Deterministic Versus (s,S) Model in Unreliable MRP System

We consider an finite horizon MRP system with dynamic demands in an environment where machines are up or down with some probability and thus induce uncertainty in the system. Such systems could be modeled as deterministic systems with an adjustment for machine breakdowns or as probabilistic systems with optimal settings of parameters in an adaptive (s,S) like policy. We develop a simulation of such a system to test the effects of these policies on expected costs. We hypothesize that introducing even a small amount of randomness in the system would cause simple adaptive policies to dominate deterministic policies.