Nodal Displacement variations 1

[hashanw007]

Hi,
i simulated a similar structure (actually more like a plate) with different element types
1 3D solid element
2. plane stress element
3. plane strain elements
4. shell elements

forces were applied in such a way that its sigma y is applied at the center of the structure.
i observed that maximum displaced node is the exact center node. BUT the nodal displacements alone the vertical & horizontal center line is different for diffeerent element types.
For vertical Disp -
plane strain element's nodes have maximum displacements, then comes 3D solid,plane stress and the minimum displacements has occurred for the shell elements.

Does anyone know the reason for this change?? its the same structure with same dimensions for all the element types.
(i have used reduced integration and Quadratic for element types)

 

[rb1957]

in-plane forces ?

 

[hashanw007]

forces acting in the negative Y direction only and there's a roller as well.. for all elements it is the same force same dimensions... everything is the same...
i have attached the graph

 

[prost]

Are all computed displacements numerically converged?

How long is this thing? My approach here is to determine if your end boundary conditions are affecting the results? What are the boundary conditions on the ends? Simply supported? Fixed? Are all boundary conditions on all models the same?

From your description, it appears you are doing the simple problem of a beam (or plate) loaded with a constant normal traction (in beam nomenclature, a 'constant distributed force'). Say this is a case of a rectangular loaded with constant pressure (or distributed force)--the plates/shells and 3D elements would have 4 boundaries to apply boundary conditions--are all four fixed? Or simply supported? Perhaps you applied simply supported to just 2 of the edges?

Perhaps if you have a Roark's, you could tell us precisely the table and solution number of the structure, loading and boundary conditions you are trying to simulate?

 

[detandj]

You are really mixing up everything. 2D models are supposed to be applied when the loads and deformations are merely working in the plane itself.

2D plane stress is used for very thin parts: the stress components orthogonal to the deformation plane are zero. The orthogonal strains aren't.

2D plane strain is used for infinitely long parts: de strain components orhtogonal tot the deformation plane are zero. The orthogonal stresses aren't.

It is very obvious that you get other results since the used element equations differ. Otherwise there would be no reason to distinguish between them.

A shell works primarily as a curved plate that acts as a membrane (uniform stress) combined with bending stress. This is a 3D (curved) plate with a small thickness. The element equations behave completely different from a 2D situation.
When the thickness is small in comparison to the length or width, you can use a shell model.

For a flat an thin plate, the results of 2D plane stress and shells must be almost the same.

A SOLID gives a good solution when the plate is not too thin. Thin plates make the meshes to much deformed. Deformed meshes do not give very good results. You should use in that case wedges or bricks or use a very fine mesh, but this increases calculation time enormously.

 

[GBor]

hashanw007,

Looking at the thread that you started in the mechanical engineering forum, and this thread, I believe you need to read a book before asking this forum to supplement your education. These questions are relatively basic for an engineer with any time in an FEA class. Self-study will do a great deal so long as you understand some reasonable level of math and some fundamentals of mechanics of materials.

There are plenty of books recommended in this forum if you do a search. One that comes up often is A First Course in the Finite Elements Method by Logan

I don't mean to be rude...I just want to save all of us some time and frustration.