Abstract:
This thesis addresses the problem of scheduling jobs on two stage unrelated parallel machines with job sequence dependent setup times so as to minimize the defined objective function in copper-wire industry. The study of the problem stemmed from a real copperwire manufacturer company, Erbakır Elektrolit. The mathematical representation of the problem considering the industry specific parameters, decision variables and constraints is presented. Since the presented problem is NP-hard due to several factors, a heuristic algorithm is proposed and tested with real life data sets. The proposed heuristic algorithm is coded in company’s live system using SAP’s internal programming language ABAP. The heuristic is composed of two stages, which are most difficult to schedule in Erbakır Elektrolit in real life. At first stage, simulated annealing technique is implemented after identifying an initial solution which also considers the properties of the problem. Parameters of the simulated annealing approach are tuned with one set of data and several test runs. At the second phase, the second stage of the problem is solved again with another heuristic approach given the input of the solution in the first stage. Heuristics are run with two sets of comparison data and the results are compared with the planners plan in the system. Results of the proposed heuristic overcome the result of the planner.
