| This thesis is composed of two parts. Part I, including Chapters 2-3, develops models that integrate inventory planning with project management in project-driven supply chains. Part II studies multi-echelon inventory models with batch ordering policy and compound Poisson demand.;Project-driven supply chains (PDSCs) are combinations of project activities and their supporting material supply chains where the demand for the latter is driven by the material requirement of the former. In Chapter 2, we present a modeling paradigm to help firms make material supply and project decisions jointly in PDSCs under the guaranteed service-time assumption. We develop an optimization model to simultaneously determine the optimal inventory levels, activity durations, and project schedule to find a balance between inventory cost and project costs. For tree networks, we present a joint optimization algorithm based on dynamic programming. Using examples, we demonstrate that isolated project planning that ignores material supplies can lead to significant losses compared to joint optimization. Even if the project activities cannot be expedited, coordinating the project schedule with inventory decisions alone can result in sizable savings. We also discuss material customization and develop insights on how the savings are generated by the joint optimization and when they are significant.;In Chapter 3, we relax the guaranteed service-time assumption to allow for stochastic service-times in which the material lead times are stochastic. Numerical results suggest that joint planning leads to a substantial cost reduction especially when lead times are uncertain.;In Chapter 4, we study multi-echelon tree-structure supply chains in which external demand follows a compound Poisson process, and each stage employs a continuous-time (R, nQ) policy. It is assumed that transit times are sequential, exogenous, and stochastic. We present an analytical framework to characterize the probability distributions of inventory levels and stockout delays. We propose procedures to approximate the internal demand processes and lead time distributions. A dynamic programming algorithm is developed to search for the optimal reorder points (subject to approximations). A numerical study is conducted to demonstrate the efficiency and accuracy of the approximations. |
