I lost most of yesterday crashing Pro/E and Mechanica while trying to run a contact analysis in 2D Plane Stress mode.
Apart from some accuracy issues with meshing, and limitations of 2D mode such as not being able to apply a torque about the Z-axis, I eventually discovered that contact interfaces not only can't be created unless the edges are very close together, but can actually be deleted by a model shape update, causing the analysis to exit with a fatal error.
This morning I changed to running a 3D analysis, but cutting a slice through the model that is thinner (0.5 mm) than the typical element size. This allows application of moments as well as contact interfaces between any chosen pair of surfaces; but not only that, it seems more robust and runs faster than the 2D model!
I've already given up on 2D Axisymmetric, as it doesn't handle torque along the axis; so now it seems that a thin-slice 3D model is always a better bet than a 2D analysis.
Can you please file a Technical Support case with your 2D model and a detailed explanation of what went wrong? Regarding the moment on the z-axis, is it possible that it is not allowed because it will violate the assumptions of axi-symmetry?
Sorry Christos, but I've moved on from where I was (and purged) and I can't easily recreate the model. Specifically, I've created a volume region at assembly level, which is actually created with the part - and I cannot find a way to delete it! I've created call 10790853 about that instead...
Regarding moments in 2D, I can just about see that 2D Axisymmetric can't handle a torque along the axis because the shear component would be in the 3rd dimension (although it would not break axisymmetry); but for 2D Plane Stress, I see no reason not to be able to apply a moment rotating the component in-plane.
Regarding contact interfaces: in 3D they can be created between any two surfaces. In 2D they can only be created between edges that are nearly touching; if this criterion is broken during a Sensitivity Study (I was analysing gears and using Sensitivity to rotate them by updating an assembly parameter) then the analysis fails.
Okay, call 10792589 created. Can't even get it to start running (it used 67 hours of CPU time over the weekend, but now produces a fatal error) but I think it's caused by the contacts being deleted.
Applying a torque around the symmetry axis in a 2D axisymmetry model violates the 2D-axisymmetry assumption. Such a torque is not rotationally symmetric. A pie-piece and cyclic symmetry constraint can be used instead. Not only geometry, but also loads, constraints, material properties, and consequently, also the deformed shape, needs to be rotationally symmetric, for the 2D-axisym assumption to be valid. A planar cross section needs to remain planar after deformaion. Same goes for plane stress, plane strain.
Thanks for the reply. I frequently use "pie-slice" models with cyclic symmetry, and I also often have models where the loads are not rotationally symmetrical so I have to use a full model.
I'm just struggling to understand how a shaft in pure torsion isn't rotationally symmetrical - a cross-section through the axis at any angle would be identical, both before and after deformation. I assume it's really the planar deformation that's the issue (although I don't fully understand the maths).
So what can axisymmetric be used for? Pure centrifugal force; pressure in or on a cylindrical vessel; and pure axial loads? None of those are very useful to me.
Yes, 2D axisymmetry means that load, constraints, properties and also displacements, are all rotationally symmetric. So it's not very common, but very helpful when applicable. I particular in nonlinear problems, contact analysis etc. making use of plane theory is very useful.