In this paper, the performance of a radial turbine working under pulsatile flow conditions is computed with two different modeling approaches, time resolved 1-dimensional (1-D) and 3-dimensional (3-D) CFD. The 1-D modeling approach is based on measured turbine maps which are used to compute the mass flow rate and work output from the turbine for a given expansion ratio and temperature at the inlet. The map is measured under non-pulsatile flow conditions, and in the 1-D method the turbine is treated as being a quasi-stationary flow device. In the 3-D CFD approach, a Large Eddy Simulation (LES) turbulence approach is used. The objective of LES is to explicitly compute the large scales of the turbulence while modeling the effects of the unresolved scales.
Three different cases are considered, where the simplest case only consist of the turbine and the most complex case consist of an exhaust manifold and the turbine. Both time resolved data, such as pressure ratio, temperature and shaft torque and time mean data from the two different modeling approaches are compared. The results show that the computed time mean shaft power differs between the two different modeling approaches with as much as 100%. Since the considered operation point for the engine in this study is 1500 rpm with wide open throttle, the turbine operates in an area where the turbine map is extrapolated. Only by using a few operation points from CFD to extend the map, an improvement is achieved for the 1-D results, but still the deviation is large. Also, the pressure ratio and temperature drop over the turbine differs for the used modeling approaches. The causes for the deviations are assessed and discussed to get a better understanding of eventually limitations of the 1-D modeling approach.