Using it as a precursor could work indeed......no other way to know then testing it I guess...

Using bodyforce term look recurrent in LES. I'm trying to implement a basic LES model including a wall treatment from Schumann (1975) and I noticed that the system of equations from Thomas & Williams (1999) has a kinematic body force term 'bi' (per unit volume) included. Since the top BC is a rigid impermeable stress-free lid imposed at a height z=H, it create a region driven by body force bx(z) acting in the streamwise direction and allowed to vary with height. I tried to find some documentation in order to implement this term in OF but I didn't find much. Do you have some suggestions, documentation that I could read related to this technique implemented in OF? Thank you very much.

I'm starting to study turbulence !

Does someone know from which theory come the (6) ?

It seems to me that we took the strain rate to the other side, and then since we cannot really "divide" by such a complex quantity, we do something like a "multiply and divide" by S(u).

To get a denominator that can act as a divisor

Okay thanks, this equation is weird.

Weird and elegant

https://en.wikipedia.org/wiki/Favre_averaging o_0 Never heard of this.

Hi guys, is there anyone with experience in using SSG and/or LRR turbulence models?

Yes, Whats the question about ?

I'm simulating a liquid metal flow through a rod bundle. I first run a simulation with the kOmegaSST model and all went good. Then, since I'm interested in resolving secondary flows and streamline curvature, I set up a run with the SSG model. I calculated R and epsilon at the end of the kOmega run using the "turbulenceFields" functionObject (I'm using OF v9), and copied them into the 0 folder of the SSG run. Here come the problems: in my 1st attempt I changed the BCs at the wall for both R and epsilon into "fixedValue" (with very small numbers, since R must approach zero at the wall) and "epsilonWallFunction", respectively. The simulation converged well but the wall shear stress was overestimated (about the double with respect to the kOmega run and to empirical correlations). In a 2nd attempt, I did not change the wall boundary condition for epsilon, leaving the one that was present in the original epsilon file calculated at the end of the kOmega run (I attach a screenshot for clarity). In this case, the simulation did not converge, velocities kept oscillating, but the wall shear stress magnitude is closer the expected one (although itself oscillating, of course). I also tried to simulate a pseudo transient to help convergence, but I did not reach it. Am I missing something in the procedure to run the SSG model, particularly in the BCs? Thank you in advance.

you don't have to use fixedValue for R and epsilon, use kqRWallFunction and epsilonWallFunction also for low-Re (suppose your mesh is y+ < 1, which it should be)

try using upwind schemes to start with and then switch to linearUpwind or similar

what's your

nut

BC?

Yeah I forgot to say that I'm interested in a wall resolving approach (y+ is <1 everywhere). I did not use kqRWallFunction since it is suitable for high-Re flow, as it is written in the source code guide. This is also why I selected fixedValue as wall boundary condition for R (basically, no wall functions are used and the equations for R components are integrated up to the wall). On the other hand, the epsilonWallFunction is suitable for both low- and high-Re approaches (but I think that also a zeroGradient for epsilon at the wall would work in a wall resolving approach). For nut I used nutLowReWallFunction

but, I mean, if you tell me that kqRWallFunction works fine also with a low-Re approach, I'll for sure have a try

well, I wouldn't argue that your wall setup is not correct but am not sure

I'm not sure either. In my kOmegaSST run (again y+<1 everywhere), I used kLowReWallFunction, omegaWallFunction (which is suitable for both low and high-Re approaches) and nutLowReWallFunction for k, omega and nut respectively and all went fine. Hence, I really cannot understand why the wall shear stress is greatly overestimated with the SSG model. Also, I put 25 inflation layers, which should be enough to fully resolve the profiles of the transported quantities

I'll figure it out in some way 🙂

I try to use the right

delta

for my IDDES simulation and I would really appreciate to have some suggestions about it. I've seen that the one normally use is

cubeRootVol

. But

cubeRootVol

doesn't seem to work with IDDES simulation. Plus, because my mesh has a ratio of around 8.8 (probably because of the cells from the layers) and

delta

is mesh related, I'm looking for the most appropriate

delta

to use. Here is what I got so far:

You need to use IDDESDelta

Pretty sure if you just delete that IDDESDeltaCoeffs part it'll just use the defaults and should work

I'm trying to initialise an les simulation with a rans solution. I ran a steady rans simulation and copied the solution to the 0 folder of the les setup. However, in the rans solution, a directory called uniform is also generated, containing some files on error estimates. When I start the les simulation along with this directory in the 0 directory, I get some warnings/errors about the uniform directory in 0. Can I remove the uniform directory or should it be incorporated into the setup in some other way?

Yes it's safe to delete, or just ignore the warning.

Thanks!

Thank you very much for replying! The simulation run with the parameters shown in this file though. But, I'm not sure what exactly IDDESdelta does or what is the equation of this filter? Since I'Il need to justify the parameters/filters for a paper eventually, I try to find some documentation on openfoam website and there is not much of it. Do I need to take a look at the code at this point? Plus, because I use the

DEShybrid

div scheme for the moment, I guessed that it was needed to specify the parameter

hmax

under

IDDESDeltaCoeffs

(didn't find documentation that these two are related though, but I noticed in some tutorials that when

hmax

is use in

DEShybrid

,

hmax

is specify in

IDDESDeltaCoeffs

). About the

smooth

section, I've seen somewhere that its use to help having better results for the cells close to the region where two different levels in the mesh coexist. I would certainly appreciate if someone could confirm that. About the different filter, it would be nice to have something similar to this for each one: https://caefn.com/openfoam/filter-width-maxdeltaxyz

If you want to study the guts, then see the docs: https://www.openfoam.com/documentation/guides/latest/api/classFoam_1_1LESModels_1_1kOmegaSSTIDDES.html

There's usually a reference to a paper:

Gritskevich, M.S., Garbaruk, A.V., Schuetze, J., Menter, F.R. (2011)
Development of DDES and IDDES Formulations for the k-omega
Shear Stress Transport Model, Flow, Turbulence and Combustion,
pp. 1-19