Model and data synthesis of the tire-road system (long-term forecast) taking into account polymorphic uncertainty

Subproject B04 deals with the development of a simplified, efficient model of the roadway for the long-term prediction with a high number of rollovers within a time period of several years. The basis for this is a detailed, multiphysical tire-road model from the short-term with one to a few rollovers in the millisecond to second range. The long-term prediction is achieved by developing and using a time homogenization method that allows simulation and prediction for different time scales, including phenomena such as material degradation, material healing, cracking and dynamic effects from the tire-road interaction.

For the long-term model, the methodological approach of model reduction is applied to be able to derive 2D and finally 1D substitute models for selected target variables such as pavement deformation of the tire-pavement model. The identification and comparison of different methods of model reduction as well as first numerical analyses with identification of data uncertainty are performed in cooperation with the subprojects B05, B06 and INF.

As a further level of abstraction, the results of the multiphysical 2D and 1D models will be mapped by means of artificial neural networks, which perspectively will serve as a basic solution for the execution of fast long-term calculations to enable statements on fatigue, longitudinal and transverse evenness of the road and its structural condition. The uncertainty in the input variables such as in wheel load, wheel sequence, pavement layers, material properties and temperature boundary conditions are considered. The application and evaluation of uncertain input variables requires their preparation, the quantification of their uncertainty and their representation with data uncertainty models.