Multipoint Optimisation of Radial Compressor Using Computational Fluid Dynamics and Genetic Algorithm

Stefan Tuechler, Zhihang Chen, Colin Copeland

Research output: Chapter or section in a book/report/conference proceedingChapter in a published conference proceeding


Radial compressors applied in automotive turbochargers require high efficiency and pressure ratio over a wide operating range. However, due to the large number of design parameters, iterative design techniques and manual tweaking easily become arduous and unfeasible In order to account for the performance requirements and achieve the desired aerodynamic capabilities, automatic optimisation techniques have found wide application.
This paper introduces a strategy for a three-dimensional, multipoint aerodynamic shape optimisation and seeks to optimise an automotive radial compressor. In detail, the method couples the computational fluid dynamics (CFD) solver of ANSYS CFX with a genetic algorithm in MATLAB to maximise isentropic efficiency at a target operating point near surge, while ensuring no significant reduction in performance is incurred near the choke margin.
A CFD model data including inlet pipe, single flow passage, diffuser as well as volute was validated using experimental data and used for optimisation considering 13 geometrical parameters. The numerical results indicated an efficiency improvement of around 1.9 percentage points in the target optimisation region with negligible decrease in performance near the choke line. Experimental testing of the optimised compressor confirm an increase in efficiency of around 1-2 percentage points with respect to the baseline model, while peak efficiency is increased by 3.0 percentage points with the choke line remaining virtually unchanged.
Original languageEnglish
Title of host publicationGPPS-2018
Publication statusPublished - May 2018


Dive into the research topics of 'Multipoint Optimisation of Radial Compressor Using Computational Fluid Dynamics and Genetic Algorithm'. Together they form a unique fingerprint.

Cite this