Hi everyone,
This question really confused me, and I think the ABAQUS user manual does not give a clear answer.
To make it clear at the beginning, I am using traction-separation law. The following command:
*COHESIVE SECTION, RESPONSE=TRACTION SEPARATION
ABAQUS user manual (version 6.12) 32.5.6 said "The features described in this section are primarily intended for bonded interfaces where the interface thickness is negligibly small. In such cases it may be straightforward to define the constitutive response of the cohesive layer directly in terms of traction versus separation." in overview. It means the thickness of the element is very very small (zero? not zero?). If I understand this right, it also means the material response of cohesive element is Traction-Separation, but not Traction-Strain.
However, in the same section, it also give the equations in the "Linear elastic traction-separation behavior" subtitle. The equation is clearly a Traction-Strain relation but not Traction-Separation relation. Because there is an original thickness of element T_0 in the equation. Is this conflict with what has been claimed in the overview?
Actually, the thickness of cohesive element makes difference in the simulation, because it increase the compliance of the whole model.
Please check the following link for the corresponding section of Abaqus user manual.
http://abaqus.software.polimi.it/v6.12/books/usb/default.htm
If you have an answer, please help me! Thank you!
Best wishes
Jiadi
Hi Dac Thuong
Thank you for your response! May I ask further question?
I think the entered parameter K_nn, K_ss, K_tt should be the Traction-Strain stiffness. Please check the *Elastic command manual in the following link.
http://abaqus.software.polimi.it/v6.12/books/key/default.htm?startat=ch05abk03.html#usb-kws-melastic
They are using the same notation with the Traction-Strain equation in section 32.5.6.
If as you said, the initial entered K is calculate by E/T_0, K is changing during the calculate. How does it change? Does Abaqus measure the opening of cohesive element every step?
I may do some test to verify what has been told in user manual, and keep you update.
Best wishes
Jiadi
Hi Dac Thuong
Thank you for your response. I have done some theoretical and numerical study of the mesh dependence in cohesive element model. I might share it when I finish writing it up.
" 2 sets of parameters (T_0=1, K=E) and (T_0=2, K=E/2) should yield the similar result."
-> I want to share some of my comments. I use Richo code to insert the cohesive element between each solid element. The zero thickness element is geometrically zero thickness but with an artificial thickness T_0. For a given geometry and meshed model, increasing the artificial thickness increase the artificial compliance of the entire model. If the effective of the entire model with cohesive model is E_eff. The Young's modulus of solid element is E. The modulus of cohesive element is K. Then in the following three conditions, E_eff yields to E.
1. Mesh size is big, which means decrease the numbers of cohesive element.
2. When T_0 is very small, which means decrease the artificial thickness of cohesive element. But this will increase the computational time is we are undertaking explicit dynamic simulation.
3. Increasing K. But this will influence the stress wave in the cohesive element, and might influence the crack propagation in the explicit dynamic simulation.
Best wishes
Jiadi
- Badges
- Science topic