I am performing a nonlinear shrink fit (press fit) analysis in Autodesk Nastran and am seeing a significant discrepancy between a quarter-symmetry model and a full 3D model.
My geometry is a steel cylindrical block with a hole of ~1in depth. I have a carbide sleeve that is press fitted into the hole with a max dia interference of 0.003" (0.0015" radial).
Hand calculations estimate about 10.5ksi contact pressure with 0.0011" of sleeve ID displacement.
I ran a quarter model with symmetry and a full model with about the same setup except for the symmetry constraint.
Quarter Model (with symmetry):
- 1/4 model using symmetry boundary conditions
- Bottom of cylindrical block is fixed
- Surface contact between mating cylindrical surfaces
- Contact face mesh of 0.01" (need it to be scalable to an even bigger model)
- Shrink Fit/Sliding
- Friction 0.5
- Contact stiffness = 1
- Max allowable penetration = 0.0001"
- Max Activation Distance 1.2 x contact face mesh
- ~10 load increments
Result:
- Contact pressure ~14ksi and max displacement 0.0012"
Full Model (same setup, no symmetry):
- Same contact definition and parameters
- Only bottom of cylindrical block is fixed
- No constraints on inner sleeve
Result:
- Contact pressure ~100-400psi psi and max displacement ~0.0012-0.0018"
I suspect the symmetry constraint helped with getting close to analytical values. I have tried the following with the full model with no luck:
- adding/removing other contacts (ie. bottom of sleeve separation)
- rigid connector to restrict sleeve rotation
None of these worked.
I'm hoping someone can help me answer these questions:
- Why would a quarter-symmetry model produce close to theoretical contact force/pressure results while the full model (with identical settings) produces extremely low contact force? I suspect the symmetry constraint difference is behind this.
- How could I replicate the model stability from the quarter model to the full model?
- If displacement appears to be reasonable, can displacement results be considered reliable even when contact force/pressure results are not? What does the calculation chain look like for Shrink fit analyses in Nastran? Does it do calculate displacement -> force -> pressure?
Any input is welcome and appreciated!
