User login

Navigation

You are here

Post-Fracture behavior XFEM

Hi
iMechanica,

 

For my
graduation project I need to model the cracking behavior of a concrete beam in
bending (To simulate delamination of Steel Reinforced Polymer).

However I’m
relative new to FEM, I’m using Abaqus 6.10 (and have 6.11 available).

To get used
to the modeling of cracking behavior I started with a simple example, as described
in:

 

Alfaro, M.
Suiker, A. (2010) Transverse Failure Behavior of Fibre-epoxy Systems, Journal
of Composite Materials, Vol. 44, No. 12/2010.

 

Here a
section of fibre and epoxy are loaded in tension and cracking will occur.

 

I modeled
the geometry as described, see attached image, and modeled the fibre-epoxy
interface with a cohesive surface. This seems to work fine.

I did model
the cracking of the epoxy with XFEM: In the material behavior, I used Maxps
Damage and Damage Evolution (type=energy). Since complete cracking is expected
viscosity and automatic stabilization is added.

The sample
is loaded, by placing a displacement on the right edge.

 

However
when I run the calculation the post-cracking behavior is completely different
then expected (see attached image). I would expect that after complete cracking
of the epoxy has occurred, no stresses would be possible to act in the left
hand side of the sample. Since on this part there is no load acting. Also in
the part on the right I would expect no stresses since this part is after
complete cracking able to get a free displacement.

 

Does anyone
now how to model past-cracking behavior in a good way in Abaqus with XFEM? Thus after
complete cracking of the sample that no stresses would be able to be transferred.
I would expect an behavior as Figure 7 in Alfaro, M and Suiker, A. (2010)
(after an initial stress peak, reduction in transfarable stresses).

 

 I must add,
that the calculation is not completed, since my laptop is not able of
performing the complete analysis. It will abort the calculation after free body
displacement occurs. (see attached image 2).

I added my .cae file in case anyone would want to take a look at it 

 

 Thank you
all in advance for you help!

Steven

 

Image 1Image 2 

rgates's picture

Questions I would ask myself:

 Is it possible that plastic deformations of the round part induce elastic deformation within the surrounding material?

If the simulation was aborted at time of rigid body translation the last available load step should be the one immediately before the final detachment occurs. This is the boundary of the validity of such an analysis and has nothing to do with your laptop.The stiffness matrix should become singular if I am not mistaken. 

Hi Rgates,

 

Thank you for the reply and taking the time to look into the problem.

however there are no plastic deformations in the cricular part (which is the fibre in the fibre-epoxy interface). I only moddeled it with elastic materialproperties. Thus it should not be able to present plastic deformations. Also when i look into the strains and displacements (see attached picture) both of them are zero in the loaded situation.

This is what maybe is most unclear to me: The occuring displacements are zero (for the left part), however there are still acting strains and stresses in the model. Which would make no sense to my opinion.  

 

About the rigid body translation, i did add stabilisation and viscosity to the model, to  make a rigid body displacement "possible". This seems to work, since a couple of iterations are made after complete cracking has occured (STATUSXFEM=1.0).

 

I'm lookin forward to any more suggestions!

Thanks in advance!

Steven 

 

 Displacements in modelActing strains in model

Andrealina's picture

Hi:

   Can you tell me the parameters in your cohesive surface? You can try how the deformation occurs without the interface property

mmalekan's picture

Dear Steven:

It is obvious that when you using cohesive surface at specific surfaces, the damage would initiate from those surfaces not the other part of model. But, when you using the XFEM approach, the damage can be initiate from every part of model (when damage criteria meet at some points). Here is my questions:

1- Are you sure about your loading type and BCs?

2-  Are your damage parameters correct?

 

M. Malekan

 

Hi,
Thank you for the reply's.

 Before creating this model, i first moddeled examples of a model with
only the cohesive behavior and a seperate model with only the XFEM behavior.
Both of the models are showing
behavior as expected.


I did model the cohesive surface with the following parameters: 

**
INTERACTION PROPERTIES 

** 
*Surface Interaction, name=Cohesive
1., 
*Cohesive Behavior
 0.1, 0.1, 0.1 
*Damage Initiation, criterion=MAXS
 2.5e-05, 2.5e-05, 2.5e-05 
*Damage Evolution, type=ENERGY, mixed mode behavior=POWER LAW, power=1.
 
0.0005,
0.0005, 0.0005 
*Damage Stabilization
1e-05
** 

 

The boundry conditions and loading is really straight forward. Simple
roller supports are added as BC's. The load is added as a displacement on the
right edge of the sample.

 

As for the damage parameters: To my opinion these should be correct, as
they are derivated from the literature (
Alfaro,
M. Suiker, A. (2010)
. And are also tested in a
seperate model, where the behavior was as expected.

 

I did upload an input file and the python script i use to start the
calculation (to avoid "Unexpected system error 193 IPC_error"
 which is caused by a wrong communication between the Abaqus GUI and
Abaqus calculation kernel
). 
I would be very thankfull if you would be willing to take a look at this.

The .inp and .py files can be found here:

http://files.engineering.com/getfile.aspx?folder=7335e7cf-5d04-45cb-b897-62a81fe7303a&file=XFEM-solve-problem.zip 

 

 Again thank you all for already havind a look at the problem!

 Steven

 

Hello Steven,

 

can I contact for more detail about your work. I am working actually on interface cracks for coating application.

Regards

Subscribe to Comments for "Post-Fracture behavior XFEM "

Recent comments

More comments

Syndicate

Subscribe to Syndicate