I am experiencing the following problem:
I have a cantilever beam, with a force acting on the free end. The simple and classical problem.
The elemental stresses are very close to teh theoretical ones, while the nodal ones (even averageg) are two times bigger!!! And that even if I reduce the element size. I am using Chexa20 elements.
When I use Ctetra10, the differences (for the same element size) are smaller, but stil significant (40%). IN the same time, the results produced by the analysis are less precise that those obtained with Chexa (40% greater than the theoretical ones.
I am confused.
A cantilever beam is not only a classic engineering mechanics problem, it is also a classic example of stress discontinuity in a FEA analysis.
I assume that you have solid meshed a rectangular prism to represent a beam, then fixed all of the nodes on one end face to make it cantilevered. If so, then the plane containing the fixed nodes has an infinite stiffness. The remaining mesh volume has a finite stiffness based on the supplied material properties.
As you refine the mesh, the stresses and stress gradients through the first layer of elements (connected to the fixed nodes) continues to increase, approaching infinity asymtotically.
Thank you for your answer. I think I got your point. Thank you.
The two attached pictures present another model. It is a small assembly that comprises a steel plate, and a on top of it, a compozite enforcement. In the plate a crack has been induced (the red line in model.jpg). The other picture (half model.jpg) is the half model I used in NX10, with the symetrical conditions suggested in the picture. If I correctly understood you, it is normal to get significant differences between elemental and nodal stresses in the circled areas.
Am I right?