Ambient turbulence intensity impact on wake loss

[cedricdallozzo]

Hi Floris team,

I'm not sure about the real impact of ambient turbulence intensity (TI) on wake losses using FLORIS for EmG and Jensen models. A common approach in the industry is to use TI as a proxy to modify wake dissipation downwind for instance using a relationship between TI and wake decay for Jensen model. Is there a default feature to include / imply ambient TI in wake losses for EmG or Jensen model? Is there a consideration of ambient TI in EmG wake model formulation?

Thank you in advance.

[misi9170]

Hi @cedricdallozzo ,

Thank you for your question!

For the Jensen model, there is not a way to do this, and we do not currently have any plans to implement a TI-dependency in the Jensen model. That being said, if you are aware of an established model for the inclusion of TI in the Jensen model, I would be happy to support you in adding that in.

For the empirical Gaussian (EmG) model, the story is a bit more complicated. There is currently a method for including a TI dependency via the atmospheric_gain_ti field on the floris input file; however, we have not tested it and are not at all confident in it's formulation. It is therefore set to 0 as default. However, if you would like to try playing around with it, you are more than welcome, and we'd be interested in what you find! But again, this is an untested feature and we don't vouch for it in any way.

That being said, we have plans to spend more time developing a TI dependency in the EmG model---see #841, #838 . We have not gotten very far down this path at this stage, as we don't have any current projects where that is a focus, but it is on the to-do list at some point in the future. In the end, this may simply be finding an appropriate value for the existing atmospheric_gain_ti field, but in my mind it's more likely that the eventual solution will be a more thorough formulation for how TI affects wakes at different distances downstream, in keeping with the general form of the EmG model.

For now, when considering TI with the EmG model, we tend to use different tunings for the wake expansion parameters for high and low turbulence conditions. This allows us to capture TI in the wake expansion parameters indirectly---slightly less elegant than a direct parametric dependency, but also gets us away from having to postulate the form of the dependency. I should also mention that the dependency of the other models (e.g. GCH, CC) on turbulence_intensities can yield some mixed results and should also be treated with some caution---it works fairly well in lower TI cases (say, less than 10%), but we don't necessarily trust the results when TI is high. Again, more testing could help here.

Happy for anyone else to jump into this discussion!

Misha

[misi9170]

Also linking #1050

[cedricdallozzo]

Hi @misi9170,

Thanks a lot for your answer.

Regarding Jensen model, one approach consists in linking wake decay to TI by a simple factor A wdc= A x TI. This factor depends on the TI levels (https://help.emd.dk/knowledgebase/content/windPRO4.1/c3-UK_windPRO4.1_ENERGY.pdf, https://www.emd.dk/files/windpro/June-22_Revised%20offshore%20PARK%20WDC%20recommendations_v0.pdf). This modification leads to important bias in terms of wake loss depending on TI.

Regarding atmospheric_gain_ti, I will try to test it when I have a bit more time. It seems very interesting. Regarding EmG model do you have some recommendations to tune the wake expansion parameters based on TI level?

Thanks again @misi9170

[misi9170]

Thanks!

Generally speaking, if the simulation cases were split between "low" and "high" TI, the wake expansion rates will be smaller for the low TI case than the high TI case. The current defaults of [0.023, 0.008] reflect a general case across all turbulence intensities, so it seems reasonable to assume that using values lower than these will be more representative of low TI conditions and values higher than these will be more representative of high TI conditions, although that's probably not always going to be the case for every site. It's also possible that you could adequately account for TI by changing only one of the wake expansion rates (my gut feeling is that changing the first one, default 0.023, may suffice).

I'll also let @paulf81 chime in here because he has been thinking about this a bit more recently.

Misha

[cedricdallozzo]

Thanks again!