Non-Uniform Stress Distribution along curved surface

The graphical user interface of Elmer
Post Reply
andrew_rafalowitz
Posts: 10
Joined: 29 Dec 2014, 20:25
Antispam: Yes

Non-Uniform Stress Distribution along curved surface

Post by andrew_rafalowitz »

Hello,

I am trying to simulate a pin bearing stress along the interior surface of a hole. Half of the hole will have the stress applied to it and the stress distribution is sinusoidal in nature (magnitude of stress is 0 at theta=0 and theta=180 degrees.This hole also has depth(cylinder) since this is a 3D step file I am working with.

I would like the stress to remain constant along the z axis as well.

Please see attached image of bearing stress distribution for reference.http://blog.3dvision.com/wp-content/upl ... bution.png

I am using ElmerGUI but have found that there is no option to apply bearing stress(only normal stress and F-x/F-y/F-z.) as a boundary condition. I am trying to formulate a user defined function using MATC. I would like to construct the force components as functions of the angle. The domain of the angle would be from 0 degrees to 180 degrees. I am having difficulty generating a function for the angle that is required for the force function(f=F*sin(theta)).

Any help setting up a user defined function and/or help with creating a bearing stress distribution would be greatly appreciated.

This link may also help with how the force distribution should be applied http://help.solidworks.com/2011/english ... bution.htm
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: Non-Uniform Stress Distribution along curved surface

Post by mzenker »

Hi,

for creating user functions, see the chapter "basic programming" in the Solver manual. There are also many examples for ufd's here in the forum.

HTH,

Matthias
biocca
Posts: 21
Joined: 09 Jul 2014, 19:28
Antispam: Yes

Re: Non-Uniform Stress Distribution along curved surface

Post by biocca »

Hi,

You need to use a MATC language to define this boundary condition. The easy way is defining a normal stress like a bondary condition. The sif file should look as follows:

Code: Select all

Boundary Condition 1
  Target Boundaries(1) = 1    ! 1 is the number of surface (for 3d problem) or line (2d problem)
  Name = "non-uniform stress distribution"
  Normal Force = Varible Coordinate; Real MATC "function(tx(0),tx(1),tx(2))"
End
Where tx(0), tx(1), tx(2) are x, y and z respectively. You need to write theta like a function of x and y, using atan(y/x), suppose F=10 in your notation,

Code: Select all

 Normal Force = Varible Coordinate; Real MATC "10*sin(atan(tx(1)/tx(2)))"
This should work for writing this stress profile.

Regards. Nicolás.
andrew_rafalowitz
Posts: 10
Joined: 29 Dec 2014, 20:25
Antispam: Yes

Re: Non-Uniform Stress Distribution along curved surface

Post by andrew_rafalowitz »

Nicolás,

Does the coordinate axis need to be at the center of the hole for this to work?

For simple geometry, shifting the coordinate axis to the center of the area of applied loading is not a problem.

When the geometry is not as simple or if there are multiple regions of applied loading that don't have the same x & y coordinates, does the force distribution get applied differently?
biocca
Posts: 21
Joined: 09 Jul 2014, 19:28
Antispam: Yes

Re: Non-Uniform Stress Distribution along curved surface

Post by biocca »

Hi,

Yes, it is necessary place axial coordinate in the center of the hole, only for the expression previously mentioned. If the hole position is not centered, you need to write the stress distribution as a function of your system reference x-y-z, wich can be more o less difficult depending on the geometry, but it is not impossible, i think that if you have the expression written for a hole centered if you only need to translate the hole in the same plane you need to make a change of variables. Suppouse you have an expresion for a centered hole, F(x',y'), if the hole is centered at xo and yo, making a change of variable x' = x - xo and y' = y - yo should work in this case.

Regards. Nicolás.
andrew_rafalowitz
Posts: 10
Joined: 29 Dec 2014, 20:25
Antispam: Yes

Re: Non-Uniform Stress Distribution along curved surface

Post by andrew_rafalowitz »

Nicolás,

Where in the Solver Input File (SIF) would I perform the change of variables for the new coordinate axis that is actually centered at the hole?
biocca
Posts: 21
Joined: 09 Jul 2014, 19:28
Antispam: Yes

Re: Non-Uniform Stress Distribution along curved surface

Post by biocca »

Hi,

See Elmer Solver Manual chapter II, the section is called boundary condition, basically:

Code: Select all

Boundary Condition 1
  Target Boundaries(1) = 1    ! 1 is the number of surface (for 3d problem) or line (2d problem)
  Name = "non-uniform stress distribution"
  Normal Force = Varible Coordinate; Real MATC "10*sin(atan(tx(1)/tx(2)))"
End
Making the next transformation in the above expression

Code: Select all

x' = x - xo
y' = y - yo
Where x' and y' are centered in the hole, x and y are the system reference of yor model, xo and yo are the position of the centered hole measured on x-y. In the sif file this means changing:

Code: Select all

tx(1) = tx(1) - xo
tx(2) = tx(2) - yo 
I hope to be clear enough.

Best regards. Nicolás.
andrew_rafalowitz
Posts: 10
Joined: 29 Dec 2014, 20:25
Antispam: Yes

Re: Non-Uniform Stress Distribution along curved surface

Post by andrew_rafalowitz »

Nicolás,

The transformation for the coordinates helped out a lot. I was able to get results out of Elmer that matched the analytically obtained answers.

Thanks again.

Sincerely,

Andrew
Post Reply