|SUBJECT:||M.S. Thesis Presentation|
|TIME:||Wednesday, May 2, 2012, 1:00 p.m.|
|PLACE:||MARC Building, 114|
|TITLE:||Supporting Multidisciplinary Analysis Using System Architectures in SysML|
|COMMITTEE:||Dr. Chris Paredis, Chair (ME)
Dr. Bert Bras (ME)
Dr. Judy Che (Ford)
To develop competitive vehicles with ever increasing complexity, automotive designers need to improve their ability to explore a broad range of system architectures efficiently and effectively. Whereas traditional vehicle systems are based upon IC engines, today’s environmentally conscience vehicle manufacturers must consider alternatives, such as hybrid or electric systems. It would be ideal to start from a base vehicle architecture, and reuse the architecture for multiple alternatives. It is necessary for each of these system architectures to be analyzed from a variety of perspectives including performance, fuel economy, etc. Creating the necessary analysis models for each architecture would be time-consuming, expensive, and could be error prone. We have developed an approach for supporting the generation and integration of analysis models to aid in overcoming these challenges. The approach is based on formally modeling the system architecture in the Systems Modeling Language and then using model transformations to generate stubs for corresponding analysis models in Modelica and Simulink. In this manner, we assist designers in managing large systems with multiple analyses and enable the reuse of generic architectures through specialization and redefinition. The starting architecture is called the Vehicle Model Architecture (VMA), in which key subsystems are modeled. In addition, we have created a generic template that is a specialized version of the VMA that can be adapted to represent a specific vehicle program. In addition, pre-defined, generic analysis templates can be redefined for the specific vehicle under analysis. The VMA system model is translated through two transformations into languages Simulink and Modelica for analysis. By automating these translations and reusing architecture templates, this approach reduces the complexity and modeling time. In this thesis, the vehicle program “C100” and a 0-to-100 km/h analysis test are used as examples.