Learning Forum

[Elia Battistini]

Good morning,

I am trying to simulate a wear process (simple 2D plain strain analysis, asymmetric contact).

The analysis was successful in the case of bare base metal. In this case, I defined a single frictionless contact in WB and I added a command snippets to define wear properties. In LS 1 I solve a simple static structural problem. From LS 2 on, I keep constant loads and the analysis is driven by the wear.

At this time, I would like to define two different contacts on the same surface, with different wear properties (coating and base metal). The aim is to switch from one contact to the other based on calculated displacement to simulate the coating consumption. This will be based on contact element birth and death.

I defined two identical contacts sharing the same contact and target surfaces and I kill one of them (say, the coating) before solving LoadStep 1. What I expected was to get the same results I already have from the pure base metal analysis. Note that in this trial I just killed (EKILL) the coating contact elements, never to reactivate them (EALIVE). But this did not happen: the analysis shows, for every LS, constant displacements equal to those of LS1. If I suppress the contact instead of killing the contact elements, the analysis works just fine, as expected.

What is the problem? How can I get around it?

Thanks

[John Doyle]

You describe "...two identical contacts sharing the same contact and target surfaces...".  What happens if you define two asymmetric pairs with scoping that is equal and opposite to each other?

[Elia Battistini]

 

 

Hi John, thanks for your suggestion.

I did the modification you suggested and what I see now is that wear works effectively. The order of magnitude of displacements is now as expected and the wear-induced displacement growth is progressive, whereas I had constant displacements before the modification.

What I cannot understand is that as soon as I define the second contact and I kill it at LS1, I get a higher wear rate, and consequently a bigger final displacement (max from 4 mm to 7 mm). Needless to say, I use the same constants to define the wear behaviour as in the case of single contact analysis. In addition, even if I set the wear constant of the second (killed) contact to 0, I get bigger displacements.

 

EDIT: the increased wear rate is due to increased contact pressure. I do not see the reason for this pressure increase, though. The second contact is killed before LS1 “SOLVE” command.

 

[John Doyle]

Not entirely sure why results might be trending in a direction opposite of what you expect.  It might be worth mentioning that each contact pair is acting independent of the other.  They will influence each other, but they do not "see" each other.

[Elia Battistini]

Actually they are not be trending in an opposite direction. Compared to the same model with just one contact pair defined, wear is just faster (because contact pressure is higher) if a define a second contact and I kill it before I solve LS1. That is really weird and I cannot justify in any manner. 

Any idea about it?

 

[John Doyle]

oh I see.  Yes, that does seem strange.   The killed elements are still in the model, they are just not contributing anything to the analysis and no results are saved for killed elements.  Are all the initial contact properties exactly the same?  Maybe we are calculating a different default property for the one pair that remains alive.