Hi all,
I wanted to analyse the capabilities of the cohesive elements, for this i thought of conducting a simple analysis with one element. I took 1 Element (shown in figure), the bottom plane is fixed completely and the top plane is deformed to form a wedge like shape. (Top right edge is displaced upwards and top left edge to bottom by an equal amount)
I have performed geometrical linear and non-linear analysis on cohesive (COH3D8) and continuum (C3D8) elements for this kind of deformation.
In the case of geom linear analysis, both cohesive and continuum cases have shown same results. But in the non-linear case the results have deviated by huge factor. Additionally, to my surprise the strains in the case of cohesive element are different from the strains in continuum case, even though i have deformed them in similar manner.
Can anyone explain this weird behaviour of cohesive elements ? And are cohesive elements suitable for geometrical non-linear analysis ?
Interesting case.
Did you use *COHESIVE SECTION, RESPONSE=CONTINUUM in the definition of cohesive behavior?
And what about displacements? How do they compare?
The displacements are same in both cases, (please find the attached figure). And, yup i did use the continuum response for cohesive element.
I have also attached the input files of the cohesive and continuum elements. Please take a look!
I think the problem lies in the use of *COHESIVE SECTION, RESPONSE=CONTINUUM. According to Abaqus' manual, this use of keywords correspond to the case of an adhesive layer of finite thickness, which is implemented with very particular kinematic assumptions.
From the Abaqus 2016 Users' manual:
"For adhesive layers with finite thickness it is assumed that the cohesive layer is subjected to only one direct component of strain, which is the through-thickness strain, and to two transverse shear strain components (one transverse shear strain component for two-dimensional problems). The other two direct components of the strain (the direct membrane strains) and the in-plane (membrane) shear strain are assumed to be zero for the constitutive calculations. More specifically, the through-thickness and the transverse shear strains are computed from the element kinematics. However, the membrane strains are not computed based on the element kinematics; they are simply assumed to be zero for the constitutive calculations. These assumptions are appropriate in situations where a relatively thin and compliant layer of adhesive bonds two relatively rigid (compared to the adhesive) parts. The above kinematic assumptions are approximately correct everywhere inside the cohesive layer except around its outer edges."
I think you've pointed out a very interesting quirk of Abaqus.
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training