Optimisation of industrial value chains Software tools are widely used for the design of value-adding chains according to the strategic and tactical objectives of a company. Within the AMETIST project significant steps have been made towards identifying the solution which actually optimises the operations involved and decreases costs. Digital Economy © Shutterstock Problems occur in the many and varied industrial domains. The AMETIST project sought to contribute to solutions by providing mathematical tools for the detection of errors. Moreover, the foundation for a unifying framework for modelling and manipulating time-dependent systems was laid. In particular, the scientists sought to translate state-of-the art research into sophisticated automatic tools that would allow real-time scheduling of production lines in factories. Project partner Axxom Software AG provided a value chain optimisation problem which represented typical resource allocation and scheduling tasks as they occur in lacquer producing industries. Each component of a pipeless batch plant serving as benchmark was modelled as a system with well-defined dynamics. All that could take place during batch processing was expressed in terms of behaviours; these were generated by the dynamical systems and constituted the semantics of the problem. The scheduling problem comprised operational costs per amount of product, as well as costs for the delayed finishing of jobs and storage expenses for early completion of an operation sequence. Optimisation-based methods were applied for the verification of the system structure, which was modified to achieve correct and optimal behaviours. Valuable experience gained with the Axxom case study showed that the application of model checking techniques for scheduling is promising. Researchers at the Universität Dortmund did not only provide extensions of the basic models. Their modelling efforts provided fundamental knowledge on how to address a great diversity of phenomena with the use of the so-called timed automata family of models. Timed automata were not found to be a push bottom technology to be applied without modelling and solution strategies specifically developed for the case under study. The generation of libraries for typical configurations appeared, however, as a path towards more widespread and easier application for non specialist users.
