Periodic-review order policy for a two-echelon inventory problem with seasonal demand

This dissertation presents methodologies to determine ordering policies for a 2-echelon inventory system with multiple replenishing modes under seasonal demand.  The system operates on periodic review basis using reorder point and order-up-to point or (R,s,S).  An Echelon stock concept is applied where each location makes decisions on its own inventory information and the information of all locations downstream.  We decompose a problem into two phases and propose a methodology based on mixed-integer programming models to determine ordering policies for a system with one replenishing mode.  Then, another methodology based on the genetic algorithm and binary search is developed to solve the problem with shorter computational time.  The later methodology can solve problem with 97.49% on average less computational time in the first phase and 95.99% on average less time in the second phase.  Afterwards, a problem is extended to a system with a regular replenishing mode as a main replenishing mode and two special modes to prevent stockouts.  Under seasonal demand, special modes could have either a static policy which applies one policy on every period or a dynamic policy which applies a different policy for each period in the demand cycle.  Special replenishing modes can eliminate the periods with unsatisfied service level with slightly higher holding cost.  A dynamic policy tends to give smaller number of special orders.  However, in case of high deviation demand and short demand cycle, for ease of use, the static policy can substitute the dynamic policy.