A probabilistic dynamic programing model for determining optimum inventories

Abstract

The occurrence and size of snowfall, and therefore, the level of demand for salt, are uncertain for each week. Therefore, a probabilistic inventory model is required. This will be a periodic review model - the inventory of salt at the beginning of each week will determine the optimal quantity to be purchased to satisfy expected demand. Dynamic programming with a backward recursion is utilized to determine weekly optimal costs/inventory level.