| Summary: | In today's competitive market place, all the components of a supply chain must be well coordinated to achieve economic and service goals. This dissertation is devoted to the modeling, analysis, and development of solution approaches for some logistics problems with emphasis on coordination of various supply chain components and decisions. Specifically, we have addressed four problems in this domain that span various decision levels.
The first problem deals with integrated production and shipping scheduling for a single manufacturer and multiple customers. We develop an optimum-seeking algorithm and a fast heuristic, both of which exploit structural properties of the problem. The second problem is a joint production and delivery scheduling problem in which a single vendor supplies goods to a single buyer over a finite horizon. We model this multi-period problem by using a dynamic programming framework and develop an effective Lagrange multiplier method for the solution of the single-period problem, which is then used to solve the multi-period problem. We show that the optimal shipments in each period follow a pattern of geometric-then-equal sizes except for the last shipment, which may be of a larger size. We also show that an optimal solution for the infinite horizon problem can be derived as a special case of our finite horizon approach. In addition, we propose two fast heuristic methods, which, as we show, can obtain almost optimal solutions. We also address the design and logistics operation of biomass feedstock supply chain. To that end, we consider two problems. The first of these problems arises in the context of delivering biomass sorghum to a biorefinery. We propose multi-period, mixed integer linear programming models, which prescribe the strategic and tactical logistics decisions. Our aim is to investigate different logistical configurations available in a sorghum biomass feedstock logistics system. The second of these problems further allows sharing of loadout equipment among storage facilities. We develop an efficient Benders decomposition-based algorithm, and also, two heuristic methods that are capable of effectively solving large-scale instances. We also show the advantage of using mobile equipment. === Doctor of Philosophy
|
|---|
