Session22 Paper3
Session22 Paper3
*ELEMENT_SHELL_COMPOSITE keyword
Summary
Because of their superior mechanical properties in combination with a relative low density Fiber
Reinforced Composites (FRC) are of great potential in the area of lightweight structures respectively
applications. Nevertheless usage and acceptance of FRC highly depends on the fact, whether or not it
is possible to prescribe and predict their structural behavior using the Finite Element Method.
LS-DYNA has a great potential to simulate the behavior of composite structures since many years. But
the model generation of complex layered structures with many different plies in different directions
may become very expensive. Starting with LS971 Revision R5.1 LS-DYNA offers the new keyword
*ELEMENT_SHELL_COMPOSITE, which is a generalization of the already known
*PART_COMPOSITE. This new keyword allows a very comfortable description of composite layers
using a ply-based concept. Within one part elements may have different number of through thickness
integration points as well as different material angles in each layer. Because of that the model must
not be split up in many parts, which makes pre- and also postprocessing much more practical.
The ANSYS Composite PrepPost enables a comfortable way to generate layered composite
structures using a ply-cased = manufacturing-based concept and export these data to LS-DYNA in the
new format. It offers a common composite definition for both ANSYS and LS-DYNA simulations.
1 Introduction to Composites
Since beginning of 20.th century there are serious efforts to investigate and to use the large potential,
which results from the use of new composite materials. From the technical production of high-quality
fibers as well as from the development of new production techniques various application possibilities
for fiber reinforced composite materials resulted. In most of these application areas beside the weight
reduction, the high specific stiffness and strength are exploited and used.
Composite materials are made of at least two distinct materials with different material parameters for
each phase. Fiber reinforced composites (FRC) are composites where one material component (fiber)
is used as a reinforcing material for the matrix. Particle reinforced composites consist of particles of
one material dispersed in a matrix of a second material. Particles may have any shape or size, but are
generally spherical, ellipsoidal, polyhedral, or irregular in shape.
In the current article we will restrict ourselves to long-fiber reinforced composite laminates even
though there are numerical solutions for short-fiber reinforced composites like LS-DYNA and Digimat
available.
Long-fiber reinforced laminates used in real structure are featuring several challenges to the user:
typically there is a huge number of plies in a large structure to handle. Additionally, plies are only
located in those regions where they are absolutely needed for deformation, stiffness and strength.
Transition between those regions is done using an extensive tapering of plies, i.e. a gradual change of
number of plies through the thickness. The latter results in a numerical model where the number of
thickness integration points changes throughout the structure even though it is one body or in LS-
DYNA language one PART.
An additional need results from the fiber or ply placement onto curved structures. Because of the
curved natures of the structure a reorientation of the fibers occurs, which needs to be accounted for a
proper representation of stiffness and strength. Draping simulation is the keyword for those kinds of
predictions. Unfortunately, this leads to a relative fiber angle change in the laminate from element to
element. Consequently an efficient preprocessing tool is needed to calculate those corrections.
Modeling of layered composites with shell elements has a long history in LS-DYNA. Twenty years ago
it was done by the use of *INTEGRATION_SHELL, a user defined thickness integration, with the input
of relative thickness coordinates and weighting factors.
Starting with LS971 in 2006 the keyword *PART_COMPOSITE enables a simplified method for
defining composite layups. No *SECTION and no *INTEGRATION_SHELL is needed. The
*PART_COMPOSITE only refers to *MAT definitions for different material data, together with the
thickness of the layer and a material angle, relative to the material direction of the element.
For simple geometries the fiber direction may be easily defined by setting of a suitable AOPT value.
But for complex structures with free-form surfaces the fiber direction may be curved over the surface.
The keyword *ELEMENT_SHELL_BETA can be used to define arbitrary material direction on each
single element. This is still a powerful tool, but don’t fulfill real composite needs, because of each layer
in a composite structure may have arbitrary curved fiber orientation, so only one material direction per
element is not enough. The BETA angle in *PART_COMPOSITE describes the material direction for
each layer separately, but it is constant for the whole part.
To describe an arbitrary composite layup, a ply-based concept is much more realistic instead of a
zone-based concept, because of ply-based means manufacturing-based. In the ply based concept it is
simple to add or delete plies at any location. It is the concept of modern composite pre/post-
processing software.
The following picture shows a simple plate made by four plies. Each ply has different size and
orientation, particularly layer #4 is diagonally oriented. Therefore the plate has to be divided into 8
zones with different thickness and different fiber direction across the thickness.
Using the classical modeling in LS-DYNA, this plate must be divided in 8 parts, each having its own
layer definition with *PART_COMPOSITE.
If one additional ply is inserted or if one existing ply should be changed in size, then the whole zoning
may change. This means that the division into parts may change and also the number of necessary
parts may change.
The next two pictures are showing a situation where 0-degree and 90-degree layers are drapped with
the same starting point and direction. Therefore the relative angle between the layers is still 90
degrees. This can be modeled with the existing *ELEMENT_SHELL_BETA.
In the next two pictures the 0-degree and 90-degree layers are draped with different starting points
and direction. Therefore the relative angle between the layers is permanently changing from element
to element. This can’t be modeled with the existing *ELEMENT_SHELL_BETA.
The material ID, thickness, and material angle for each through-thickness integration point of a composite shell
are provided below (up to two integration points per card). The integration point data should be given
sequentially starting with the bottommost integration point. The total number of integration points is determined
by the number of entries on these cards. The thickness of each shell is the summation of the integration point
thicknesses. Define as many cards as needed.
Card 1 2 3 4 5 6 7 8
Type I F F I F F
Cards
Type I F F I F F
Here all elements will have the same number of layers and therefore the same number of through
thickness integration points (NIP=4, with each having a thickness of 0.6mm). But because of draping
effects the fiber direction from layer to layer may vary in each element. This is here the reason why to
use the new *ELEMENT_SHELL_COMPOSITE keyword, otherwise each element must have its own
*PART_COMPOSITE definition.
Here the number of plies will change inside one part, therefore the number of through thickness
integration points will change (from NIP=4 in the first element to NIP=5 in the second). To have one
even surface on the shell lower surface, the _COMPOSITE option is used in combination with the
_OFFSET option. An arbitrary order of these two options is allowed.
Again, without using the new *ELEMENT_SHELL_COMPOSITE keyword, each element must have its
own *PART_COMPOSITE definition.
There are many technical features provided through ANSYS Composite PrepPost, which ease life of
engineers dealing with composites. Just to list of view of those:
- Easy and efficient definition of material reference direction
- Tapering of layer (layer drop-off)
- Analyze ability to drape with fiber correction and calculation of flat wrap
- Shaped cores using CAD-surfaces for variable core thicknesses
- Section cuts with full control on visualization of modeling, production and analysis plies
- And many more…