Aerodynamic Optimization of an Axial Turbine Blade and Diffuser

Abstract This paper presents a multi-component optimization approach for improving overall performance of the axial turbine blade and axial-radial diffuser of a marine turbocharger. Blade throat-to-pitch distribution and diffuser passage shape are concurrently optimized, exploiting coupling between blade exit flow and diffuser pressure recovery to boost overall efficiency. Algorithmic approaches are used for shape generation of both blade and diffuser. For the blade, a spanwise twisting and re-staggering approach fixing turbine mass-flow is adopted. For the diffuser, a NURBS-based process allowing for non-axisymmetric passages is developed by visual programming. Steady RANS CFD is used for aerodynamic evaluation and optimization is performed by a parallelized hill climbing method for both axisymmetric and non-axisymmetric passages. The best design from concurrent optimization with axisymmetric passage achieves overall efficiency improvement of +0.49% (abs) compared to the base case. This performance is 40% higher than the combination of best blade and diffuser optimized individually. The best concurrently optimized design consists of a blade with triple-peak, W-profile throat-to-pitch distribution with more uniform mid-span flow and accelerated tip-side flow and swirl combined with a diffuser with 14% higher passage area ratio and 9% higher pressure recovery without sacrificing separation margin. Over 70% of efficiency gain is due to the blade, far exceeding the effect of passage shape. This design also retains its performance improvement under overloaded operation, showing viability for actual use. Best design with non-axisymmetric passage achieves +0.04% (abs) further improvement but shows little difference in terms of flow mechanisms. This result shows the utility of this approach in exploiting coupled mechanisms between turbomachinery components. In future, this approach could support existing part-by-part design methods with greater understanding of the influence of interdependencies between components on overall performance.