Abstract
This study used Markov modulated Poisson process (MMPP) to model supply systems under various uncertainty constraints. First the uncertainty factors embedded in a supply system were assumed to be unreliable server with or without unreliable repairman. The market demand may be non-stationary random process. Steady state performance measures under different uncertainty scenarios were investigated by using an algorithmic procedure to solve the underlying quasi-birth-and-death (QBD) process. The optimal resource design of the supply systems under various uncertainty constraints was then solved as the trade-off among costs of inventory holding, customer waiting, server repair and normal operation. Numerical studies of the proposed models investigated the impact of uncertainties on the performance of a multi-echelon supply system under a make-to-order (MTO) supply policy. The results are satisfactory and provide managerial insights on strategic supply decision.