Modeling and optimization of Advanced Planning and Scheduling (APS)

Abstract

Production planning and control systems based on the Material Requirements Planning (MRP) logic have been extensively implemented in the manufacturing industry. Despite its widespread use, MRP ignores capacity constraints, assumes that lead times are fixed, and does not consider operation sequences of items. All of these create many problems on the shop floor for later production. Unquestionably, MRP and operations scheduling are closely interrelated and intertwined together. Consequently, they should be integrated together to generate realistic production schedules for the shop. This integration leads to the problem of Advanced Planning and Scheduling (APS) and this thesis mainly focuses on the modeling and optimization of APS. In this thesis, a Mixed Integer Programming (MIP) model for the APS, with the objective of minimizing cost of both production idle time and tardiness or earliness penalty of an order, is formulated. The proposed mathematical model explicitly considers capacity constraints, operation sequences, lead times and due dates in a multi-order environment and generates production schedules with operation starting time and finish time for the shop floor. Numerical results indicate that the established APS model can favorably produce optimal schedules. Since the APS problem has been proved to be NP-hard, a genetic algorithm (GA) is built to tackle it more efficiently. A series of computational tests demonstrate that the suggested GA approach is satisfactory in creating effective production plans and schedules. In order to cope with the Dynamic Advanced Planning and Scheduling (DAPS) problem where new orders arrive on a continuous basis, both the MIP and the GA are further extended by incorporating a periodic policy with a frozen interval. The objective of the offered methodology is to find a schedule such that both production idle time and penalties on tardiness and earliness of both original orders and new orders are minimized at each rescheduling point. The provided dynamic mechanism is confirmed to be capable of improving the schedule stability while retaining efficiency. Furthermore, a prototype of the advanced planning and scheduling decision support system is designed to assist production planners to make effective decisions. Finally, a real APS problem arising from a specialist light source manufacturing company is illustrated to validate the applicability of the developed methods and system.