English

This dissertation considers the joint optimization of production and outbound distribution decisions on a detailed scheduling level for manufacturers that rely on third party logistic providers. To accommodate customer preferences many manufacturers, offer a high degree of product variability that is produced in a make-to-order or build-to-order fashion to minimize long-term inventory carry. In practice such pull type systems foremost rely on using third-party logistics to timely distribute produced products to widely separated customer locations. This context is modeled by integrated production and outbound distribution scheduling problems with fixed delivery departure dates which is an emerging research field that still offers many research opportunities. Particularly, the presented study analyses the so far neglected aspect of minimizing inventory cost and batch transportation cost of finished products in an environment with hard production and delivery time-windows in static and dynamic optimization settings. This work provides mathematical formulations for the considered problems and proposes operations research methods to generate high-quality schedules for practical applications. It proposes an exact branch and bound procedure and a heuristic greedy randomized adaptive search procedure for the static problem formulation. The dynamic problem with dynamic order arrivals is tackled by a specifically designed real-time control approach that schedules tasks in real-time. All developed methods were implemented and computationally tested on specifically designed problem instances. The proposed branch and bound algorithm dominates a standard solver approach without specific problem knowledge in terms of solution quality and computing times. The heuristic is shown to quickly provide a large sample size of high-quality solutions which makes the heuristic especially suitable for the dynamic application. The real-time control approach performed well in terms of dynamic order acceptance rate and scheduling costs by utilizing a specifically designed pro-active component. Overall, this work provides insights into the application of implementable methods for the considered problem and forms a starting point for research concerning related or more complex applications.