Tabular Damage Evolution Softening Curve (Trapezoidal cohesive law)

[sandmonstersandy]

I am running a model simulating composite delamination using a cohesive interface. The model delaminates correctly using energy based damage evolution, however when I try to use displacement based, tabular data to define the softening curve, it does not delaminate. I am trying to implement a trapezoidal cohesive law. I have defined the tabular displacement by equating the area under the softening curve to the Fracture toughness (1.0755 N/mm) and am using MPa and mm.

Does anyone know why this is not delaminating? Or have experience implementing a trapezoidal cohesive law?

Thanks


** MATERIALS
**
*Material, name="Material 1"
*Elastic, type=ENGINEERING CONSTANTS
159000.,8490.,8490., 0.32, 0.32, 0.32,4610.,2780.
2780.,
**
** INTERACTION PROPERTIES
**
*Surface Interaction, name=Bonding
1.,
*Cohesive Behavior
3.52e+06, 2.78e+06, 2.78e+06
*Damage Initiation, criterion=MAXS
10000., 128.5,10000.
*Damage Evolution, type=DISPLACEMENT, softening=TABULAR
0., 0.
0.989, 0.0042
1., 0.0125
*Damage Stabilization
1e-05

 

[compabaq]

Hi sandmonstersandy,

I have been having the same problem, as my tabular cohesive law is also not breaking. The documentation on tabular cohesive laws is very poor. This is what I have understood and applied:
- The damage variable is defined as 1-K/K0, with K0 the stiffness at initiation and the K the stiffness at other points in the diagram.
- The total/plastic displacement is the crack opening displacement (COD) minus the COD at initiation.

Using that interpretation however does not work at all. My cohesive simply does not break for some unknown reason. The traction keeps on increasing.

I have created a two-element model to test a lot of different parameters real quickly, but nothing that I have tried works...

Anyone else have any clue what could be causing this?

Comp

 

[compabaq]

I think that I have found the problem. I was using about a 1000 equally spaced points to define the softening curve. It seems that I needed to have more points towards the end of the softening curve. That region shows an exponential increase (if you plot the table you entered), and the only way to capture that properly is by having a huge amount of points in that region. I don't really understand why Abaqus makes this so difficult and does not mention this in the documentation, but it resolved my issue.

I hope this helps!

Comp

 

[sandmonstersandy]

Hi comp,

Thanks for the advice! By using 10,000 data points it finally decohered.

A further complication came up. I tried to replicate the linear, energy based option using tabular, but this gives a slightly different force delta response. Using the tabular damage evolution it fails earlier than the energy based method.

The logic I was using was
For Gc=1.0755 N/mm
K=1000000
max traction=128.5mpa
therefore onset of damage occurs at 0.000128mm (maxtraction/K)
delta critical occurs at 0.0167mm (2*Gc/maxtraction)
Input into the tabular table (10,000 points):
- interpolation of D from 0 to 1 (1-k/ko)
- interpolation of delta minus delta onset from 0 to 0.01661 (delta critical-delta onset)

I've attached the excel with my working

Have you managed to replicate the results between energy and displacement? Do you have any ideas?

Thanks

Sandy