Performance Evaluation of Real-time Scheduling Approaches in Vehicle-based Internal Transport Systems

This paper studies the performance of static and real-time scheduling approaches in vehicle-based internal transport (VBIT) systems, which can be found in manufacturing and warehouse facilities. We propose three heuristic approaches for static VBIT problems (insertion, combined and column generation), extend them to a dynamic, real-time setting and compare their performance over a rolling time horizon. This time horizon can be seen either as a fixed-time interval in which advance information about loads’ arrivals is available, or as a fixed number of loads which are known to become available in the near future. We also propose two dynamic assignment approaches: with and without look-ahead, respectively. Performance (primarily average load waiting time) of the above five dynamic scheduling approaches is compared with two nearest-vehicle-first rules (with and without look-ahead), which are the best vehicle dispatching rules known from literature and which are commonly used in practice. Experimental results show that, if sufficient prior information is available, our dynamic scheduling approaches consistently outperform vehicle dispatching rules. Results also reveal that guide-path layout, load arrival rate and variance, and amount of load pre-arrival information have strong impacts on the performance of vehicle control approaches. Column generation or the combined heuristics are recommended in small or medium-scale VBIT systems, whereas for large scale VBIT systems, both the combined heuristic and the dynamic assignment approach with look ahead perform best.