This dissertation focuses on the average cost and service performance models in the shipment consolidation setting, which is treated as an application of stochastic clearing models. Specifically, we consider generalized control policies, generalized demand pattern, multi-item systems, and alternative performance criteria, where various techniques in stochastic analysis and stochastic optimal control are applied. By using stochastic impulsive control technique, we prove that, in the single item shipment consolidation model with drifted Brownian motion demand, the optimal quantity-based policy achieves the least average cost in the long run, among the admissible policies. In multi-item shipment consolidation model, we propose a (Q+? ) policy and an instantaneous rate policy. We prove that among all (Q + ? ) policies, either a quantity-based policy or a time-based policy is optimal in terms of average cost. Furthermore, we demonstrate that the optimal instantaneous rate policy would dominate the optimal (Q + ? ) policy in terms of average cost. In terms of service performance criteria, we propose average order delay in the single-item case and average weighted delay rate in the multi-item case. From a martingale point of view, we provide a unified method to calculate the service measures. Moreover, by revealing new properties of truncated random variables, we provide comparative results among different control policies in terms of the service measures. Finally, we provide an analytical integrated inventory/hybrid consolidation model, and give comparative results in the integrated inventory/shipment consolidation models in terms of service measures and average cost.
