Closed lawrenceccheung closed 1 year ago
Hmm the circulation term should account for this. Looking at the code it is being derived from the angular velocity. I think we should pull up the paper and take a closer look. A quick test, does the sign flip when you switch it in the angular velocity? Also is this with or without tip correction?
This is with the tip and root correction turned on. Yeah, we should look at the paper, maybe they assumed a counter-clockwise turbine rotation? We can also try it with the rpm schedule flipped to negative, and see if that changes anything.
This issue is stale because it has been open 30 days with no activity.
Ping to keep open
Sorry I was tied up and only able to get back to this just now. Using a negative rpm, in the hopes of flipping the rotor spin direction, actually causes it to blow up fairly quickly:
==============================================================================
Step: 1 dt: 0.1 Time: 0 to 0.1
CFL: 1.82811 (conv: 0.36 diff: 0 src: 1.63825 )
WARNING: fixed_dt does not satisfy CFL condition.
Max. CFL: 0.5 => dt: 0.0273507; dt_inp: 0.1
Godunov:
System Iters Initial residual Final residual
----------------------------------------------------------------------------
MAC_projection 9 6.642375679 3.898922851e-12
velocity_solve 2 2.339986401e-05 4.078515303e-12
Nodal_projection 7 27.81969413 1.229763825e-10
WallClockTime: 1 Pre: 0.00836 Solve: 5.498 Post: 2.28e-05 Total: 5.507
Solve time per cell: 6.701e-05
==============================================================================
Step: 2 dt: 0.1 Time: 0.1 to 0.2
CFL: 4.06122 (conv: 3.01855 diff: 0 src: 2.05779 )
WARNING: fixed_dt does not satisfy CFL condition.
Max. CFL: 0.5 => dt: 0.0123116; dt_inp: 0.1
Godunov:
System Iters Initial residual Final residual
----------------------------------------------------------------------------
MAC_projection 8 27.93489851 3.320632658e-11
velocity_solve 2 0.0002104789075 4.786215868e-11
Nodal_projection 7 74.72594784 1.370289906e-10
WallClockTime: 2 Pre: 0.0132 Solve: 5.581 Post: 9.76e-06 Total: 5.594
Solve time per cell: 6.802e-05
==============================================================================
Step: 3 dt: 0.1 Time: 0.2 to 0.3
CFL: 10.711 (conv: 10.1722 diff: 0 src: 2.40238 )
WARNING: fixed_dt does not satisfy CFL condition.
Max. CFL: 0.5 => dt: 0.0046681; dt_inp: 0.1
Godunov:
System Iters Initial residual Final residual
----------------------------------------------------------------------------
MAC_projection 7 5261.589596 2.372345342e-08
velocity_solve 2 331013.7244 0.08343505859
Nodal_projection 5 6.085379949e+10 0.009047065436
WallClockTime: 3 Pre: 0.00963 Solve: 5.135 Post: 2.18e-05 Total: 5.145
Solve time per cell: 6.259e-05
==============================================================================
Step: 4 dt: 0.1 Time: 0.3 to 0.4
CFL: 1.04332e+10 (conv: 1.04332e+10 diff: 0 src: 3.39202 )
WARNING: fixed_dt does not satisfy CFL condition.
Max. CFL: 0.5 => dt: 4.79239e-12; dt_inp: 0.1
with the input
Actuator.T0.rpm = -5.500000000399841 -5.7503730274604 -6.924733009169061 -7.950466035772244 -8.820804709636782 -9.530006866155707 -10.073393458373337 -10.447379427563192 -10.649497356086282 -10.678413746477254 -10.82288967359941 -11.139804720841314 -11.627068002870239 -12.28146474128283 -13.098677474494233 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643 -13.141137992834643
Not sure why that would be, but it's definitely unexpected.
Lawrence
Hopefully solved in https://github.com/Exawind/amr-wind/pull/848
There might be an issue with the way that the swirl (or the turbine rotation direction) is oriented in the Joukowski ADM model. Normally for clockwise rotating turbines, the vertical velocity in the wake should be positive on the right hand side of the rotor disk (if you're upstream of the turbine looking downstream), or on the negative y region as correctly shown in this top-down image:![image](https://user-images.githubusercontent.com/15526007/231549872-592909a0-a286-4e5d-938e-92049ba57857.png)
Running a Joukowski model gives the opposite result:![image](https://user-images.githubusercontent.com/15526007/231550420-f7477678-670e-414b-9cd9-22552365cf6e.png)
It might be as simple as flipping the sign in the
f_theta
computation here; https://github.com/Exawind/amr-wind/blob/7891377a4342436d109090f0e24525aa877bf665/amr-wind/wind_energy/actuator/disk/Joukowsky_ops.H#L236-L237 Although I could also imagine situations where we'd want to specify if the turbine is spinning clockwise vs counter-clockwise.Lawrence