Boundary conditions

[kotawsu]

I have a question regarding the boundary condition for a plate which is simply supported on all four edges. How do i simulate this condition in a FE software. what i mean is what are the d.o.f that i would have to constrain considering each node has 6 d.o.f..Thanks.

 

[GBor]

Fix the three translations at each node (Tx, Ty, and Tz)...I'm still thinking about your core shear issues.

Garland E. Borowski, PE

 

[crisb]

If the plate is in the xy plane I would fix all the edges in the z direction and then fix just enough points in the x and y direction to prevent rigid body movement.
As the edges are simply supported no rotational restraints are required.
If you fix all the edges in xyz direction then membrane stresses will be induced which I assume you dont want as you said its simply supported.

 

[airmail]

As far as I know, simply supported conditions imply only vertical movement null. The rest of movement and rotations are allowed... So, only Tz=0

 

[feadude]

if you fix translations on all 4 edges and deflections are large you could get stress stiffening (in plane stresses) if this is the case and you do not want stress stiffening, then I recommend releasing translations (Tx, Ty) in x and y directions for three edges only.

 

[GBor]

I generally read "simply supported" as "pinned" for plate theory, but I've always needed the membrane stresses for my purposes...perhaps I made a bad assumption.

Garland E. Borowski, PE

 

[crisb]

feadude

Personally, I would only fix one point in xy direction and one other point in x direction.

In large dispacement analysis I think that the edges will try to move towards the centre of the plate, this distance will be greater at the middle of the edges (than at the corners).

 

[johnhors]

To avoid stress stiffening, then you need to apply a minimal set of restraints. Crisb has described this very well with his last post. To have one complete edge still totally fixed in all translations is over restraining the model and will create false stresses.

 

[Denial]

Don't forget that, depending upon your load configuration, the corners of a simply-supported, rectangular plate will want to move upwards. If they are restrained against upwards movement as well as downwards movement, fine. But if they are free to lift you should allow for this tendency in your model.

 

[airmail]

This is the continuation of my first contribution in this tip. Obviously, only Tz=0 restriction is indetermined in processing analysis (e.g. NASTRAN). So you have to impose (apart form Tz=0 in all borders) in two points of the plate's diagonal the next conditions (to avoid rigid body movement):
Tx=0 (one point)
Tx=0 and Ty=0 (the other point)

You will see that loads in these SPC's are negligible, so your results will be reliable.

 

[feadude]

crisb
you are right. One of the replies had restrained on all 4 edges and I did not think about over restraining the edge 4 if you remove extra restrains from the other 3 edges. I guess I was thinking about beam theory not plate.
Thank You

 

[kotawsu]

thanks everyone for ur inputs..i had used the bc as stated by airmail aleady and it worked perfectly fine. thanks once again

 

[GregLocock]

Well, by worked perfectly I'm assuming you mean it solved.

Does it agree well with the test data?

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[kotawsu]

hey greg..yes it matches well with the test data.thanks