Evaluation of inventory control policies for perishable products

For inventory management, it is very important to meet customer demands on time and exact quantity. Especially when the product is perishable, this importance is increasing even more for the companies. The studies in the perishable inventory field began to be matched to real life problems by going out of mathematical modeling. We have examined an (s, S) continuous review inventory model for perishable items, one of the most common problems in the literature. Firstly, a comprehensive literature review was conducted for perishable items. Two different demand classes for one perishable item were examined. It was assumed that the demands for both classes occur according to Poisson process and the customer demands were met according to the on-hand inventory level. Accordingly, it was assumed that when the on-hand inventory drops to s, only the priority customer demands were met whereas the demands from ordinary customers were lost. And also, it is assumed that the demand occurring stock out periods was lost. The system is characterized by Markov process, besides lifetime and lead time have an exponential distribution. The stationary probabilities were derived and the expected total cost formula was provided. The pseudo-convexity of the expected cost function was proved. A numerical study was performed with MATLAB. Differences were illustrated between the inventory management of the perishable products and non-perishable products. As a result, perishable items cause a constant cost increase, it is important to model perishable items in the most appropriate way. The novelty of this study is to provide a model and a solution to a perishable problem with two demand classes, in which both the lead time and the lifetime is stochastic. At the same time, this dissertation may lead to other studies that have a different distribution of lead time or lifetime. By characterizing the model with different distributions and different processes, the system can be even closer to the real life problem.