Production planning and quality of service allocation across the supply chain in a dynamic lead time model

In today's highly competitive marketplace, lowering the overall supply chain (SC) inventory holding cost while providing short product lead time and high customer quality of service (QoS) guarantee is significant. Minimizing some SC costs without actively controlling warehouse and facility inventory distribution is inadequate and may result in inefficient inventory distribution and lower the SC's competitiveness. Yet, state of the art SC production planning approaches only model facility WIP with static predetermined QoS requirements resulting in sub optimal production plans.;This thesis proposes a synergistic SC production planning methodology modeling and minimizing inventory costs under QoS guarantees. Non-linear SC production facility WIP and Inter Facility Inventory (IFI) levels are estimated in each period of the planning horizon using explicit models of stochastic production dynamics. A minimum inventory costs production plan is obtained subject to desired QoS guarantees. This methodology provides a practical and indeed tractable algorithm for solving the SC planning problem in an iterative manner. More specifically, it employs time scale driven decomposition of the original problem to (i) a multiple long time period (e.g. week) planning master problem that determines tentative targets for each SC production facility and inter-facility QoS levels; and (ii) short time period (e.g. hour) facility-specific WIP and QoS horizontal coordination (QoS-HC) sub-problems that determine WIP and IFI levels and necessary sensitivity information based on the tentative master problem targets. Outer linearearization constraints are constructed with sub-problem generated sensitivity information and appended to the master problem iteratively until the master problem representation of the non-linear relationships is sufficient to allow the generated targets to converge to the optimal solution.;This optimal scheduling methodology is implemented based on effective analytic approximations of SC performance evaluation and proposed stop-and-go SC operation protocol. The analytic approximations include SC decomposition, Large Deviations asymptotics, G/G/1/K approximations, Inverse Gaussian distribution approximation, and Monte-Carlo-simulation based calibration of SCV describing functions. Extensive computational experience and Monte-Carlo-simulation verification of the accuracy of the proposed SC production planning methodology is provided to demonstrate its effectiveness and document its superiority relative to industry practice.