I have never found a free floating boundry condition. Try analyzing 1/2 or 1/4 of the model and then adding Symmetry in Advanced fixtures. Sometimes this works for me.
Restraints in thermal models can be tricky.
You want to be sure that they don't induce artificial stresses.
Off hand you could try this a couple of ways:
1. As Randy suggests, cut the model into 1/4ths and use 2 symmetry
restraints. Fix one of the vertices on the floor. This will allow the model
to grow in every direction and not leave the "floor" -- what it does not do
is keep your model from rotating.
You could try activating "soft spring" and see how that works in a thermal
2. the classic way to do this is think through each and every one of your
constraints. Constrain vertices only. The ones that touch the "floor" of course.
In one corner (any corner), FIX the vertex. This will reduce your degrees of
freedom per above.. it can't translate anywhere, but it can rotate.
to keep it from rotation add a "Reference geometry" constraint to the next
vertex over (not the one diagonally opposite). Make that so that you limit only
one direction.. ie the vertex can move towards or away from the other fixed vertex
only. This will fully constrain your model (it can no longer spin) -- but keep in mind
all of your deformations will be AWAY from these vertices (the model won't "grow"
from the center when heated).
Hope that helps.
Thanks for the good feedback...which were kind of things that I have tried. It's very good to know you are suggesting similar. I will try further as you commented.
By the way, when you say "fix one vertice on the floor"...does this mean to fix one vertice all three direction? I wonder if there is kind of "floor" in the option of restraints.
In oder to make the model grow from center of the mode, can I simply choose the one point at the center of model instead of one corner vertice?
If I fix only one vertice, as I remember, the solution becomes "error" not even go though....due to rotation issue as you mentioned. It's good to know and I am trying to fix reference plan in addition to one fixed vertice....by the way, in this case....I think that I still can not prefernt the model move up and down.....while I could prevent the rotation....
There is no real "floor" option.
The only time you might need a "floor" is if you are also considering
gravity and your model could droop.
Have a look at attached image. This should work for you.
THERMAL.jpg 35.0 KB
Thanks for the fantastic advice and drawing....which I confirmed works! I could generate nice symmetric results.
By the way, would you confirm if I can use the exactly same boundary conditions if I use the 1/2 model ...instead of 1/4 model?
Also, I wonder if you can advise and confirm if there is a way to make or display the results for full model though we did simulation with 1/2 or 1/4 model? I couldn't find...though I think that there should be the way....
I don't believe this is a way to show results for the full model.
Yes you could use the same boundary conditions for a 1/2 symmetry model.
In fact, if your FEA solves quickly, you could run the entire model with the same
restraints. You will have to use some split lines to break the model into 4 pieces.
Split the "bottom" face. The only thing the split allows you to do is to select
vertices for the same restraints you used in the 1/4 model.