Scheduling model in foundry

Abstract

This thesis concerns the development of production scheduling model in automotive foundry. The production system in this foundry is a continuous process with batch production system. The batch size is identified from customer order. At present, the schedules are planned by the production manager, so that, they are totally relied on the production manager’s experience and skills. This inconsistent system is leading to delivery problem - delay delivery and high % missing shipment which is directly impact to customer satisfaction and business opportunity. The developed scheduling model is developed from dynamic programming technique and backward scheduling technique. The dynamic programming technique is applied into the front process, while the backward scheduling technique is applied into the back end process. The dynamic programming is selected for this development because the priority setting is one of the requirements from customer. To evaluate the efficiency of the developed mode, Johnson’s law is used as the comparator along with current result. The result from the developed method shows significant improvement in both mean tardiness and % missing shipment. The % missing improves from 10% to 3.5 % averaging from six months. Based on 95% confidential, this improvement is significant. Even though the mean tardiness of current system cannot be quantified due to too many incomplete shipments, the cumulative plot clearly shows an improvement on the developed method. Comparing with Johnson method, the developed method, also, clearly show significant improvement based on 95% confidential. Thus, regarding all improvement mentioned above, the developed method is recommended. However, it has a major constrain in sample size limitation and long calculation time that the company has to continuously develop for further improvement and for supporting the market growth.