Secondary wake effects

[prith-gs]

Hello,

I have been trying to understand what enable_secondary_steering, enable_yaw_added_recovery and enable_transverse_velocities does.

I have ran 4 cases as you can see in the table below.
For all cases, I have set enable_secondary_steering to True and active_wake_miing to True.

Each of these 4 cases have 2 more cases depending on whether the freestream turbine has any yaw offsets or not.

The plot below shows the wake profile at 6.5D downstream from the freestream turbine, T0. T1 just refers to the location which 6.5D from the freestream turbine.

The Turbine has no yaw offset

Turbine has 20-degree yaw offset.

I see that when there's no yaw offset, all the 4 cases give the same exact wake profile.
When the Turbine has a 20-degree yaw offset, case 111 alone has a different profile compared to the rest. Case 111 has a lower wake loss.

What is happening here?

1. Why did case 110, 100, 101 give the same wake profile?
2. looks like enable_yaw_added_recovery and enable_transverse_velocities have to be both enabled or disabled to see an effect?

Looking forward to your response.

Best,
Pritheesh

[misi9170]

Hi Pritheesh,

Thanks for the question. So that I understand your setup better, are you running with the velocity_model and deflection_model both set to "gauss"? If that's the case, the "enable_active_wake_mixing" flag won't affect results---that's only used in the "empirical_gauss" model.

I'd recommend using (at least) 3 turbines in a row to look into this. Because there are "secondary effects", some of the effects won't appear until the second wake forms, which means they'll only be observed on the third turbine in a row. I've made a short script to run all of the combinations (with zero yaw angles) and look at the power output of each turbine in the row. Note that you'll need to be running Floris v4.3 or greater to use fmodel.show_config(full=True):

from floris import FlorisModel

fmodel = FlorisModel("../inputs/gch.yaml")
fmodel.set(
    wind_directions=[270.0],
    wind_speeds=[8.0],
    turbulence_intensities=[0.06],
    layout_x=[0.0, 500.0, 1000.0],
    layout_y=[0.0, 0.0, 0.0],
    turbine_type=["nrel_5MW"]
)
fmodel_dict = fmodel.core.as_dict()
fmodel.show_config(full=True)

# CODE:
# enable_secondary_steering . enable_yaw_added_recovery . enable_transverse_velocities

# Default config is TTFT
fmodel.run()
pow_TTT = fmodel.get_turbine_powers()
print("Turbine powers default (TTT):", pow_TTT)

# Loop through configurations and collect
pows = {}
options = [a+b+c for a in ["F", "T"] for b in ["F", "T"] for c in ["F", "T"]]

for config in options:
    fmodel_dict = fmodel.core.as_dict()
    fmodel_dict["wake"]["enable_secondary_steering"] = (config[0] == "T")
    fmodel_dict["wake"]["enable_yaw_added_recovery"] = (config[1] == "T")
    fmodel_dict["wake"]["enable_transverse_velocities"] = (config[2] == "T")

    fmodel = FlorisModel(fmodel_dict)
    fmodel.show_config(full=True)
    fmodel.run()

    pows[config] = fmodel.get_turbine_powers()

new_result = {}
i = 0
for k, v in pows.items():
    if str(v) not in new_result:
        new_result[str(v)] = i
        i += 1
    print(f"{k}: {v} ({new_result[str(v)]})")

The final printout is then:

FFF: [[1753954.45917917  354598.09795631  336751.2426415 ]] (0)
FFT: [[1753954.45917917  354598.09795631  336751.2426415 ]] (0)
FTF: [[1753954.45917917  354598.09795631  336751.2426415 ]] (0)
FTT: [[1753954.45917917  354990.76412434  339432.47753403]] (1)
TFF: [[1753954.45917917  354598.09795968  336751.24264444]] (2)
TFT: [[1753954.45917917  354598.09795968  336881.55665176]] (3)
TTF: [[1753954.45917917  354598.09795968  336751.24264444]] (2)
TTT: [[1753954.45917917  354990.76412771  339561.53631283]] (4)

I'll point out here that all of these flags activate if statements in the sequential_solver, so my following suggestions are based on the code there.

Based on my understanding:

• enable_transverse_velocities = True means that the solver will compute v and w components of the flow. If this flag is False, all v and w components will stay at their initialized values of 0. However, v and w components are only used inside the enable_secondary_steering and enable_yaw_added_recovery parts, so if these are both False, enabling calculation of v and w will not change the results over the "baseline" configuration (all secondary effects off, FFF above). This explains why FFT is equivalent to FFF.
• FTF is also the same as FFF because enable_yaw_added_recovery only makes a difference if there is a nonzero yaw value. Since no yaw values are set, and enable_secondary_steering is False (so there is no "effective yaw"), there is no yaw added recovery and the results are the same.
• Other than that, the only results that are the same as each other are TFF and TTF, with the difference between the two being whether enable_yaw_added_recovery is on or off. As in the first bullet, if calculate_transverse_velocities is False, there are no v and w components so the yaw_added_turbulent_mixing computes a zero output. This is to say that 'enable_traverse_velocities' is required to see an effect from enable_yaw_added_recovery; but enable_yaw_added_recovery is not required to see an effect from enable_transverse_velocities, because enable_transverse_velocities will also be involved in the effective yaw angle computed in the enable_secondary_steering if statement (i.e., compare TFT to TTT).

Sorry that this is a bit of a complicated answer---I hope it helps some!

Misha

[prith-gs]

Hi @misi9170

FFF and TFF are both pretty much same as well.
The differences in the third turbine are at the 5th decimal point.

Although they are not the same, I believe without the enable_transverse_velocities set to true --> without v_wake and w_wake not being calculated, the added yaw offset because of secondary steering is very very minimum.

The avg_v in the wake_added_yaw is zero.

So, TFF is only from the v_core which is really small in magnitude.

Looks like enable_transverse_velocities is the biggest contributor here. enable_transverse_velocities should be set to true to see an effect of both secondary_steering and yaw_added_recovery. Do you agree with this?

Best,
Pritheesh

[misi9170]

Hi @prith-gs ,

Good point, agreed on all fronts. Yes, you should use enable_transverse_velocities if you want to use secondary steering and yaw added recovery. Functionally, we almost always (maybe always) have them either all true (the full "Gauss Curl Hybrid" model) or all false (the "legacy" Gauss model).

Misha

[prith-gs]

Thanks for the response!

[prith-gs]

Hi @misi9170 !

Thanks for the response. This does make sense.