help on contact definition .. 1

[bagarbilla]

i want to simulate a component compressed by two surfaces .. one moving in and second stationary .. i defined contact b/w the surface and the compressed body ... i want to know if there is any recommended practice or procedure in application of contact problem and to apply loads on it.. and how to trouble shoot if solution not converging ?? ..

 

[imagitec]

There are many considerations. I'll mention one to start.

There are three contact models that you can choose from. In increasing order of stiffness and accuracy, they are:

1. Pure penalty
2. Augmented Lagrange
3. Normal Lagrange

With higher stiffness, however, the solution is more likely to "chatter", with contacting surfaces bouncing off each other. It will take longer to converge and is more likely to not converge. Not only do the general models vary in stiffness, but you can vary parameters of each model that affect stiffness.

The pure penalty model permits the most penetration of the contact surfaces in the converged result; it's reduced in Augmented lagrange; and it's near zero in Normal Lagrange. Keeping in mind that FEA is an approximation in many ways, ask whether your problem truly requires no penetration (e.g., predicting the preload in a ball bearing). Even if it does, consider beginning with Pure Penalty or Augmented Lagrange to work out other aspects of the model first.

Evaluate the amount of penetration in the final result to determine if it is reasonable. You can do a stiffness study, varying the contact model parameters, to see if the results change radically with increased stiffness.

Rob

Rob Campbell, PE