The f2*Om term basically has the effect of lowering the eddy viscosity when there's an adverse pressure gradient as I understand it

So in the situation without a pressure gradient, SST should reduce to baseline I'd have thought, but Schmidt number is different so it doesn't

But maybe when there is an adverse pressure gradient, Schmidt number should be 1 or something, and then 0.85 is a compromise between the two

Idk

According to Wilcox

sigma_star in Wilcox nomenclature corresponds to sigma_k1

sigma_star in Wilcox nomenclature corresponds to sigma_k1

^wilcox's formulation is also known to be sensitive to BC, so there are additional developments on top of it to get models like SST kw

Another intersting difference between the BSL and SST model is the coefficient sigma_k and a_1:

Well those are the inverse quantities (in BSL it's 1/2 = 0.5, and in SST it's 0.85)

My comment above adds no value

Well my advice to implement the GEKO model in OpenFOAM. The model is designed to be re-calibrated for different types of flows with constants that make much more sense for each type of flow

https://www.ansys.com/content/dam/amp/2022/march/quick-request/Best%20Practice%20RANS%20Turbulence%20Modeling%20in%20Ansys%20CFD.pdf From page 11: *Note that the GEKO model is not fully published – this could lead to issues in case users want to publish their results in a scientific journal. *

Hi all, I have a question that might be obvious/dumb, but I just wanted to double check: I have an LES simulation in which I've used the UPrime2Mean function to sample the resolved part of the Reynolds stress tensor. That gives the u'u' etc components in the x,y,z directions. I want the components aligned with my flow, which is travelling at 45 degrees for instance. I take it that it's fine to just multiply the stress tensor with a 3x3 rotation matrix to achieve this? There's no turbulence weirdness that makes this invalid?

basically I've done a simulation of the atmospheric boundary layer, but due to coriolis forces the wind direction isn't constant with height - I want my turbulence statistics as along-wind and cross-wind, not x and y

I'm pretty sure just rotating it is fine?

@Zino remember you have to pre AND post multiply by the rotation matrix

Ah great point, I'm so used to rotating vectors that I forgot that particular part of tensor rotation, thanks 🙂

Hey @Zino , When we calculate the Uprime2mean value, it is dependent on the Umean calculation, right? And the Umean is subject to the window size I specify. So, wouldn't that make stuff sensitive to my choice of averaging window? Is there a some guideline on which window to consider (eg maybe close to the integral timescale?), so that I the statistic remains meaningful

Yeah that's right, from my notes:

Unfortunately I don't have a good answer for you RE the appropriate windowing, it is indeed sensitive

This is just a formula for calculating rolling average.. I have done this by hand and through some Python scripts to confirm the equivalence.

Okay good to know, I know the formula for a cumulative average is like this:

Right. In that case, my alternative would be to get a few variations of these two quantities ..Let's say incrementing the window to 10 times.. but the fieldAverage object has hardcoded names for the fields so i am not sure whether we can do multiple of those.

But I hadn't had the chance to sub in variance and expand it

I'd be surprised if there was a hard and fast rule for window size though