FEM Coincident Elements Bad? 2

[timthereaper]

Hi all,

I talked with an analyst recently about modeling up a composite preform joint, and after all the case studies he's done he told me that the nearest to actual results he got was to use coincident shell elements with different properties. After doing some searches, I wasn't able to find anything on coincident elements being an acceptable or unacceptable practice. I'm somewhat new to FEA, but I've taken a couple classes on the mathematical theory. After thinking about it for a while, I can't say it's necessarily wrong to do it that way from the theory, but my engineering "spider sense" tells me something's not quite right. Has anyone had experience with this or can anyone elaborate/correct me about this method?

Thanks!

Tim

 

[rb1957]

i don't think there's anything Unacceptable about it ... codes will check for coincident elements in case you doubled them by mistake.

Quando Omni Flunkus Moritati

 

[timthereaper]

So it's not really a bad practice? I can see duplicating them by mistake is bad, but doing it on purpose is acceptable? I'm just trying to understand if it's something I should get used to seeing/doing.

 

[rb1957]

the "problem" will be in understanding the results ... i think all you can get are strains (which is probably all you want for composites)

Quando Omni Flunkus Moritati

 

[jhardy1]

There's nothing inherently "wrong" with using coincident elements (as long as they are not accidental duplications!) As rb1957 says, the complication is in interpreting the results.

If you have two or more coincident plate elements, for example, the total moment at an edge (or membrane force, or whatever) will be the sum of the moments in each of the coincident elements. If total force and moment resultants are of interest to you, you may need to write some sort of post-processor to find and sum the individual coincident plate element resultants.

Some FEA software systems have a "Composite" module which does the hard work for you - you specify your laminate build-up, and the software sums the individual lamina stiffness contributions to compose a single plate / shell element stiffness matrix. The structure is then solved using a single element, and then the post-processor takes the element force and moment resultants and back-substitutes to get the individual lamina force resultants.

http://julianh72.blogspot.com

 

[timthereaper]

Thanks rb1957 and jhardy1. That's great feedback. Now I think I understand how this works. I've been told the reason other analysts I work with use this method is because it's a 3D composite preform. It's really just one big tangled weave, but we're using a ply-based approach to model it. I'm not sure if I looked at the elements separately it would correspond too well to the actual interactions of the preform anyway.

 

[rb1957]

why not use laminate properties ?

Quando Omni Flunkus Moritati

 

[jhardy1]

If I'm modelling "thin-walled" structures that can be modelled sensibly using plate-shell elements, then I would use my FEA software's built-in "laminar composites" functionality to define the lamina properties, build a laminar-composite element property that incorporates each of the lamina properties, and then extract the individual lamina stress / strain / force / moment results using the built-in laminar composite post-processor.

However, if I was looking at something which is "thick-walled" and / or too irregular to model sensibly using plate-shell elements to capture the full geometry, I would typically model the "skins" using laminar-composite plate-shell elements, and fill the "core" with a mesh of solid elements. (You might even model "discrete" reinforcing components such as wire / cable, bar or strip pre-forms using linear "beam" elements.) You need to make sure you use a "compatible mesh" at the junction between the solid core elements and the "skin" plate-shell elements (and "beam" elements if applicable), to ensure full composite behaviour between the various components of the build.

Hope this helps!

http://julianh72.blogspot.com