When a real-world scheduling problem is considered, it is often solved with substantial unrealistic simplifications, mostly without taking into account the scheduling task of the production planner. If the scheduling problem is not studied together with the scheduling task of the production planner, important issues, such as small system setups, that may create operator nervousness will be unidentified or remain unsolved. At the same time, the gap between theory and practice will probably remain if the real-world problem is not considered at an appropriate level of detail. In this paper, a real-world scheduling problem of a hybrid flowshop in an automotive manufacturer, equipped with parallel machines at multiple production stages with multiple objectives and constraints, has been studied. A discrete event simulation model has been built that represents the production system in details. The model was developed to support not only different simulation-based scheduling approaches, namely, direct, indirect and hybrid, but also the Internet-based real-time scheduling system, OSS, to aid both the work of the production planner and the operators on the shop floor to re-generate feasible schedules when required. In other words, both predictive and reactive scheduling approaches to handle uncertainty are supported by OSS. In contrast to other theoretical scheduling studies, the following important issues have been considered during the development of OSS: (1) avoiding small setups (nervousness); (2) the problem of deploying the new schedule when disturbances like machine failures have happened whilst the optimization process is running; (3) the need for a forecast simulation to quickly see the consequences with the updated problem status of the system; (4) support the monitoring process graphically for a quick holistic overview and more detailed quantitative data for deeper analyses and (5) rescheduling tool that supports the work of the scheduler to re-generate new schedules to cope with uncertainty. Initial results have shown that the application is successful in finding good solutions according to the fitness function, within a limited computation time. Initial tests have shown that the scheduling application manages to generate better schedules when compared to the current scheduling practice used in the system. The remaining work consists of testing the whole system both offline in a virtual environment and online in the real-world production with more production scenarios.
A simulation-based scheduling system for real-time optimization and decision making support
2011
10 Seiten, 7 Bilder, 2 Tabellen, 21 Quellen
Conference paper
English
Decision support for real-time, dynamic decision-making tasks
Online Contents | 2005
|Surrogate Modeling Approach to Support Real-Time Structural Assessment and Decision Making
Online Contents | 2015
|Transportation Research Record | 2015
|