Composite shell with soft layer at core

[xyont]

hi, is it appropriates using soft layer at core in shell element with composite options? all result shown large discrepancy compared to solid model, An example of plate is 100mm x 100mm dimension, thickness is 10mm, three side fixed and one is free support. Surface load applied partially at center with 10N/mm2 values. Outer layer has 1mm thickness and material is steel, core layer is plastic(ABS) material,

*Material, Name=S235
*Elastic
210000, 0.28
*Material, Name=ABS
*Elastic
2000, 0.394

below result using the same material as steel for all layer, both model of shell and solid given identical results.

top view,

bottom view,

however, result are significantly different when using plastic(ABS) at core layer.

*Shell section, Elset=Compound-1, Composite
1.,,S235
4.,,ABS
4.,,ABS
1.,,S235

top view,

bottom view,

and concrete material at core layer for comparison, shown better in stress result.

*Material, Name=Concrete
*Elastic
25000, 0.2

[vicmw]

I don’t think so because the shear strain would be very different in each material and a shell element wouldn’t be able to capture that well with only linear strain (quadratic displacement) through the thickness. Perhaps if you look at the deformed shape of the section, you can see if it’s kind of Z-shaped when it’s in shear? I expect not.

[xyont]

indeed, it seems composite shell element still have limitation due to shear transverse. Varying stiffness of elastic modulus can not be too large, maybe about 5 to 10 still acceptable but case dependent. Probably it works best for in plane loads rather than out of plane.

[rsmith]

Most of my work with CalculiX involves composites (often sandwich structures). I prefer to explicitly create the differeny layers, mesh them separately and connect them with TIE using NEIGH. Like this:

# vim:fileencoding=utf-8:ft=cgx
# All values in SI units: m,s,kg,N,Pa,J
# (length,time,mass,force,stress,energy).

rot -z
rot d 30
rot r 30

valu L 0.1
valu divL 20

seto bot
pnt ! 0 0 0
swep bot new tra L 0 0 20
swep bot new tra 0 L 0 20
setc bot

seto botskin
swep bot IB1 tra 0 0 0.001 2
setc botskin

seto core
swep IB1 IB2 tra 0 0 0.008 8
setc core

seto topskin
swep IB2 topsur tra 0 0 0.001 2
setc topskin

# node 1 X Y Z
elty all he20r
mesh botskin
mesh core
mesh topskin

# Node sets for fixation and rigid body load.
seta nodes n all
enq nodes fix1 rec _ 0 0 1e-6
seta fix n fix1
enq nodes fix2 rec 0 _ 0 1e-6
seta fix n fix2
enq nodes fix3 rec _ L 0 1e-6
seta fix n fix3

# Show the mesh and selected nodes.
frame
zoom 0.8
view elem
plus n fix r 8
#plus n load b 8

valu output no
# Write data
if output eq yes
# Tie together different meshes, when not using “mesh all”.
neigh all 1e-5 abq tie
# This creates DCF* and ICF* files. Combine them with:
# cat DCF* ICF* > ties.sur
# rm -f ICF* DCF*
send all abq
send botskin abq nam
send core abq nam
send topskin abq nam
send fix abq nam
endif