HAL Id: hal-01552114
https://hal.archives-ouvertes.fr/hal-01552114
Submitted on 30 Jun 2017
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
A 4pt Bending Bond Test Approach to Evaluate Water
Eect in a Composite Beam
Armelle Chabot, Ferhat Hammoum, Manitou Hun
To cite this version:
Armelle Chabot, Ferhat Hammoum, Manitou Hun. A 4pt Bending Bond Test Approach to Evaluate
Water Eect in a Composite Beam. European Journal of Environmental and Civil Engineering, Taylor
& Francis, 2017, Mechanisms of Cracking and Debonding in Pavements: debonding mechanisms in
various interfaces between layers, 21 (s1), pp.54-69. �10.1080/19648189.2017.1320237�. �hal-01552114�
A 4pt Bending Bond Test Approach to Evaluate Water Effect in a
Composite Beam
Considering that water may cause a separation of interfaces between layers of
pavement structures, specific test on bi-layer specimens is performed in a water
bath. For the study of the bond between layers made of cement concrete overlay
on bituminous material, four-point bending results show a competition between
different failure mechanisms. Actually a very good bond resistance between
layers compared to the fracture tension resistance of the cement concrete layer is
preliminary observed in dry conditions. In this work, first results of the water
effect on the behaviour of such a material interface are presented. The final
fracture length of the specimen and the corresponding curve of force-
displacement highlight the influence of water immersion on the debonding failure
mode. The field displacement measurement obtained by Digital Image
Correlation (DIC) is used to improve the understanding of the fracture scenario.
Keywords: crack, debonding, four-point bending test, water effect, DIC,
heterogeneous material
1. Introduction
Composite pavement systems, made of concrete/asphalt and asphalt/concrete structure,
show good potential for being an interesting rehabilitation option for urban pavements
that exhibit structural deterioration (De Larrard et al., 2005). However, due to shrinkage
phenomenon of cement materials, the existing vertical crack combined to environmental
and traffic loading lead to failures that may develop as delamination between layers
(Pouteau et al., 2004) (Chabot et al., 2008). Typically, such debonding initiates and
propagates under the combined influence of normal and shear stresses (Tran et al.,
2004) (Chabot et al., 2004, 2005, 2007). The presence of water, whatever its phase is,
adds to this complex phenomenon an irreversible damage that cannot be ignored
(Vulcano-Greullet et al., 2010) (Vandenbossche et al., 2011) (Mauduit et al., 2010,
2013) (Raab et al., 2012) (Mateos at al., 2016). The damage can be driven with two
mechanisms: (a) loss in strength and durability of materials due to the presence of water
in the pore of asphalt concrete (b) loss of mechanical behaviour of bond between layers.
Evaluating the effect of water on the interface behaviour is complex.
In order to investigate the mixed mode characterisation of such a bond between
bi-layers in the pavement, an existing four-point bending test (4PB) has been adapted
(Figure 1) (Hun, 2012). Two types of bi-layer interface are made with the same material
layers (cement concrete and bituminous concrete). The first experimental results tested
under controlled static conditions (0.7mm/min, ambient temperature), have shown that
this approach is able to give a first idea on the interface resistance of the bond between
two layers and of its treatment (Hun et al., 2012). The advantage of such a test is to be
able to investigate the mechanical properties of interfaces under mixed mode conditions
without using any supports nor applying any loads directly on the bituminous material.
In the case of the UTW (UltraThin White-topping pavements) bi-layer
specimens made of a cement concrete overlay (noted layer 2 on Figure 1) on bituminous
material (noted layer 1 on Figure 1), first analyses at a macro scale level have lead to
the conclusion that the UTW bond has a very good resistance in such a test compared to
the fracture tension resistance of the cement concrete layer. Several final fractures of the
specimens have occurred (Figure 2) in a similar way than for materials used in
reinforced concrete (RC) beams or slabs from where this test is used (Teng et al., 2003).
For the final fracture configuration of Figure 2c, it is difficult, by visual observations
and mechanical response expressed by the load displacement curve, to determinate if
the debonding phenomenon occurs first or not compared to the final bending crack of
the cement concrete layer (Chabot et al., 2013a).
In this paper, to improve the knowledge of the fracture scenario of bi-layer
materials of such a 4PB test, the Digital Image Correlation (DIC) method is used
(Sutton et al., 1983) (Roux et al., 2009). The work also aims to understand if water
favours or not the debonding phenomenon at the interface between UTW pavement
layers. The phenomenon of the cement shrinkage or other effects on bonding between a
cement substrate and a layer such as studied in (Pérez et al., 2009) (Pan et al., 2010) for
cement concrete materials are not considered here.
In the following sections, the theoretical elastic characterization of the bond
during the 4PB test is first briefly presented. Secondly, the characteristics of the
materials tested are given and the preparation of the specimens is presented. Then
several information of the experimental device is described before showing the result
curves for the different test conditions. The fourth paragraph, with the help of the
several DIC investigations and basic calculations, aims to propose a final fracture
scenario of the specimens tested before concluding.
2. Theoretical Characterization of the bond during the 4PB test
A mechanical analysis of the design of the specimen has been performed (Hun et al.,
2012). The calculation has been realized with the help of a multi-particular modelling
approach developped for studying edge effect leading to the delamination of composite
materials (Naciri et al., 1998) (Chabot 1997) (Chabot & Ehrlacher, 1998). In such a
layer-wise modelling of the structure, so-called M4 for multi-particle model of
multilayer materials, the stress fields are defined by through polynomial approximations
in the vertical direction for each layer i. The M4 offers the advantage of defining the
out-of-interface plane normal
v
i,i+1
and shear stresses
τ
i,i+1
at interface between layers i
and i+1. The interface is supposed so to be a surface (that is to say with no thickness).
These interface stresses have a physical meaning and represent the exact out-of-plane
3D stresses calculated at the interface between two layers. The evaluation of the
mechanical fields is then obtained by using the Hellinger–Reissner variational principle
(Reissner, 1950). It reduces the dimension of the problem that is especially convenient
for computing 3D pavement solutions (Tran et al., 2004) (Chabot et al., 2004, 2005,
2007) (Nasser et al., 2016). As opposed to other classical models, the main interest of
this modelling is to yield finite stresses at a free edge or crack tip at the interface point
location of two different layers (Chabot, 1997). For the M4-5n with five kinematic
fields per layer (n: total number of layers i) used here, the membranar stresses in each
layer are written as first-order polynomials and shear and normal stresses are then
obtained by integrating the 3D equilibrium equations. In the M4-5n, the multilayered
structure is considered as a superposition of Reissner–Mindlin plates linked together by
the interfacial stresses.
Applied to the 4PB test under 2D plane strain assumptions with homogeneous,
elastic and isotropic materials, the M4-5n solutions are given in a previous work with
parametric calculations. Depending on ratio value of the Young modulus ratio of the
two materials, the simulation indicates that a competition exists between tensile stress at
the bottom of the cement concrete in layer 2 (between points B and C) and interface
normal and shear stresses (at points A and D) (Figure 1). In order to obtain the
debonding phenomenon between the two layers onto only one side, a non-symmetrical
geometry of the specimens has been finally proposed. A higher geometrical value of a
1
length compared to the a
2
length one favours delamination on the side a
1
first (Figure 1).
This anti-symmetrical specimen allows using the digital image analysis on one side of
the specimen only (see §4). The interested reader should refer to (Chabot et al., 2013a)
for a complete description of all the calculations.
On the following we denote b as the width of the beam. e
i
, E
i
and
υ
i
are
respectively the thickness, Young’s modulus and Poisson’s ratio of each layer i. Figure
3 illustrates, for a total load F of 4.2N, the M4-5n stress distribution in the bi-layer
specimen made of the pavement materials tested in this study around 20°C
(E2/E1≈17.5;
υ
1
=
υ
2
= 0.35
= 0.30, , L
Total
= 480 mm with L=420 mm between supports,
e1 = e2 = 60 mm, b=100). The Figure 3 gives the M4-5n interface distributions of the
normal stress ν
1,2
(x) and the shear stress τ
1,2
(x) between the cement concrete layer 2 and
the bituminous layer 1. Depending on the ratio of modulus of the two material layers
and the loading value, the distribution of the normal stress at the interface may become
negative around the corresponding bending force locations of points B and C. The use
of an anti-symmetrical specimen increases the chances to get cracks (such as a
debonding) around point A before point D. But, depending on the mechanical resistance
of the interface material and the material of layer 2, debonding may occur between the
layers or not as it will be presented in the following sections for the UTW type of
interface.
In addition, to study crack initiation problems with help of the M4-5n, two
delamination criteria in the angle-ply laminates have been proposed (Chabot, 1997,
2000) (Caron et al., 2006). The first one is based on the maximum value of interface
stresses. Using the virtual crack closure technique (VCCT) (Bui, 1978) (Moutou Pitti et
al., 2008), the second criterion is based on an analytical calculation of the individual
strain energy release rates (G
I
, G
II
, G
III
). At a given crack length, the VCCT yields an
expression involving interfacial forces and relative displacements near the crack tip. For
the 4PB test with a debonding crack length “a”, the analytical expression of strain
energy release rates is given in (Eq. 1). Only G
I
is a pure quadratic function of the
υ
1
=
υ
2
= 0.35
normal interface stress. G
II
takes into account the combined terms containing the shear
interface stress, and the generalized out-of-plane shear stress resultants
Q
1
i
of each layer
i.
(1)
As explained and illustrated in details in (Chabot et al., 2013a, 2013b),
according to the principles of linear elastic fracture mechanics, from the sign of the
derivative of energy release with respect to the crack length, the crack is expected to
propagate initially along the interface in a unstable way before this propagation
becomes stable from a given crack length value (function of the values of the materials
and the geometry of the specimen). The instable part of this propagation has been
specifically illustrated very recently on two specimens only, by means of a high-speed
digital photography (Mulian & Rabinovitch, 2016).
Before testing the bi-layer specimen made of pavement materials, a benchmark
procedure has been used on homogeneous materials (Aluminium and PVC) chosen for
their equivalent ratio of Young modulus to those studied in the case of Ultra-Thin
White-topping pavement (UTW) made of a cement concrete overlay on bituminous
material at ambient temperature. The reference specimen, made with an aluminium
(Alu) homogeneous layer 2 sticks with epoxy glue on PVC homogeneous layer 1, has
been tested (L=420mm; b=125mm; a1=40; a2=71; e1=30.6; e2=40.6 mm; Young
modulus ratio: E2/E1≈22.4;
υ
1
= 0.30 ;
υ
2
= 0.34
). The benchmark procedure has
confirmed successfully the simulation results of this brittle fracture mechanics
approach. As expected, the debonding phenomenon is coming very rapidly from the
expected edge until the half-length of the specimen (between the two static loading
forces) (Figure 4) (Hun, 2012).
Using the Digital Image Correlation (DIC) method (see the details given in §4),
the crack length “a” at the interface is estimated. For each crack length, the crack
opening
δ
z
and sliding
δ
x
displacements are measured (Figure 1). Analytical
expressions (Eq. 2) from the literature provides in 2D the corresponding elastic energy
release rate G of each mode I and II (Hutchinson et al., 1987), where α and β are the
Dunder’s nondimensional constants of a bi-material structure (Dunders, 1969).
G
T
a
( )
= G
I
(a) + G
II
(a)
with
G
I
(a) =
1
2b
.
13 e
1
E
2
+ e
2
E
1
( )
35E
1
E
2
ν
1,2
a
( )
( )
2
G
II
(a) =
1
2b
.
4 e
1
1+
υ
1
( )
E
2
+ e
2
1+
υ
2
( )
E
1
( )
15E
1
E
2
τ
1,2
a
( )
( )
2
−
1
2b
.
1+
υ
1
( )
5E
1
Q
1
1
a
( )
+
1+
υ
2
( )
5E
2
Q
1
2
a
( )
⎛
⎝
⎜
⎜
⎞
⎠
⎟
⎟
.
τ
1,2
a
( )
⎧
⎨
⎪
⎪
⎪
⎪
⎩
⎪
⎪
⎪
⎪
