Radial-turbine mistuning

One of the common failure modes of the diesel engine turbochargers is high-cycle fatigue of the turbine-wheel blades. Mistuning of the blades due to the casting process is believed to contribute to this failure mode. A laser vibrometer is used to characterize mistuning for a population of turbine wheels through the analysis of the blade-response to piezo-speaker induced noise. The turbine-wheel design under investigation is radial and is typically used in 6-12L diesel engine applications. FRFs and resonance frequencies are reviewed and summarized. The study includes test results for a paddle wheel that represents a perfectly tuned system and acts as a reference. A discrete mass-spring model is developed for the paddle wheel and the model suitability is tested against measured data. Density randomization is applied to model mistuning in the turbine wheels. Frequency mistuning and relative amplitude modelling for blade modes is found in good agreement with the data, however the mass-spring model over-predicts amplitude-amplification factors for a population of radial-turbine wheels, especially with regard to hub-dominant modes. A continuous twisted-blade model is developed in Matlab using finite-element techniques. Experimental data is shown to have good agreement with the twisted-blade model. Whitehead’s maximum amplitude-amplification prediction using RMS value for a tuned amplitude value is calculated, and the turbine-wheel response is found to fit within the theoretical limit. Different mistuning patterns are studied using the twisted-blade model. Maximum and minimum response patterns are identified and recommended.

Name dc:type.qualificationname

Master of Science (MSc)

Level dc:type.qualificationlevel

Masters

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2017

Author dc:creator

• Futoryanova, Valentina

Advisor dc:contributor.advisor

• Hunt, Hugh

• radial-turbine
• mistuning
• turbocharger

Licence

• https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/65eee3cf-3070-4c30-a1d0-893364348b53/download
• https://www.rioxx.net/licenses/all-rights-reserved/

Language dc:language

en

DOI dc:identifier.doi

https://doi.org/10.17863/CAM.17032

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/270194