Discontinuous Galerkin for convective heat transfer problems

Numerical methods and mathematical models of Elmer
Post Reply
vencels
Posts: 26
Joined: 20 Sep 2016, 17:05
Antispam: Yes
Location: Latvia
Contact:

Discontinuous Galerkin for convective heat transfer problems

Post by vencels » 13 Aug 2019, 11:14

Hi,

I am trying DG for simple convective heat transfer problem - curved pipe, 2D
Image

Result (temperature dstribution):
Image

Expected result:
Image

Any ideas why this could happen?
Attachments
hexInflation2D.zip
Mesh
(21.25 KiB) Downloaded 22 times
heatSolver-Elmer.sif
Case file
(3.04 KiB) Downloaded 25 times

kevinarden
Posts: 319
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Discontinuous Galerkin for convective heat transfer problems

Post by kevinarden » 13 Aug 2019, 15:23

temperature.png
temperature.png (84.98 KiB) Viewed 452 times
case.sif
(3.36 KiB) Downloaded 32 times

vencels
Posts: 26
Joined: 20 Sep 2016, 17:05
Antispam: Yes
Location: Latvia
Contact:

Re: Discontinuous Galerkin for convective heat transfer problems

Post by vencels » 13 Aug 2019, 17:43

Although it is not Discontinuous Galerkin solution, your solution helped me to find a bug in my viscosity value. I was using kinematic instead of dynamic. Thanks!

annier
Posts: 1123
Joined: 27 Aug 2013, 13:51
Antispam: Yes

Re: Discontinuous Galerkin for convective heat transfer problems

Post by annier » 14 Nov 2019, 05:55

Juris,
I would say it is difficult to get convergent solutions for such a low value of viscosity for transient simulations (asssuming the original kinematic viscosity magnitude in the order of 10^{-7} supplied into the N-S equation is interpreted as dynamic viscosity by ElmerSolver ). In your case you have performed steady state simulations, and so the solutions were obtained without difficulty.
Yes, the viscosity in Elmer is always supplied to solver input file as dynamic viscosity (Pa s) , and not as kinematic viscosity (m^{2}/s).
The unit of mu (Div(velocity)) must return the unit of Div(stress), and this is true for the unit of viscosity (mu) in [Pa S] or [ Ns/m^{2}].

Yours Sincerely,
Anil Kunwar
Anil Kunwar
Department of Materials Engineering, KU Leuven, Belgium

Post Reply