Simulation-based solution for a dynamic multi-crane-scheduling problem in a steelmaking shop

Here, we present a simulation-based solution for a multi-crane-scheduling problem derived from a steelmaking shop. This problem features non-conflict constraint between cranes, station-capacity constraint, and jobs with inaccurate release times and different temporal scheduling objectives. The predictive-reactive rescheduling strategy was applied to solve the problem. The problem was modelled considering different temporal objectives for the jobs and workload objective for the cranes and the model was solved by a heuristic. In the simulation, the jobs were not directly given but generated by a job-prediction method. The cranes' moving behaviours were controlled by a designed crane trajectory solution. Experimental tests were conducted using data from the site and the results show that the proposed crane-scheduling solution provided better scheduling results than both the exhaustive method and the method that is used in the production field. The best predictive spans for the jobs in this specific crane-scheduling problem were found to be 7-14 min. The real-time performance of the crane-scheduling solution is demonstrated to highly guarantee its practicability.