Programmatic deletion of a link between two nodes

[fheinrich]

Dear all,

I linked two nodes (node 55 and node 121) in a simulation with the following command:

Code: [Select]

LINK
  55, 121, 0, 0, 0, 0
Now I would like to programmatically delete this link with the source-code during run-time
as soon as a certain condition is reached.

The question:
Could someone give me a hint where to implement such a deletion?

As far as I have figured out, the main routine dealing with the links is
"plink.f". Also, I figured out that the two nodes "55" and "121" can be found
in the variables n1, n2, m1 and m2 in the routine "plink.f".
Now I am looking for the right spot to tell the code: "Delete the link once the
condition is reached".

Many thanks in advance!

Best regards

PS: The closest thread I found in this forum is http://feap.berkeley.edu/forum/index.php?topic=1017.msg3079#msg3079 - unfortunately this does not apply to my problem as I want to implement the deletion myself.

[FEAP_Admin]

Have you tried making a second call to LINK:

Code: [Select]

LINK
  55 121 0 0 1 1


[Prof. R.L. Taylor]

The attached file does a solution when linked, then unlinks the node and does another solution.  You could use LINK instead of ELINk too.

[fheinrich]

Dear FEAP_Admin, Dear Professor,

many thanks for your hints which are indeed correct!
However, I am not sure whether this approach solves my problem.

I attached an image which hopefully makes it clearer what I intend to do:
1. I want to link two surfaces to each other (i.e., the nodes of the surfaces).
This is what I did with the command "LINK ..."

2. I want to apply a force, as shown in the attached image, preferably in many
load steps. As soon as a critical stress (or rather force) is detected at a certain
linked node-pair, I want to delete the link. In this way I want to model the break-up
of the interface. Think of two plates which are glued together. The interface then fails
at a certain stress.

For that reason, I think I cannot solve this problem only from the input-file. I assume
I rather have to detect the actual force after each load step within the code and then delete
the link somewhere (this is the question) in the code.

Many thanks for your efforts!

Best regards

[Prof. R.L. Taylor]

You should be able to use the 'gapnd.f' element in the directory /elements/frame as a prototype for the method you describe.  You may have to modify the routine to consider the force states you intend to simulate.

[fheinrich]

Dear Professor,

many thanks again for your hints and patience!

I am currently having a closer look at the 'gapnd.f'-routine. It looks very helpful, although
it does not seem to be trivial to get the routine running.

In the meantime, I would like to ask a general question which maybe makes my problem
clearer to others:
Does FEAP currently support the simulation of cohesive- or adhesive interface-failure? (Such as
the break-up of two bonded geometries at the interface. In particular, can FEAP somehow model
something like this (a cohesive zone model or something similar) : https://youtu.be/qBdI9D4fNjQ?t=296 ?)

As I did not find anything like this in the manual, I assumed that there is no "out-of-the-box" feature
to accomplish this. Therefore I am using the above approach to build this feature myself.

Best regards

[Prof. R.L. Taylor]

At present the program does not have any cohesive element.  One was started several years ago but never finished -- see attached for beginning.  Maybe this helps a bit.

[fheinrich]

Dear Professor,

many thanks for the clarification and the files!
I will have a deeper look at them and will post an update in case
they help to solve the problem.

Best regards

[Yaakov]

Dear Prof. Taylor,

Thanks very much for your cohesive element!

If I want to better understand  FE-Theory part of this user element (Residual + Tangent Matrix FORM for cohesive element),

Which Articles or Chapiter of Books should I read ?

Warm regards,
Yaakov 

[Prof. R.L. Taylor]

Yaakov,

Probably the best thing to do is Google "Cohesive Zone Models".  You should find several report and thesis publications you can download to study.  From a brief review of some of these it is evident that you will need to understand the application you have to know which of the many behavior models is appropriate for your application.