How to resolve boundary layer in OF ?

Blue8655 · May 3, 2018, 17:01

Hi,

I tend to use high Reynolds wall functions in my simulations, but I'm in a case where it seems more appropriate to resolve the boundary layer. However, I don't know how to do this in OpenFOAM, I just don't know what boundary condition to use for solid walls. At first I thought I should use:

type fixedValue;
value uniform 0.01;

for k, epsilon and nut, but the simulation crashes or provides poor result. Moreover, I don't understand how some authors seem to avoid the use of wall functions, since (Launder and Spalding,74) say that the k-epsilon model is not defined near walls... An example of authors who solve the boundary in OF would be: " (Limane et al. 2015) Thermo-ventilation study by OpenFOAM of the airflow in a cavity with heated floor". They don't mention the BC used though.

Some people** talk about low-Re turbulence models but (Limane,2015) used buoyantSimpleFoam and other "regular" models (BSF, BPF, and BBPF) while keeping y+<11. So I'm quite confused right now!!

Any help appreciated! Pointing me to an older post is good too. I use OF-3.0.1. Thanks you very much.

**Solving boundary layer with k-e model

piu58 · May 4, 2018, 00:47

A wall is connected to wall functions, a patch is not. If you use patches at the boundaries, you may left out the wall functions without problems.

For an accurate result you need to resolve the boundaries down to the region y+=1, combined with very small time steps. Or in other words: You need to calculate a DNS simulation.

Another way would be to work with LES models and wall functions. If you use finer and finer meshes, more and more of the turbulence is calculated directly. But you have the wall function as a kind of safeguard.

Blue8655 · May 4, 2018, 10:36

Thank you for reply Uwe. If I get you right... you are telling me that if in polyMesh/boundaries, I use 'patch' instead of 'wall' for those solid surfaces, my simulation will behave much better ? I never would have thought about this! I will try it soon.

P.S. Stupid question, but how can I search for an exact expression in the CFD-online forum ? (e.g. "wall function")

piu58 · May 4, 2018, 13:06

'Much better' in the sense 'close to that what you wish.

It may be however, that wall functions give a amore reliable result (if you cannot afford the spatial and time resolution you need).

Blue8655 · May 6, 2018, 18:59

I tested your suggestion Piu58. It works. Using y+<1, flow behavior is coherent.
At first, OF complained about a wall function still present for alphaT.
( type compressible::alphatWallFunction; )
I used "calculated" instead, as for inlet and outlet. Worked.
I agree with your comment though that in most cases I will continue to use wall functions to limit computational requirements.
Thanks for your help. Much appreciated.

May 3, 2018, 17:01	How to resolve boundary layer in OF ?	#1
Blue8655 New Member Join Date: Jun 2015 Location: Montreal Posts: 11 Rep Power: 11	Hi, I tend to use high Reynolds wall functions in my simulations, but I'm in a case where it seems more appropriate to resolve the boundary layer. However, I don't know how to do this in OpenFOAM, I just don't know what boundary condition to use for solid walls. At first I thought I should use: type fixedValue; value uniform 0.01; for k, epsilon and nut, but the simulation crashes or provides poor result. Moreover, I don't understand how some authors seem to avoid the use of wall functions, since (Launder and Spalding,74) say that the k-epsilon model is not defined near walls... An example of authors who solve the boundary in OF would be: " (Limane et al. 2015) Thermo-ventilation study by OpenFOAM of the airflow in a cavity with heated floor". They don't mention the BC used though. Some people talk about low-Re turbulence models but (Limane,2015) used buoyantSimpleFoam and other "regular" models (BSF, BPF, and BBPF) while keeping y+<11. So I'm quite confused right now!! Any help appreciated! Pointing me to an older post is good too. I use OF-3.0.1. Thanks you very much. Solving boundary layer with k-e model

May 4, 2018, 00:47		#2
piu58 Senior Member Uwe Pilz Join Date: Feb 2017 Location: Leipzig, Germany Posts: 744 Rep Power: 15	A wall is connected to wall functions, a patch is not. If you use patches at the boundaries, you may left out the wall functions without problems. For an accurate result you need to resolve the boundaries down to the region y+=1, combined with very small time steps. Or in other words: You need to calculate a DNS simulation. Another way would be to work with LES models and wall functions. If you use finer and finer meshes, more and more of the turbulence is calculated directly. But you have the wall function as a kind of safeguard. __________________ Uwe Pilz -- Die der Hauptbewegung überlagerte Schwankungsbewegung ist in ihren Einzelheiten so hoffnungslos kompliziert, daß ihre theoretische Berechnung aussichtslos erscheint. (Hermann Schlichting, 1950)

May 4, 2018, 13:06		#4
piu58 Senior Member Uwe Pilz Join Date: Feb 2017 Location: Leipzig, Germany Posts: 744 Rep Power: 15	'Much better' in the sense 'close to that what you wish. It may be however, that wall functions give a amore reliable result (if you cannot afford the spatial and time resolution you need). __________________ Uwe Pilz -- Die der Hauptbewegung überlagerte Schwankungsbewegung ist in ihren Einzelheiten so hoffnungslos kompliziert, daß ihre theoretische Berechnung aussichtslos erscheint. (Hermann Schlichting, 1950)

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
sliding mesh problem in CFX	Saima	CFX	46	September 11, 2021 07:38
[snappyHexMesh] Help with Snappy: no layers growing	GianF	OpenFOAM Meshing & Mesh Conversion	2	September 23, 2020 08:26
Radiation in semi-transparent media with surface-to-surface model?	mpeppels	CFX	11	August 22, 2019 07:30
Problem in setting Boundary Condition	Madhatter92	CFX	12	January 12, 2016 04:39
Convective Heat Transfer - Heat Exchanger	Mark	CFX	6	November 15, 2004 15:55

May 4, 2018, 10:36		#3
Blue8655 New Member Join Date: Jun 2015 Location: Montreal Posts: 11 Rep Power: 11	Thank you for reply Uwe. If I get you right... you are telling me that if in polyMesh/boundaries, I use 'patch' instead of 'wall' for those solid surfaces, my simulation will behave much better ? I never would have thought about this! I will try it soon. P.S. Stupid question, but how can I search for an exact expression in the CFD-online forum ? (e.g. "wall function")

May 6, 2018, 18:59		#5
Blue8655 New Member Join Date: Jun 2015 Location: Montreal Posts: 11 Rep Power: 11	I tested your suggestion Piu58. It works. Using y+<1, flow behavior is coherent. At first, OF complained about a wall function still present for alphaT. ( type compressible::alphatWallFunction; ) I used "calculated" instead, as for inlet and outlet. Worked. I agree with your comment though that in most cases I will continue to use wall functions to limit computational requirements. Thanks for your help. Much appreciated.