Information updates and inventory decisions for supply chain management

We have studied various aspects of a supply chain in this research. First, we have investigated the effect of accuracy of lead-time information on supply chain costs. We have illustrated that the assumption of availability of accurate information may impair optimal decision-making, especially when demand and lead times are stochastic. We have utilized two types of models for this purpose: two-station and two-time decision inventory models to study the effects of different levels of information accuracy on total costs. More accurate decisions regarding inventory will be made if information on inventory availability is updated, thus, inventory costs will decrease. However, information update costs will occur. On the other hand, if information is not updated, then excess holding cost or lost customer cost may occur because less accurate decisions on inventory may be made. We have formulated this as optimization problems for both types of models and have studied the expected total cost function in terms of update decisions, order size and minimum accepted demand amount.For the sequential two-station inventory model, we have provided convexity conditions of the expected total cost function under various inventory policies. When the convexity conditions are met, one can minimize the cost through derivatives. When the convexity conditions are not met, we suggest that direct search be conducted to find the minimum cost solution. For the two-time decision inventory model we found that the expected total cost function is too complex to be solved analytically. We have utilized computational methods, in particular a genetic algorithm, to study the behavior of the total cost for this case. The example problems for both types of models showed that accuracy of information is important for minimizing costs in inventory decision making.Lastly, we have illustrated a framework for a new modeling and optimization management tool (MOMT) for large-scale supply chain networks. A graph-based method is used to automatically generate and update optimization models for large-scale supply chain networks. MOMT increases accuracy and flexibility of modeling and information processing in large-scale SC models. The tool applications, advantages and disadvantages are described by real world examples from Motorola Inc. SC.