Statistical Model Verification and Validation Concept in Automotive Vehicle Design

The automotive design process is highly complex and is getting more diverse. With increasing possibilities in modeling, computing and analyzing, simulation driven design is essential to support this process. It has the advantage to reduce costs, time and the risk of wrong decisions. To achieve these benefits verification and validation is crucial, as they define the trustworthiness of simulation models. In the automotive domain, modeling and simulation is growing rapidly, while efforts for the theoretical examination of verification and validation are decreasing. There is a lack in describing the models' credibility, because conventional validation neglects uncertainties. Since uncertainty in vehicle production is mostly documented especially in times of Industry 4.0, this potential is not used. To fill the gap, this paper develops a concept of a statistical framework, which integrates uncertainties in the verification and validation process of total vehicle simulations with Monte-Carlo sampling. Uncertainties of parameters due to measuring error and calibration, numerical and model uncertainties are considered. The application of the statistical framework in a longitudinal consumption simulation proves that uncertainty can be introduced through statistical validation. The evaluation of the framework demonstrates that it improves the reliability quantification of models and their simulation results. The increased knowledge about the product characteristics supports the simulation driven design process, reduces the risk of wrong decisions and raises the sustainability of products. It provides the basis for statistical validation in vehicle simulation to support transparent decision-making in the automotive design process.