CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > OpenFOAM Running, Solving & CFD

Symmetric 3D Cylinder solver doesn't converge

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Display Modes
Old   February 27, 2013, 14:11
Default Symmetric 3D Cylinder solver doesn't converge
  #1
New Member
 
Join Date: Feb 2012
Posts: 28
Rep Power: 5
Matt_B is on a distinguished road
Hi Foamers,
I'm experiencing a problem with my solver. I have a 3D cylinder mesh, filled with air, with a rotating top lid at a constant angular velocity. I set my solver starting from pisoFoam, which is an incompressible fluid solver, and then I modified it, introducing all needed for a compressible fluid. Before I run it on the whole cylinder mesh and everything worked well; but after I decided to decrease the calculation effort by means of using just 1/4th of the cylinder cutting it by the 2 available symmetry planes. So, in brief, now I'm running on a 90 part of a cylinder.
Since I changed the geometry, the calculation blows up very soon, with just a few time steps, either with bigger or smaller timesteps and independently on the mesh size. I think I defined in a proper way the two simmetry plane with a simmetryPlane patch and every paramether in "0" folder has this defined.
Going deeper into the calculation results, I understand that the U velocity calculation is responsible for this and it seems that the BC defined as symmetryPlane are considered as a wall instead; I see this because I have z components of U along the cylinder height which are not expected at all.
I'm wondering if there is something else I should define also for a symmetric calculation which I don't know, or if there is some known problem with this issue.
Ah, the turbulent model is k-epsilon.

Thanks in advance!

Matteo
Matt_B is offline   Reply With Quote

Old   February 28, 2013, 13:29
Default
  #2
New Member
 
Join Date: Feb 2012
Posts: 28
Rep Power: 5
Matt_B is on a distinguished road
Has somebody experienced my same troubles with a symmetric volume?
I attach here below a picture of Uz component rising after one timestep (deltat=0.05 s) and of U. I'm expecting not having such a Uz component.
I also attach U dictionary file with the starting boundaries values.



Code:
FoamFile
{
    version     2.0;
    format      ascii;
    class       volVectorField;
    object      U;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

dimensions      [0 1 -1 0 0 0 0];

internalField   uniform (0 0 0);

boundaryField
{
    top
    {
        type            rotatingWallVelocity;
        origin (0 0 10);  //  l'asse di rotazione rispetto all'origine degli assi
        axis (1 1 1);  // deve rimanere sempre uguale a (1 1 1)
        omega constant 1;
    }

    side
    {
        type            fixedValue;
        value           uniform (0 0 0);
    }

    bottom
    {
        type            fixedValue;
        value           uniform (0 0 0);
    }
    oneSym
    {
        type            symmetryPlane;
    }
    twoSym
    {
        type            symmetryPlane;
    }
    invisible
    {
        type            zeroGradient;
    }
}
Matt_B is offline   Reply With Quote

Old   February 28, 2013, 13:58
Default
  #3
ngj
Senior Member
 
Niels Gjoel Jacobsen
Join Date: Mar 2009
Location: Rotterdam, The Netherlands
Posts: 1,592
Rep Power: 24
ngj will become famous soon enoughngj will become famous soon enough
Hi Matteo,

A symmetry plane does exactly what the name suggest, namely that it keeps the flow variables symmetric across the patch. For all scalar quantities this results in a zeroGradient condition, whereas the component of a vector normal to the patch most be zero, because that it the only possibility.

I believe that what you are after is to specify cyclic boundary conditions on the two patches. I have never used them for that type of layout, but the boundary condition do perform a transformation of vectors due to changes in direction of the normal vector for the two patches, thus it should work.

An alternative, again something I have never tried, is to consider whether you could benefit from the wedge-type boundary condition. Many post are available on this topic on the forum.

Good luck

Niels
ngj is offline   Reply With Quote

Old   March 1, 2013, 04:42
Default
  #4
New Member
 
Join Date: Feb 2012
Posts: 28
Rep Power: 5
Matt_B is on a distinguished road
Quote:
Originally Posted by ngj View Post
Hi Matteo,

A symmetry plane does exactly what the name suggest, namely that it keeps the flow variables symmetric across the patch. For all scalar quantities this results in a zeroGradient condition, whereas the component of a vector normal to the patch most be zero, because that it the only possibility.

I believe that what you are after is to specify cyclic boundary conditions on the two patches. I have never used them for that type of layout, but the boundary condition do perform a transformation of vectors due to changes in direction of the normal vector for the two patches, thus it should work.

An alternative, again something I have never tried, is to consider whether you could benefit from the wedge-type boundary condition. Many post are available on this topic on the forum.

Good luck

Niels
Hi Niels,
thank you for you reply. Now, everthing is clearer to me. I just don't understand how and when using symmetryPlane by the way....I thought it should be implemented for symmetric volume, diving it by means of symmetry axis and mirroring the result of a slice of the volume on the whole volume.
What you say about zeroGradient of scalar and vector quantities makes sense to me for my actual results.
As suggested by you, I will try to define cyclic or wedge patches, but I have to investigate more about their implementation.
Thank you again for now!

Matteo
Matt_B is offline   Reply With Quote

Old   April 21, 2014, 23:51
Default
  #5
New Member
 
Chad
Join Date: Sep 2011
Posts: 16
Rep Power: 5
chaolian is on a distinguished road
Quote:
Originally Posted by Matt_B View Post
Hi Niels,
thank you for you reply. Now, everthing is clearer to me. I just don't understand how and when using symmetryPlane by the way....I thought it should be implemented for symmetric volume, diving it by means of symmetry axis and mirroring the result of a slice of the volume on the whole volume.
What you say about zeroGradient of scalar and vector quantities makes sense to me for my actual results.
As suggested by you, I will try to define cyclic or wedge patches, but I have to investigate more about their implementation.
Thank you again for now!

Matteo

Hi, Matteo,

I have exactly the same issue. Do you have any idea how to do with symmetryPlane in OpenFOAM?
Thanks in advance.

Chao
chaolian is offline   Reply With Quote

Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
OpenCL linear solver for OpenFoam 1.7 (alpha) will come out very soon qinmaple OpenFOAM Announcements from Other Sources 4 August 10, 2012 11:00
different results between serial solver and parallel solver wlt_1985 FLUENT 10 April 11, 2012 15:25
inviscid 2d computation for M=0.3 flow past cylinder on fine mesh can not converge boubalos Main CFD Forum 3 March 20, 2011 05:26
Working directory via command line Luiz CFX 4 March 6, 2011 21:02
Getting too many iterations by velocity solving (aborting). Changing U - Solver? suitup OpenFOAM Running, Solving & CFD 0 January 20, 2010 08:45


All times are GMT -4. The time now is 19:50.