Académique Documents
Professionnel Documents
Culture Documents
1. Abstract
Composite laminate and Sandwich plates are widely used in aerospace and other industries due
to its specific strength and specific stiffness characteristics. In a typical composite sandwich
plate construction, face sheets take majority of the bending and in-plane loads and the sandwich
core takes the shear loads. The sandwich core materials are having very low allowable
compared to that of the face sheets. Core is expected to fail under shear mode, due to through
thickness transverse shear stress.
An attempt is made through series of ANSYS simulation to understand the good modelling
approach to predict the transverse shear stress in the core of a sandwich plate. A study is
conducted here to understand the sensitivity of the number of elements used in the through
thickness direction. A comparison is also made between solid and shell element modelling
approaches. This study addresses three questions 1) Why is the shear stress constant in the
core? 2) Is solid or shell elements good to model sandwich structures for transverse shear
prediction? 3) If modelled using solid elements how many elements required through
thickness?
2. Nomenclature of the composite sandwich construction
A typical composite sandwich plate consists of face sheets (composite laminate) at top and
bottom and a low density core in between. The core is boned to face sheets with adhesive
material as shown in Figure 1.
Face sheets can be fabricated by arranging the plies at designed orientation as shown in Figure
2 to achieve the desired directional properties.
Core material exhibit 3D anisotropic behaviour. The core material has very low Youngs
modulus in the in-plane direction and has considerable Youngs modulus in through thickness
direction.
Sandwich plate as a structure can fail due to the failure of the skin and / or failure of the core.
Honeycomb core under transverse loads can fail due to core shear and core crush.
The face sheets are considered as made from carbon fibre fabric with following typical material
properties
E1 = E2 = 62000 MPa, E3 = 8500 MPa
G12 = G23 = G13 = 4200 MPa
12 = 0.05, 23 = 13 = 0.35
3
5.1. Study-1
It is assumed that both face sheets and core are made from same isotropic material and
transverse shear stress Sxz at the middle of the plate (location A shown in Figure 6) is plotted
in Figure 7. The purpose of this study is to make sure that the number of elements required in
the thickness direction to capture the through thickness transverse shear stress for an isotropic
material.
The variation of the transverse shear stress Sxz through thickness is parabolic as expected.
Refining the mesh at the core from 4 to 10 elements through thickness, the parabolic variation
is captured more smoothly. However, in practical point of view using 4 elements through
thickness is sufficient.
Having established that the 4 elements through thickness is sufficient to capture the parabolic
through thickness transverse shear stress distribution with reasonable engineering accuracy
further studies are done using 4 elements through thickness.
For graphing the purpose, the transverse shear stress is normalized with average shear stress.
Sxz-ave = F/A = 1000/ (50x (10+1+1)) = 1000/ (50x12) =1.67 MPa
Figure 7: Transverse shear stress in Isotropic material for 4 and 10 elements through thickness
5.2. Study-2
The stiffness difference between the face sheets and core is introduced still keeping the material
model as isotropic. The purpose of this study is to understand the impact of differential stiffness
between face sheet and core.
The Youngs modulus of the core is scaled by the ratio between the shear modulus of the face
sheet and core material.
=
Figure 8: Transverse shear stress in Isotropic material: Differential stiffness for face sheets and
core
From Figure 8, it can be seen that core is taking most of the shear load and the face sheets take
no shear. An important point to note here is that the transverse shear stress distribution is
changed from the parabolic shape to almost constant value due to the differential stiffness
values considered between the core and the face sheets.
For graphing purpose, once again the transverse shear stress is normalized with the average
shear stress
Sxz-ave = F/A = 1000/ (50x (10+1+1)) = 1000/ (50x12) =1.67 MPa
Having understood the impact of differential stiffness on transverse shear stress distribution,
further studies are done on the actual sandwich material model.
5.3. Study-3
The face sheets and core are modelled using 3D orthotropic material properties. The purpose
of this study is to capture the transverse shear stress distribution in actual core material using
the solid elements.
The distribution of transverse shear stress is shown in Figure 9.
From Figure 9, it can be seen that the transverse shear stress in the core is constant, no more
parabolic. All the shear load is taken only by the honeycomb core.
For graphing purpose, the transverse shear stress for orthotropic case is normalized with the
average shear stress. Note that only half of the face sheet thickness is considered for shear area,
similar to text book calculation.
Sxz-ave = F/A = 1000/ (50x (10+0.5+0.5)) = 1000/ (50x11) =1.82 MPa.
The reason for constant transverse shear stress in core is looked at 3D material constitutive
model point of view. The stress tensors in a material under 3D stress state is shown in Figure
10.
Under the absence of body forces, the equilibrium of forces due to internal stress system can
be written as;
;
+
+
=0
+
+
=0
+
+
=0
For 3D orthotropic material model for core; the in-plane stiffness of the core in normal and
shear direction are assumed to be zero in the 3D orthotropic material input. Which implies that
= = = 0
;
+
+
=0
+
+
=0
+
+
=0
is zero, meaning
5.4. Study-4
The face sheets and core are modelled using shell elements in this study. The material model
used is 3D orthotropic. The purpose of this study is to compare the solid and shell modelling
approaches.
It can be seen from Figure 11 that the Shell element modelling (shown with thickness
visualisation) of the sandwich plate exactly predicts the transverse shear stress as that of 3D
solid modelling. Shell model has 6300 active DOFs to be solved, whereas the solid model has
8580 model to be solved. The ratio of the active DOF can be significant in large models and
thus the solution time. However, when the solid model is meshed with one element through
thickness there would not be much difference in DOFs.
It is recommended that shell element modelling is sufficient to capture the transverse shear
stress in the core.
6. Conclusion
The suitable FE modelling approach to predict the transverse shear stress in the composite
sandwich plate is studied through series of ANSYS simulation. This study focused only on
predicting transverse shear stress in the core material. The shear stress on the core is constant,
can be effectively predicted through shell element modelling. One or two elements through the
thickness is sufficient to capture the constant transverse shear stress in the core when there is
requirement to model using solid elements.
7. References
1. Michale C.Y. Niu, Airframe stress analysis and sizing, second edition, 2001.
2. Petras, M.P.F. Sutcliffe, Failure mode maps for honeycomb sandwich panels,
Composite Structures 44 (1999) 237-252.