GUST-WING INTERACTION
5.5 Interaction with heaving plate gusts
5.5.3 Discussion
The heaving plate successfully generated vortical gusts that interacted with the downstream airfoil, and resulted in repeatable forces. These forces were associated with two factors: the primary vortex and the wake of the plate.
When the plate did not pass the midline, the lift in the early vortex-wing interaction was adequately modeled by the estimates from the UPM or semi-analytic models.
The quasi-steady method’s inaccuracy was due to the fact that it lacks a wake model, and so it responded without the smoothing that the Wagner function provides. This suggests that the initial lift peak was an inviscid effect of the vortex, moderated by the wake of the airfoil.
In the experiments where the plate passed the tunnel’s midline, its wake resulted in unwanted changes in the forces both before and after the vortical gust arrived. It reduced the amount of vortex shedding for the α = 5° airfoil, increased the lift on
Figure 5.15: Mean-subtracted CL due to gusts from the heaving plate moving at S = 0.1 impacting the test article atα = 0°, scaled by the quasi-steady estimate of the lift peak amplitude. For the yupstr eam = 0 case, the lift is only scaled by S. For comparison, the results of the different models are presented. Each row is a different ypeak.
theα=5° andα =10° airfoils, and modified the lift on theα=0° airfoil. Though these effects were transient, they were significant.
The multiple timescales for the return of the flow to its original state suggest multiple causes. The rapid recovery of the forces forα = 0° is consistent with the Wagner function. When the vortex and wake were far from the airfoil, the change in forces was similarly rapid, even at the higher angles of attack. This is because the flow was not grossly perturbed, and so could easily return to its initial state. When the gust strongly interacted with the airfoil atα=10°, the flow reattached, and required additional time to re-develop its separated flow. Similarly, theα=5° airfoil needed
Figure 5.16: Mean-subtracted CL due to gusts from the heaving plate moving at S = 0.1 impacting the test article at α = 5°, scaled by the quasi-steady estimate of the lift peak amplitude. For the yupstr eam = 0 case, the lift is only scaled by S.
For comparison, the results of the different models are presented. The left column moved initially in the−ydirection, and the right in+y. Each row is a differentypeak.
Figure 5.17: Mean-subtracted CL due to gusts from the heaving plate moving at S = 0.25 impacting the test article at α = 5°, scaled by the quasi-steady estimate of the lift peak amplitude. For the yupstr eam = 0 case, the lift is only scaled by S.
For comparison, the results of the different models are presented. The left column moved initially in the−ydirection, and the right in+y. Each row is a differentypeak.
Figure 5.18: Mean-subtracted CL due to gusts from the heaving plate moving at S = 0.1 impacting the test article at α = 10°, scaled by the quasi-steady estimate of the lift peak amplitude. For the yupstr eam = 0 case, the lift is only scaled by S.
For comparison, the results of the different models are presented. The left column moved initially in the−ydirection, and the right in+y. Each row is a differentypeak.
an intermediate amount of time to recover from strong perturbations. The recovery time of the vortex shedding was approximately 15 tc for the α = 5° airfoil. This appeared to be unrelated to the return to normalcy of the low-pass-filtered forces.
As with the pitching gust generator, the models were unable to properly predict the wake of the heaving plate, the spatial extent of the vortex, or the behavior of the boundary layers on the airfoil. The over-prediction of the lift in theS =0.25 case is likely a result of two factors: the large size of the generated vortex and the difference between its predicted and true circulation. The larger vortex, and its associated structures, would result in a more drawn-out vortex interaction, as compared to the point-vortex used in the models. The over-prediction of the circulation simply resulted in larger estimated forces.
Since all of the numerical models lack viscosity, they were unable to account for the effects of the oncoming wake. Thus, they were inappropriate for modeling the cases where the plate passed the midline. This lack of viscosity also prevented predictions of the drag, which changed due to the velocity deficit in the wake of the plate, or due to the separated region on the airfoil being swept away by the gust. The different numerical models with wakes performed similarly, again confirming that the initial part of the vortex-wing interaction is an inviscid effect. The unsteady panel method and extended Tchieu-Leonard models were slightly more accurate than the model using the Wagner function in thin airfoil theory. The simplicity of the W-TAT model again recommends it, however, due to the relative expense of the other methods.
The heaving plate gust generator is imperfect. Although its wake only interacted with the test article for a finite time, and only when the plate reached the midline of tunnel, the resulting variations in the forces were of the same magnitude as those from the vortex interactions. This suggests that it is inappropriate to use for vortex generation when it would pass in front of the test article. Unfortunately, this constrains the polarity of the generated vortices to gusts with negative circulation above the airfoil, and positive circulation below it.