Crack propagattion and Cohesive Elements in Abaqus

[eispiata]

Hi everybody,

I read up more on "crack propagation" and "cohesive elements" in Abaqus. But i am still struggling to make it run and frankly i am running out of time. I have got only two more days to get the model running...

I recall the problem that i want to simulate:

A piece of rubber (hyperelastic neo-hokean) material is subjected to static uniaxial tension untill it fractures. As you can see in the attached picture, there is initially no crack in the geometry. However there is a singularity that is likely to be the crack initiation point. And the crack is going to propagate along the horizontal dotted line.

I am using Abaqus Standard to simulate static crack propagation.

To begin with i sketched the geometry in the Part module and then i partitioned the geometry to create an interface that is going to play the role of the crack propagation line. My idea is that by making use of "cohesive elements" in the interface i could be able to get a crack propagation.

So in the Property Module i created a "hyperelastic neo-hookean material" to account for the rubber material outside of the interface. I then created two sections:
- a "solid homogeneous" section to account for the hyperelastic material outside of the interface (top and bottom regions)
- a "cohesive section" with the same material properties (hyperelastic material) as rubber to account for the crack path...

In the Interaction Module i didn't impose any constraint on the interface faces. I don't really understand whether or not i should modify or create something in this Module.

Indeed, i should probably try to impose a "bond" constraint on the interface faces and then set a fracture criterion to progressivly debond the top and bottom regions throughout the deformation of the model...??? But how could and shoud il do that..?

Anyway, then i got to the Mesh Module and tried to mesh the interface. But when i tried to run the analysis i got an error message like "some elements have no property". And it turns out these elements correspond exactly to the elements lying in the interface.

Please let me know what was wrong in my model and should i proceed in the different Modules (Part, Property, Assembly, Step etc...).

Regards,

Malik

 

[eispiata]

Here is a picture of the model with an interface sketched to account for the crack path.

Regards,

Malik

 

[iwb]

Hi,

as far as I know, if you define a solid section as crack path using a material damage law is only
possible in ABAQUS/Explicit. In your case, I would use a
traction-separation-law. This is explained in section

26.5.6 Defining the constitutive response of cohesive elements using a traction-separation description

of the users manual. Have a look at the subsection damage modeling!

A good source for information is the paper of

G. Alfano and M. A. Criseld
Finite element interface models for the delamination
analysis of laminated composites: mechanical and
computational issues
Int. J. Numer. Meth. Engng 2001; 50:1701-1736.

You can follow the steps below as a starting point to get
the model running:

-Define a material with material behaviour elastic, type
traction-->define damage initiation crit. --> define
damage evolution law
-Create a section other-->cohesive, choose your material,
response: traction-separation
-asign the section to your cohesive zone

Hope this helps.

Marco

 

[eispiata]

Hi,

I took your advice.

For the rubber part i defined an elastic behaviour (small strain problem) and a Solid Homogeneous section that i assigned to the bottom and top parts.

Then i defined a second material:
- Elastic --> Traction (E=200 000 Abaqus Unit System, Poisons's ration = 0.45)
- Damage for Traction-Separation Laws-->Quads Damage

Then a cohesive section associated with this second material was assigned to the thin layer accounting for the crack path.
I used (static, General) Explicit solver. Should i use Dynamic Explicit instead?

I skipped the Interaction Module because i believe there is nothing to define or modify in this Module. And then i defined the boundary conditions (Uniaxial Tension) in the Load Module.

I meshed the whole model with the appropriate mesh. So the mesh itself shouldn't be a problem.

The analysis completed successfully.

However no crack propagation occured as you can see in the picture.I am deeply suprised in the sense that i set quite a low value for the Nominal stress. Hence the crack should propagate unless there is something that i missed in the process.

Regarding the picture, let me know what you think about the "no crack propagation occurs" issue.

I am looking forward to your comments and remarks!

Malik

Thanks,

Malik

 

[eispiata]

Hello,

I have been working for two days on my model and i made some progress in the process of understanding "crack propagation" modeling in Abaqus but i didn't get what i was looking for. Indeed no crack propagation occured ...Frown



The idea is to define the crack tip as the zone where ?22 (stress component in the y-direction) is maximum (see attached picture).

I expected the crack to propagate along the horizontal line made up of cohevsive elements by defining the normal nominal stress...

The "upper" and "bottom" material is linear elastic (isotropic) and accounts for a rubberlike material in the case of small strain.

The "interface" is defined as linear elastic (traction) associated with a Damage for Traction-Separation law.

I am trying to figure out why no crack propagation occurs.

Here are the values that i set for in the Material Module

- the elastic properties of rubber: Young modulus = 2 , Poison's ratio = 0.48

- the elastic properties of the cohesive zone: E=1000, G1=1 and G2=1

- the damage for Traction Separation law (cohesive zone): Nominal stress _normal mode =1

Nominal stress _first direction =1000, Nominal stress _2nd direction =1000.

I am doing it in a hurry so i probably missed some important points. Moreover i am not familiar with FEM crack propagation so my questions might seem silly to some of you...

I am looking forward to your comments and remarks.

Regards,

Malik

 

[iwb]

Hi Malik,

unluckily I'm out of office , so
I can't check your settings with ABAQUS.
There are two things you can check with your
model. First, there is an output field variable
called sdeg that shows the degree of damage in
the cohesive element. You can also check for
quads (or something like that) to see if your
damage initiation criterion works. Second,
there is a benchmark example from alfano for
cohesive elements. I would test the model with
this parameters, because they are definitely
reasonable. Your plot looks reasonable. I just guess
your damage initiation crit. is not met.
Maybe you should use a finer mesh along the crack
path.

Good luck
Marco

 

[eispiata]

Hi,

According to you the damage criterion is not met.
I am going to work on this using SDEG.

My last question is more about the geometry itself.

Indeed i just created one single geometry that i partitioned to account for the interface. So i basically created one single part.

Should i create separate parts instead and tie them with a TIE constraint?

Thanks,

Malik

 

[eispiata]

Hi Marco,

Could you send me the article ??

Thanks,

Malik

 

[iwb]

Hi,

actually this article is copyrighted by wiley press.
I can not post it here.
But you can find it in

International Journal for Numerical Methods in Engineering
Volume 50 Issue 7, Pages 1701 - 1736
Published Online: 26 Jan 2001

It should be available at every technical university library
or you can order it online.

To your question above, I think it's not the worst thing to
model your specimen as one part and partition it. If you can get a feasible meshing this way it's perfect. Using several parts and couple them with a tie constrained is realized usually by introducing Lagrange multiplier functionals which adds additional computational cost to your system.

Greetings
Marco