Comparing models describing temperature evolution on railway vehicle brake discs with regards to accuracy and computational effort

Abstract

With more powerful computers, simulations are becoming increasingly important in science and industry. However, a simplified model is sometimes preferred for complex systems, for example when doing system simulations. This report presents four different models for thermal analysis of railway vehicle disc brakes. The models investigated are a lumped capacity model, a friction zone model, and two one-dimensional finite cell models with and without the brake pads. They are developed using Dymola as an environment for the Modelica programming language. Temperature models are fitted using experimental data from a test rig. The experimental data represents three different runs of multiple braking cycles, each with a different braking force. The models are evaluated on experimental data. The accuracy of the mean disc temperatures are evaluated using absolute deviation and integrated squared deviation. The CPU-time is also compared in order to evaluate the computational effort of the models. It is shown both by initial calculations and in the simulations that the lumped capacity assumption is not a valid assumption for this case. The most accurate model with respect to the absolute temperature deviation is the friction zone model, while the most accurate with respect to integrated squared temperature deviation is the friction zone and finite cell models. Including the brake pads into the model showed very limited difference in the brake disc temperature. The friction zone model is found to be the computationally most efficient, using half the CPU-time of the finite cell models.