A comparison of predicted performance values from a 1D model and CFD simulations to measured data for an industrial high-speed centrifugal compressor
Klüver, Merle (2023)
Diplomityö
2023
School of Energy Systems, Energiatekniikka
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe2023050340467
https://urn.fi/URN:NBN:fi-fe2023050340467
Tiivistelmä
This thesis compared two different performance prediction tools for an industrial high-speed centrifugal air compressor through comparison of their predictions to experimental measurements. This comparison is done for the design rotational speed of 461 rotations per second both at the design point as well as at off-design mass flow rates. Performance values to be primarily considered are pressure ratio and isentropic efficiency. The examined tools were a one-dimensional in-house developed meanline code (CentriFlow) used at LUT, and an established tool for the 3D numerical analysis of turbomachinery (Ansys CFX).
The results reveal that both methods have their strengths and limitations. CentriFlow showed solid accuracy below the design point, while demonstrating a further decline in prediction precision beyond it. The computational fluid dynamics (CFD) simulations on the other hand, overestimated performance parameters, but did so in a consistent manner and for comprehensible reasons. The overall shape of the performance curves based on CFD results closely resembled the experimental data.
In conclusion, CentriFlow turned out to be more suitable for the use as a design tool during the first phases of the design process, than as an exact performance prediction tool per se. Used for preliminary design purposes, CentriFlow still provides results that leave room for improvement regarding their accuracy but persuades with its ease of use and great speed regarding result generation. As expected, the CFD simulations confirmed to be a more precise prediction tool, additionally providing a wider and more detailed range of information. A limitation of the numerical method is the extensive amount of time necessary to become acquainted with the software, configure the simulations, execute them, and analyse the outcomes.
The results reveal that both methods have their strengths and limitations. CentriFlow showed solid accuracy below the design point, while demonstrating a further decline in prediction precision beyond it. The computational fluid dynamics (CFD) simulations on the other hand, overestimated performance parameters, but did so in a consistent manner and for comprehensible reasons. The overall shape of the performance curves based on CFD results closely resembled the experimental data.
In conclusion, CentriFlow turned out to be more suitable for the use as a design tool during the first phases of the design process, than as an exact performance prediction tool per se. Used for preliminary design purposes, CentriFlow still provides results that leave room for improvement regarding their accuracy but persuades with its ease of use and great speed regarding result generation. As expected, the CFD simulations confirmed to be a more precise prediction tool, additionally providing a wider and more detailed range of information. A limitation of the numerical method is the extensive amount of time necessary to become acquainted with the software, configure the simulations, execute them, and analyse the outcomes.