Dear Vinay,

How are you orienting the elements? As you must know, cohesive elements ask for a orientation (definition of normal and longitudinal direction). In a first try, I would recommend orienting each element at the curve region individually. In the past, when I applied cohesive elements in a curved region I created a polar reference system, but it did not work in my case.

Wish you luck,

Bruno.

Dear Vinay Mautage, these links will help you

http://ijame.ump.edu.my/images/Volume_7/18_Huzni%20ei%20al.pdf

https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20070018344.pdf

Dear Bruno,

Thank you very much for your insight. I see that the each cohesive element should be assigned with the orientation. But, my model is 3D and with the fine meshing, the work of assigning orientation will be huge.

But, You said absolutely right, cohesive elements should be treated separately. 

Can you tell me How did you finally solve this trouble? or Have you chosen some alternative?

Thank you so much,

Vinay

Dear Sir Al-Sherrawi, 

Yes, the one document actually contains the 'L" shape model, very close to my real problem. I will build & simulate the way they showed in the paper. (By attaching another document of NASA) are you also suggesting me to use the shell elements for the parts & cohesive elements as interaction? 

As I found in this NASA's article that, "The formulation of an 8-node cohesive element for modeling delamination between shell elements on the basis of a 3D cohesive element. This formulation takes into account the rotational degrees of freedom at the shell nodes and ................. and DOES NOT REQUIRE COINCIDENT NODES ALONG AN INTERFACE nor does it require the use of rigid."

links."

Thank you so much,

Vinay