Hello all,
I am trying tò simulate an interference fit between two bodies through *CLEARANCE with negative value equal to value of interference.
I saw that there are some differences than Abaqus in terms of Von Mises stress.
What do you think about this method?
I got the dame result by using *clearance and without, Just using higethe penalty behavior
Do you use other suggestions or other method?
I simulated interference fit in CalculiX without using the *CLEARANCE keyword: https://www.youtube.com/watch?v=sos-6ilPIZc
It doesn’t have to he used for this purpose in Abaqus either.
Yes normally clearance it is used to adjust the geometry artificially, but It can be used also as interference because create pre-stress in Abaqus.
Lets assume that I have a CAD formed by 2 components in contact themself with zero penetration and I wanna give and interference fit. In other words the mesh of two component are zero-zero and I wanna five a interference without modify the mesh.
In abaqus generally the clearance is used for this application. What about Calculix?
This should be doable with *CLEARANCE in CalculiX too, according to its description. But why don’t you want to model the parts with an initial interference as it’s typically done in such cases ?
it occurs to me that it can be very interesting if one has to try different values without having to modify geometry and meshing.
Because a lot of times, either the CAD are nor perfect or if we wanna change the value of the interference to understand what happen in terms of stress it is very useful to model the component zero-zero varying the value of the interference.
For a curved surface like a pin in a hole, with different meshes, there will be some discretization error in the meshed interference/clearance which could easily dominate the tiny interference you’re trying to model. *CLEARANCE eliminates that error by zeroing it away.
@Iacobellis88 don’t worry about differences in stress unless you can tell that it’s actually wrong. You can do a mesh study separately in both programs to see if they converge to the same value.
Yeah, I know the benefits of this approach. I was just wondering why the OP wants to use it in CalculiX. The traditional approach with mesh overlap is still very common (if not the most common) in Abaqus.
Unfortunately, there are no example files for this keyword mentioned in the CalculiX documentation but you could take a simple benchmark problem from Abaqus and try to use it in CalculiX with the same settings. The results may still be different but they shouldn’t differ too much. And you can check them analytically for basic cases.
1 Like
I didn’t know about *Clearance KEYWORD and a priori it seems to have many advantages.
I have taken a look and applied it to one of my models which I know has very good agreement with the theory and if I’m not doing something wrong it is not the same.
*Clearance applied to a component with zero-zero mesh ends up with an open GAP in-between the two surfaces instead of approaching them together.
Initial geometric interference fit closes the two surfaces (with more or less penetration depending on the contact Stiffness).
Seems a different problem.
The final pressure distribution across the curved surface is not as clean as the one for the model starting with geometric interference.
Did you specify a negative value (indicating overclosure) for *CLEARANCE ?
At the end, the surfaces may appear separated but the stresses should be correct (from Abaqus):
When an initial overclosure is specified, the two bodies will deform during the step to resolve the prescribed overclosure and will appear to be separated by the specified amount at the end of the step.
Also, in Abaqus, this only works with small-sliding contact and there are more ways to specify the initial overclosure value.
Yep,
Oh, so what I see is not real. Thanks for pointing that.
That seems incompatible in ccx. *CLEARANCE only applies to face-to-face contact while SMALL SLIDING is only available for Node to surface.
In any case I can’t explain the difference in results. The Contact pressure range is wider with clearance.Problem is mostly the same as in your tutorial but with different numbers.
My theoretical result was 308.11 MPa. I have rescued my final model beside the *CLEARANCE OPTION
1 Like
I have tried another approach.
I have set up the surface behavior as TIED. (No penalty parameter to worry about 
).
The result is amazingly uniform but unfortunately the final model clearance and consequently the Contact Pressure is wrong. (Not sure if this could be a bug).
Scaling the input clearance according to the result delivers
a contact pressure across the surface with a variation of less than +/- 0.03% and a deviation from the theoretical value of 0.62%.
I’m considering three options:
1-Linear / Hard contact affects the previous stablished *clearance. Something similar to what happened with pretension if it wasn’t frozen.
2-There is some bug with the magnitude of the clearance . (Direction seems right)
3-I’m doing something wrong in my set up.
Apart from that:
I don’t have clear how the friction could be introduced here or if it even has any effect when *Clearance is used.
The surfaces in regular contact problems are fighting against friction to accommodate and find their final positions. I’m not sure if that’s the case when using *Clearance or ccx resolving an initial interference fit.
Seems like surfaces avoids all that extra work jumping directly to their final positions.