Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 13

Damage Models and formulations:

1. The intralaminar damage model:


a. Initiation of intralaminar damage
Layers are the assembly of their main constituents: the fiber and matrix, and therefore the
criteria for their rupture are specific to each of these elements. During an impact, the fiber
and matrix are subjected to bending stresses that are generated by the combination of tensile
and compressive forces. Figure 1-33 shows a diagram illustrating the modes of damage in the
folds resulting from these tensile and compressive stresses.

To predict the onset of intralaminar damage, a model was utilized that is based on Hashin's 2-
D theory [28, 29]. This theory assumes that the composite material being analysed is linearly
elastic prior to damage initiation. Hashin's damage model includes four distinct types of
damage mechanisms that account for the initiation of intralaminar damage in the
unidirectional fibre sub-plies. These damage mechanisms correspond to tensile and
compressive fibre failures, as well as tensile and compressive matrix failures. The damage
criteria for each of these mechanisms in Hashin's approach are formulated as follows:
- For fiber Failure in tension, (σ11 ≥ 0):

( ) ( )
2 2
σ 11 σ 12
F ft = +¿ α ≥1
Xt S12
- For fiber failure in compression, (σ11 < 0)

F fc =
( ) σ 11 2
Xc
≥1

- For matrix failure in tension (σ22 ≥ 0)

( ) ( )
2 2
σ 22 σ 12
F mt = +¿ ≥1
Yt S12
- For matrix failure in compression (σ22 < 0)

( ) [( ) ] ( ) ( )
2
σ 22 2 Yc σ 22 σ 12 2
F mc = +¿ −1 ≥1
2 S 23 2 ST Yc SL
The equations presented above contain variables labelled F ft , F fc, F ct and F cm which
correspond to the four kinds of damage: tensile fiber failure, compressive fiber failure, tensile
matrix failure, and compressive matrix failure. When F is equal to or greater than 1, failure is
expected to happen. The parameters X T and XC represent the strengths of the laminate in the
longitudinal fiber direction for tensile and compressive stress, respectively. Y T and YC
indicate the strengths of the laminate in the transverse direction for tensile and compressive
stress, respectively. The shear strengths of the laminate are denoted by S L and ST = YC/2 in the
longitudinal and transverse directions to the fibers, respectively. The variables σ 11,σ 22 and σ 12
refer to the normal and shear components of the effective stress tensor, which are used to
evaluate the above criteria.
b. The evolution of intralaminar damage
Once damages have been initiated in a structure, they propagate and cause a decrease in its
stiffness until it eventually collapses. Damage mechanics is used to describe the elastic
behavior of a damaged structure. The damage evolution model is based on Hashin's theory,
which defines damage initiation criteria. To capture the damage caused by tensile fiber
failure, compressive fiber failure, tensile matrix failure, and compressive matrix failure, four
damage variables, namely d tf , d Cf , d tm and d cm, were incorporated in the model. When an
element is undamaged, these variables have a value of 0, but when fully damaged, they have
a value of 1. The damage variable, d, takes on a general form once a particular damage
mechanism is initiated, as described by [25]:
ε f ( ε −ε 0 )
d=
ε ( ε f −ε 0 )

The equation for the damage variable, d, involves the applied strain, ε , and the strain values
corresponding to the initiation of damage, ε 0, and final failure, ε f .
Once damage occurs and evolves, the response of the material is obtained using equation 9
and the damaged elasticity matrix C(d), as defined by equation 8. The values of the damage
variables reflect the current state of fibre, matrix, and shear damage and can be used to
compute the degraded stresses and strains in an element for a given time step using equations
13 to 14.
{ σ }=[ C (d ) ] { ε }
Where [C(d)] is the elasticity matrix which is written as:

[ ]
(1−d f ) E1 (1−d m )(1−d f )❑12 E 2 0
¿ (1−d m)(1−d f )❑12 E 2 (1−d m )E 2 0
0 0 D(1−d s )G1

Where
D=1-(1−d m)(1−d f )❑1221
The three damage parameters d f ,d m andd sthat reflect fiber rupture, matrix rupture, and shear
rupture, respectively, are determined using the damage variables (d ft ,d fc , d mt∧d mc ) by these
equations:

{
t
d f , σ 11 ≥ 0
- Fiber failure :d f =
d cf , σ 11 < 0

{
t
d ,σ ≥0
- Matrix failure :d m= mc 22
d m , σ 22<0

- Shear failure : d s =1-(1−d tf )(1−d cf )(1−d tm)(1−d cm )

2. Modeling of Interface Damage (Delamination).


One of the most important modes of damage within a composite material is delamination.
This damage appears at the interface between two plies [86]. Composite laminates are highly
sensitive to the initiation and propagation of delamination along the interfaces, which can
lead to a decrease in strength and stability in the plane and even catastrophic failure of the
structure [87]. Interlaminar damage refers to the formation and growth of delamination
between the plies of a composite laminate,
Inter-laminar stresses are influenced by the fiber orientation within the composite material.
Delamination propagation at the interface involves three main fracture modes (Figure 1-34)
[88]:
- Mode I (opening mode) corresponds to tensile stresses generated by displacement
perpendicular to the fracture plane.
- Mode II (sliding mode) corresponds to shear stress in the fracture plane generated by
displacement perpendicular to the crack front.
- Mode III (tearing mode) corresponds to shear stress generated in the plane parallel to
the crack front [89].

Delamination can be modelled using two different approaches: cohesive elements and
cohesive surfaces.
- Cohesive elements: Cohesive elements are implemented between the layers by adding
additional degrees of freedom.
- Cohesive surfaces: Separation between two plies is provided by contact surfaces
without any intermediate element.
In finite element codes, interfaces can be modelled using cohesive laws. These cohesive laws
can be applied to surfaces (cohesive surfaces) or elements (cohesive elements). These laws
are based on the description of an attraction law between two nodes of two distinct surfaces

initially in contact. This attraction law is governed by an inter-laminar stress which is a


function of the relative displacement between these two nodes. Figure 1-35 illustrates the
principle of this law.

The initially coincident nodes are represented in red. We can then define the relative
displacements d n and d t , which respectively represent the relative displacement in the normal
direction (mode I) and in the sliding direction (mode II), as well as the associated stresses.
As in the case of criteria related to plies, the traction-separation law is generally bilinear as
shown in Figure 1-36. It consists of a first ascending part reflecting the reversibility of the
contact for a stiffness K0. Once the admissible interlaminar stresses S_n,t are reached, the
stiffness of the interface is degraded until the contact is broken and consequently, the
delamination propagates. This second portion of the curve is called the evolution law, and the
area under the curve corresponds to the energy required to break the contact, which is equal
to the interlaminar toughness for the considered mode [VAN11].

As with plies, the modelling of delamination in the interfaces of a composite material is based
on the delamination initiation criterion and the delamination propagation criterion.
a. Delamination initiation criteria:
The principle of fracture criteria is based on determining the stress and strain field required to
create a crack [88].
The first and most commonly used criterion is the maximum nominal stress criterion. As long
as the stress in a given direction has not reached its limit value in that direction, there is no
initiation [87]:
σt
max
⟨ σn ⟩ SL ;
T max ;
S max
σ if σ n> 0
σ s =1 ; ⟨ σ n ⟩= n
σ n if σ n< 0 {
The quadratic nominal stress criterion can also be used and allows coupling of the different
modes of loading [99]:

( )( )( )
2
⟨ σn⟩ σt 2 σs 2
+ + =1
N max T max Smax

b. Delamination propagation criteria:


These criteria are based on fracture mechanics and take into account the energy release rates
G. Most of these criteria compare the release rates in different directions to their allowable
value G in the different fracture modes. Due to the complex loading modes applied to
structures, cracks propagate in mode I, II, and III. The pure mode is the simplest and easiest
to implement criterion as it does not take into account any coupling.
GI GII GIII
c =1 ; c =1 ; c =1
G I G II GIII

Another simple criterion to implement is Reeder's linear criterion [100] and its quadratic
evolution proposed by Chen [101].
GI GII
c + c =1
GI GII

( ) ( )
α β
GI GII
c + c =1
GI GII

The delamination propagation can also be modelled by criteria that take into account the
coupling between modes such as the Power Law [97].
The power law criteria, which considers the energy release rates in mode I and II as shown in
Equation (3.6) [42], is commonly used under mixed mode loading. However, to better
incorporate the changes in fracture toughness for mixed mode loading, a more suitable model
was proposed by Benzeggagh-Kenane, known as the B-K criterion. This model combines the
contributions of pure modes into a single mode and is illustrated in Figure 3 16 [43].

( )( )( )
α α α
GI GII G III
c
+ c
+ c
=1
GI GII G III

where the coefficient α in the equation needs to be experimentally determined.


Benzeggagh-Kenane [102] proposed a new energy-based formulation that couples modes and
integrates the shear energy release rate GShear with the total energy release rate GT.

( )
α
G s h ear
G I + ( G II −G I )
c c c
=¿1
GT

Where: Gs h ear =G II +GIII


GT =GI +G shear

Most of these criteria are implemented in finite element analysis software that can predict the
impact behavior of composite structures.
The Abaqus simulation employed an embedded cohesive surface law, which requires the
interlaminar fracture energy, Gc, as input. The value of G c was determined using the energy-
based Benzeggagh-Kenane (B-K) mixed-mode propagation criterion. This criterion has been
extensively used in the literature to study delamination growth in composite materials.
The embedded cohesive surface law used to analyse interlaminar damage requires a value of
the interlaminar fracture energy, Gc. This represents the area under the bilinear law shown in
Figure 11. The energy-based Benzeggagh-Kenane (B-K) mixed-mode propagation criterion
was employed to derive a value for Gc. This criterion takes into account both Mode I
(opening tensile) and Mode II (in-plane shear) interlaminar fracture energy, denoted by G Ic
and GIIc, respectively, as well as the B-K mixed-mode interaction exponent, g. The values of
these terms can be experimentally measured. The current Mode I and Mode II energy release
rates, GI and GII, respectively, can be calculated from the FE code by multiplying the relevant
local stress by its conjugate displacement.

1. Finite element modelling


This work describes an attempt to model the results of impact tests on a CFRP composite
specimen. The goal is to develop a validated model that can be used for predictive studies.
The impact event is modelled using a finite element code called Abaqus/Explicit 2018. The
impactor is modelled as a spherical, rigid surface with a lumped mass of 3.2 kg or 7.1 g
depending on the test. The composite specimen is defined using continuum shell elements,
and only in-plane material properties are needed for the numerical model.
Cohesive surfaces are defined at the 0/90 interfaces to capture the interlaminar damage. The
global contact is governed by a general contact algorithm, and friction coefficients are set for
the metal/composite ply and composite ply/composite ply interfaces. The computational
accuracy is set as "double procession" to reduce error accumulation. The FE model is run
with individual time steps, and the computations are performed using 32 CPUs on a Linux
cluster with a run time of 12-15 hours. The material coordinate system for the CFRP laminate
is defined with the lengthwise direction of the specimen as the 11-direction and the width
wise direction as the 22-direction. The fibres are aligned in different directions for the 0 and
90 plies.
All experimental tests were simulated using ABAQUS software to predict the mechanical
behavior and damage mechanisms of the studied composites. The elastic and failure
properties used in the FE model were obtained from the experimental results presented in
chapter 1 and mechanical characterization tests published by Mr. Tarfaoui et al. (25). The
mechanical properties required for the developed FE models are provided in Tables 4.1.
Modules de Young(MPa) Coefficients de Poisson Module de cisaillement(MPa)
E 11 E 22 E 33 ν 12 ν 13 ν 23 G 12 G 13 G 23
Composite stratifié 48.16 11.21 11.21 0.27 0.27 0.096 4.42 4.42 9
l'âme en PVC 0.077 0.077 0.11 0.3 0.3 0.3 0.029 0.029 0.029

Les contraintes admissibles longitudinales et transversales Les contraintes admissibles en


(MPa) cisaillement(MPa)
Xt Xc Yt Yc S 12 S 13
1021 978 29.5 171.8 70 70

Longtudinal
96 96 192 192

The model was based on an asymmetrical laminate of 12 layers with ply angles of [0/90].
The virtual composite specimen was discretized using continuum shell elements (SC8R), and
the loading tool and supports were modelled as analytical rigid bodies (R3D3). Various mesh
configurations were considered for convergence analysis, and for the optimized model, there
were a total of 31834 continuum shell elements (SC8R) and 13506 analytical rigid elements
(R3D4). The boundary conditions had six degrees of freedom of the supports fixed, while the
loading nose was allowed to move only in the direction of the laminate thickness, and the
composite specimen was allowed to move in any direction.
In this study, boundary conditions were applied according to the experimental conditions of
the tests. The damage criteria used here are Hashin criteria for plies, quadratic stress for
initiation of cohesive surface damage, and the Benzeggagh-Kenane energy criterion for
delamination propagation, as presented in Section 2. These failure criteria are implemented in
Abaqus code without the need to define any special elements or user-specific code. The
cohesive surface solution was used for the interfacial contact between composite plies to
simulate interlaminar damage. The cohesive behavior based on the surface provides a
simplified model of cohesive connections with negligible interface thickness. The traction-
separation response defined the properties of the cohesive surface during the simulation. The
elastic properties of the interface and fracture parameters used in the model are given in
Table 1. The onset of delamination damage follows a quadratic nominal stress criterion, and
once the starting condition is satisfied, it progresses according to the Benzeggagh-Kenane (B-
K) criterion [26]. The interaction between objects is simulated by Abaqus general contact
algorithm with the kinematic constraint method for normal behavior and Coulomb friction for
tangential behavior.
Composite type stratifié Composite type sandwich
Les peaux
La mousse de PVC stratifiées

A 3D finite element model was created to simulate damage in composite laminates under
low-speed static indentation using the configuration shown in Figure 4-4b and 4-4c.
Boundary conditions were applied in accordance with the test conditions described in section
3.2 of Chapter 3. The indenter was considered as a discrete 3D rigid body and modeled with
3D rigid elements. The center of gravity of the indenter was chosen as a reference point to
define the velocity and boundary conditions. The clamped boundary condition was
implemented in the plate to replicate the experimental conditions, as shown in Figure 4-4a.
This condition imposes constraints on the translation and rotation degrees of freedom around
the edge of the composite laminate during indentation. Once the fiber and matrix damage
variables reach a maximum value, the element is removed from the mesh to account for the
possibility of penetration. A displacement load was applied to the center of the plates, and the
incident indentation velocity in the transverse direction was set to v = 1.2 mm/min.

La plaque du Impacte
composite ur
Encastrement
To simulate the progressive damage of composites, the built-in Hashin Damage model in
ABAQUS was used to define the damage model in the material definition. The damage
initiation was modeled using Hashin’s failure criteria, which employs four damage initiation
mechanisms, including tension, fiber compression, matrix tension, and matrix compression.
The stiffness degradation and material softening can give rise to convergence problems, so a
viscous regularization scheme was employed to facilitate solution convergence. In this
regularization scheme, a viscous damage variable of 4E-5 was defined for each damage
model.

2. Results and discussion:


Single quasi-static indentation impact
To assess the validity of the numerical simulation, a comparison of the experimental and
numerical characteristics of the response of laminated and sandwich composites to impact,
including force-displacement curves and damage mechanisms, is illustrated in Figures 11 to
15. The desired responses are reported and discussed below.
Load-displacement responses
Figure 4-11a shows the numerical and experimental evolution of impact force as a function of
impactor displacement for the sandwich composite. Points (A-E) have been marked on the
curve to demonstrate the sequence and progression of damage and rupture of the laminated
composites during the impact event (Figure 4-11b).
Figure 4-11a shows that the numerical simulation is able to reasonably predict the key aspects
of the experimental curve, and there is good agreement between the initial slope of the curves
and the maximum force. The load evolution response is primarily decomposed into two
phases. The first phase is elastic and linear (Zone I), corresponding to an elastic and linear
behavior where there is a linear increase in impact force until the peak, at which the
maximum load is reached. A non-linear phase follows (Zone II), characterized by a decrease
in load until rupture and complete perforation due to the initiation of damage.
The results shown in Figure 4-12a provide a detailed view of the behavior of the sandwich
composite during an impact event. The load-displacement curves for both experimental and
numerical data were compared, and points (A-E) were marked on the curve to track the
damage and failure sequence of the composite layers.
The load response was analyzed in three stages. In the first stage, the force increased
nonlinearly until it reached the first peak. This behavior corresponds to the initial elastic
deformation of the composite layers. In the second stage, the force dropped to a valley,
indicating a transition to a nonlinear behavior. This was due to the foam core becoming the
primary component to support the additional impact load. Finally, in the third stage, the force
dropped rapidly due to tearing of the back face of the sandwich composite.
The small differences between the experimental and numerical curves are likely due to the
presence of damage related to material uncertainties or manufacturing defects that are not
accounted for in the FEM model. Figure 4-12b provides a cross-sectional view that illustrates
the overall damage footprint, which can help researchers better understand the failure
mechanisms of sandwich composites during impact events. The extent of damage steadily
increases until a perforation appears on the impacted side of the composite laminates.

The figure 4-13 shows the comparison between the experimentally measured maximum load
and the corresponding numerical results for the laminated and sandwich composites. The
maximum load was predicted with reasonably good accuracy of 1.43% and 6.45% for the
laminated and sandwich composite, respectively.
Figures 4-14 and 4-15 provide visual comparisons of the damage morphology resulting from
the laminated and sandwich composites after quasi-static perforation impact testing. The red
areas on the figures represent the regions where the fracture conditions have been met
according to Hashin's damage criteria, which is a widely used criterion for predicting
composite material failure. The colors within the red zone represent different degrees of
damage, with the rainbow colors indicating the varying levels of damage intensity.
Conversely, the blue areas represent undamaged regions.
Observations revealed that both the laminated and sandwich composites suffered significant
damage, including permanent indentation, matrix cracking, fiber fractures, and delamination.
Delamination was found to be the most prominent type of damage, and it was concentrated
around the point of impact, presenting a circular shape with a slight elliptical tendency.
For the sandwich composite, the impactor penetrated the PVC foam, causing significant
damage that mainly consisted of delamination. The lower face sheet contracted as a result of
the impact, leading to permanent indentation and fractures. The core suffered significant
damage in the form of plastic deformation of the foam. These findings highlight the
importance of considering the impact of core materials in designing sandwich composites for
impact resistance. The results also indicate that the FEM model used in the simulation was
effective in predicting the extent and nature of the damage to the composites, though with
some differences from the experimental results due to material uncertainties and
manufacturing defects.
Based on the results of the finite element simulation, the damage diagrams of all composite
laminates obtained by numerical simulations correspond well to those obtained by
experiments.
Expérimental Numérique

Vue
en
face

Vue
en
coupe

Expérimental Numérique

Vue
en
face

Vue
en
coupe

Conclusion:
This work has discussed a study on the impact behavior of laminated and sandwich
composites. The study used both experimental and numerical simulation methods to
investigate the extent and nature of the damage caused by quasi-static perforation impact.
The results indicate that both laminated and sandwich composites suffered significant
damage, including permanent indentation, matrix cracking, fiber fractures, and delamination.
Delamination was found to be the most prominent type of damage, and it was concentrated
around the point of impact, presenting a circular shape with a slight elliptical tendency.
The FEM model used in the simulation was effective in predicting the extent and nature of
the damage to the composites, though with some differences from the experimental results
due to material uncertainties and manufacturing defects. The findings also highlight the
importance of considering the impact of core materials in designing sandwich composites for
impact resistance.
Overall, the study provides valuable insights into the behavior of laminated and sandwich
composites under quasi-static perforation impact and can inform the design of more resilient
composite structures in various applications.

You might also like