FEAP contact algorithm

[Tao Jin]

Dear all,

I am trying to use feap to solve some contact problems.
According to the feap manual, basically we have two solution procedures for a contact problem.
The first option is:
LOOP, contact, n-check
  CONTact CHECk
  LOOP, newton, n-iters
    TANGent,,1
  NEXT newton
NEXT contact

The second option is:
LOOP, newton, n-iters
  CONTact CHECk
  TANGent,,1
NEXT newton

I think the differences between these two options is that, the first option will check the contact status every N iterations and the second option will check the contact status every iteration.
To my problem, if I apply the first option, the solution will converge. But if I use the second option, which means that I check the contact status every iteration, it is very difficult to converge.

So my question is that:
Do we need to check the contact status every iteration? How should we choose between option 1 and option 2?

Thanks

[Prof. R.L. Taylor]

You do not need the contact check in the second approach -- by default the check will be done (usually).

Generally, the first approach does converge better but is more expensive due to extra iterations.  Describe your problem a little more and additional comments may be appropriate.  Is it 2 or 3d, friction? etc

[Tao Jin]

Thank you, Professor.

I am trying to simulate the contact problem between hydrogel and the rounded-corner steel support. It is a 3D contact problem without friction. I think one prominent issue is that the contact happens between a very soft surface(hydrogel) and a very rigid surface(steel).

I applied to two different FEM approaches.
In the first approach, I applied RIGId surface to represent the rounded-corner steel support. The in the definition of PAIR, I applied NTOR (Node to Rigid).
In the second approach, I established mesh for the rounded-corner steel support and assigned the material properties of standard steel to it. Then in the definition of PAIR, I applied NTOS (Node to Surface).
The material properties of hydrogel is a user-defined material coded by myself.

As I mentioned above. I guess one challenge is that the difference of stiffness of the two contact surfaces is very huge.

Thank you for your time, Professor.

You do not need the contact check in the second approach -- by default the check will be done (usually).

Generally, the first approach does converge better but is more expensive due to extra iterations.  Describe your problem a little more and additional comments may be appropriate.  Is it 2 or 3d, friction? etc