Learning Forum

[Aksh]

Hi everyone,

I am doing a 3-point bending analysis of composite lamina (GFRP-0/0/90/0/0). I want to implement the Continuum Damage Mechanics for damage evolution of lamina. For that reason, I created the new material in Engineering Data with separate mechanical properties, which I got from the industry.

My problem—Whenever I tried to give the value to “Energy dissipated per unit area from tensile/compressive matrix damage” rather than 0 N/mm-1 (it should be 4 to 8), my simulation process will not converge after 1 mm deflection. But with the value 0, it works perfectly. In short, it works only with 0 value, and I want a realistic value (4-8).

I tried the following possibilities,

·      Varying the number of steps and sub-steps

·      Fined the Mesh

·      Decrease the contact stiffness

·      Change method of contact elements

So, can anyone have an idea, of how can I overcome this problem? Any advice or help would be really appreciable. Here, I also attached my unsolved file “CDM_with0value.wbpz” for your kind reference.

CDM_with0value.wbpz

[Sean Harvey]

Hello Thanks for your query. When you specify zero, you say it works perfectly. Can you clarify? The model converges is what you are referring to, correct? Meaning you can get your desired displacement, but once you have a non-zero Gc it fails to converge.
With damage mechanics it can be very challenging to converge, so you can try some viscous regularization. Also, you mention you get to 1mm deflection, is your loading force-controlled or displacement? If you have contacts in the model, convergence can also be coming from the contact and not the material. What I might suggest is to isolate the material behavior from the contact. Try to damage the material with displacement control and some simple B/Cs to see if the issue is with the contact nonlinearity or the material. Once we can validate the material is damaging and can converge that with displacement control, move to force control, then include contacts. Keep in mind that for this simple layup, there won't be much ability to redistribute the loading so when damage occurs, it can be sudden and with force-controlled, the solver will not be able to withstand the force. I just point out that with more complex layups with more plies, damage does not affect the overall stiffness reduction as great and all at once.
Let us know if this helps.
Regards Sean

[Aksh]

Thank you very much Sean for your kind suggestions. I will try the tips you mentioned and then tell you about the progress.
For clarify my question here I added the figures.
1- My model with BCs (A-Fixed support, B- Punch displacement in negative Y-axis (30mm), C- Line constrained the model in X and Z-axis )
2- Material properties for Damage evolution
(With this zero value, there is no force convergence problem till 25 mm displacement of punch but with non-zero value it will go just 1 mm and got force convergence problem)
3- Solution output (I gave 150 steps with 1000 sub-step through program control, but it goes smoothly till 6 step only with above properties )