Usage of RSTM model in lowRe simultation possible?

sebonator · July 22, 2015, 03:32

Dear all,

I searched around the forum now for about 3 hours, but I did not find a clear answer for my question:

I would like to use one of OpenFOAMs RSTM models in a incompressible turbulent flow very comparable to a backward facing step and compare its results to results of a k-omega-SST and a realizable-k-epsilon simulation in the sense of a turbulence model study.

For the 2-equation models I used a lowRe meshing approach (y+ ~ 1) and of course applied the lowRe wall boundary conditions kLowReWallFunction, epsilonLowReWallFuntion, omegaWallfunction and nutLowReWallFunction in the corresponding simulations.

My problem is now the setup of the RSTM models (I don't care if LRR or LaunderGibsonRSTM at the moment). I can see that for the Reynolds stresses field R only a highRe wallfunction (so a real wallfunction in the original sense) called kqRWallFunction is available in OpenFOAM 2.4.

Does this really mean that I can not perform a RSTM simulation with a lowRe mesh (y+ ~1) with OpenFoam 2.4? I am wondering because in my mind, a fixed Value uniform 0 boundary condition should be OK for the R-field if you mesh down into the viscous sublayer! But I know that e.g. for some k-epsilon models additional "damping functions" are necessary to enable their usage in a lowRe-case and I am wondering if that might be the case for the RSTMs implemented in OpenFOAM 2.4 as well?

Unfortunately, the "Descriptions" inside the .h files of the RSTM models in $FOAM_SRC/turbulenceModels/incompressible/RAS are very short so I can not find my answer there either... :-(

I try to sum up my questions again:
1) Are the RSTM models implemented im OpenFOAM 2.4 capable of being integrated down to the viscous sublayer in a y+~1 mesh (so are they "lowRe-capable"?) and only a corresponding wall boundary condition is missing in OpenFOAM 2.4 or are these models themselves only "highRe-capable" and it is really a must to use them on a y+~30 mesh?
2) Is my statement R=fixedValue uniform 0 (tensor with zeros of course) a correct lowRe-wall boundary condition for the R-field? This is of course only relevant if the answer to 1) is that the RSTMs themselves are lowRe-capable.

Thank you very much for your help!
Best regards in advance
Sebastian

P.S.: I also downloaded the papers by Launder, Reece and Rodi respectively Gibson and Launder where the models are described, but I have to admit that the theory in these papers is to complicated for me as a simple CFD user... I want to use CFD as a tool to create scientific outcome in an engineering application and I think I understand enough about CFD to use this tool CFD in a competent manner. But please understand that I am not capable of understanding every piece of theory or code about CFD out there.

canopus · November 19, 2015, 05:30

Dear Sebastian
Did you find answers to your questions and can you share them?
Thanks

sebonator · November 24, 2015, 15:41

Dear canopus,

sorry, I haven't been able to answer for a while.

Unfortunately, I did not find any answer myself. In the meantime, OpenFOAM3.0.0 was released but I just checked the source code: there is still no alternative wall function for the R-field than the kqRWallFunction and for this wall function still the restriction "high Reynolds number flow" is written in the .h file.

I decided to leave the RSM models out of my study for that reason, what was very disappointing, though...

Sorry that I can not help in this case. If you are able to find anything about the RSM usage in lowRe-cases, please let me know!!!

Best regards,
Sebastian

Roal-Yr · November 28, 2015, 09:49

Hello.
I am currently working with a mesh which balances at y+ 10-50, and my case is a cyclone-like domain where I want to run a RSTM model.
At the moment I am facing the same issue when I have to decide between performance and correctness of my calculations.

I have an idea on how to check on whether RSTM are low-re capable or not, and it is, simply put, just a comparison of two options:
1). Purely high-re flow;
2). Purely low-re (on a very fine mesh, which I am afraid of calculating)
3). Something in between, when the y+ is at logarithmic sub-layer, but somewhere is getting to low for purely high-re simulation.

To be honest I myself lack clarity on the matter of "what is happening below y+ <30". Seeing that non-functional BCs are applied at y+ >1, I stuck at understanding what is meant for y+ 1-30.

I have limited amount of time to perform my calculations, so I will reply soon.

kaw · June 3, 2016, 06:36

A low Re model with integration into the wall does not appear to have been implemented for the two-equation models. This means that the length scale equation for the RSTM is not strictly applicable for meshes with y+ < 1 or so. However, this can be corrected. The low Re BC for k is straight forward and omega/epsilon at the wall can be constructed by a blending approach, such as suggested by Menter and, in the same spirit, Knopp. I think that the length scale equation is the only issue with extending LRE treatments into the RSM models. Since the first grid point can be assumed to be at the wall, the other BCs should be reasonable trivial.

July 22, 2015, 03:32	Usage of RSTM model in lowRe simultation possible?	#1
sebonator Member Sebastian Lang Join Date: Aug 2009 Posts: 47 Rep Power: 16	Dear all, I searched around the forum now for about 3 hours, but I did not find a clear answer for my question: I would like to use one of OpenFOAMs RSTM models in a incompressible turbulent flow very comparable to a backward facing step and compare its results to results of a k-omega-SST and a realizable-k-epsilon simulation in the sense of a turbulence model study. For the 2-equation models I used a lowRe meshing approach (y+ ~ 1) and of course applied the lowRe wall boundary conditions kLowReWallFunction, epsilonLowReWallFuntion, omegaWallfunction and nutLowReWallFunction in the corresponding simulations. My problem is now the setup of the RSTM models (I don't care if LRR or LaunderGibsonRSTM at the moment). I can see that for the Reynolds stresses field R only a highRe wallfunction (so a real wallfunction in the original sense) called kqRWallFunction is available in OpenFOAM 2.4. Does this really mean that I can not perform a RSTM simulation with a lowRe mesh (y+ ~1) with OpenFoam 2.4? I am wondering because in my mind, a fixed Value uniform 0 boundary condition should be OK for the R-field if you mesh down into the viscous sublayer! But I know that e.g. for some k-epsilon models additional "damping functions" are necessary to enable their usage in a lowRe-case and I am wondering if that might be the case for the RSTMs implemented in OpenFOAM 2.4 as well? Unfortunately, the "Descriptions" inside the .h files of the RSTM models in $FOAM_SRC/turbulenceModels/incompressible/RAS are very short so I can not find my answer there either... :-( I try to sum up my questions again: 1) Are the RSTM models implemented im OpenFOAM 2.4 capable of being integrated down to the viscous sublayer in a y+~1 mesh (so are they "lowRe-capable"?) and only a corresponding wall boundary condition is missing in OpenFOAM 2.4 or are these models themselves only "highRe-capable" and it is really a must to use them on a y+~30 mesh? 2) Is my statement R=fixedValue uniform 0 (tensor with zeros of course) a correct lowRe-wall boundary condition for the R-field? This is of course only relevant if the answer to 1) is that the RSTMs themselves are lowRe-capable. Thank you very much for your help! Best regards in advance Sebastian P.S.: I also downloaded the papers by Launder, Reece and Rodi respectively Gibson and Launder where the models are described, but I have to admit that the theory in these papers is to complicated for me as a simple CFD user... I want to use CFD as a tool to create scientific outcome in an engineering application and I think I understand enough about CFD to use this tool CFD in a competent manner. But please understand that I am not capable of understanding every piece of theory or code about CFD out there.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Coupling RPI wall boiling model with population balance model in Fluent	softice2006	Fluent Multiphase	1	April 19, 2023 02:09
coalChemistryFoam: "Foam::error::printStack(Foam::Ostream&) at ??:?"	musabai	OpenFOAM Running, Solving & CFD	2	February 20, 2015 14:07
Is it possible to model natural convection in a 2D horizontal model in fluent	caitoc	FLUENT	1	May 5, 2014 13:32
Features of a good mesh for RSTM model	soroushsaberi	Main CFD Forum	0	November 5, 2013 20:11
Problems bout CFD model of biomass gasification, Downdraft gasifier	wanglong	FLUENT	2	November 25, 2009 23:27

November 19, 2015, 05:30		#2
canopus Member SM Join Date: Dec 2010 Posts: 97 Rep Power: 15	Dear Sebastian Did you find answers to your questions and can you share them? Thanks

November 24, 2015, 15:41		#3
sebonator Member Sebastian Lang Join Date: Aug 2009 Posts: 47 Rep Power: 16	Dear canopus, sorry, I haven't been able to answer for a while. Unfortunately, I did not find any answer myself. In the meantime, OpenFOAM3.0.0 was released but I just checked the source code: there is still no alternative wall function for the R-field than the kqRWallFunction and for this wall function still the restriction "high Reynolds number flow" is written in the .h file. I decided to leave the RSM models out of my study for that reason, what was very disappointing, though... Sorry that I can not help in this case. If you are able to find anything about the RSM usage in lowRe-cases, please let me know!!! Best regards, Sebastian

November 28, 2015, 09:49		#4
Roal-Yr New Member Roman Join Date: Nov 2015 Posts: 2 Rep Power: 0	Hello. I am currently working with a mesh which balances at y+ 10-50, and my case is a cyclone-like domain where I want to run a RSTM model. At the moment I am facing the same issue when I have to decide between performance and correctness of my calculations. I have an idea on how to check on whether RSTM are low-re capable or not, and it is, simply put, just a comparison of two options: 1). Purely high-re flow; 2). Purely low-re (on a very fine mesh, which I am afraid of calculating) 3). Something in between, when the y+ is at logarithmic sub-layer, but somewhere is getting to low for purely high-re simulation. To be honest I myself lack clarity on the matter of "what is happening below y+ <30". Seeing that non-functional BCs are applied at y+ >1, I stuck at understanding what is meant for y+ 1-30. I have limited amount of time to perform my calculations, so I will reply soon.

June 3, 2016, 06:36		#5
kaw New Member Dr. K.A. Weinman Join Date: Apr 2012 Posts: 1 Rep Power: 0	A low Re model with integration into the wall does not appear to have been implemented for the two-equation models. This means that the length scale equation for the RSTM is not strictly applicable for meshes with y+ < 1 or so. However, this can be corrected. The low Re BC for k is straight forward and omega/epsilon at the wall can be constructed by a blending approach, such as suggested by Menter and, in the same spirit, Knopp. I think that the length scale equation is the only issue with extending LRE treatments into the RSM models. Since the first grid point can be assumed to be at the wall, the other BCs should be reasonable trivial.