This report details the numerical and experimental investigation of the performance characteristics of a conventional radial turbine compared with a new back swept design for the same application. The blade geometry of an existing turbine from a turbocharger was used as a baseline. A new back swept blade was subsequently designed for the rotor, which departed from the conventional radial inlet blade angle to incorporate a 25° inlet blade angle. A comparative numerical analysis between the two geometries is presented. Results show that the 25° back swept blade offers significant increases in efficiency while operating at lower than optimum velocity ratios (U/C). Improvements in efficiency at off-design conditions could significantly improve turbocharger performance since the turbine typically experiences lower than optimum velocity ratios while accelerating during engine transients. A commercial CFD code was used to construct single passage steady state numerical models. The numerical predictions show off-design performance gains of 2% can be achieved, while maintaining design point efficiency. A finite element stress analysis was conducted to show that the nonradial inlet blade angle could be implemented without exceeding the acceptable stress levels for the rotor. A modal analysis was also carried out in order to identify the natural blade frequencies, showing that these were not significantly changed by the implementation of backswept blading. A prototype backswept rotor was produced and tested in a direct comparison with the baseline rotor geometry. Experimental performance results showed strong correlations with those obtained numerically, and verified the predicted performance gains at off-deign velocity ratios. This numerical and experimental study has shown that it is feasible from both an aerodynamic and structural point of view to improve the performance characteristic of a radial turbine at lower than optimum velocity ratios through the implementation of back swept blading.
Assessing the Accuracy of Performance Modelling in Software Defined Networks
