Global warming is a climate phenomenon with world-wide ecological, economic and social impact which calls for strong measures in reducing automotive fuel consumption and thus CO2 emissions. In this regard, turbocharging and the associated designing of the air path of the engine are key technologies in elaborating more efficient and downsized engines. Engine performance simulation or development, parameterization and testing of model-based air path control strategies require adequate performance maps characterizing the working behavior of turbochargers. The working behavior is typically identified on test rig which is expensive in terms of costs and time required. Hence, the objective of the research project “virtual Exhaust Gas Turbocharger” (vEGTC) is an alternative approach which considers a physical modeled vEGTC to allow a founded prediction of efficiency, pressure rise as well as pressure losses of an arbitrary turbocharger with known geometry. The model is conceived to use smallest possible number of geometry as well as material parameters. Thus, conventional expensive and time-consuming application processes can be countered and test rig as well as in vehicle measurements can be reduced. Furthermore, the vEGTC model enables the prediction of different turbocharger behavior caused by geometry variations.
Within this paper it is shown in which way the radial compressor as a representative modeling component can be described by zero-dimensional equations: in order to simulate the working behavior of the compressor the geometry, the thermodynamic state of the inlet-air and the turbocharger speed are assumed to be known. The loss mechanisms are devised using analytical and semi-empirical loss correlations. In order to validate the compressor efficiency the heat transfer from the turbine to the compressor is considered. Finally, the simulation output is compared to manufacturer maps of three different turbochargers pointing out the reliability of the model. Thus, a comprehensive validation of the vEGTC model is yielded. The object-oriented language Modelica is used for modeling and the simulations are provided by the Dymola solver.