Integrated design optimisation and engine matching of a turbocharger radial turbine

Abstract

This paper presents a novel methodology for engine tailored optimisation of turbocharger turbine design. Both the turbine rotor and volute geometries for a turbocharger radial turbine were parameterised in order to enable CFD calculations for variations of predefined design parameters. The results of this analysis where used to develop and validate two approaches for computationally efficient and reliable prediction of radial turbine performance maps, quantified by total-to-static turbine efficiency and mass flow parameter. The first method utilises a meanline model which was calibrated to experimentally validated CFD data using a genetic algorithm. The second method makes use of an artificial neural network which was trained using the same CFD approach, to predict turbine performance as a continuous function of design and operating parameters. The modelling accuracy of both approaches was evaluated and compared. Finally, the meanline model was integrated into the calibrated 1D engine model of a turbocharged 1.6 litre gasoline engine. The meanline model was used to generate maps for a latin hypercube sample of four meanline design parameters. Five steady-state operating points and one transient operating point were simulated for each point in the sample, allowing the selection of optimised designs on the basis of fuel consumption and transient performance as objectives. Due to the use of a design of experiment approach, the impact of turbine design parameters on the engine performance could also be evaluated separately. Finite Element Analysis of the turbine wheel was conducted simultaneously for the assessment of stress in individual turbine geometries. Three optimised turbine designs were selected to cater to different engine operating scenarios: ecological, sustainable and sport driving. The presented investigation clearly displays the methodology and benefits of engine integrated turbocharger design optimisation.