Accurate predictions of the mean flow field entering the stator are required for aerodynamic and tone-noise calculations. Estimates of the turbulence field and space-time correlations are required for broadband noise calculations. These predictions require detailed understanding of the mean flow and turbulence structure downstream of the fan. This understanding is particularly important because of the long distances involved. The axial separation between fan and stator may be as large as four axial tip chord lengths. The flow, evolving along a helical path, may cover a distance of more than twice this between fan and stator. To calculate the evolution of the mean flow over this distance requires detailed calibration of a turbulence model, supported by understanding of the turbulence stress fields, of the balance of turbulence kinetic energy, and of the main turbulence-energy generating mechanisms. An understanding of the eddy structure contained within the turbulence is required to determine the space-time correlations needed for broadband noise calculations. The purpose of this study has been to provide this information. We have chosen to focus on the flow downstream of a linear cascade producing a large scale, easily studied, tip leakage flow. We have simulated and examined the effects of the important relative motion between blade tip and casing, by using a moving endwall extending both upstream and downstream of the blade row. We have made detailed single and two-point three-component turbulence measurements in these flows over a range of positions extending from the trailing edge to almost 3 axial chordlengths downstream, for a range of tip gaps.
The first part of the study concentrated on
measurements of the tip leakage flow with stationary end wall. Three
component velocity and turbulence measurements are used to reveal the
evolution of the flow as a function of distance downstream of the blades for
a tip gap of 1.6% chord, and as a function of tip gap from 0.8% to 3.3%.
Overall, these measurements reveal much of the structure of a tip leakage
vortex wake, the manner of its decay and mechanisms of turbulence production
and its relationship to the two-dimensional parts of the blade wakes.
While the vortex is a region of coherent rotating motion, we find that its
dynamics are dominated by the streamwise mean-velocity deficit it produces.
Mean velocities associated with the deficit are more than twice as strong as
those associated with the rotating motion, and decay more slowly with
downstream distance. Turbulence kinetic energy in the vortex is produced
almost entirely by velocity gradients associated with the deficit. Turbulent
activity is thus centered in an arc-shaped region above the vortex center
where these gradients are large.
In the second part of the study, the effects of relative motion between the blade tips and endwall on the flow downstream of the cascade were studied. Endwall motion was simulated by using a 0.25-mm thick Mylar belt propelled over a sliding surface beneath the tips of the cascade blades. Three-component mean velocity and turbulence measurements made in cross sections downstream of the cascade reveal that the wall motion flattens and shears the turbulence and mean velocity distributions of the vortex. Mean helicity density plots also show that endwall motion smears the vortex center from a single point (when seen in cross-section) into a ribbon that makes an angle of some 30 degrees with the endwall. Despite these effects, many critical features of the vortex are almost unaffected by the endwall motion. The vortex produces almost the same magnitude of streamwise mean-velocity deficit, and this deficit still dominates both the mean velocity field and the production of turbulence. Thus, while endwall motion distorts and displaces the leakage vortex it does not fundamentally alter the mechanisms that govern the development of its mean flow and turbulence structure.
In the third part of the study two-point turbulence measurements were made in the leakage vortex. The two-point measurements show no correlation between turbulent motions in the wakes or leakage vortices shed by adjacent blades and that the leakage vortex is not subject to low-frequency wandering motions. They also show that the tip leakage vortex turbulence is highly anisotropic, and characterized by elongated eddies inclined at about 30 degrees to the vortex axis. The presence of such structures, which produce no clearly identifiable footprint in velocity spectra, appears consistent with the helical structures seen in DNS simulations of a line vortex with unstably large streamwise velocity defect. This suggests the same mechanism is behind the generation of turbulence in this vortex. Examination of the correlation data from the point of view of a hypothetical stator blade suggests that, because of the anisotropy in turbulence structure, velocity correlations seen by a stator in a real engine are likely to be a strong function of engine geometry and operating point.
This work was performed using the Virginia Tech Low Speed Cascade Tunnel. The authors acknowledge the support of NASA Langley, in particular Dr. Joe Posey, for their support under grant NAG 1-1801. Some of the analysis of this data was also performed with the support of ONR, under grant N00014-99-1-0294, administered by Dr. Ki-Han Kim. Single point experimental data from this study is available for download here.
Please direct any questions or comments concerning the data or other aspects of this work to William Devenport at devenport@aoe.vt.edu.