Comments

• Victor

  January 2018

  A rule of thumb for slope of pressure-overclosure curve is
  

  (Young's modulus / thickness) * 10

  
  That should converge OK, then you can increase it by a factor of 10 at a time if penetration is too high. Too low and you get large penetration. Too high makes the convergence slow or fail. I'm not sure if there's also a too-high region where it leads to errors in stress though.
  
  The stick slope is similar in that a higher value is more accurate by can cause convergence failure if it's too high. I don't use it much but I think a similar value to the slope of pressure-overclosure curve would be a reasonable starting point. It's probably easier to start with 0 for the friction parameters until the contact is working OK, then add it on if you need it.
  
  Stress at the contact surface is notoriously noisy. Mesh refinement should improve it. Have you looked at CCX's contact stress output? That might be better quality than the element stress which is extrapolated from the integration points inside the elements rather than right on the surface.
  
• Roli

  January 2018

  Very thanks for your answere. It was good to know to take the mesh into account.
  
  I did various tries on the mesh and geometry (it is a circular bolt in a bush). It worked after I took quadratic elements with mid size nodes. It seems the element type was the issue to get the high stresses in the contact zone.