The paper assumes that the accident occurrence time is stochastic and emergency supplies is perishable, then two replacement stochastic models based on remaining lifetime and remaining quantity are first proposed. In order to identify the effectiveness of replacement strategy, two replacement-based stochastic models are compared with the general stochastic model that is non-replacement, measured by inventory level and total costs. A discrete-event simulation model is developed to demonstrate effects of occurrence time uncertainty, replacement ratios and distributed functions in occurrence time and demand. Sensitive analysis shows that the optimal decision is more sensitive to remaining quantity ratio as compared to remaining lifetime ratio. The paper shows that when decision-makers ignore occurrence time uncertainty and limited warehousing time, they may significantly miss better decisions. Further, simulation results demonstrate that different distributed functions in both occurrence time and demand lead to different inventory strategies.

No abstract