Help Delamination Modeling with cohesive Elements

[wazulu]

Hi All,

I am new to this forum and fairly new to Abaqus. I'm using Abaqus 6.7 to model a carbon fiber prepeg plate (6"x2"x0.05") with an initial delamination 0.5" from one of its edges. The plate will be fixed at one end with a concentrated load at the opposite end.

I want to build a 2D and a 3D model. I'm using cohesive elements to model the delamination. These are the steps I used to build the model:

1. Create plate geometry.
2. Extrude cut from plate geometry the delamination region
3. Create delamination geometry.
4. Create material properties for plate and cohesive elements.
5. Assigned properties to plate and cohesive regions.
6. Create assembly, two instances, one for the plate, one for the cohesive region.
7. Create contact between plate and cohesive section.
8. set up steps
9. Set uo B.C's
10. Set up loads

After I run the job I get mane warnings saying that there are negative igenvalues and then there is an error.

The error says:

***NOTE: THE SOLUTION APPEARS TO BE DIVERGING. CONVERGENCE IS JUDGED UNLIKELY.


***ERROR: TIME INCREMENT REQUIRED IS LESS THAN THE MINIMUM SPECIFIED

When creating this model, I have followed the delamination example from the Abaqus Benchmarks Examples from the Abaqus documentation.

I have tried unsucesfuly to debug the model and to find the cause of the error but I have had no luck.

Any help will be much appreciated

Attached is the cae model.

Thanks in advance for your help.

 

[rstupplebeen]

Eng-Tips seems to be having problems with CAE files. Could you try zipping it up and reposting?

Rob Stupplebeen

 

[rstupplebeen]

Correction the file did transfer well. Sorry I spoke too soon. The file size was reported by Explorer as 100kb and I have had that as an issue recently.

Rob Stupplebeen

 

[rstupplebeen]

If you look at your deformed geometry the cohesive layer has flipped inside out. I believe that issue can be solved by increasing the element through the thickness. You really need at least 3 elements to accurately model bending. When I seeded it to have 3 elements through the thickness and a global size of 0.3 the deformed results were much more inline with my imagination. The bond did not however fail. That however is another issue. Try this out and post back. I hope it helps.

Rob Stupplebeen

 

[rstupplebeen]

Here is a picture of the original and modified models.

Rob Stupplebeen

 

[amubashar]

I would like to add a few comments to the suggestions given above. First is that your out of plane thickness for cohesive elements and plane stress/strain thickness for continuum elements are different. This means that the dimensions of the cohesive layer and adherends are different in Z direction, which I think is not the case.

Secondly, in step incrementation control, I would use a small initial increment size and also keep the maximum increment size small. Nonlinear geometry should also be used.

And finally a general observation regarding the use of cohesive zone elements. Cohesive zone elements are sensitive to the material values and it is generally difficult to obtain convergence by using a set of values not properly calibrated. There is a lot of published literature regarding the calibration of cohesive zone law.

 

[wazulu]

Thanks rstupplebeen for your comments and advice. I'm going to implement the changes you suggested to see if the problem its resolve.

Did you get any error messages or warnings when you made those changes??? I would like to get an idea of how small should I reduce the step increments. Will this get rid of the negative eigenvalues???

I will post back with my new results.

Thanks once again

 

[wazulu]

Hello amubashar, thanks for your comments.

In regards to the thickness of the cohesive elements, aren't they suppose to have almost negligible thickness??? Or should the thickness be the same as the continuum elements?

As far as the step increment, do you have any recommendations in how small they should be?

I appreciate your help and thanks for your suggestions

 

[amubashar]

When defining cohesive zone elements in a 2D problem, there are two thicknesses that can be defined. The thickness of the cohesive zone element, that represents the opening direction of the cohesive element, should be zero or very small. This thickness is the initial thickness in the edit section dialog box. You can leave it as analysis default, which is one in Abaqus. The out of plane thickness represents the thickness of the cohesive element in the Z-direction, similar to plane stress/strain thickness for continuum element.

 

[wazulu]

Hello rstupplebeen,

I tried the changes you suggested but I did not get near your results (as far as the deformed shape). I created a new model exactly as the one I attached an seed it the cohesive region with 4 elements through the thickness and a global size of 0.3. I also reduced the step increments as suggested by amubashar.

The good news is that the previous error messages are gone, however I got a new error message:

***ERROR: TOO MANY INCREMENTS NEEDED TO COMPLETE THE STEP

I also get tons of warnings such as these:

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_1_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_2_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_3_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_4_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_5_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_6_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_7_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_8_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_9_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_10_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_11_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_12_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_13_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_14_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_15_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 3 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_16_1_1.

The system matrix has 1 negative eigenvalues.

Solver problem. Zero pivot when processing D.O.F. 2 of 1 nodes. The nodes have been identified in node set WarnNodeSolvProbZeroPiv_2_1_17_1_1.

The system matrix has 1 negative eigenvalues.

I seriously don't know what is wrong with this model, I believe it has something to do with the cohesive section, however, I am unsure if this has something to do with the model itself or with the material properties used for the cohesive section. I don't know if the cohesive section mesh should be much more fine than the plate section for this to work.

Any suggestions and advise will be greatly appreciated.

Attached is the new cae file

Thanks again

 

[schurik01]

Hey wazulu, did get it to solve your Problem? If yes, could you post a cohesive zone model that works?

I have the same problems

Thanks