The study starts from a typical vehicle routing problem with time windows and then considers a real-time vehicle routing and dispatching optimization model with different delay mitigation strategies during the distribution processes. We solve a small case problem to check the feasibility with the optimization software LINGO, and then further apply Genetic Algorithm to solve a large-scale network problem. System performances based on four types of delay mitigation strategies are analyzed, such as: (1) S0: Do nothing (i.e. simply ignore the upstream delay); (2) S1a: Selected nodes with higher demand first served, and S1b: Selected nodes with tighter time windows first served; (3) S2a: Increasing homogenous delivery vehicles in service, and S2b: Increasing heterogeneous delivery vehicles in service; (4) S3a: A hybrid strategies with increasing homogenous vehicles and selected nodes with tighter time windows first served, and S3b: A hybrid strategies with increasing heterogeneous vehicles and selected nodes with tighter time windows first served.
Findings in our numerical examples show that the S1a strategy could not easily reach a feasible solution, and the solutions with strategies S2a and S3a outperform other strategies in the most studied cases. But if the upstream delay keeps increase, the strategy S3b becomes the most effective approach due to the more complexity environment.
|
|---|
