compression only support on curved face

rhoka

hi All,

looking from another forum.

http://forum.freecadweb.org/viewtopic.php?t=7224

MECWAY can solve this case. i reproduced a similar case with simple models, but failed to convergence.

attached my input files. what's wrong with my model?

thanks you

VMH

rkoka, the compression only support is a non-linear type support and we should limit the convergence limit to a number that would be acceptable to you. I usually limit mine to 0.001 or 0.0001.

However, there was another issue when I attempted to run the analysis: your meshes weren't connected. See a quick video by clicking on link below. If you restraint at least one node in the Y-direction along the load neutral axis, your results would be exactly symmetric. With only compression only support being used, it wanted to rotate a little (See this video at 4:22 ).

rhoka

thank you for explanation in your videos.

yes, i forget to merge nodes create by mirror commands.

try to following you in setting converge as 0.001 the problem solved.
but result expectation not the same as you shown in video.

the deformation shown the plate is moving outward, i just apply frictionless support at both side of plane then running. problem case solved an the result shown as expected, convergence faster than before.

is this such a procedure in MECWAY to run compression only support on curved faces?

next time i'll try in a different case to study the behavior.

VMH

rhoka, I'm not sure why your results came out different from the video since I used the same model you provided.

Since this is a 3D analysis, you also need to restraint the Z-direction. Once you do that, you will have all three directions properly restraint and the results should come out as how we would expect them. If you do a 2D analysis, only two directions (X and Y) need to be properly restraint. This should hold true for any software not just Mecway.

You don't have to use frictionless supports on both sides of the plate to restrain the Z-direction. Probably one or two nodes restrain in the Z-direction along the neutral axis of the load would be enough for the program to solve.

rhoka

yes, the result are not the same. plate moving out of plane. i don't know what's a problem with my pc.

in my work using other FE software this additional restraint at some nodal required for numerical stability purpose (e.g. plate rested on spring foundation) but in MECWAY for many case it's not a mandatory.

previously i did not found how to add node restraint, since right click on selected node did not show any menus. this program behavior are differs comparing to selected face. now i was found in tree's menu.

restraining both plane surface make another problems in reaction force.

i try another case: only apply symmetric boundary condition and models did not have any restraint in Z direction. the problem solved with convergence without issues.