Re-defining the inbound logistics strategy

The project was conducted at the Procter and Gamble (P&G) plant located in Amiens, France, and it specifically deals with the Material Management Organization Department's operations, especially Packing Materials supply. New brand launches and production line installations forced drastic growth in volume and complexity of the goods flows at the Amiens plant. Moreover, the current operation strategy, i.e. Produce-to-Demand, which means that any stock-keeping unit (SKU) can be produced on any day, puts high pressure on inbound logistics operations. To provide a flexible and efficient supply chain network that is capable of coping with increased and volatile demand, the current Material Inventory Management System has to be redesigned. The main outputs of the system are reduced to more expensive finished product inventory, sufficient material supply to provide efficient functioning of the Produce-to-Demand concept, and minimum of overage inventory. Analysis of current operations reveals that there is an imbalance between the current material requirements pattern and stock levels. On one hand, the imbalance leads to high overage inventory for certain categories of SKU's. On the other, rescheduling actions caused by a lack of packing material occur frequently. In addition, providing adequate and balanced stock levels brings significant benefits, because current inbound onsite storage and external inbound warehouse Baron have only limited available space. The project assignment is to design a conceptual decision-making system that analyses the impact that the materials' inventory model settings have on stock level. The parameters behind each setting that have the most impact on inventory performance are defined and analyzed. The core of the system is software, written in Visual Basic, that simulates the inventory control system known as Min-Max, the (s,S) system. Inputs to this system are demand distribution, inter demand time distributions, and control characteristics of material inventory management, such as order sizes, service level, and lead times. Based on these input parameters, the system computes the expected stock behavior. The system's state is tracked at all times, so that accurate performance measures, such as shortage probability and duration, can also be calculated. The two distribution families are considered in this report, the normal and gamma distributions. Although modeling demand with the Normal distribution is a common practice, it has some serious disadvantages in this usage context, namely that negative values are allowed and needless symmetry is assumed. In such situations, data tends to skew right when modeling real-life environments, so utilizing the gamma distribution seems appropriate here. Another option is to use modified, truncated Normal distributions with required properties. The choice between the two alternative models heavily depends on the item demand properties. A Normal distribution must be used where the coefficient of variation is relatively small, otherwise the gamma demand model is appropriate. Total savings in reduced safety stocks provided by studied concepts depending on SKU characteristics may vary from 20% to 30%. The difference in resulting service levels from current material inventory model settings is insignificant and is equal, depending on SKU characteristics may vary from two to five percent