Thank you for comments. Could you please share some UMAT code for this issues.It would be very grateful for my research . 

You have local problem because you used a softening behavior law but I'm surprise that you have this problem with a composite. Non local or gradient approach is then the unique solution but it's not easy to have a umat as you need to take the other elements into account which is a problem with abaqus as is limited to block of 128 elements.

Franck lauro : Yes I have this problem when I am using classical Tvergaard and Needleman failure criterion in the model. Do u have any other alternatives method to overcome this mesh dependency issue in the case perfectly bonded composite .

Hi Mr. Pandi, 

The GTN model can be assembled in ABAQUS using UMAT if you want to play with GTN further. About the UMAT code I have attached an article which has codes for basic types of plasticty. All you have to do is change yield function and add equations of GTN in plastic convergence part of Isotropic Hardening code in the article. 

As yours is a composite you may use USDFLD with UMAT to defiine matrix, fiber matrix and other failure criteria. One tip if you want to do element deletion preferable would be VUMAT, however you can degrade the elements in UMAT instead of deleting them. 

All the very best.

I have checked the code, it works well.

Hi Abishek

Thank you for your documents. The important problem is converting local GTN into Nonlocal GTN

Dear Mr. Pandi,

I am also working on that; to make GTN non-local, one type of work has been done already with Rosselier's Model, you may find it here http://www.sciencedirect.com/science/article/pii/S1877705813006401

by making  a damage parameter a field variable. try the concept, you may succeed before me :)

Dear Mr. Pandi, 

The answer to your question might be subroutine UEL. 

From UEL one can call UMAT for local GTN do be done in each element, and then an extra DOF based on weight factor and loacl GTN can be calculated and stored.

but I need a sample UEL for it for normal 3d elements.

With regards,

Abhishek

Thank you

you can use the modified GTN damage model by Gatea et al (2017).

here are the VUMAT and UMAT subroutine;

http://en.banumusagr.com/shop/gtn-damage-abaqus-vumat/

Goodluck.

We have released the implementation of a nonlocal GTN model for Abaqus/Standard (UMAT) and Abaqus/Explicit (VUMAT) at http://www.tu-freiberg.de/en/nonlocalGTN . It does not require an UEL but uses the analogy to the heat equation and is thus easy to use.