Simulation-based Validation and Certification of Vehicle Tests and Designs

Principal Investigators
Michael Kokkolaras, Zheng-Dong Ma, Greg Hulbert (U. of Michigan)

University Researchers
Matt Reed, Panos Papalambros (U. of Michigan)

Industry
Ren-Jye Yang, Saeed Barbat (Ford Motor Co.)
Jinzhong Wang (MSC Software Corp.)

Government
Wesley Bylsma, Mark Brudnak (TARDEC)

Students
Harshit Sarin, Geunsoo Ryu (U. of Michigan)

Simulation models enable virtual engineering; i.e., creating designs that can be delivered for production with minimal prototyping or testing. But how much, if at all, can we trust computer simulations in the absence of physical prototyping and testing, and in the presence of approximations and uncertainties? Much highly valuable work has been performed to date in verification and validation (V&V) within the broader CAE community. Significant effort is being expended to establish V&V standards within the professional societies and practitioners worldwide. This research builds a new simulation-based validation and certification methodology that links simulation model validation with the designer’s intent and the necessary cost for obtaining the information required to perform the validation. The research also supports ISO committee work for creating standards for validation of simulation tools and establishing the process of electronic design certification.

The developed methodology will provide the Army with the ability to validate and certify tests, simulations and ultimately, proposed vehicle designs with a minimum amount of information and/or real prototype testing in order to assess various performance metrics with minimal cost. It will utilize the vast amount of data that are already available and provide guidelines and standards for future tests to be optimized with respect to information acquired and cost expended. Finally the developed validation metrics will be extended to monitor the condition of various systems and components by comparing real-time data gathered by diagnostics sensors to knowledge-base data to predict remaining useful life and dictate whether maintenance is necessary or not.