Hybrid Constraints in Automated Model Synthesis and Model Processing

Both parametric design tasks and analysis tasks of technical systems have a similar problem setting: The structure of the system to be configured or analyzed is defined already. Within the parametric design task unknown values for component geometries have to be determined, while within the analysis task the system has to be completed with respect to missing physical quantities. The tasks mentioned form hybrid constraint satisfaction problems, which may be solved by a generic procedure. However, if the no-function-in-structure principle holds, i.e., if the behavior of the entire system can be derived from the behavior of its parts, engineering semantics of model synthesis and simulation apply. As a result, not only domain knowledge can be exploited to solve the constraint satisfaction problem efficiently, but also instances of both types of problems can be tackled by the same problem solving approach: a sequence of intertwined model synthesis and simulation steps. The paper in hand introduces this problem solving approach as a cycle comprising five generic steps and presents case studies of real life problems from the field of hydraulics, which illustrate its successful application.