A Bi-level Decomposition Algorithm for Multi-factory Scheduling With Batch Delivery Problem

This paper introduces a multi-factory scheduling with batch delivery problem. A novel mixed-integer programming model is proposed to minimize the sum of total tardiness, holding and batching costs. A bi-level decomposition algorithm (BLDA) is developed involving two sub-problems: scheduling problem in the upper level and batching problem in the lower level. Four versions of the BLDA are created by combinations of CPLEX and simulated annealing in both levels, which interactively collaborate until the algorithm converges to a solution. The BLDAs are examined on several random and real-life test instances. A statistical analysis is performed by comparing the BLDAs’ solutions with the exact minimum and lower bound values of the total cost. The results indicate that about all versions of the developed BLDA provide high quality solutions for real-world zinc industry problems as well as generated instances in a reasonably short time. Finally, some managerial insights are derived based on sensitivity analysis.

Multi-objective multi-factory scheduling

This paper introducesa multi-factory scheduling problem with heterogeneous factories and parallel machines. This problem, as a major part of supply chain planning, includes the finding of suitable factory for each job and the scheduling of the assigned jobs at each factory, simultaneously. For the first time, this paper studies multi-objective scheduling in the production network in which each factory has its customers and demands can be satisfied by itself or other factories. In other words, this paper assumes that jobs can transfer from the overloaded machine in the origin factory to the factory which has fewer workloads by imposing some transportation times. For simultaneous minimization of the sum of the earliness and tardiness of jobs and total completion time, after modeling the scheduling problem as a mixed-integer linear program, the existing multi-objective techniques are analyzed and a new one is applied to our problem. Since this problem is NP-hard, a heuristic algorithm is also proposed to generate a set of Pareto optimal solutions. Also, the algorithms are proposed to improve and cover the Pareto front. Computational experiences of the heuristic algorithm and the output of the model implemented by CPLEX over a set of randomly generated test problems are reported.