Low value of Turbulent heat fluxes in Natural convection

Questions and remarks about Code_Saturne usage
Forum rules
Please read the forum usage recommendations before posting.
Post Reply
kunalck
Posts: 15
Joined: Fri Sep 21, 2018 2:06 pm

Low value of Turbulent heat fluxes in Natural convection

Post by kunalck » Mon Apr 08, 2019 3:23 pm

Hello Everyone,

I am running Rayleigh Benard natural convection case with Rayleigh number to be 630000 and Pr as 0.71. I am using AFM as turbulent heat flux model. For boussinesq implementation, I am adding boussiesq(rho*beta*(T-t_ref)*g) source term in momentum equation.
I am facing an unusual problem regarding the turbulent heat flux magnitude. The turbulent heat fluxes for this case are incredibly low value(1e-38 to 1e-36). Whereas the temperature profiles are predicted well when compared to DNS results.
Also the turbulent kinetic energy (i.e. rij values) is also incredibly low.

Code_saturne v 5.1.3.
Turbulence model : RSM-EB
turbulent heat flux model : AFM
The attachment afm_dns.png is no longer available
afm_dns.png
Kindly suggest few changes.


Regards
Kunal
Attachments
63e4_afm_data0.csv
(2.15 MiB) Downloaded 26 times

C0st4s
Posts: 42
Joined: Fri May 11, 2018 12:21 pm

Re: Low value of Turbulent heat fluxes in Natural convection

Post by C0st4s » Mon Apr 08, 2019 9:43 pm

Hi,

May I asked how did you implement Boussinessq?

i.e cs_user_source_terms.f90 under ustsnv or cs_user_physical_properties.f90 under usphyv?


Cheers,
Costas

kunalck
Posts: 15
Joined: Fri Sep 21, 2018 2:06 pm

Re: Low value of Turbulent heat fluxes in Natural convection

Post by kunalck » Tue Apr 09, 2019 9:00 am

Hello,

I used ' cs_user_source_terms.f90' under ustsnv for Boussinesq approximation.

Regards
Kunal

C.FLAG.
Posts: 20
Joined: Fri Apr 08, 2016 2:19 pm

Re: Low value of Turbulent heat fluxes in Natural convection

Post by C.FLAG. » Tue Apr 09, 2019 10:29 am

Hello,

I have started a case with natural convection under 6.0-beta, Rij-SSG + AFM, the temperature / turbulent heat fluxes have a very high amplitude, 10e50. Switch back to 5.0 if you need results quickly. Otherwise, wait for fixes.

P.S.: tests performed with the temperature flagged as buoyant in cs_user_parameters.c and the SIMPLEC time scheme. Switching to PISO and nterup = 3 did not improve the situation.

Code: Select all

void
cs_user_parameters(cs_domain_t   *domain)
{

  /* Declare the temperature as a buoyant scalar */
  cs_field_set_key_int(cs_field_by_name("temperature"),
                       cs_field_key_id("is_buoyant"),
                       1);
}
Best regards,
Cédric

Jean-Francois Wald
Posts: 8
Joined: Fri Apr 27, 2018 2:33 pm

Re: Low value of Turbulent heat fluxes in Natural convection

Post by Jean-Francois Wald » Tue Apr 09, 2019 12:54 pm

kunalck wrote:
Mon Apr 08, 2019 3:23 pm
Hello Everyone,

I am running Rayleigh Benard natural convection case with Rayleigh number to be 630000 and Pr as 0.71. I am using AFM as turbulent heat flux model. For boussinesq implementation, I am adding boussiesq(rho*beta*(T-t_ref)*g) source term in momentum equation.
I am facing an unusual problem regarding the turbulent heat flux magnitude. The turbulent heat fluxes for this case are incredibly low value(1e-38 to 1e-36). Whereas the temperature profiles are predicted well when compared to DNS results.
Also the turbulent kinetic energy (i.e. rij values) is also incredibly low.

Code_saturne v 5.1.3.
Turbulence model : RSM-EB
turbulent heat flux model : AFM
afm_dns.png
63e4_afm_data0.csv
Kindly suggest few changes.


Regards
Kunal
Thank you for your feedback on the AFM model. Your temperature profile looks good probably because the GGDH part of the AFM works well whereas the additional terms ( bouyancy, velocity gradient,...) might be bugged.

Can you post your test case so that I can have a look ?

C0st4s
Posts: 42
Joined: Fri May 11, 2018 12:21 pm

Re: Low value of Turbulent heat fluxes in Natural convection

Post by C0st4s » Thu Apr 25, 2019 10:55 am

Hi,

Thanks for the clarification. Another question if you do not mind. If you use Boussinesq through cs_user_source_terms.f90 under ustsnv what did you do with the gravity vector in the generation rate of turbulence (Gk)?

Cheers,
Costas

kunalck
Posts: 15
Joined: Fri Sep 21, 2018 2:06 pm

Re: Low value of Turbulent heat fluxes in Natural convection

Post by kunalck » Mon Apr 29, 2019 2:21 pm

Jean-Francois Wald wrote:
Tue Apr 09, 2019 12:54 pm
kunalck wrote:
Mon Apr 08, 2019 3:23 pm
Hello Everyone,

I am running Rayleigh Benard natural convection case with Rayleigh number to be 630000 and Pr as 0.71. I am using AFM as turbulent heat flux model. For boussinesq implementation, I am adding boussiesq(rho*beta*(T-t_ref)*g) source term in momentum equation.
I am facing an unusual problem regarding the turbulent heat flux magnitude. The turbulent heat fluxes for this case are incredibly low value(1e-38 to 1e-36). Whereas the temperature profiles are predicted well when compared to DNS results.
Also the turbulent kinetic energy (i.e. rij values) is also incredibly low.

Code_saturne v 5.1.3.
Turbulence model : RSM-EB
turbulent heat flux model : AFM

afm_dns.png
63e4_afm_data0.csv
Kindly suggest few changes.


Regards
Kunal
Thank you for your feedback on the AFM model. Your temperature profile looks good probably because the GGDH part of the AFM works well whereas the additional terms ( bouyancy, velocity gradient,...) might be bugged.

Can you post your test case so that I can have a look ?
Hello Jean-Francois,

Thank you for your feedback. I have also run the same case with unsteady solver(CS_v4.0) and I could not prevent the laminarization. Although, temperature profile was like previous case(predicting well compared with DNS).
u_rbc1.xml
(10.66 KiB) Downloaded 26 times
Rbc_3_mesh.cgns
(1.2 MiB) Downloaded 26 times

kunalck
Posts: 15
Joined: Fri Sep 21, 2018 2:06 pm

Re: Low value of Turbulent heat fluxes in Natural convection

Post by kunalck » Mon Apr 29, 2019 2:26 pm

C0st4s wrote:
Thu Apr 25, 2019 10:55 am
Hi,

Thanks for the clarification. Another question if you do not mind. If you use Boussinesq through cs_user_source_terms.f90 under ustsnv what did you do with the gravity vector in the generation rate of turbulence (Gk)?

Cheers,
Costas
Hello,
I gave the value of gravity as a negative vector in GUI. And when using Boussinesq through cs_user_source_terms.f90, then the I would use gy for gravity.

Post Reply