Prefabricated construction is widely adopted in the current bridge construction project, especially offshore bridges. The realistic requirements of large quantities of prefabricated parts and tight delivery schedules make it extremely challenging to develop optimal scheduling. We address a new customer order scheduling on a serial-batch machine (COS-SBM) to reduce the sum of inventory holding costs of finished jobs and tardiness costs of orders in precast bridge construction. In the COS-SBM problem, all jobs with incompatibility in orders need to be divided into batches, which are then scheduled for processing on a serial-batch machine. We develop a mixed-integer linear programming model to formulate this new problem. Since the COS-SBM problem is NP-hard, we propose a genetic algorithm based on a novel batch sequencing and forming encoding method (GA-BSFE), which makes the scheduling and batching decisions simultaneously to enhance its exploration. Moreover, we design an efficient three-stage heuristic based on the order weighted modified due date rule and batch weighted longest processing time rule. The three-stage heuristic is introduced into the initiation of GA-BSFE to enhance its exploitation. Finally, a set of instances generated based on the realistic production of precast girders is tested to validate the effectiveness of GA-BSFE. The performance analysis suggests that GA-BSFE is the most appropriate for the COS-SBM problem.
Read full abstract