On-line versus Off-line Control with Multi-load Vehicles

Abstract

Computer controlled vehicle-based internal transport systems can be found in many warehouses, manufacturing plants and terminals. The vehicles are usually dispatched on-line, since information on load release times, origins and destinations is usually available only at the last moment. In the theoretical case of off-line control, all load origins, destinations, release instants and transportation times are known in advance. In this case, a mixed integer-programming algorithm can be used to optimize the performance, and heuristics can be used to quickly find good and sometimes optimal solutions.

In this paper, we compare the performance gap between off-line control and on-line dispatching for the case of vehicles with multi-load capacity. The performance is measured as a function of load throughput time, that is, the time a load has to wait when it has been released for transport until it arrives at its destination.

We will show that the performance gap between on-line dispatching and off-line control depends mainly on the throughput. In low throughput environments, vehicles can become idle and park when dispatched on-line. With off-line control, this idle time is used to travel to the next load transportation assignment, hence, compared to on-line dispatching, reducing load waiting times. However, on-line dispatching can outperform off-line control when the actual release times deviate slightly from those that were used to calculate the vehicle routes off-line.