how do we define speed multiplier in heterogeneous flow?

[kaibingc]

Hi,
I am looking to simulate real world wind conditions using heterogeneous flow and I am wondering how would a realistic configuration look like?

I understand from the documentation that:
"The heterogeneous_inflow_config dictionary is defined as below. The speed ups are multipliers of the ambient wind speed,
and the x and y are the locations of the speed ups.

heterogeneous_inflow_config = {
'speed_multipliers': [[2.0, 1.0, 2.0, 1.0]],
'x': [-300.0, -300.0, 2600.0, 2600.0],
'y': [ -300.0, 300.0, -300.0, 300.0],
}"

How do we define which coordinates should speed up occur at and what the speed multiplier should be? Is there a model that this is based on? Would appreciate if you could share some examples too, thank you!

[misi9170]

Hi @kaibingc ,

Heterogeneous inflows are very site-specific. For instance, wind farms that has a forest on one side and open space on the other may exhibit different "free stream" wind speeds across the site. In this example, perhaps the speed multiplier on the forest side could be 0.9, while the speed multiplier on the open side could be 1.1. Because speed multipliers are so site-specific, we don't have any general rules about what they should be.

However, your suggestion of modeling is a good one. If you have access to some sort of model or data for the free stream wind speeds at the site you are interested in modeling, you could use that to compute the speed multipliers. For instance, if your data/model suggests that at a wind direction of 270 degrees and site-average wind speed of 8 m/s, the actual wind speed at a certain point (x, y) is in fact 7 m/s on average, then the speed multiplier for that point (and at that wind direction/wind speed combination) can be applied as 7/8 = 0.875.

If you don't have any access to a model for the free stream wind speed variations at the site, or any data for the site, then I would recommend not applying heterogeneity.

One last point about the coordinates that you should choose: if possible, it's best to select a set of coordinates that completely cover the turbine locations, plus some buffer to cover the rotor radii. The reason for this is that the speed multiplier will be interpolated between (interior to the convex hull) the defined coordinates, but a speed multiplier of 1.0 will be applied outside the convex hull since the speed multiplier cannot be inferred there.

Hope this helps,
Misha

[kaibingc]

Thank you very much for your reply! To elaborate further,

1. We intend to utilize the heterogenous map feature for the purpose of representing wind speed and direction variation across the x-y surface of the wind farm.
2. We have several wind masts at different points within the wind farm for capturing wind speed and direction data
3. We intend to use these to establish the heterogenous map for wake steering optimization.
4. However, we are unclear how to translate these data into the inputs for the heterogenous map module - i.e., convert wind speed and direction into speed multipliers.
5. Finally, we would like to check if any information is lost during the conversion process - i.e., speed multipliers only represent wind speed variation but not direction variation (across the surface of the farm).
6. Related to 5, can the heterogenous map feature represent variation in wind direction across the farm?

[paulf81]

Hi, I have a few responses to offer:

1. I wanted to note there is also a higher-order class for working with heterogeneity, HeterogeneousMap, here is the docs: https://nrel.github.io/floris/heterogeneous_map.html and some examples are https://nrel.github.io/floris/examples/examples_heterogeneous/002_heterogeneous_using_wind_data.html

2. With respect to computing speed-up, we typically assume one point to be providing the inflow condition, and then other points are computed as relative to that point. This is like what is presented in this paper: https://iopscience.iop.org/article/10.1088/1742-6596/2265/2/022008/meta (see figure 2)

3. Note that FLASC includes some methods to automatically compute heterogeneity from SCADA data: https://github.com/NREL/flasc/blob/main/examples_artificial_data/03_energy_ratio/05_estimate_heterogeneity_from_energy_ratios.py

4. heterogeneity of wind direction is not currently in FLORIS but is under development: Wind direction heterogeneity #954