Hey:D Im trying to run a case where I simulate a jet entering a stream of parallel, slower fluid. So far so good but I strugle way to hard to get it to converge for the apparent simplicity of the case. I am running rhoRheactingFoam with reactions turned of because I also want to include the transport of a species.

Basically this is the case. And my goal is to check a. what "domain" gives good results, so is a wedge sim acceptable, should we run a quater of the domain and then use symmetrie boundaries or do we have to run a full case. and b. to check the performance of turbulence models, mainly 2Equation Models compared to Reynolds Stress Models. https://cdn.discordapp.com/attachments/598785402391887872/1352221420289134652/image.png?ex=67dd39d5&is=67dbe855&hm=c0eb4b2ba707299460107376adba8526db09f9b4ba45433ee7eeb033aa169731&

Now so far I have only been able to obtain some results running rhoSimpleFoam, but that allready was a huge struggle. And from these results it seams that the fully3D case is the "easiest to converge"

Now I want to get the Fully 3d Case running, so far only the kOmegaSST model had any chance to converge, for any of the RSM Models the simulation crashes rateher quickly.

My mesh should be fine I think, the checkmesh returns

Checking geometry... Overall domain bounding box (-1 -2 -2) (5 2 2) Mesh has 3 geometric (non-empty/wedge) directions (1 1 1) Mesh has 3 solution (non-empty) directions (1 1 1) Boundary openness (-2.46429e-16 -1.33335e-17 3.12533e-16) OK. Max cell openness = 3.13521e-16 OK. Max aspect ratio = 139.177 OK. Minimum face area = 1.01647e-06. Maximum face area = 0.00931218. Face area magnitudes OK. Min volume = 2.51891e-08. Max volume = 0.000689479. Total volume = 75.361. Cell volumes OK. Mesh non-orthogonality Max: 38.6464 average: 2.7922 Non-orthogonality check OK. Face pyramids OK. Max skewness = 1.01348 OK. Coupled point location match (average 0) OK. Mesh OK.

the Max aspect ratio is somewhat large, but I lack the experience to tell if that actually is a problem because visually the mesh looks good.

To start the RSM simulation I use the results of a previous kOmegaSST Simulation run on the same mesh whilest calculating the turbulent quantities with the turbulentFields funcObject.

Does anybody have any clue why such a simulation should blow up?

And I have talked to people working on similar setups and they have struggled as well to get a stabil solution

ah and im running version v2406

Hello guys, I'm trying to run a rather large LES model with rotating mesh, should I first try RAS (RANS or URANS) before doing LES to find possible errors or is it ineffective?

Not sure if anyone else knows of a different common practice, but starting with a RANS run and then using that to calculate the integral length scale and the cell size of your mesh to resolve at least 80% of the TKE in LES is a good place to start. Aidan goes over it in this video:

are there any y+ insensitive epsilon wall functions?

I am very confused do I just have to enable the lowReCorrection flag?

You'll also have to take care about first cell's y+ for that to work well.

could you elaborate on that?#

from https://www.openfoam.com/documentation/guides/latest/doc/guide-bcs-wall-turbulence-epsilonWallFunction.html it looks like with that option turned on it should addapt the initial guess based on the y+ of the mesh

usually I see a y+ < 6 for resolved bl and 100 < y+ < 300 for unresolved

do I have to avoid the area in between?

or region might be a better word

Aaa mm okay, yeah it seems like you're already doing research on it. That's what I meant, to confirm the y+ requirement. I think there is some y+ around 10-30 where standard wall function fitting used to fail, but then there are enhanced models which fit the transition point by point and blend it further into a two-layer approximation. y+<5 I've seen for kOmega but elsewhere people also recommend y+<1. I've never explored that close to the wall myself, so I am not the right person to make recommendations. But I'm sure you'll see advice pop up if you search the comments above.

okay thank you Ill look into it 😄

Hi, can anyone please tell me how to extract Reynolds stress as an outcome from a simulation? I'm validating an experiment and couldn't figure out how to extract it. I'm using RANS K-epsilon and K-Omega SST turbulence models

This is a job for the ``turbulenceFields`` function object

this is .com branch, but the same function exists in .org branch if this is what you are using. You might just want to check the documentation related to your version to make sure you use the proper syntax

Thank you for this, I understand it is a function object, but I'm not familiar with .org and .com branch, if you don't mind can you please explain what it is? I'm using OpenFOAM v2406

I'll take a look at this now and update you

OpenFOAM-v2406 is the OpenCFD branch (openfoam.com)