Üretim planlaması ve tedarikçi sevkiyatlarının koordine edilmesi için üretim parçalarının tedariğinin eniyilenmesi

It has been recognized that coordinating the shipments of manufacturing parts from the suppliers with the production schedule takes important roles on the smooth continuation of production in mass production systems. In this thesis, we investigate two problems that we encountered while analyzing the sourcing operations of a leading coach bus manufacturer in Turkey. The first problem addresses unorganized delivery operations while the second considers unreliable supplier capability. We optimize the parts ordering decisions under these two settings.In the first part of the thesis, we study the multiple-item lot-sizing problem for a manufacturer that sources parts from a single supplier over a multi-period planning horizon. In order to operate more efficiently, the manufacturer controls its suppliers' delivery process in addition to its own parts ordering process. Since transportation costs are charged to the manufacturer, the manufacturer optimizes the ordering and shipment decisions. We consider the option of delaying transportation of a less-than-full truckload to the next period by allowing the use of items in the safety stock of the manufacturer. We develop a mixed integer programming model that minimizes the sum of transportation and inventory holding costs incurred to the manufacturer under the proposed policy. We investigate the effects of delaying shipments on both cost and service levels under stochastic environments by numerical experiments. The results indicate that the proposed policy is especially effective in reducing cost when frequent shipments with small sizes arise without creating much stock-out risk.In the second part of the thesis, we study the dynamic lot-sizing problem under random supply where the supplier's shipment behavior is represented by a model that assumes a random portion of the current order is shipped in every period. To improve the sourcing process, we propose a method that enables the manufacturer to obtain more information about the supplier reliability throughout its ordering process. For this purpose, we develop a dynamic programming model with Bayesian Updates of supplier capability. There is no information available about the supplier capability at the beginning of the planning horizon. We try to estimate the supplier's capability by using information on previous orders' ordered and received amounts. In this method the ordering decisions are optimized by considering the available information until that point. We then compare the proposed algorithm with the cases under Perfect Information as well as the case with No Information on supplier capability. In the Perfect Information case, the optimal sourcing decision is found by assuming that the supplier ships the given order with a binomial distribution and the supplier capability (reliability) parameter is known by the manufacturer. In the No Information case, the supplier shipment behavior is again binomially distributed but the reliability parameter is not known. By computational experiments, we show that the Bayesian Update approach provides significantly better expected total cost values than the No Information case. Furthermore, the optimal expected costs found by the Bayesian Update approach are close to those found in the Perfect Information case. With the proposed approach, the state space grows faster compared to the Perfect Information and No Information cases with problem size and input data magnitude. For this reason, problems with only moderate size can be solved in reasonable time with this approach. To overcome this computational difficulty, we develop an approach that reduces state space and solves larger problems approximately in reasonable solution time.