Flexible Transport Process Planning in Volatile Environments and the Adaptation of a Cost-Based Objective Function

The formation of efficient and reliable processes to be executed in modern complex and highly volatile logistic systems requires computational support. An important prerequisite of computer based planning is the representation of the real world problem into a formal decision model. Automatic decision algorithms are then able to derive suitable solutions of such a model and one solution is selected to be executed as a process in the real world system. For decades, researchers have defined models assuming that the planning data do not change (or that changes do not require any consideration) until the once proposed solution is completely executed. Since two decades, different and independent efforts have been made to deal with problems in which it is not possible to disregard changes of the problem data between the creation of a process and its completion. In this contribution, we investigate a decision problem from transportation logistics that requires the update of a process under execution due to the appearance of additional problem data. Furthermore, the process planning is compromised by an unpredictably varying system load that leads to undesirable process disturbances. To overcome such situations, we propose to adjust the decision model to the current situation instead of solving the same generic problem in response to a major variation of the planning data. Section 2 introduces the motivating decision problem. Section 3 describes a solver for single static instances of this decision problem. Section 4 is dedicated to the description of the online decision strategies to manage the variation of the problem data and Section 5 reports about the results of a comprehensive simulation study for assessing the proposed models and algorithms.