Performance Evaluation of Dynamic Scheduling Approaches in Discrete Internal Transport Systems
Discrete or vehicle-based, internal transport (VBIT) systems are used widely in many industrial facilities such as manufacturing plants, warehouses and transhipment terminals. Due to the dynamic and stochastic environments in VBITSs, simple and intuitive online dispatching rules, such as the nearest-vehicle-first (NVF) rule, dominate the scheduling of internal vehicles (or automated guided vehicles-AGVs). However, with the increasing application of computer-aided technologies in VBITSs, it becomes possible to dynamically schedule vehicles in real-time, which potentially outperforms dispatching rules (even with only little look-ahead information of arriving loads). Although real-time scheduling has been researched for external transport, no systematic comparison between scheduling and dispatching approaches for internal transport has been carried out. We are particularly interested how much pre-arrival information must be available in order to profit from real-time scheduling and which other factors (layout, utilization, variability) play a role. This paper proposes five scheduling approaches, three of which are adaptations of static heuristic approaches (insertion, combined heuristic, and column generation heuristic) to dynamic scheduling by using a rolling time horizon. The other two are dynamic assignment approaches with and without look-ahead, respectively. We evaluate and rank the performances of these five dynamic scheduling approaches and two NVF rules (with and without lookahead). We show that dynamic scheduling approaches consistently outperform the NVF rules. The column generation heuristic is recommended for small or medium scale VBIT systems. For large scale VBIT systems, the combined heuristic and the dynamic assignment approach with look ahead are preferred. Results also reveal that the performances of dynamic scheduling approaches are greatly influenced by availability of pre-arrival information, warehouse layouts, load arrival rate and variance.
vehicle-based internal transport dynamic scheduling dispatching logistics
Tuan LE-ANH René B.M.De KOSTER Yu YUGANG
RSM Erasmus University,Rotterdam,The Netherlands
国际会议
北京
英文
2007-05-30（万方平台首次上网日期，不代表论文的发表时间）