WARNING: DIV(V)-QTL too large!

[Ambrish Pandey]

Hello All,

I am simulating convection flow in a two-dimensional horizontally-periodic domain and I am using the temperature-dependent thermal diffusivity, i.e., the variable properties option in version19. However, I am getting a warning 

WARNING: DIV(V)-QTL too large!

which persists for a few (hundred) time steps before it disappears and then reappears. This goes on and off during the simulation. I am attaching two steps from the output of my simulation in the following. Could someone please tell me the causes of this warning and suggest some workarounds?

Thanks a lot in advance.

Step 219842, t= 1.3543491E+02, DT= 3.4546896E-05, C=  0.556 9.6905E+04 6.0066E-01

             Solving for Hmholtz scalars

     219842  Hmholtz TEMP     118   9.5694E-06   7.8401E-01   1.0000E-05

     219842  Scalars done  1.3543E+02  5.1468E-02

             Solving for fluid

     219842  PRES gmres       200   2.2458E+00   2.2458E+00   1.0000E-05   1.9120E-01   4.0124E-01    F

     219842  Helmh3 fluid     165   9.3139E-06   4.1988E-01   1.0000E-05

             L1/L2 DIV(V)           2.1400E-15   3.1305E-01

             L1/L2 QTL              0.0000E+00   0.0000E+00

             L1/L2 DIV(V)-QTL       2.1400E-15   3.1305E-01

             WARNING: DIV(V)-QTL too large!

     219842  Fluid done  1.3543E+02  5.4160E-01

219842 Nusselt  1.354349E+02  1.482339E+01  5.596478E+00  2.953501E+03

Step 219843, t= 1.3543495E+02, DT= 3.4546896E-05, C=  0.559 9.6906E+04 6.0298E-01

             Solving for Hmholtz scalars

     219843  Hmholtz TEMP     118   9.6770E-06   7.8408E-01   1.0000E-05

     219843  Scalars done  1.3543E+02  4.9909E-02

             Solving for fluid

     219843  PRES gmres       200   2.2545E+00   2.2545E+00   1.0000E-05   1.9129E-01   4.0191E-01    F

     219843  Helmh3 fluid     165   9.2503E-06   4.1984E-01   1.0000E-05

             L1/L2 DIV(V)           2.2357E-15   3.1437E-01

             L1/L2 QTL              0.0000E+00   0.0000E+00

             L1/L2 DIV(V)-QTL       2.2357E-15   3.1437E-01

             WARNING: DIV(V)-QTL too large!

     219843  Fluid done  1.3543E+02  5.4190E-01

219843 Nusselt  1.354349E+02  1.482342E+01  5.596443E+00  2.953511E+03

Best regards,
Ambrish

[Stefan K.]

Something is wrong here. The pressure solver doesn’t converge (iteration cap is 200).

On 24 Feb 2021, at 04:30, Ambrish Pandey <ambri...@gmail.com> wrote:

Hello All,

--
You received this message because you are subscribed to the Google Groups "Nek5000" group.
To unsubscribe from this group and stop receiving emails from it, send an email to nek5000+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/nek5000/FC6F28D2-A1A0-45CF-A86F-349245BFEC73%40gmail.com.

[Ambrish Pandey]

Hello Stefan, 

Could you please suggest some solutions to overcome this issue?

Thanks.

Best regards,
Ambrish

--------------------------------------
Dr. Ambrish Pandey
Post-Doctoral Associate
Center for Space Science
New York University Abu Dhabi
PO Box 129188, Saadiyat Island, Abu Dhabi
United Arab Emirates
--------------------------------------

[Stefan K.]

It's hard to come up with any suggestion without any additional
insight. Can you please share your setup?

[Ambrish Pandey]

Dear Stefan,

Thanks for your prompt response. Some details the flow, which I am trying to simulate are following:

a) I am trying to solve non-Oberbeck-Boussinesq (NOB) thermal convection flow in a two-dimensional rectangular domain of aspect ratio two with no-slip and isothermal horizontal walls and periodic sidewalls.

b) The NOB effects arise due to the use of a temperature-dependent thermal diffusivity \kappa(T), whereas the viscosity is kept constant.

c) I am using the variable thermal diffusivity as a polynomial in temperature T, i.e.,

\kappa(T) = \kappa_{top}(1+149T+350T^3).

d) I am prescribing the thermal diffusivity at the top plate \kappa_{top} and viscosity \nu such that Pr = 12.73 and Ra = 1e10 at the top plate. These two parameters then have fixed values also at the bottom plate due to constant temperature boundary condition but vary in the flow.

e) The vertical temperature profile is not symmetric with respect to the midplane and we observe a temperature stratification near the top plate.

f) We thus use an asymmetric grid in the vertical direction with a lot more grid points near the top plate to adequately resolve the asymmetric temperature field.

g) I start my simulations with random initial perturbation and set the temperature, pressure, and velocity tolerance as 10^{-5}.

Please let me know if you need any further details.

Best regards,
Ambrish

--------------------------------------
Dr. Ambrish Pandey
Post-Doctoral Associate
Center for Space Science
New York University Abu Dhabi
PO Box 129188, Saadiyat Island, Abu Dhabi
United Arab Emirates
--------------------------------------

[nkul...@ncsu.edu]

Hi Ambrish,

Could you please share how to setup variable properties (temperature dependent viscosity and thermal diffusivity)? Can it be somehow applied to a non-dimensional simulation?

Thank you

Nilay

[Smitten Clark]

good question，I also meet the same problem about how to setup variable properties