We are trying a shrink fit of a sleeve in an end cap under internal pressure but we are getting some strange results in a couple of places. See attached pictures, anyone seen this behavior before?
Works fine w/o the shrink fit, but when I apply it to the two faces with 0.24mm overlap the base deflects in the wrong direction and there is some strange deformation of the flange near the bolt hole..
Cheers.
Are you change the deformation scale of model to 1 or true scale? Do you use shrink fit? (Types of contact)
Satit Nilyai wrote:
Are you change the deformation scale of model to 1 or true scale? Do you use shrink fit? (Types of contact)
The scale is 1:1, we are getting large displacements, about 50mm in the centre. Yes, it is a shrink fit contact set between the outside face of the sleeve and the inside face of the cap. If I suppress the shrink fit contact then the study runs as expected, no large displacements.
It correct. you suppress the shrink fit contact. "In pair surface is bonded"---> No reality, So you going to option Large displacement. "Right click study name/option/large displacement"
It not correct and nothing is bonded, all is no penetration. I only suppressed it to test the model for accuracy. My interference is very small (0.24mm) hence large displacements should not be used. The problem is most likely caused because the interference is less than 0.1% of the outside diameter of the larger part (about 0.39mm). This falls within the discretization error range, which is most likely causing the problems. However.. this type of interference is very common so why are the tools that are meant to solve these problems flawed for the applications for which they are designed..
Has anyone found a way to run these problems accurately?
In this case interference is very small. The effect large displacement, you can checking static problem. "shrink fit on static study"
But When the model have to increase interference? As you have to run on non linear problem. "shrink fit on nonlinear study"
However you attach file on web. I'm find solution to you.
