Tower and nacelle/hub implementation

[danrhouck]

Tower:
I have the tower "turned on" with the following parameters:
itower = 1
tower_x = 0
tower_ybot = 0
tower_ytop = 2.3704
tower_D = 0.1481
tower_CD = 0.87
When I look at outputs, though, I actually see a speed up in the flow where the tower should be. I've attached two images of results averaging the last 10 timesteps of 41 revolutions with 20 timesteps per revolution. The first is the centerline vertical-streamwise plane and the second is the ground level horizontal-streamwise plane. They're both the average streamwise velocity (not normalized or just induced).

Nacelle/hub:
This may or may not be related, but, because there is no nacelle/hub, I get a speed up in the middle of the wake that persists for about 5 radii downstream. I'm concerned that this could affect the wake decay. Could there be a way to implement an effective blockage in the middle similar to the tower? I'm going to try to essentially extend the circular section of the blade shaft all the way to the axis to see how that affects the results, but that is also potentially not realistic.

[whophil]

I never used the tower model, but the details of the tower-induced velocity can be found here:
https://github.com/SNL-WaterPower/CACTUS/blob/dev/src/mod/tower.f95#L81

Nacelle/hub:
This may or may not be related, but, because there is no nacelle/hub, I get a speed up in the middle of the wake that persists for about 5 radii downstream. I'm concerned that this could affect the wake decay. Could there be a way to implement an effective blockage in the middle similar to the tower? I'm going to try to essentially extend the circular section of the blade shaft all the way to the axis to see how that affects the results, but that is also potentially not realistic.

I don't think extending the circle to the axis of rotation will do anything, as the blade is treated as a lifting line and has no thickness. However, it is easy to model a nacelle or spinner using the solid boundaries feature. This allows you to model arbitrary "walls," so long as they can be specified in Plot3D grid format. See this video as an example where a cube was placed in the wake.
https://www.dropbox.com/s/l4cw2c0qlqud19q/cube.avi?dl=0

The grid file used for this is: cube.xyz.zip
The snippet you need in your .in file is:

&CaseInputs
    ...
    ! Wall Geometry
    WallMeshPath = './cube.xyz'
/End

[jcokemurray]

I don't recall if the tower model velocity increment is actually added into the wake velocity output you are plotting, or if it's just used in the blade loads calculation. It's an empirical wake model for a cylinder structure, and isn't constructed in the same manner as the blade wake. This could be sorted by looking through the code... However, the high velocity region seen near the rotation axis I think is typical, and results from the fact that the blade spanwise loading drops at the hub. The resulting "tip vortex" shed by each blade at the hub contributes to the wake deficit on the blades, and conversely produces a jet behind the hub... If you are using the ground plane model, that likely accounts for the high velocity region between the rotor and the ground. The flow is getting locally squeezed between the rotor wake and the ground, and is accelerating accordingly. Jon

…

On Mon, Jul 20, 2020, 4:59 PM danrhouck ***@***.***> wrote: Tower: I have the tower "turned on" with the following parameters: itower = 1 tower_x = 0 tower_ybot = 0 tower_ytop = 2.3704 tower_D = 0.1481 tower_CD = 0.87 When I look at outputs, though, I actually see a speed up in the flow where the tower should be. I've attached two images of results averaging the last 10 timesteps of 41 revolutions with 20 timesteps per revolution. The first is the centerline vertical-streamwise plane and the second is the ground level horizontal-streamwise plane. They're both the average streamwise velocity (not normalized or just induced). [image: 1] <https://user-images.githubusercontent.com/67019759/87994232-ae40fd80-caa9-11ea-90e9-a5592065133e.jpg> [image: 2] <https://user-images.githubusercontent.com/67019759/87994233-af722a80-caa9-11ea-8ad2-f56c71f33c34.jpg> Nacelle/hub: This may or may not be related, but, because there is no nacelle/hub, I get a speed up in the middle of the wake that persists for about 5 radii downstream. I'm concerned that this could affect the wake decay. Could there be a way to implement an effective blockage in the middle similar to the tower? I'm going to try to essentially extend the circular section of the blade shaft all the way to the axis to see how that affects the results, but that is also potentially not realistic. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#30>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABC2ZXORKRTD6SVWFOKJUH3R4TD63ANCNFSM4PC4KJDA> .

[danrhouck]

I'm not a coder, really, but, in the link that Phil provided above for the modification of velocity due to a tower, it appears that the modified tower velocity is not being "written", i.e., that line is commented out. If that were simply uncommented, would it change the output of induced velocities?

[danrhouck]

I think this is the tower. It shows up when plotting the freestream.

I believe this suggests that, to calculate the velocity field, the freestream must be added as a function of space as opposed to a simple constant. So streamwise velocity = induced velocity*uinf+freestream (not +uinf).