Convergence

[QuentinG]

Hello,

I am doing a quick calculation of a steady laminar flow over a plate with code saturne.
I want to show that the boundary layer thickness is equal to d=4,92.x/[(Rex)^1/2]
In the same time I want to show that the thermal boundary layer thickness is : d_T=d*Pr^(1/3)

I just did a rectange 10*20*100 cm, (with the top face as a symmetry), the fluid is air ;
the inflow velocity is 1m/s, inflow temperature 40°C;
the wall temperature is 20°C;
100 iterations (0.01 pseudo step time)

But my results are wrong : I find a temperature going to 321 °C, the velocity does not show any boundary layer..

1) Does anyone has an idea why this is not converging ?

2) What is the main criteria (in the listing file) of convergence (I saw both Norm. Residual, Derive, and Rhs Norm) ?

Thank you,

P.S. : my Norm residual for velocity pressure and temp is approximately 10e-3, whereas my derive is of 10e-3 for the velocity, 10e-1 for pressure and 10e3 for temperature (seems really high).

[Yvan Fournier]

Hello,

What type of mesh are you using ? Could you post the log file types listed in the forum recommendations section ?

Derive is the main convergence indicator, but there is no automatic convergence test, and plotting probes is a recommended way to check the convergence behavior.

Best regards,

Yvan

[QuentinG]

The mesh is a CGNS (2.5) file produced using Hypermesh, code saturne check mesh is telling me that the type is "MIXED"

extract from the check_mesh :
Zone 1: "cfd_model"; type "Unstructured"

85753 vertices; 419557 cells

(cartesian coordinates, "RealDouble" precision, unit "Null")

Section 1: "Fluid";

(indices 1 to 419557, type "MIXED")

Section 2: "Inflow";

(indices 419558 to 421853, type "MIXED")

Section 3: "Outflow";

(indices 421854 to 424149, type "MIXED")

Section 4: "Wall";

(indices 424150 to 457943, type "MIXED")

Section 5: "Symmetry";

(indices 457944 to 480373, type "MIXED")

[Yvan Fournier]

Hello,

This only provides me limited information. Could you please also post the info which is listed in the forum usage guidelines ?

Regards,

Yvan

[QuentinG]

Ok sorry,
My version of C_S is the 4.0
I do not have any file called compile.log (is it strange ?)
I send you the mesh preprocessor.log and listing if you need anything more tell me

Thank you

[Yvan Fournier]

Hello,

I don't see anything obviously wrong with your time step/CFL values, but as long as you only post part of the recommended information, I can't guess at if/where you may have an error in our setup.

Regards,

Yvan

[QuentinG]

Hello,

I'm sorry I just saw :

In case the recommendations above are not enough, providing the following details when posting to this forum will help others answer:
• version of Code_Saturne,
•the compile.log,
•the listing and preprocessor.log (or listpre for old versions),
•in parallel, the error_n* files if they are present.

If it's possible:
• user sources and xml file,
• meshes.

But as I told you I have neither compile.log, nor error files.
Could you tell me what file is missing ?

Regards

[Yvan Fournier]

Hello,

How about the setup (xml and/or user subroutines) ?

Regards,

Yvan

[xingjian]

Hello,

What type of mesh are you using ? Could you post the log file types listed in the forum recommendations section ?

Derive is the main convergence indicator, but there is no automatic convergence test, and plotting probes is a recommended way to check the convergence behavior.

Best regards,

Yvan

Hello Yvan,

How to plot probes with values of derive in the listing? In order to see if it converges.

Thanks very much

Kinds Regards

XING Jian

[Luciano Garelli]

Hello,

As Yvan mention, you could use monitoring (probe) points inserted through the GUI (Calculation Control -> Output control -> monitoring points) in different spatial location in order to check the time/iteration evolution of your variables (pressure, velocity, etc), but not the derive indicator. If you want to check the derive of any of the variable you can parse the listing file during the simulation with this command for example for the pressure

Code: Select all

tail -f listing |  grep  "c  Pressure"

Or after/during the simulation with

Code: Select all

cat listing |  grep  "c  Pressure" | cut -c52-64

which will cut columns

Regards

Luciano