Elastomer-metal Interface stresses

[biomedman]

I am modelling a elastomer bonded to a metal in ABAQUS. An ideal bond between them (using tie constraints) is assumed. When I ran the analysis, I got high stresses at the interace and upon refining the mesh, the stress values continued to rise. My questions are

1) How believable are the stress values at the interface
2) What is a good method to check for convergence

Regards
Biomedman

 

[pjhype]

Biomedman,

What you have "discovered" is a known singularity in the mathematical theory of linear elasticity that occurs at the interfaces of dissimilar materials (boundary layer effect). The singularity is due to the fact shear stress is being transferred between the two materials, however, shear stress cannot be supported on the free edge. Therefore, the mathematical theory tends to infinity near the free edge. High stresses actually happen in real life, however, when the yield stress is reached, linear elastic theory is no longer valid, and the stresses are not infinite. As long as you are using a linear solution, the stresses will continue to climb with increasing mesh density.

pj

 

[biomedman]

PJ, Thanks for your feedback. Since i am getting high stresses, it probbaly means the real life application will also see high stresses, but my analysis cannot predict what happens once it gets into the non-lineriaties.

At the material discontinuities, the node values are way higher than those at the element integration points. What does that mean and which do I look at?

Biomedman

 

[pjhype]

You should look at the stress gradients to determine which stresses look more reasonable. Keep in mind that the nodal stresses are average stresses obtained through the interpolation of the stress field defined by the element integration points.

In real life, there are plenty of applications where elastomers are bonded to metals where failure of the adhesive layer does not occur. Linear finite element analyses for such applications should therefore be interpreted carefully, as some of the assumptions in linear elastic theory may not be directly applicable for elastomeric materials.

pj

 

[GregLocock]

Poor profiling of the fillet where an elastomer is bonded to the metal is a very common cause of failure on prototype bonded rubber or PU bushes. Failure of the bond itself is also possible (the difference is that in one case the elastomer remains stuck to the metal, and tears away above it, in the other case it peels cleanly away from the metal).

I'm sorry I can't tell you how to solve this problem, I'd like to know myself, but (a) it CAN be solved, using FEA (one of our suppliers used to have a consultant who analysed our designs and he was very very good) and (b) when you learn how to do it you will have a very worthwhile skill.



Cheers

Greg Locock

 

[ParaDoc]

...in real life, on the interface between the metal and the rubber part, there is a small layer of "material" in which (on the rubber side!) the density of "molecular linkages" is much higher than in the rubber bulk itself. That means the stiffness of the rubber increases rabidly when approaching the mentioned interface. With other words, modelling appropriately, you may use a constitutive model which allows for this "gradient".

This would automatically decrease the stresses at the interface, as is always the case in "real-life"...