Scheduling parallel serial-batch processing machines with incompatible job families, sequence-dependent setup times and arbitrary sizes
研究了金属加工行业中并行串行批处理机的调度问题，考虑工件族不兼容、序列相关准备时间和任意尺寸，提出了混合整数规划和启发式算法以最小化加权总延迟，实验表明方法能有效提升交货可靠性。
The scheduling of (parallel) serial-batch processing machines is a task arising in many industrial sectors. In the metal-processing industry for instance, cutting operations are necessary to fabricate varying metal pieces out of large base slides. Here, the (cutting) jobs have individual, arbitrary base slide capacity requirements (sizes), individual processing times and due dates, and specific material requirements (i.e. each job belongs to one specific job family, whereby jobs of different families cannot be processed within the same batch and thus are incompatible). In addition, switching of base metal slides and material dependent adjustments of machine parameters cause sequence-dependent setup times. All these conditions need to be considered while minimising total weighted tardiness. For solving the scheduling problem, a mixed-integer program and several tailor-made construction heuristics (enhanced by local search mechanisms) are presented. The experimental results show that problem instances with up to five machines and 60 jobs can be tackled using the optimisation model. The experiments on small and large problem instances (with up to 400 jobs) show that a purposefully used batch capacity limitation improves the solution quality remarkably. Applying the best heuristic to the data of two real-world application cases shows its huge potential to increase delivery reliability.