HOME BOOKMARKS SEARCH

Simulating the Impact Behaviour of Composite

Aircraft Structures

Erkan Kirtil, Dieter Pestal, Alexander Kollofrath, Nils Gänsicke, Josef Mendler
EADS Military Aircraft

81663 Munich - Germany

Abstract: Lightweight composite structures are widely used in aircraft industries. Within
certification procedures the impact absorption capability of affected structural parts have to be
proved usually via time and cost consuming certification tests.
Although authorities do not accept a mere theoretical approach for impact-proof , simulation
techniques of different impact scenarios gain increasing significance , especially within
development phases of aircraft structures.
If those theoretical approaches are finally validated by sufficient test results , the simulation
techniques can also be applied on similar structures in order to certify the structure under impact
resistant considerations.
Under this background bird impact analyses on aircraft CFRP leading edges are illustrated using
ABAQUS/Explicit. Phenomenological micro-mechanics will be discussed in order to focus on
high-velocity impact behaviour of brittle structures. In addition, the application of the user-
defined subroutine VUMAT within ABAQUS/Explicit will be demonstrated.
Finally, the report deals with the development of a special impact resistant composite design for
typical aircraft leading edges , combining the flexibility of aramid fabric with the stiffness of
CFRP shell structures.

1. Introduction

Composite materials gain increasing significance in their application within aircraft structures.
Regarding the impact caused by foreign objects (e.g. bird, tire debris, engine fragments, etc.) no
catastrophic flight situation after the damage may occur.
It is essential during a development phase of aircraft structures to simulate the (high and low
energy) impact events using theoretical approaches in order to identify critical impact scenarios
and to initiate adequate structural improvements.
The theoretical approaches have to be verified by certification tests and are often substantiated in
an early project status via development tests. This procedure helps also to reduce the number of
necessary tests to the most critical events which are figured out by mere impact analyses.

2003 ABAQUS Users’ Conference 1

 HOME BOOKMARKS SEARCH

In addition, those activities require detailed knowledge of the failure mechanism of the affected
structure but also the engineering capability to simplify the complex impact response behaviour of
composites within theoretical approaches.
This report deals with the analyses of a bird impact on a vertical stabilizer CFRP leading edge.
The FE-code ABAQUS/Explicit is used in combination with the user-defined subroutine VUMAT
in order to perform single ply analysis during short term impact simulations. Based on classical
laminate theory (CLT) prerequisites linear shell elements are used to model monolithic design, but
also to investigate a specially developed impact resistant design combining CFRP with fabric
aramid. The report illustrates also the usage and effect of available contact algorithms within
ABAQUS/Explicit impact analyses and provides recommendations and discussions of special
parameters which determine significantly the impact response of composite structures.

2. Modelling the impact behaviour of composites

2.1 Material Models

There are different possibilities to define composite materials in the finite-element program
ABAQUS/Explicit.
One possibility is to define a linear elastic orthotropic material in plane stress with the definition
of the LAMINA–parameter in the ELASTIC–option. Elements, associated with the LAMINA–
parameter can not fail in a simulation. In this case failure of elements will be just indicated by a
failure index. This means, that elements are still able to transfer loads within subsequent time
steps.
In order to let elements fail, another possibility is to define a linear elastic and isotropic material
behaviour with the ISO–parameter in the ELASTIC–option. Elements, associated with the ISO–
parameter can fail in a simulation. This means, that elements are not able to transfer loads within
subsequent time steps.
The third possibility is the definition of a user–defined material behaviour in the subroutine
VUMAT, which allows to use user–defined material laws and failure criteria. Elements, associated
with the user subroutine can fail in a simulation.

2.1.1 Orthotropic Material (LAMINA)

This approach can be used when the composite structure can be considered as an linear elastic and
orthotropic material in plane stress. The definition of an orthotropic material allows the input of
the elastic modulus in 1– and 2–direction, whereas the 1–direction is assumed to align with the
fiber direction of the material. Furthermore the poisson`s ratio and shear modulus have to be
defined. In addition, the actual lay–up definition of the composite structure can be defined in the
SHELL SECTION–option using the COMPOSITE–parameter. Therefore, each single ply is
described by a correlating lamina dataset. The density value of one layer has to be added in the
material block using the DENSITY–option.

2 2003 ABAQUS Users’ Conference

 HOME BOOKMARKS SEARCH

2.1.2 Isotropic Material (ISO)

This approach can be used when the composite structure can be considered as a linear elastic,
homogeneous and quasi-isotropic material. The definition of an isotropic material allows the input
of just one elastic modulus and one poisson`s ratio. In this case the resultant material data of a
composite structure, which depend on the elastic data and the lay–up of the structure, are
calculated with the CLT and an average elastic modulus of the whole laminate has to be defined.
The density value corresponds to the resultant density of the entire composite structure.

2.2 Default Failure Criteria in ABAQUS / Explicit

2.2.1 LAMINA – FAIL STRESS / FAIL STRAIN

Elements, associated to the LAMINA–option can not fail during a simulation. One possibility is to
indicate failure with the FAIL STRESS– and FAIL STRAIN–option, which can be added to the
material block. The FAIL STRESS–option provides four stress–based failure criteria. For these
failure theories stress–based tensile, compressive and shear limits have to be defined. The failure
criteria provided are:

• Maximum stress theory

• Tsai–Hill theory

• Tsai–Wu theory

• Azzi–Tsai–Hill theory
The FAIL STRAIN–option provides one strain–based failure criterion. For this failure theory
strain–based tensile, compressive and shear limits have to be defined. The failure criteria is:

• Maximum strain theory

2.2.2 Shear Failure

For an isotropic material element failure is controlled by the SHEAR FAILURE–option, which
describes the equivalent plastic strain at failure. This option needs in addition the PLASTIC–
option, which defines the yield stress. Due to the fact, that composite materials, such as CFRP or
GFRP, do not have significant plastic deformation, the allowed equivalent plastic strain at failure
of the structure has to be kept low.

2.3 VUMAT Approach

This approach can be used to define a user–defined material behaviour within the user subroutine
VUMAT. Unlike to the usage of the predefined LAMINA–option, the usage of this subroutine has
the advantage of being able to define an orthotropic material behaviour including element failure.
If including the VUMAT into an analysis by the USER MATERIAL–option, the entire material
behaviour and failure criteria have to be defined within the subroutine itself. Therefore, the CLT
for plane stress was implemented in the subroutine VUMAT, which is valid for linear elastic and

2003 ABAQUS Users’ Conference 3

 HOME BOOKMARKS SEARCH

orthotropic composite structures. In addition, the actual lay–up definition of the composite
structure can be defined in the SHELL SECTION–option using the COMPOSITE–parameter.
Therefore, all material values in the material block defined by the USER MATERIAL–option
refer to a single ply analysis based on lamina input data. In addition, the DEPVAR–option has to
be defined in the material block in order to include a solution–dependant state variable. The
DELETE–parameter of this option defines the state variable, which controls element failure during
a simulation. The Tsai–Wu theory and a modified Maximum–Stress theory are the applied failure
theories, which control the value of the state variable. These theories are also implemented within
the subroutine. When using the subroutine VUMAT, the finite element program
ABAQUS/Explicit is not able to calculate the transverse shear stiffness of the material. Therefore,
it has to be defined by using the TRANSVERSE SHEAR STIFFNESS–option, which has to be
added to the SHELL SECTION–option. An example of a possible ABAQUS/Explicit–entry is
given below.
*SHELL SECTION, ELSET=STRUCTURE, COMPOSITE, ORIENTATION=OID1
0.625, 3, CFRP, 0.
0.625, 3, CFRP, 90.
0.625, 3, CFRP, 0.
*TRANSVERSE SHEAR STIFFNESS
10000., 10000., 0.
*MATERIAL, NAME=CFRP
*DENSITY
1.8E-9,
*USER MATERIAL, CONSTANTS=9
136000., 3600., 2100., 0.3, 1360., 110., 800., 80.,
45.
*DEPVAR, DELETE=1
1

3. Contact Algorithm

In impact analyses two different contact algorithms can be used. The following chapters point out
the fundamental differences of the available algorithms.

3.1 Kinematic Contact

With the Kinematic algorithm the contact penetration of a slave node into a master surface is
eliminated at the end of each time increment i. After prediction of the penetration distance dpred an
acceleration correction is applied to the nodes to result in a compliance of the slave nodes with the
master surface. The contact force is calculated by the product of mass m times acceleration
dpred/∆t² (Figure 1). Since the nodes come to lie exactly on the master surface the Kinematic
contact can be physically interpreted as a local plastic contact.

4 2003 ABAQUS Users’ Conference

 HOME BOOKMARKS SEARCH

Slave Master
fi i
i+1 predicted
fi = m*(dpred/∆t²) corrected dpred

Figure 1. Kinematic contact algorithm

3.2 Penalty Contact

The Penalty contact algorithm permits small penetrations in order to generate contact forces. The
contact penetration is minimized by inserting a virtual spring between the slave node and the
master surface. The contact force is then equal to the product of the automatically calculated
spring stiffness k times the penetration distance d (Figure 2). The Penalty contact can be
physically interpreted as a local elastic contact.

Slave Master
i
k i+1
fi+1 = k*d fi+1
d

Figure 2. Penalty contact algorithm

3.3 Comparison of contact algorithms regarding bird impact

Comparing energy time histories the most significant difference exists in the results for the work
of external forces ALLWK. Figure 3 shows the results related to bird impact analyses: While the
energy term ALLWK remains zero throughout the Penalty contact analysis, it becomes strongly
negative in the Kinematic contact. It represents a numerically induced kinetic energy loss to
ensure compliance of slave nodes (bird) and master surface (structure). In the Users Manual it is
recommended to minimize the effect by refining the mesh in the contact area in order to reduce the
mass of contacting nodes and consequently the kinetic energy loss involved. Due to a continuous
failure of bird elements all bird nodes will finally contact the structure which will produce an
inevitable high negative external work.
For most problems the conversion of some bird energy into external work will have negligible
influence on the absorbed strain energy of the structure ALLIE. Especially for very stiff structures
(e.g. rigid wall) the decrease of external work can result in less energy absorbed by the structure.
This effect can be supported by a contact algorithm-dependant failure behaviour of the bird.

2003 ABAQUS Users’ Conference 5

 HOME BOOKMARKS SEARCH

KINEMATIC CONTACT PENALTY CONTACT

Figure 3. Energy history of bird and VTP leading edge regarding bird impact

Regarding the bird behaviour the Penalty contact analysis revealed an insufficient failure of bird
elements probably due to a low stiffness of the virtual springs (Figure 4). Especially for leading
edges of high stiffness and high curvature the bird can not be cut into two pieces by the structure.
The bird will stick to the structure and will continue to absorb energy and to apply a contact load
in the proceeding analysis.

KINEMATIC PENALTY

Figure 4. Intact bird elements at t=2ms

6 2003 ABAQUS Users’ Conference

 HOME BOOKMARKS SEARCH

The advantages and disadvantages of both contact algorithms are listed in Table 1.

Table 1. Features of Kinematic and Penalty contact

Kinematic Penalty
+ Default setting of Abaqus + External work equal to zero
+ Failure of bird elements + Improved numerical stability
+ No penetration of slave nodes - Reduced failure of bird elements
- Negative external work
- Less absorbed strain energy

4. Bird Impact Simulation

The following chapters show examples of bird strike analyses for typical composite vertical
stabilizer leading edges. The details described in chapter 2 and 3 have been applied within these
analyses.

4.1 Monolithic CFRP Leading Edge Design and FE Model

For certification of the CFRP design, a damage of the leading edge is accepted, as long as the
kinetic energy of the bird elements penetrating the skin is low and the damaged structure is able to
carry all occurring loads.

Impact Zone

A
A

CFRP t=4,51mm
CFRP t=1,44mm
3 3
2 1 2 1

Figure 5. Design and FE-Model Vertical Tail Plane

The vertical stabilizer leading edge skin consists of CFRP with a basic wall thickness of 1.44 mm.
The leading edge nose is reinforced and has a thickness of 4.51mm with a single ply thickness of
0.205mm. The ribs are made from aluminium 2024T3.

2003 ABAQUS Users’ Conference 7

 HOME BOOKMARKS SEARCH

Section A-A: Lay Up CFRP

Outside Inside
4.51 mm 0/45/0/45/0/45/0/45/0/0/45/0/45/0/45/0/45/0/45/0/45/0/45/0
45/0/45/0/45/0/45

1.44 mm

Figure 6. Lay Up CFRP Leading Edge

The impact position at the leading edge is located between two ribs.

Table 2: Material Data for CFRP1,2

E11 E22 G12 ν12 σ1 t σ2 t σ1 c σ2c τ12 t ρ
[MPa] [MPa] [MPa] [-] [MPa] [MPa] [MPa] [MPa] [MPa] [kg/mm³]
CFRP 62000 62000 4100 0.05 740 740 825 825 120 1.6E-6

pl
E σy ν ε f ρ
[MPa] [MPa] [-] [-] [kg/mm³]
2024T3 72390 350 0.33 1. 2.8E-6

Definition of shell offset

As different wall thicknesses are used, the parameter OFFSET is necessary in order to model in
correlation to the loft line.
Input deck entry for a CFRP shell element
*SHELL SECTION, ELSET=SHELL, COMPOSITE, ORIENTATION=OID22, OFFSET=SPOS
Boundary Conditions
The structure of the vertical stabilizer is fixed at the skin nodes linked with the front spar (not
modelled). Degrees of freedom DOF 1 and 3 are set to zero in the global coordinate system. Nodes
of the lowest and highest rib, connected to the spar, are fixed in 1-, 2- and 3-direction.
Furthermore, the top and bottom edges are restrained in global 3-direction.
Contact definition
The analyses are done with the kinematic as well as the penalty contact algorithm. Possible effects
on the calculation results are principally described in chapter 3.

1
Letter t describes a tension value, letter c a compression value.
2
In order to model the correct 1 and 2 direction in ABAQUS, a cylindrical coordinate frame is
used.

8 2003 ABAQUS Users’ Conference

 HOME BOOKMARKS SEARCH

Failure Criteria
The failure criteria for those analyses are based on Tsai-Wu-theory and Maximum-stress theory
which have been implemented in VUMAT (ref. chpt. 2).
FE Model BIRD
For bird strike simulations onto tail planes, an 8 lb bird impact must be considered for empennage
structures, acc. to FAR §25.631. The bird is modelled within a Lagrangian approach as a cylinder
with hemispherical ends [1]. With an average element length of l=7,8 mm the model results in
11711 solid elements (C3D8R). The bird nodes are charged with an initial velocity
vBIRD=295kt=152m/s. As failure criterion the tensile failure is used.

l=288 mm

D=144 mm

Figure 7. Lagrangian FE Model BIRD

Syntax for material BIRD:

*MATERIAL, NAME=BIRD
*DENSITY
9.343E-10,
*ELASTIC, TYPE=ISO
100., 0.49
*PLASTIC
0.5
*TENSILE FAILURE
10.

2003 ABAQUS Users’ Conference 9

 HOME BOOKMARKS SEARCH

Results

No damage occurs in the monolithic CFRP structure. To check the plausibility of the results,
transient energy graphs are regarded.

ETOTAL
ALLPD BIRD
ALLAE BIRD
ALLKE BIRD
ALLSE BIRD
ALLIE BIRD

Energy [mJ]

Time [s]

ALLKE LE
ALLIE LE
ALLSE LE
ALLPD LE
Energy [mJ]

Time [s]

Figure 8. Energy vs. time of monolithic CFRP leading edge analyses

The external work ALLWK and the total energy ETOTAL stay at constant levels. The energy
absorbed by the leading edge structure ALLIE is low, compared with the high kinetic bird energy.
As the leading edge exists mainly of CFRP components, they have no plastic deformation. The
little value of ALLPD is caused by low plastification of the aluminium ribs. In addition, the
recoverable strain energy ALLSE remains at low levels during the impact simulation.

10 2003 ABAQUS Users’ Conference

 HOME BOOKMARKS SEARCH

Reaction forces at the leading edge attachment to the spar box

The calculated reaction forces of the leading edge attachment area result in bearing stress at the
joint positions of about 270 MPa. This is lower than the allowable bearing strength of CFRP of
about 480 MPa. Therefore, no failure in the attachment area is expected.

4.2 CFRP-Aramid-Leading Edge Design

This design is developed in order to design lightweight composite structures as impact resistant:
An outer CFRP shell covers a dry woven fabric aramid which is again sheltered by a thin CFRP
shell. At the edges the CFRP and the aramid are co-cured [3]. The design philosophy allows the
bird to penetrate the exterior CFRP-surface. The remaining kinetic energy will be absorbed by the
aramid woven fabric. The dry fabric catches the bird and destroys the inner CFRP shell within its
large deformation. The geometry and boundary conditions are the same as described in chpt. 4.1.

4.2.1 Geometry

A A

CFRP CFRP
ARAMID (dry) ARAMID
CFRP CFRP
(co-cured)

Figure 9. Cross section of CFRP-Aramid-Leading Edge

border
ARAMID „CFRP“ inner
A AL-RIBS
A

X
Y

X = “CFRP” outer side

Y = “CFRP” outer forward

Figure 10. FE Model

2003 ABAQUS Users’ Conference 11

 HOME BOOKMARKS SEARCH

Table 3. Thickness distribution

Section t [mm]
“border” 3.845
“CFRP” outer side 1.435
“CFRP” inner 0.41
“CFRP” outer forward 2.255
Aramid 2.0

4.2.2 FE-Modelling

The whole model is designed with S4R elements in three layers. Special effort with the modelling
of this design is put on the aramid elements and their contact conditions. The border elements are
one layer and the connection is created as follows:

border elements CFRP (S4R)

(S4R)
ARAMID(S4R)
CFRP (S4R)

Figure 11. Element modelling at the edge

Material data
Tables 4 and 5 give an overview of the material and allowable data of the ARAMID fabric. The
allowable data for ARAMID for compression and shear are set to very high values in order to
prevent respective failure. The carbon data are described in chpt. 4.1. All material values refer to a
single ply.

Table 4. Material data for Aramid

material t [mm] E11 [MPa] E22 [MPa] G12 [MPa] ν12 [-] ρ [kg/mm³]

ARAMID 0.2 120000 120000 3200 0.15 0.625E-06

Table 5. Allowable data for Aramid

material σ11t [MPa] σ22t [MPa] σ11c [MPa] σ22c [MPa] τ12 [MPa]

ARAMID 2800 2800 10000 10000 10000

12 2003 ABAQUS Users’ Conference

 HOME BOOKMARKS SEARCH

Table 6. Material data for border elements

material E11 [MPa] E22 [MPa] G12 [MPa] ν12 [-] ρ [kg/mm³]

CFRP/ARAMID 88000 88000 4000 0.095 1.1628E-06

The stress limits for the border elements are the same as for the carbon layer, which is supposed to
be a conservative assumption. The material data for the border elements are shown in Table 6.
The stacking sequence at the impact centre is shown in Table 7.

Table 7. Stacking sequence of the laminate in the impact area

laminate sequence outer < ------ > inner

C45 C0 C45 C0 C45 C0 C45 C0 C45 C0 C45 n*Adry C0 C45

where, C0 CFRP in 1-direction

C45 CFRP in 1,2-direction
n*Adry n ARAMID-layers

Contact definition
Amongst the leading edge layers the contact definitions are defined as master slave balanced
contact. The contact with the bird is defined as a pure master slave contact.

Table 8. Contact pairs

SLAVE MASTER
BIRD LAYER_1
NODE
BIRD LAYER_2 /
SURFACE
BIRD LAYER_3

LAYER_1 LAYER_2 SURFACE

/
LAYER_1 LAYER_3
SURFACE
LAYER_2 LAYER_3 (balanced)

where: Layer_1 → outer CFRP layer (t=2.255mm)

Layer_2 → ARAMID layer (t=2.0mm)

Layer_3 → inner CFRP layer (t=0.41mm)

2003 ABAQUS Users’ Conference 13

 HOME BOOKMARKS SEARCH

4.2.3 Results

Figure 12. Outer CFRP layer

Figure 12 shows the penetrated outer CFRP layer. Figure 13 depicts the ARAMID layer. No
element has failed.

Figure 13. ARAMID layer

The mass reduction potential is supposed to be in the range of 10…15% [2] in comparison to
monolithic CFRP design (chpt. 4.1).

14 2003 ABAQUS Users’ Conference

 HOME BOOKMARKS SEARCH

5. Conclusion

This report illustrates the application of ABAQUS/Explicit within the impact analysis of
composite aircraft structures.
One emphasis is put on specific features and techniques which have been used within those types
of analyses:
Material models were discussed applying the “lamina”-entry, which does not include the
failure of elements within subsequent transient process simulations. In addition, the
typical input of the VUMAT-approach is given, which enables the user to consider
element failure.
Contact algorithms, as “kinematic” and “penalty” are described with respect to their
significant influence on the results of impact simulations.
Furthermore, this article focuses on the example of a bird impact simulation on a typical vertical
stabilizer CFRP leading edge.
Typical output results, based on transient energy graphs are given, which contain the first
plausibility checks of the FE-analysis. Additionally, the main input data, like geometrical and
material data, boundary conditions, contact definitions and applied failure criteria are given,
including the Lagrangian FE-Bird-Model.
As an outlook the development of an impact resistant composite design is illustrated using mere
analysis results. Based on those theoretical data, a weight benefit of 10 .. 15% is supposed to be
reached in comparison to the CFRP-design of a leading edge with a wall thickness corresponding
to the ballistic limit.

6. References

1. Betz S., Investigation of Bird Material Substitutes considering Bird Impact Simulations,
Dornier Luftfahrt GmbH and GhK Kassel, 2000
2. Hartung, S, Vogelschlagsimulation in Anwendung auf Flugzeugstrukturen aus
faserverstärkten Kunststoffen, Thesis at Fairchild Dornier GmbH and Hochschule der
Bundeswehr München, 2002
3. Mendler, J., Strukturelement für ein Luftfahrzeug, PCT/EP02/06859, 2001

2003 ABAQUS Users’ Conference 15