This thesis focuses on designing and enhancing a turbocharger's radial turbine through a custom automated optimization workflow. This multidisciplinary method employs various cost functions as objectives or constraints, including polar moment of inertia, wheel mass and center of gravity, stresses, blade eigenfrequencies, response to forced excitation, turbine efficiency, and delivered torque. By utilizing gradient information alongside standard function evaluations, the optimization process is significantly accelerated, especially in high-dimensional design spaces. The gradient calculation involves determining surface sensitivities, providing insights into how geometric changes impact the relevant cost functions. Notably, the study examines the forced response of the wheel to harmonic excitation with aeroacoustically motivated damping, marking a novel contribution to gradient-based radial turbine optimization. The method is computationally efficient, avoiding the need for fluid simulation coupling. Several successful optimization studies demonstrate the effectiveness of this gradient-based workflow in addressing complex design challenges, revealing valuable insights into radial turbine design while being cost-effective. Its industrial relevance is highlighted by the integration of commercial software.
