Unsteady aerodynamics in an axial flow compressor

Current methods for designing compressor blades in axial turbomachines assume that the flow through each blade row is steady. However, interactions associated with the relative motion of neighbouring blade rows are known to produce a disturbance field with both random and periodic components. Despite a growing amount of research into the influence of these disturbances on the downstream row in a rotor-stator or stator-rotor blade row pair, these effects are still not generally accounted for at the design stage. Recent advances in Low Pressure Turbine blade design have shown that incorporating unsteady effects can lead to increases in blade loading beyond the loading limits inferred from steady flow calculations. The current experimental work investigates the unsteady flow behaviour in the neighbourhood of the outlet stator in a 1.5 stage axial flow compressor using thermal anemometry. The aim is to provide a base for more accurate unsteady modelling, and facilitate the development of compressor blade designs which gain maximum benefit from unsteady effects. High-speed data acquisition with synchronised sampling was used to acquire data ensembles for a specific set of rotor wakes, and estimate the periodic and random components of the stator inflow disturbance field. The stator inflow disturbance flow field was altered by clocking of the inlet guide vane row relative to the stator row, and by changing the rotor-stator axial blade row spacing. The interaction of inlet guide vane and rotor wakes was examined using hot-wire measurements downstream of the rotor row. The interaction process was shown to restrict the relative motion of rotor wake fluid and produce local accumulations of low energy fluid on the suction side of the inlet guide vane wakes. Significant circumferential variations in both time-mean velocity and the periodic disturbance component were observed. Clocking of the downstream stator row relative to the inlet guide vanes altered the mid-span stator blade boundary layer behaviour. Hot-wire measurements performed downstream of the stator were used to evaluate the influence of blade row clocking on the stator mid-span viscous losses. The magnitude of periodic fluctuations in ensemble-average stator wake thickness was significantly influenced by IGV wake-rotor wake interaction effects. The changes in time-mean stator losses were marginal. The periodically unsteady laminar to turbulent transition of the stator blade boundary layer was examined using a hot-film surface array mounted on both the suction and pressure surfaces. Observations were made for stator blade loading or incidence cases near stall, design and maximum flow, and for a range of relative axial and circumferential blade row positions. Ensemble average plots of turbulent intermittency and relaxation factor (extent of calmed flow following the passage of a turbulent spot) are presented for a range of inflow disturbance cases. These show the strength of periodic wake-induced transition phenomena to be significantly influenced by incidence, clocking and blade row spacing effects. The periodic, wake-induced, transition in separation bubbles was also altered by changes in blade row spacing. Significant differences between suction and pressure surface transition behaviour were observed, particularly with regard to the strength and extent of calming. Subsequent collaborative work (not reported within) evaluated the quasi-steady application of conventional transition correlation to predict unsteady transition onset on the blading of an embedded axial compressor stage. The viscous/inviscid interaction code MISES was used to calculate the blade surface pressure distributions and boundary layer development. Predictions of the temporal variation in transition onset based on the measured temporal variation of inflow turbulence were compared with the transition onset observations from the compressor stator. Computations for both natural and bypass transition modes indicated that the natural transition mode predicted by a modified e\\(^n\\) method tended to dominate on the compressor blades. The success of the MISES transition onset predictions provided strong circumstantial evidence for the importance of natural transition mechanisms in strongly decelerating flow and provided the impetus further experimental investigation. Transitional flow data from the surface film gauges were also analysed using wavelet techniques. The primary use of the wavelet analysis was to facilitate identification of transient instability phenomena in the complex periodic transitional flows present in the stator blade boundary layer. Wavelet analysis and high-pass filtering techniques revealed significant wave packet activity in the unstable laminar flow regions. An algorithm was developed to identify instability waves within the Tollmien-Schlichting (T-S) frequency range. This was combined with a turbulent intermittency detection routine to produce space~time diagrams showing the probability of instability wave occurrence prior to regions of turbulent flow. The implications for transition prediction in decelerating flow regions on axial turbomachine blades are discussed.