Local Forcing Mechanisms of Centrifugal Compressor Blades Excited by Impeller-Diffuser-Interaction
- Doctoral Thesis
Centrifugal compressors allow the realisation of compact and robust designs which provide high-pressure ratios for a single stage, compared to axial compressor designs. Hence centrifugal compressors are found in a wide range of applications from aerospace propulsion to process industry. The demand for high efficiencies, pressure ratios, and flow rates are increasing the mean stress levels in the material. Due to this, the material becomes more vulnerable to cyclic stress, induced by vibrations and leading to high cycle fatigue failures. The reliable prediction of stationary and cyclic stress levels in mechanical structures during the design phase is essential to ensure mechanical integrity and to assess component lifetimes correctly. Vibrations in turbomachines, unfortunately, cannot be avoided and require a careful assessment during the design process. Various excitation sources exist, caused by the mechanical interaction of components or unsteady fluid-structure interaction. Research in the field of forced vibrations aims at a better understanding of the systems damping properties and excitation mechanisms. A correct prediction during the design phase would allow reducing the development cost and time since the number of design iterations can be reduced. However, the correct assessment of high cycle fatigue is challenging due to a lack of representative experimental data. The goal of this work was to investigate variations in the local impeller blade excitation mechanisms with changing operating conditions. Therefore, a state-of-the-art industrial design compressor from a turbocharging application has been integrated into the high-speed centrifugal compressor facility 'RIGI', located at ETH Zurich. The excitation source of the impeller blades is the unsteady fluid-structure interaction originating from the vaned diffusers potential field effect. An experimental approach has been developed to investigate the blade forcing function resulting in resonant response conditions. The experiments are based on in-house developed approaches. To investigate the forcing, an experimental approach was required to extend the variations of the forcing function while minimising the effects of damping avoid changes in modal response properties. This assumption requires leaving the compressor geometry unchanged but instead change the thermodynamic conditions of the stage by changing the working fluid properties. A CO2 injection and control system has been developed and integrated into the closed loop of the facility which allows changes in the circumferential Mach numbers for selected resonance crossings. Experimental techniques to measure the vibrational response of the stage are applied to separate the contribution of blade damping and the forcing function from the resonance amplitudes. The work has been complemented with a numerical investigation of the time-resolved 3D flow field, validated with flow measurement techniques and, by applying a finite element response model, validated against response measurements. The flow simulations are used to determine the local excitation mechanisms on the blade by applying a generalised force approach at resonance. Evaluation of the resonance response measurements revealed changes in response amplitudes reach up to a factor of 4, between air and air-CO2 mixtures and up to a factor of 2.5 with flow rate setting. Changes in the damping of the system were found partially responsible, but changes in the forcing function are the main contributor. The experimental investigation outlines potential limitations arising from low amplitude resonances which increasingly occur at part load conditions and low inlet pressure settings, required for the experimental approach. The numerical FE model to simulate the resonance response amplitudes, using measured damping and the simulated unsteady blade pressure distribution as boundary conditions, provided results within the measurement uncertainties for mode 4 resonance crossings. Some cases for mode 3 resonances showed increasingly insufficient prediction due to flow vortex emerging from part load operation. The local blade forcing distribution found deflective anti-nodes to contribute most to the excitation while high-pressure unsteadiness at the rigid impeller trailing edge area is neglectable. Changes in the forcing function caused by mass flow variations at a given resonance crossing speed are driven by scaling effects of local pressure unsteadiness amplitudes. The variations on flow rate have a minor effect on the flow Mach number distribution, which influences the pressure wave propagation. Hence it has a minor impact of the upstream travelling wave pattern. Variation of the gas properties by CO2, for a constant resonance crossing, lead to changes in the flow Mach number distribution. This has an impact on the spatial distribution of the pressure wave pattern. The phase relation between pressure and mode shape changes considerably as well as the resulting force. Ultimately the forcing not only depends on the local amplitude of pressure unsteadiness but also on how the wave pattern matches the mode shape. A high dependency on the operating condition was found and must be addressed in the design process. The applied numerical approach allows determining the local blade forcing on unsteady flow simulations only without the requirement of computationally expensive response simulations. This provides an efficient way to estimate critical operation conditions in an early design phase. Show more
External linksSearch via SFX
ContributorsExaminer: Abhari, Reza S.
Examiner: Song, Seung J.
Examiner: Zemp, Armin
SubjectCentrifugal compressor; Blade Resonance; blade damping; blade forcing; resonant response; generalised force; generalized force; Compressor Technology; rotor-stator interaction; impeller-diffuser interaction; MODAL ANALYSIS (VIBRATIONS); CFD and experiment; Strain gauge measurements; FRAP; fast response aerodynamic probes; data acquisition; rotating frame data acquisition; CO2; Carbon dioxide; Carbon dioxide capture and storage; Turbocharging; Turbocharger
Organisational unit03548 - Abhari, Reza S. / Abhari, Reza S.
MoreShow all metadata