Rationalizing safety stock settings within NXP Semiconductors

Context NXP operates in a highly dynamic and globalized semiconductor market. The NXP supply chain control is like it peers in semiconductor business quite complex. To manage the supply chain and to deal with the market characteristics, six so called Business Renewal II objectives were launched as part of the one page strategy. On one hand the BR II objectives target for a lower break-even point and on the other hand it aims for operational excellence. For supply chain management this translates in lower stock targets on one hand and better supply chain performance on the other hand. Because of the existing link between these two objectives, this research will support the inventory management project team within the Global Supply Chain Management Competence Center by providing insights in the dynamics between inventory levels and the perceived service levels. With these insights, reducing inventories in an uncontrolled way will be avoided and customer service will be increased. Problem formulation The current tools (supply chain configuration tool and zero based inventory budgeting) within NXP do not give support in making a trade-off between inventory costs and service levels. Hence there is a need to: 1. Support planners in deciding where to place the customer order decoupling point and how much safety stock to keep in all the available stock points in order to support NXP's business in the most effective way. 2. Consider an integral supply chain network for safety stock optimizing instead of optimizing local stock points. 3. Perform a sensitivity analysis on the relation between the (safety) stock settings (translated into costs) and the delivery performance (service) is needed. Based on these three objectives, in combination with the decision that an off-the-shelf model (SCOpE) will be used for optimizing safety stock settings, the central question can be formulated as follows: Can SCOpE be used for defining the optimal safety stock settings over the network in reference of customer order behavior and preferences and what are the implications of doing so? Model testing and improvements made To find out whether SCOpE can be used in practice, six representative case studies within BL MMS-PM are validated by running the model under the same conditions as the real world system followed by a comparison of the model outcomes (i.e. perceived service) with the actual perceived service (i.e. RLIP%). The month under analysis is May 2007 and the results are as follows: Eleven root …