Suppose I take a pole (hollow, length>>width) and convert the CAD model into a surface model specifying a certain thickness and mesh it in a commercial CAE package such as ANSYS (in this case it'll be a surface mesh-quad)
Compare the results of the above analysis, with a similar ribbed model (not surface, solid) and do 3D meshing-hexa, tetra
Will the results be in 99.999% conformity?
Depending on the characteristics of the solid that you are studying.
The finite elements are derived from certain hypothesis that must be fulfilled in order to obtain accurate results.
For example, when studying the deflection of a beam using Timoshenko beam elements (1D), it is assumed that planar sections of the beam will remain planar once it is deformed under load. This is true for short beams, where one dimension is greater than others. If the beam that you are studying meets this hypothesis, the deflection result would be the same that (or really close to) the result of a 3D FEM model of the beam (assuming a decent FE mesh). If this hypothesis are not fulfilled, the result obtained by each model will differ.
Something similar would apply to 2D analysis. If the basic assumptions of the models (plane stress, plane strain, Midlin plates, etc) are fulfilled, results will coincide with the results of a 3D FEM model.
Hope that it helps.
Regards
In Finite Element Analysis the results you obtain depend on the type of elements and mesh refinement you use. If you use a very coarse mesh with shell elements, stress and strain values will differ from an analysis made on the same structure using a denser mesh or a different type of element, such as a 3D tetrahedral.
In very simple structures, such as yours, results can be the same, but in any structure with average complexity that's not what you should expect.
An important step in FEM analysis is carrying out a convergence test, in which the same geometry is used with the same type of elements with increasingly refined meshes in subsequent analysis. Reference values on these analysis, such as a stress at a critical point, are then taken down and compared. Usually there's a level of mesh refinement which does not show considerable differences from the previously coarser mesh, and so this should be the type of mesh used. 1% difference is in most cases acceptable, but it depends.
So probably you won't get results conforming 99.9999% using two different types of elements.
Hi,
You must make a study of sensitivity to the mesh size and the elements type on your numerical model.
you take three types : fine mashed, Medium and coarse . and see who offers a good results.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Computer Graphics and Animation
- 3D