Multi-physics and multi-fidelity simulation techniques for turbochargers

In the field of internal combustion engines (ICEs), forced induction by means of turbocharging is an industrial standard in many applications, from the automotive to the power generation sector. Turbocharging enhances an ICE’s performance with respect to a naturally-aspirated engine of the same displacement, and is currently a key technology in the development of increasingly more efficient and environmentally sound ICEs. Currently, manufacturers heavily rely on computational methods during development, devoting experimental tests, while still indispensable, to more advanced development stages. Numerical modelling techniques are indeed a robust and cost-effective tool for the development of turbochargers. The present work focuses on the main drivers for turbochargers research and development, namely, the characterization through performance maps, thermal and structural analyses, and design optimizaton strategies. In the framework of industrial cooperation with leading companies in the ICEs field, such as Ferrari S.p.A. and Yanmar Co. Ltd., different test cases and simulation techniques were investigated. The commercial software suite ANSYS® was employed for the calculations. Initially, the prediction capability of steady-state 3D CFD (Computational Fluid Dynamics) approaches was assessed on a double-entry turbine, a particular design for which the available literature findings are remarkably scarce, especially under asymmetric, partial admission operation. The study was then extended to 3D CHT (Conjugate Heat Transfer) approaches with the aim of performing transient thermo-structural analyses on a turbine wheel: specifically, a computationally efficient model was developed in order to avoid expensive, often unfeasible, transient CHT calculations. Eventually, a multi-fidelity mixed 1D-3D approach for the aerodynamic optimization of a compressor’s stationary parts (i.e. diffuser and volute) was developed with the aim of avoiding a lengthy iterative procedure for the matching with the ICE. In the development of the above-mentioned numerical models, several collateral aspects were investigated through sensitivity analyses, enhancing the scientific value of the work, moreover, the availability of experimental data for validation improves the robustness of the study.