Dear All I am trying to analyse th

[Eng.SaTa]

Dear All
I am trying to analyse the failure of a composite scarf joint. Composite adherends are bonded together by adhesive.Image of the joint is attached.
I have modeled each ply as a part instance.
I used solid elements in plies tips and in order to take advantage of Hashin criteria of ABAQUS I used SC8R for the rest of composite adherends. Shown in the picture.
I have defined a CSYS for material orientation (both for solid and shell elements.
I assigned a stack direction (orientation lay-up) for the shell elements.
To have a generalized plane strain condition, I extruded my model in 1 elm size and constrained the nodes on the rear and front face of the joints.
But the result I collected is weird.
The stress distribution through the thickness of laminate - far from the scarf region is not symmetric while my laminate is symmetric.
I have already tried different methods to define material property and orientation for the shell part of the model including shell section and composite lay-ups, none of them worked.
I'm pretty sure the issue comes from shell elements stacking direction/ composite layup definition for +/-45 plies in conjunction when they are extruded up to 1 element size. Because I have modeled an angelply laminate [45/-45]2s with 1 elm size in width direction and collected the stresses through the thickness of the laminate and I found out the behavior I already explained still exists.
There might be some tips to model 1 element width composite parts with continuum shell elements that I have missed. So, I'd be more than happy if you could share your thoughts or comments with me.

You can find the way I defined material property and orientations for solid and shell elements.

 

[FEA way]

This model is quite complex, I’d have to take a closer look at it to find out what’s wrong (at the moment I suspect boundary conditions or interactions). Can you share the cae or inp file ?

Maybe a different modeling approach will be needed.

 

[Eng.SaTa]

Thanks for your quick reply.
Sure, I am trying to attached 2 .inp files in t=current and my next post.

0-Layupdefined.inp : I have defined a composite latyup (one ply) for the shell part of my model.

0-Shellsection.inp : I have defined shell part of composites using shell section and then assigned a material ORI for them.

 

[Eng.SaTa]

Shell Section and material ORI Approach

 

[FEA way]

I would try running this with a refined mesh, especially in the through-thickness direction. And be careful with stack orientation of continuum shell elements, it should be consistent.

 

[Eng.SaTa]

Thanks for your help!
Pls take into account model works but does not give accurate result
If you check the stress distribution far from the scarf region you will see the stress through the thickness of laminate is not symmetric. That is the thing I cannot handle!
Besides. It's a failure model and running the model will take long, so the complete run is not necessary.
Again,thanks

 

[FEA way]

That's why it's especially worth trying with a few elements through the thickness. You could also increase the number of integration points in each ply used for stress output.

 

[SWComposites]

If you make a model of the laminate without the scarf joint, do you get correct stresses?

 

[Eng.SaTa]

Dear SWComposites
No, I dont. When i model an angelply laminate [45/-45]2s using the the Coordinate system an material orientation illustrated in pictures, I get the wrong result. the Stress distribution I collected along the thickness of laminate is not symmetric. I also tried a wider laminate (10 element in width) and defined composite lay-up similar to above mentioned model, stress distribution becomes correct!

And I'm obliged to use 1 element as width of my model, because of general contact definition! Also, a wider model makes the computational cost too much that I can not afford!
I'm wondered why continuum shell element is behaving like this!

 

[FEA way]

Why would a higher number of elements in the through-thickness direction be a problem for general contact definition ? Can you elaborate on that ?

It’s hard to get accurate stress representation across the thickness with one element in this direction, even if it’s a continuum shell element. That’s why those elements are often stacked to obtain better results. As I’ve mentioned before, you could also try increasing the number of integration points per layer.

 

[Eng.SaTa]

I tried shell section instead of composite lay-uo definition and used 5 integration points and the problem still persist!
I will refined my mesh through the thickness of laminate and find if it works! I am trying to verify a paper that used 1 element in thickness of a ply! so I did the same

 

[FEA way]

Can you share the title of this paper ? There might be some differences in the modeling approach.

 

[Eng.SaTa]

I have another question, will be happy if could answer it
Is there any possibility to change the sketch plane in abaqus?
the default sketch plane in abaqus is XY, can I change it to XZ or YZ?

 

[FEA way]

When you model a new part using Create Part option, the plane used for the first sketch is always the XY one. But you can rotate the part instance in the assembly module or create a part in form of a point and then manually add the actual part modeling features (extrusion for example). In such a case you are not limited to the XY plane.

 

[SWComposites]

Well, perhaps post pictures of your FEM stresses and expected stresses for the model without scarf. Also post picture of your model with loads and boundary conditions. We can’t possibly guess at what your are doing or getting.