Original Publication Date: 2009-Jul-15
Included in the Prior Art Database: 2009-Jul-15
AbstractCompanies with high-performing supply chain networks (SCN) enjoy essential competitive advantages . Unfortunately, SCN executives report that the risk has significantly risen in the past years . Both in practice and academia the topic of mitigating SCN risk receives considerable attention, i.e. , . Thereby, a deep understanding of these risks is crucial. Based on that understanding, the users need to be able to mitigate those risks both at a global and local level. In high tech supply networks like the Semiconductor industries, this problem becomes increasingly complex as the supply chain is characterized by strongly cyclical demand and a highly immature supply process due to the constant need for delivering cutting edge technologies. This results in a severely strained supply processes that is increasingly subject to number of disruptions. Traditional ERP or advanced planning systems that match supply to existing demand fail to adequately support the SCN and are often blamed for causing information distortion due to issuing of ?nervous build schedules?. Distorted demand leads to demand amplification and finally increased supply chain inventories. The point of criticism is that the direct execution of the MRP generated production schedules, might lead to unstable supply schedules and create demand fluctuations for all upstream levels in the chain. This results in a constant never ending cycle of gaps between planning and execution processes. The present invention provides a unique method for determining the production, distribution, or network flow rates in a multi echelon SCN or multiple customer-supplier pairs of operations, while providing a framework to manage the risks or variability by optimizing the total network inventory required for supporting a desired service level. This invention develops an improved technique to allow users to dampen the propagation of variability through the SCN resulting in lower total inventory buffer through out the multi echelon supply network.
Traditionally SCNs face tremendous pressures from both directions; demand and supply variability. Coupled with the complexities of long product lead times and immature manufacturing process (due to the need to maintain cutting edge technology leadership), it is an extremely complex problem to optimize in Semiconductor industries. The effects of one event in the SCN can amplify the classic problem of "Bull Whip effect", with disastrous implications to the bottom line. A traditional ERP or APS system with its deterministic capabilities often falls short in being able to represent and optimize the business and fails severely in supporting a "Supply Chain for Growth" strategy. The root cause of the problem also lies in the technique that Advanced Planning (APS) and ERP system uses a deterministic DRP- or MRP I/II-based logic. This logic explodes independent demands through the bill of material (BOM) to generate requirements and schedules at each BOM level in the SCN. Some of the "best of class" processes try to mitigate this problem by managing risks using optimization techniques around capacity constraint tradeoffs aligned to business goals. These objectives are then centrally orchestrated using consensus business rules. They still fall short of addressing the problem as these processes are still dependent on nominal and deterministic inputs and often lead to propagation and amplification of both demand and supply variability across all stock nodes. This results in a significant supply / demand miss-match. Moreover, the variability in demand patterns and supply performance cannot be considered easily if at all. The other root cause is not being able to effectively factor variability in their calculations and employ a simple lead time offset Qty per explode calculation of independent demands to determine the demands at the intermediate stocking points in the SCN. The safety stock for manufacturing is expressed in simplistic techniques like Days of Supply or Demand buffer if considered at all. The exploded demands then become the inputs for the local manufacturing sites or operations to execute according to the plan. The goal of these processes or techniques is to generate a time based schedule for each operation in the SCN, giving it very little room to make any changes without re-planning the entire SCN, a very cost and manual intensive process to run. So, most companies choose not to rerun the process. The deterministic demand-based planning system described above results in tremendous amount of "waste" as it tries to determine "what", "when", and " how much" to build based on aged data or information, constantly in a catch up mode with the plan.
Therefore, there is a need to develop an improved methodology of planning a multi echelon SCN, which understands the involved risks in a highly uncertain...