[Junrui]

Originally Posted by sinatahmooresi

Hello Junrui,

In addition to the mesh refinement adjacent to the wall, I suggest you to switch to cubic convective schemes in your fvSchemes dictionary. Using this, I could solve the problem of turbulence generation in my pipe flow case.



Sina

Hi Sina,
Thanks so much for your advice! I'm now using STARCCM+ to carry out my simulation. When I switch my scheme to MUSCL cubic(I wonder if I have the right understanding of the cubic convective scheme), the upwind mixing factor needs to be decided. May I ask the value of the upwind mixing factor in your setting?
Thanks a lot!
Regards,
Junrui

[sinatahmooresi]

Originally Posted by Junrui

Hi Sina,
Thanks so much for your advice! I'm now using STARCCM+ to carry out my simulation. When I switch my scheme to MUSCL cubic(I wonder if I have the right understanding of the cubic convective scheme), the upwind mixing factor needs to be decided. May I ask the value of the upwind mixing factor in your setting?
Thanks a lot!
Regards,
Junrui

Hello Junrui,

By cubic, I mean using a higher order scheme. <<Gauss cubic corrected>> is third-order accurate in OpenFOAM. I hope this helps.



Sina

[Junrui]

Originally Posted by sinatahmooresi

Hello Junrui,

By cubic, I mean using a higher order scheme. <<Gauss cubic corrected>> is third-order accurate in OpenFOAM. I hope this helps.



Sina

HI Sina,
Thanks a lot for your advice! Actually I use this to obtain a fully developed turbulent profile for another region. I used the central difference scheme and the velocity profile seems close to the experiment results, however the turbulent kinetic energy is small and the pressure gradient is also small. Secondly, when the velocity is mapped to the region in interest, it's not converged. Convergence may be a problem especially for higher order scheme. Have you ever come across this kind of situation?
Many thanks to your reply!
Regards,
Junrui

[Junrui]

Originally Posted by guanjiang.chen

Hi J,

I have abandoned the mapped BC. What I do is conducting the RANS simulation first and then set the RANS results as the initial set for the LES.

Regards,
Guanjiang

Hi Guanjiang,
I did conduct the RANS simulation first and set the results as the initial set for LES, however, the velocity profile still seems not like fully turbulent flow. May I ask what kind of scheme for convective term did you apply?
Thanks for your time.
Sincerely,
Junrui

[JM27]

Hi Junrui,

What RANS turbulence model did you use? My profile also looks laminar

[Junrui]

Originally Posted by JM27

Hi Junrui,

What RANS turbulence model did you use? My profile also looks laminar

Hi，
I have tried several RANS models including RKE, SST. Making sure the y+ matches the wall function may help.

[guanjiang.chen]

Originally Posted by Junrui

Hi，
I have tried several RANS models including RKE, SST. Making sure the y+ matches the wall function may help.

Hi Junrui,

The RANS results will looks time-averaged. The turbulence will occur in the converged LES results I think. The schemes I use are Gauss linear, Gauss limited linear, and Gauss LUST grad(U).

Regards,
Guanjiang

[Junrui]

Originally Posted by guanjiang.chen

Hi Junrui,

The RANS results will looks time-averaged. The turbulence will occur in the converged LES results I think. The schemes I use are Gauss linear, Gauss limited linear, and Gauss LUST grad(U).

Regards,
Guanjiang

Hi,Guanjiang
Thanks a lot for your advice. Do you think this may have something to do with the B.C. because I use the mapped B.C. in my simulation? And would you mind telling me about the y+ in your LES simulation? I wonder if making sure y+ less than 1 is enough for LES simulation.
Regards,
Junrui