Prediction of Acoustic Resonances in Core Volumes
Author(s)
de Montgros, Armel
Type
Thesis
Abstract
In numerous industrial systems, acoustic resonances are synonymous with loud
noise and dramatic structural vibrations. They take their origin from trapped weak
pressure waves, but can sometimes develop into complex instabilities. With specific emphasis on turbomachines applications, two computational methods have
been developed for the prediction of acoustic resonances in core volumes. The
first method, called the averaged response function, consists of combined time-domain and frequency-domain approaches. A Favre-averaged RANS solver is used
to model the response of a system to a controlled chirp excitation. The response
is then analysed using well-known modal analysis tools in order to extract the system’s acoustic characteristics. The second method, called the Arnoldi method, focuses on the stability of the CFD solver. The system is transformed, using a linear
Euler solver, into a classic matrix stability problem. The eigenpairs of the matrix,
corresponding to the acoustic modeshapes of the system, are then extracted thanks
to the iterative Arnoldi method. The two methods are validated and compared on
a wide range of cases such as enclosures, open geometries and flow applications.
Such a thorough study first provides the reader with a deeper insight into acoustic phenomena by considering classic acoustic examples such as the end correction
concept, the Doppler effect and the ”lock-in“ phenomenon. This study also investigates the limitations and qualities of the implemented methods which are seen to
behave very well when compared to theory and experiments. They give accurate
results in predicting the three components of the acoustic resonance: the frequency,
the damping and the modeshape. As a result, the methods implemented are considered
to be mature and can be used to study either the complete acoustic map of
the system across a wide range of frequencies or specific acoustic instabilities in a
narrow frequency range.
noise and dramatic structural vibrations. They take their origin from trapped weak
pressure waves, but can sometimes develop into complex instabilities. With specific emphasis on turbomachines applications, two computational methods have
been developed for the prediction of acoustic resonances in core volumes. The
first method, called the averaged response function, consists of combined time-domain and frequency-domain approaches. A Favre-averaged RANS solver is used
to model the response of a system to a controlled chirp excitation. The response
is then analysed using well-known modal analysis tools in order to extract the system’s acoustic characteristics. The second method, called the Arnoldi method, focuses on the stability of the CFD solver. The system is transformed, using a linear
Euler solver, into a classic matrix stability problem. The eigenpairs of the matrix,
corresponding to the acoustic modeshapes of the system, are then extracted thanks
to the iterative Arnoldi method. The two methods are validated and compared on
a wide range of cases such as enclosures, open geometries and flow applications.
Such a thorough study first provides the reader with a deeper insight into acoustic phenomena by considering classic acoustic examples such as the end correction
concept, the Doppler effect and the ”lock-in“ phenomenon. This study also investigates the limitations and qualities of the implemented methods which are seen to
behave very well when compared to theory and experiments. They give accurate
results in predicting the three components of the acoustic resonance: the frequency,
the damping and the modeshape. As a result, the methods implemented are considered
to be mature and can be used to study either the complete acoustic map of
the system across a wide range of frequencies or specific acoustic instabilities in a
narrow frequency range.
Date Issued
2011-02
Date Awarded
2011-07
Copyright Statement
Attribution NoDerivatives 4.0 International Licence (CC BY-ND)
Advisor
Imregun, Mehmet
di Mare, Luca
Sponsor
EPSRC and Rolls-Royce
Creator
de Montgros, Armel
Publisher Department
Mechanical Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)