Construction and Building Materials 53 (2014) 381–391

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Analysis of the behaviour of a bonded joint between laminated wood

and ultra high performance fibre reinforced concrete using push-out test
Georges Youssef a,⇑, Louisa Loulou b, Sylvain Chataigner a,c, Sabine Caré b, André Flety a, Robert Le Roy b,
Michel Bornert b, Christophe Aubagnac a
a
Département Laboratoire d’Autun – CETE Lyon/Bd de l’Industrie, BP 141, 71405 Autun, France
b
Université Paris-Est, Laboratoire Navier (UMR 8205), CNRS, ENPC, IFSTTAR/6-8 Av. Blaise Pascal, 77455 Marne La Vallée, France
c
LUNAM Université – IFSTTAR de Nantes – SOA/Route de Bouaye, 44341 Bouguenais, France

h i g h l i g h t s

 A test protocol using the push-out test was developed.

 The polyurethane did not provide sufficient capacity for the bonded assembly.
 The bonded assembly’s ultimate capacity did not correlate with used adhesive’s stiffness.
 We obtained adhesion properties drop of about 30% due to ageing.
 A method for measuring strains by digital image correlation was also tested.

a r t i c l e i n f o a b s t r a c t

Article history: Bonded composite constructions using timber and ultra-high performance fibre reinforced concrete
Received 25 January 2013 (UHPFRC) are investigated as highly innovative structural elements technically and economically effi-
Received in revised form 17 October 2013 cient, and having better environmental performances. Such technique could indeed offer new structures
Accepted 31 October 2013
typologies for bridges for instance associating timber for main beams with UHPFRC slabs. In timber–con-
crete-composite structures, connection is traditionally achieved with mechanical means. The research
presented herein describes the timber–concrete assembly by adhesive bonding and specially the behav-
Keywords:
iour of bonded joint between Laminated Wood (LW) and UHPFRC. This type of assembly has to be inves-
Timber–concrete joints
Bonding structural
tigated concerning mainly the durability of the bonded assembly and its ability to transmit the forces
Shear strength concentrated on edges. The present study describes the development of a test called ‘‘Push-Out’’ used
Stiffness to characterize the behaviour of the timber–concrete bonded after or prior to environmental ageing. Fol-
Push-out test lowing the development of the used test protocol, investigations are realized on four different commer-
Digital image correlation cial resins to study the influence of the adhesive stiffness on the force transfer focusing in particular on
Hygro-thermal deformations the failure mode, on the ultimate capacities and some local deformations. The mechanical behaviour of
the bonded joint is investigated in particular from a theoretical point of view and from the determination
of deformation field obtained by Digital Image Correlation (DIC).
Ó 2013 Elsevier Ltd. All rights reserved.

1. Influence of the stiffness of the resin Concrete increases the stiffness of the structure and protects
the wood from environmental effects [1]. There have been few
1.1. Context of the research investigations [2–6] on the use of adhesives for the connection
of timber beams with standard concrete slabs. Within all these
Timber–concrete composite structures have been used in cases, the transfer of shear forces through the bond was limited
many countries. These structures are becoming important by the low tensile strength of the concrete material. In this con-
because they bring many advantages when compared with tradi- text, the use of steel fibre reinforced high-, or ultra-high-perfor-
tional timber structures. Structurally, wood is a lightweight mate- mance fibre reinforced concretes seem promising, because it
rial, it is renewable and, when properly treated, it is durable. exhibits high tensile strength of the concrete material [7]. A fur-
ther advantage that results from the steel fibre reinforcement is
that the concrete slabs can be very thin because there is no need
⇑ Corresponding author. Tel.: +33 3 85 86 67 41; fax: +33 3 85 86 67 79. for conventional steel-bar reinforcement. Thereby, the weight of
E-mail address: georges.youssef@hotmail.com (G. Youssef). the composite construction can be reduced further.

0950-0618/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.conbuildmat.2013.10.074
382 G. Youssef et al. / Construction and Building Materials 53 (2014) 381–391

IFSTTAR Institute (Institut Français des Sciences et Technologies important to note that structural adhesive bonding reduces stress
des Transports, de l’Aménagement et des Réseaux) has worked concentration phenomena compared with more conventional
from many years on the development of composite timber– methods of assembly and that the stiffness of the connection de-
concrete structures to obtain new typologies of structures, techni- pends in that case on the stiffness of the used adhesive and the
cally and economically efficient, and having a better environmental thickness of the adhesive layer.
performance [8]. Up to date, the connection between the timber The connecting system exhibits a non-linear shear force-rela-
and the concrete has been achieved with mechanical means such tive slip relationship even for low load values [26,27]. Indeed, the
as screws, bolts or with a concrete indentation into the wood use of an adhesive allows distributing the shear forces uniformly
[9–13]. Since a few years, the technique of structural adhesive over the entire surface and thus limits the local force concentra-
bonding has been used and developed and is the subject of a pat- tions which are unavoidable when mechanical connections are
ent. This technique, unlike the conventional assembly techniques used. The adhesive connection is also quite slip-free, which helps
(local or discontinuous connectors), can ensure an almost-perfect to reduce the deflections.
connection between the two materials [14,15], and thus optimize In order to assess the influence of the stiffness of the adhesive
the quantity of material to be used in structures as each part on the elastic behaviour of a composite beam, numerical modelling
may contribute to the structure according to its own mechanical using finite elements FE was carried out [28]. A 10 m long, and
performance. 240 mm wide timber–concrete beam has been modelled and sub-
Although the assembly by structural bonding has been used for jected to 4 points bending. Distance between the loading points is
several years for other civil engineering applications (for example, 2.2 m and the value of each applied load is 500 kN (Fig. 2b). A rect-
in the case of adhesively bonded composite reinforcement or com- angular section is included to simplify the analysis and realize a 2D
posite steel–concrete structures [16–22]), its possible use for this calculation supposing plane strain conditions. The concrete ele-
type of application has to be investigated in particular to verify ment has a thickness of 170 mm, the element in wood has a thick-
its ability to transmit the forces concentrated on edges or to resist ness of 600 mm and the adhesive layer has a thickness of 2 mm
to environmental loading. First studies conducted at Navier Labo- (Fig. 2a). The three materials are supposed to be isotropic and elas-
ratory have investigated on large scale the process for composite tic, and perfect adhesion between the three materials is consid-
wood–concrete beams. Static and fatigue validations of the process ered. The study focused on the evolution of the mid-span
have been realized [14], and there is now an interest in the envi- displacement as a function of the elastic modulus of the adhesive
ronmental durability of such structures [23]. (Fig. 3).
In order to conduct a more consistent number of investiga- The numerical modelling shows that mid-span displacement of
tions with a more limited quantity of material, it was decided the composite structure is very little affected by the elastic modu-
to investigate the use of the ‘‘Push-Out’’ test that consists in lus of the used adhesive when its stiffness varies between 200 and
loading the two symmetrical joints in shear using a compression 10 000 MPa (Fig. 3). Most so-called structural adhesives have an
tool in order to study in particular the influence of the properties elastic modulus greater than 200 MPa and are thus within this
of the resin on the shear capacity of the bonded joint. In the fol- range of stiffness. This result is consistent with Fig. 1.
lowing section (Section 1.2), as a preliminary work, a numerical It exists therefore an important range for the designer on the
study is undertaken to confirm the importance of the choice of choice of the adhesive. It must be emphasized that the stiffness
the resin on the mechanical behaviour of a bonded composite of the used adhesive will also impact on the behaviour of the struc-
construction. The experimental program will then be described ture in response to hygro-thermal stress. This is in particular the
(Section 1.3). subject of this work.

1.2. Properties of bonded composite construction, preliminary work 1.3. Aim of the research

In composite structures, strength and deformation characteris- If in the short term, structural adhesive bonding seems ade-
tics of interface connections are very important in order to evalu- quate [14,15], it is momentous to get interested in the ageing of
ate the mechanical behaviour of the structure. The degree of such an assembly solution. More particularly, the influence of the
composite action depends on the type of connector, timber and stiffness of the used connection in the case of thermo-hygroscopic
concrete properties. loading (ageing conditions) may be an important topic to investi-
The major influence on the effectiveness of timber–concrete gate. Initial investigations on constrained shrinkage phenomenon
composite structures can be assigned to the stiffness of the con-
nection between timber and concrete. Flexible connections like
dowel-type fasteners, which were mostly used up to now, lead
to decreasing bending stiffness of the members. Using the adhe-
sive bonding technology is a way to realize a rigid compound.
Especially if fibre-reinforced concretes are applied, this technol-
ogy is reasonable because of the high tensile strength of the con-
crete material. Until now, few investigations were made on this
type of application.
Many studies got interested in the influence of the stiffness of
the connector on the stiffness of the obtained composite structure
studying different existing connector systems. These studies al-
lowed [14,24,25] to realize a pseudo-classification of the used con-
nectors: Local connectors (tips, screws, studs, long bolts, etc.) that
provide a limited stiffness to the structure, half-continuous con-
nectors (set of long bolts, expanded metal, etc.) whose properties
of stiffness are highly dependent of the distribution and geometry
of the elements, and structural adhesive bonding, that allow to en-
sure an optimal mixed functioning of the structure (Fig. 1). It is Fig. 1. Stiffness of a composite structure against the stiffness of the used connector.
 G. Youssef et al. / Construction and Building Materials 53 (2014) 381–391 383

Fig. 2. Cross-section of the studied composite beam (left). Load configuration (right).

The presented study describes in the first part the development

of this test: choice of the assembly materials, choice of sample
dimensions, realization of samples, and realization of the mechan-
ical tests. The second part of the study focuses on the results of this
test: failure mode, interpretation of the results from a theoretical
point of view (theory of Volkersen) and using Digital Image Corre-
lation (DIC). The theory of Volkersen allows to determine the stress
concentrations along the joint which may be correlated to the
strain mapping obtained by the DIC method. Digital images are re-
corded during mechanical test and processed using an image cor-
relation algorithm to obtain the displacement field from the
reference image and the deformed images [37,38]. The DIC proce-
dure consists in finding the mechanical transformation between
the reference and deformed images, by optimizing a correlation
coefficient which measures the similarity of the grey level distribu-
Fig. 3. Vertical displacement along half of the composite beam for different tion in the two considered images. From the obtained displace-
adhesive’s Young modulus. ment field, local strains are then determined.

2. Materials and testing

for this type of composite structures were realized in [29]. The
study here is complementary and is designed to determine the 2.1. Materials

influence of the stiffness of the adhesive on a composite structure

In order to realize tests to choose the most efficient resin and to study the influ-
connected by bonding with no ageing in a first step. Consecutive ence of hygroscopic thermal cycles on the resistance of the assembly, four adhesive
studies are underway on the subject of hygro-thermal loading with materials of various elastic characteristics (stiffness, strength, etc.) were used to
the same adhesive materials [30]. Four adhesive materials with make push-out investigations. The four adhesive materials, chosen in order to cover
various elastic stiffnesses are here studied using push-out tests. a large range of glue properties, are given in Table 1. The four materials are two-
component and cold-curing.
Moreover, the stiffness of the used organic materials also depends The pot life is the time for the application of the material without risk of rheol-
on the thermo-hygroscopic conditions and may change with time ogy losses. The mechanical behaviour of the resins was determined by tensile tests
[31]. Thermo-hydro-mechanical characterization methods (using realized at a rate of 1 mm/min (Fig. 4). Strain was measured either by strain gauge
a dynamic mechanical analysis for example for the thermo- in the case of the two stiffer adhesives or by laser optical strain sensor in the case of
the two softer adhesives. To determine the kind of mechanical behaviour, load-un-
mechanical issue) are therefore necessary to help choosing the
load cycles were performed. Glass transition temperature was determined using
adhesive. DSC analysis (differential scanning calorimetry) according to EN 12 614. It must
The stiffness of a connection system is determined using its slip be highlighted that the glass transition temperature is close to the maximum ser-
modulus that quantifies the resistance offered against displace- vice temperature of structural adhesives.
ment in the contact surface between concrete and timber when a The adhesive material samples were realized at the same time than the push-
out samples (three specimen for each resin), and stored under the same conditions.
load is applied to the structure. In addition to stiffness, ductility Tests were performed for all samples (resin samples and push-out samples) four
of the connection is also an important safety characteristic of the months after the beginning of the polymerization of the resins. Two series of tests
composite structure. Strength, stiffness and ductility of interface have been considered (without and with ageing). The first one was related to tests
connection are usually evaluated through shear tests. Few without RH (Relative Humidity) and T (temperature) variations (room conditions,
the moisture content MC in the wood is about 13%). Another series of tests was
researchers have worked on shear tests for adhesively bonded
made to test assemblies similar to those previously studied after one year of ageing
joints in the case of timber–concrete composite structures. In gen- in various conditions (constant temperature: T = 20 °C but variable Relative Humid-
eral, researchers use national standards such as EN 26891 [32] and ity: 50% < RH < 95% or RH > 95%, the moisture content in wood varies between 13%
ASTM D 5652 [33], both for timber joints, or EUROCODE 4 [34] for and 22%).
steel–concrete composite structures. Steel–concrete composite The four studied resins, provided by SIKA company are:

structures already have a standard shear test for joints (push- – Resin Sikadur Ò 30: bi-component epoxy resin used in the field of civil engineer-
out) [35]. As far as timber–concrete composite structures with ing for bonding of pultruded carbon plates on concrete civil engineering struc-
mechanical connectors are concerned, there exists in the literature tures (Sample named E1).
a shear tests for joints (push-out) [36]. One of the aims of this pa- – ResinÒ Sikadur 330: two-component, solvent-free, moisture-tolerant, high
strength, high modulus structural epoxy adhesive used in Structural Strength-
per reports on the development of a shear test for timber–concrete
ening Systems (Sample named E2).
composite structures using a static test (push-out) without – ResinÒ SikaForce 7550L15: two-component polyurethane sealant used in the
mechanical connectors. industrial field (Sample named PU).
384 G. Youssef et al. / Construction and Building Materials 53 (2014) 381–391

Table 1
Main properties of the four used adhesives.

Adhesive Epoxy 1 Epoxy 2 Modified Acrylic Polyurethane

Pot life at 20 °C 1h10 30 min 5 min 15 min
Short term mechanical behavior Elastic and brittle Elastic and brittle Elasto-plastic (important viscosity) Hyperelastic
Tensile Young’s modulus (MPa) 12,000 5200 200 10
Ultimate tensile stress (MPa) 32 32 3 (Yielding) <. . . . . . <10 . . . <4.5
Glass transition temperature (°C) 51 63 70 . . . <50

Fig. 4. Geometry and dimensions of samples and Specimen under tensile test in the Zwick machine.

– Resin SikaFastÒ 5215: two-component modified acrylic used for structural It is possible from these data and the stiffness of the adhesive to determine the
bonding in the industrial field (Sample named MA). value of the characteristic elastic anchorage length of the realized bonded joint Lc
according to the theory of Volkersen [41] and the development given in [42,43].
As far as concrete and wood are concerned, no characterization tests were car- The bonded joint length Lc is then given by:
ried out. Mechanical data necessary for these materials were taken from the   qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1
literature. Lc ¼   nM þ 1 þ n2 ðM 2  1Þ ð1Þ
k 1f

with k the exponential term of elastic shear stress distribution M the joint balance
factor defined as:
2.2. Dimensions of the samples rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
Gc 1
k¼ ec E1 e1
þ E21e2
The protocol presented below was developed following the realization of sev-
ð2Þ
eral series of push out tests (see Fig. 5 for the dimensions of the samples). It allows  
working on small samples easy to manipulate with [39] similarly to the work car- M ¼ Min EE21 ee21 ; EE12 ee12
ried out in [40]. In the case of the push-out test, two failure modes can occur: (i)
failure at the interface by shearing or tearing (which may occur in one of the adher- where E1, E2: respectively, the axial elastic moduli of the first and second adherent;
ents, one of the interfaces or in the adhesive), (ii) or failure in one of the adherents e1, e2: respectively, the thickness of the first and second adherent; Gc: the elastic
by compression. The size of specimen should allow obtaining a shear failure mode shear modulus of the adhesive; ec: the thickness of the adhesive layer; n: An effi-
of the assembly, and it is therefore necessary to work with samples having suffi- ciency factor representing the quantity of stress transmitted on the Lc length.
cient compressive ultimate capacities (and then sufficient cross-section). It is also Considering 100 mm-thick elements (to obtain three strips of wood in GLULAM
important to limit cleavage forces at the end of the bonded joint with a constraint element), and an adhesive thickness of 1 mm, the evolution of the anchorage length
at the edges of the sample for example to prevent from peeling failure of the assem- was plotted against the elastic stiffness of the resin and the efficiency factor n
bly. The two adherent materials are laminated Douglas wood GLULAM GL24 with a (Fig. 6). Assuming that 90% of the ultimate capacity (n = 0.9) should be transferred
characteristic bending strength of 24 MPa, and UHPFRC (‘‘Ultra High Performance during the realized push-out test, for common structural adhesives (elastic modu-
Fibre Reinforced Concrete’’ interim recommendations, AFGC, France, French/Eng- lus of about 5000 MPa), it is necessary to have a bonded length of 180 mm (Fig. 5). It
lish) with an average compressive strength of 150 MPa. One notes therefore that was thus decided to adopt this value for the bonded length of the studied samples.
among the two adherents, the wood material is less strong in compression and it It is noteworthy that this value varies with the modulus of the used adhesive. In the
was consequently chosen that within the push out samples, wood would have case of more flexible resins, the bonding length should be greater. However, for
the largest cross-section. The elastic tensile modulus of the wood is about 10 GPa. practical reasons (size of the test sample and prevention from buckling), the same
The elastic modulus of UHPFRC is approximately 60 GPa. bonding length is considered for all tests.

Fig. 5. Dimensions of the push-out samples.

 G. Youssef et al. / Construction and Building Materials 53 (2014) 381–391 385

2.4. Mechanical tests and ageing procedure

Tests were realized using a compressive test machine with a capacity of

1000 kN. An initial load of the sample is done up to 20 kN. Then, a loading ramp
is applied until 30kN (at a rate of 100 N/s). After maintaining this load for 60 s, a
second loading ramp is applied until failure at a rate of 500 N/s. The horizontal dis-
placement of the specimen at the bottom is blocked in order to limit peel stresses
(Fig. 7a).
Samples are instrumented in order to (i) follow some local displacements and
(ii) measure the local stiffness of the assembly and check the distribution of shear
stresses which is not constant along the bonded lap length. The instrumentation is
performed using displacement sensors that measure local slips as defined in [24].
To realize the measure of these slips, rigid metal supports are used and bonded
on the surface of the sample (Fig. 7b). It must be noted that it is a measure of pseu-
do-local stiffness. Four slips are recorded during the test and the results can be com-
pared to numerical calculations results using finite element for example. The
positions of the rigid metal supports are precisely measured for each specimen be-
Fig. 6. Elastic anchorage length according to Volkersen [41] as a function of the fore each test. The used sensors are LVDT type sensors having a range of ±1 mm.
stiffness of the adhesive and the efficiency factor n . Measurements are recorded at a frequency of 10 Hz.
Among the methods of optical measurement, Digital Image Correlation (DIC)
can measure the deformation fields on the surface of a sample. This measurement
2.3. Samples realization technique has already been used for different assemblies or materials (e.g.: for
glued wood joints [55,23], wood fracture mechanisms [56–58] or for reinforced
In general, there are three arrangements of test specimens used for shear tests concrete [59–61]). The measuring device is presented in Fig. 7c. Samples have been
in timber–concrete composite structures: push-out type with concrete in the centre marked with black speckle painting. This surface is observed with a high resolution
and timber on both sides [44–48], push-out type with timber in the centre and con- camera (2048  2048 pixels), and a rather standard optical setup is used to record
crete on both sides [49–52] and pure shear type [53,54]. images of the whole sample (field: 320  320 mm2, pixel size: 150 lm). Important
Carvalho and Carrasco [36] compared results obtained from shear tests con- additional lighting was necessary to ensure good quality of observed images. The
ducted with different specimen arrangements using a mechanical connector analysis of displacement and strain fields by the in-house CMV software [62] pro-
(push-out type with concrete in the centre and timber on both sides and push- vided information about strain fields on the surface of the sample, in particular at
out type with timber in the centre and concrete on both sides) and assessed the the wood/concrete interface. The used CMV software gives local strain values at
influence of this arrangement on test results. He concluded that slip modulus of each centre of a cross of the considered grid where the strain field is calculated.
type-A specimens (Wood–Concrete–Wood) was greater than type-B (Concrete– The local strain values are characterized by a specific color and averaged on a do-
Wood–Concrete) for all connector diameters. Since stiffness of a connection system main surrounding the considered point.
is estimated using slip modulus, adopting the results of type-A specimens could Few investigations have been realized concerning ageing of such assembly [63].
lead to overestimation of composite structure stiffness and a decrease in safety of It was thus decided to store additional samples under ageing conditions closed to
the design. Moreover, type-B specimens better represent the conditions of the com- the one defined in [64] (relative humidity RH superior to 60% and constant temper-
posite beam and are closer to standard tests for steel–concrete composite struc- ature T = 20 °C). For each kind of adhesives, two samples have been tested.
tures. Among the different presented test samples geometries, wood–concrete–
wood had lower strength and higher slip modulus. As in our case, the concrete
has good compressive strength; the weak element in compression is wood. This 3. Results and discussion
is the reason why we chose in this study to adopt push-out samples with concrete
in the centre and timber on both sides. 3.1. Introduction
The push-out samples are made of two elements of laminated timber with a
section of 100 mm  100 mm, bonded to an UHPFRC element having a cross section
of 100 mm  100 mm. The bonded lap length is 180 mm and 20 mm were left on
The push-out test allows focusing on three results:
both sides of the sample (Fig. 5). In reason of the importance of the symmetry of
the specimen on the result, the samples realization should be done with great care. – The failure mode that can occur in one of the adherents, in the
After 28 days of curing of UHPFRC elements (endogenous treatment), concrete sur- resin or at one of the interfaces. One can also have a mixed fail-
faces are sanded with a diamond disk. The surfaces of the laminated timber ele-
ure mode. Note that in the case of bonded assembly, an interfa-
ments are prepared using sandpaper (the wood fibres being aligned with the axis
of loading). The resin is then applied after dusting and degreasing surfaces as rec- cial failure is called «adhesive» and highlights the inadequacy of
ommended by the technical data sheet (double bonding in most cases, respect of adhesion forces. This may be due to an insufficient or inadequate
pot life). A specific wooden frame was realized to ensure correct geometry of the surface preparation. In the particular case of push-out tests with
whole assembly. A clamp is then put in place to maintain the item and to transport glued laminated timber adherents, the orientation of the annual
it more easily. All is left in place during the polymerization in laboratory conditions
rings has an influence on the observed failure mode (Fig. 8a). To
(which is conformed to the polymerization conditions recommended in the data
sheet). Then, samples bases are polished in order to get a horizontal and symmet- increase the capacity of the realized assembly, annual rings
rical support. For each kind of adhesives, three samples have been tested.

Fig. 7. (a) Sample position under the compressive tool with wedges to block horizontal displacement. (b) Push-out instrumented sample and (c) Optical set-up for DIC
measurements during mechanical tests.
386 G. Youssef et al. / Construction and Building Materials 53 (2014) 381–391

Fig. 8. Failure mode within wood ring (left) (a) and Failure mode disturbance due to a knot (right) (b).

Table 2
Main results of push-out tests for the four studied adhesives without ageing.

Adhesive Epoxy 1 Sikadur 30 Epoxy 2 Sikadur 330 Modified Acrylic Polyurethane

Failure mode 30% concrete, 70% timber 25% concrete, 75% timber 50% timber, 50% interface 50% interface concrete/adhesive,
adhesive/timber 50% interface timber/adhesive
Average ultimate capacity (kN) 242 237 177 41.1
Average ultimate shear stress (MPa) 6.7 6.6 4.9 1.1
Dispersion of the ultimate capacity (%) 43 20 13.8 42

Table 3
Main results of push-out tests for the four studied adhesives with ageing.

Sample Ultimate capacity Failure mode Average ultimate Dispersion of the Average shear
(kN) capacity (kN) ultimate capacity (%) stress (MPa)
Epoxy 1: Sikadur 30 E1-1 181.9 25% Concrete, 75% timber 167.8 24.8 4.7
E1-2 114 (asymmetric failure) 50% Concrete, 50% timber
Epoxy 2: Sikadur 330 E2-1 183.4 95% Concrete, 5% timber
E2-2 191.7 0% Concrete, 100% timber
Modified Acrylic MA-1 115.7 (asymmetric failure) 10% timber, 90% interface adhesive/timber 122.4 7.7 3.4
MA-2 129 50% Timber, 50% interface adhesive/timber
Polyu-rethane PU-1 0 100% Interface timber/adhesive 14.8 – 0.41
(failure during the initial load)

should be positioned as possible perpendicular to the bonded – Cohesive failure in the concrete very close to the interface.
surface. Similarly, presence of knots can also affect the observed – Cohesive failure in the wood (for these investigations, the
failure mode (Fig. 8b). annual rings were oriented as much as possible perpendicularly
– The ultimate capacity of the assembly that is essential for the to the bonded surface).
design of a composite structure. The ultimate capacity of the – Failure at the interface between the adhesive and the wood.
assembly may be mostly related to the ultimate capacity of – Failure at the interface between the concrete and the adhesive.
the ‘‘weakest’’ element in which the failure occurred. All the
materials, however, influence stress distributions, and it is For each sample, two failure modes may be observed. A per-
therefore necessary to take them into account during the results centage is given for each observation based on the photos realized
analysis. after testing. Though the method may not be precise, it gives good
– The stiffness (local or global stiffness of assembly) that is linked information on the weak link of the assembly.
to the used adhesive. Measurement of local stiffnesses can allow It is difficult in the last two cases to differentiate a cohesive fail-
studying stress profiles along the glued joint and this measure- ure mode that would be in the adhesive material from an adhesive
ment also at the same height on each of the two bonded laps failure mode that would indicate insufficient adhesion forces and
using displacement sensors or DIC can help to check the sym- consequently inadequate surface preparation or unadapted adhe-
metry of the test too. It is then possible using simple analytical sive. In the case of the two stiffest resins (epoxy resins) which
methods [43] to determine the global stiffness of the specimen. are also the strongest ones, weak elements are clearly the two
These methods suppose that all the materials behave elastically adherents. In the case of the modified acrylic adhesive, it seems
and neglects shear effects within the adherents and bending that the weak element of the assembly is the wood material or
moment effect due to load eccentricity. the interface between the adhesive and the wood. In the case of
polyurethane, the weak element is clearly the adhesive (the mate-
3.2. Failure mode and ultimate capacity rial itself, or the adhesion forces). After one year of ageing, we find
that the fracture surface is not affected by ageing (Fig. 9).
The results of the push-out tests for the four studied adhesives Concerning the results in terms of the ultimate capacity, several
are given in Table 2 for sample without ageing and in Table 3 for conclusions can be drawn. It is important to note that the calcu-
samples with ageing. Concerning the failure mode, four different lated dispersion consisting in the ratio of the standard deviation
cases were observed: with the average ultimate capacity was only determined using
 G. Youssef et al. / Construction and Building Materials 53 (2014) 381–391 387

Fig. 9. Photos of fracture surfaces of samples using different adhesives: (A) before ageing and (B) after ageing.

two values. This is thus clearly unsufficient from a statistical point

of view but it can help in the qualitative examination of the ob-
tained results:

– The most flexible adhesive material (polyurethane) does not pro-

vide sufficient capacity for the bonded assembly as no remaining
capacity has been found after one year ageing. It is therefore not
possible to use it for the realization of composite beams. Actually,
at room temperature, this adhesive is under its rubbery state (its
glass transition temperature being below 20 °C). This is not a
structural adhesive at normal service temperatures.
– Fig. 10 justifies the mismatch between data of the adhesive and
data of the assembly (due to Volkersen’s theory [41], and failure
modes).
– There is no linear correlation of the ultimate capacity of the
bonded assembly with the stiffness of the used adhesive; the
dispersion of the test in terms of maximum capacity seems to Fig. 10. Average shear stress for the four studied adhesives without and with
increase with the stiffness of the used adhesive. ageing.
388 G. Youssef et al. / Construction and Building Materials 53 (2014) 381–391

0.8 the most efficient resin as shown in particular for three cases (sam-
Displacement A ples E2, PU and MA):
Measured displacement, in mm

0.7 Displacement B
Displacement C – Resin epoxy, Sikadur 330 (sample E2-1): in the case of the epoxy
0.6
Displacement D
resin for the tested sample (Fig. 11b and c), more than half of
0.5
the left block of wood is solicited during the test. It undergoes
0.4 compression, with negative deformations (eyy = 0.3%) and
shear (exy = 0.3%). This phenomenon is explained by a high stiff-
0.3 ness of the adhesive and good properties of adhesion to both
0.2 wood and concrete materials. Strains in the right block are prac-
tically nil because the test is asymmetric and rupture, cohesive
0.1 here, took place as a side in the woods. There was no significant
0 shear concentrated at the both interfaces. Furthermore, contrac-
0 50 100 150 200 250 300 350 tion on the top and extension at the bottom are observed for the
Applied force, in kN strain exx because of peel stresses. In this case the horizontal
displacement of the specimen at the bottom is not completely
Fig. 11. Displacement at the bond joint of the assembly for the sample E2 (Sikadur blocked (see Fig. 7a).
330).
– Resin modified acrylic, Sikafast (Sample MA-1): In the case of
modified acrylic, axial strain can clearly be observed in the
wood, but in a less important order of magnitude than in the
– For the four adhesives, it should be noted, however, a drop of
case of epoxy resin (eyy is about 0.1%) (Fig. 12b) due to
adhesion properties of about 30% due to ageing compared to
the good rigidity of the adhesive. In contrast, the adhesion of
the case without ageing.
the glue on the wood and concrete materials is insufficient
– In the case of cohesive failure, it is expected to achieve the max-
and we observe shear strains concentrated at the interfaces
imum possible capacity (we cannot optimize capacity when we
(Fig. 12c), more important at the left interface than the right.
work on surface preparation, for example).
This test is not symmetric, and the failure occurred on the left,
at the interface wood / glue mainly. As for the sample E2, peel
While in the case of a failure in the interface, the capacity tends
stresses are observed.
to decrease.
– Resin polyurethane, Sikaforce (Sample PU-1): For the case of the
polyurethane adhesive, no strain can be observed on the timber
3.3. Displacement measurements and deformation maps adherent; only shear strains are visible at the two interfaces
(Fig. 13b and c).
The displacement data (A, B, C and D) are presented in Fig. 11.
These measurements allow to check the symmetry of the realized
testing and the balanced factor (ratio between slip at the points C 3.4. Discussion on the choice of the resin and the effect of ageing
and A and at the points D and B). The experimentally measured
balanced factor is found to be close to the theoretical factor given To detail the study of these phenomena and explain the men-
by Volkersen [41]. It can be noticed that the possible asymmetric is tioned observations, it is interesting to examine shear stress pro-
confirmed using DIC measurement. Despite this possible asymmet- files along the bonded assembly. These can be assessed in a first
ric, behaviour of the bonded joint may be characterized to choose approach using the formula given by Volkersen [41] (Fig. 14).

Fig. 12. Local strains for the sample «E2» just before the failure using DIC measurements (a) exx, (b) eyy and (c) exy.
 G. Youssef et al. / Construction and Building Materials 53 (2014) 381–391 389

Fig. 13. Local strains for the sample «MA4» just before the failure using DIC measurements (a) exx, (b) eyy and (c) exy.

Fig. 14. Local strains for the sample «PU4» just before the failure using DIC measurements (a) exx, (b) eyy and (c) exy.

The adhesive’s stiffness tends to increase stress concentrations. failure mode, this adhesion drop may be partly explained by
The increase in these stress concentrations tends to increase the the decrease of the mechanical properties of wood which is a
dispersion of the test since the results become more sensitive to function of its moisture content.
material heterogeneities and singularities (see Fig. 15). – Acrylic resin always gives good results in terms of shear stress
In addition, it helps in understanding that the notion of average of 3.4 MPa, and the fracture surfaces are similar to the case
shear stress (capacity divided by bonded surface) is not consistent before ageing (Fig. 9). As for epoxy resin, we note a drop resis-
with the observed physical phenomena in the case of stiff resins. In tance of about 30% which may be mainly explained by a
the case of the two softer adhesives on the contrary, the calculated decrease in adhesion properties of the adhesive given the failure
average shear stress may be meaningful and may help in predicting modes essentially adhesives.
the ultimate bonded joint capacity when the adhesive is the weak – The polyurethane resin has insufficient adhesion properties
link of the assembly. with a failure during the initial load.
The main results of this study are:
4. Conclusion and Prospects
– The epoxy resin presents the most interesting adhesion proper-
ties after ageing in terms of ultimate capacity with an average of In this study, a test protocol using the push-out geometry was
4.7 and cohesive fractures (Fig. 9A and B). Given the cohesive developed in order to study and characterize the assembly by
390 G. Youssef et al. / Construction and Building Materials 53 (2014) 381–391

Fig. 15. Push-out sample and shear stress profiles along the bonded joint according to the elastic theory of Volkersen [37] for four adhesives having different elastic moduli.

adhesive bonding between wood and concrete with and without technique to investigate new series of tests on the durability of
ageing. It allowed to get interested in the sustainability of the these assemblies in the longer term.
assembly subjected to variable relative humidity for four different
types of resins: a soft resin (polyurethane), an elastoplastic resin
(modified acrylic) and two elastic and stiff resins (epoxy). Acknowledgements
In the first part, a focus was made on the problems linked to on
site implementation (type of glue, pot life, surface preparation, The authors gratefully acknowledge the SIKA Company for
etc.) and the ultimate capacity of the assembly and some local stiff- gracefully providing resins and for the discussions related to the
nesses were commented. The second part of the paper presents the mechanical performances of used resins. They also would like to
realized investigations using this type of test on the influence of thank Fargeot Co. for providing the material (timber) for these test
the stiffness of the adhesive on the realized composite structures. series.
It is shown that the designer of such structures has a wide range
of choices in terms of adhesive’s stiffness that does not seem to af- References
fect much the stiffness of the composite structure. Then, the pre-
sented push-out tests allow to illustrate the use of the proposed [1] Yttrup PJ. Concrete enhanced timber. Timber building in Australia; 2005.
test protocol with four resins having different stiffnesses. Note that [2] Brunner M, Schnüriger M. Holz-Beton-Elemente mit Klebeverbund.
Forschungsbericht Nr. 2637-SB-01, Hochschule für Architektur, Bau und Holz
the use of a less rigid resin allows decreasing stress concentrations HSB, Burgdorf, Biel; 2005.
and thus reducing the sensitivities of the test to local heterogene- [3] Haiman M, Rak M, Krolo J, Herceg J, Calogovic V. EPS concrete composite
ities and singularities which may be predicted the theory of Vol- structures – lab testing and FEA modelling. In: Proceedings of the international
conference on experimental mechanics, Milan: Italy; 2004. p.733–734.
kersen [41]. [4] Negrao JH, Oliveira FM, Oliveira CL. Investigation on timber–concrete glued
The presented tests are the first investigations of a wider study composites. In: 9th World conference on timber engineering, Portland, OR,
aiming at studying the consequences of hygroscopic cycles on the USA; 2006.
[5] Pincus G. Behaviour of wood–concrete composite beams. J Struct Div, Proc Am
resistance of the assembly. It confirms that we obtained four differ- Soc Civ Eng S 1970:S2009–19.
ent cases of failure mode for samples without and with ageing: [6] Pincus G. Epoxy bonded wood–concrete and corrugated steel–concrete
Cohesive failure in the concrete very close to the interface, cohe- composite. In: International conference on materials of construction for
developing countries, Bangkok, Thailand; 1978. p. S1003–15.
sive failure in the wood, failure at the interface between the adhe-
[7] Schafers M, Seim W. Investigation on bonding between timber and ultra-high
sive and the wood, failure at the interface between the concrete performance concrete (UHPC). Constr Build Mater 2011;25:3078–88.
and the adhesive. After one year of ageing, we find that the fracture [8] Bouhaya L, Le Roy R, Feraille-Fresnet A. Simplified environmental study on
surface is not affected by ageing. The epoxy resin presents the most innovative bridge structure. Environ Sci Technol 2009;43(6):2066–71.
[9] Bathon L, Graf M. A continious wood–concrete–composite system. In:
interesting adhesion properties after ageing in terms of ultimate Proceedings on 6th world conference of timber engineering, Whistler
capacity with an average of 4.7 and cohesive fractures. (Canada), 2000.
It is therefore not possible to use the polyurethane for the real- [10] Blass HJ, Schlager M. Trag- und Verformungsverhalten von Holz-Beton-
Verbundkonstruktionen. Bauen mit Holz, Hefte 5 & 6; 1996.
ization of composite beams. This material does not provide suffi- [11] Cecotti A. Composite concrete-timber structure. Prog Struct Eng Mater
cient capacity for the glued and especially after one year aging. 2002;4(3):264–75.
There is no linear correlation of the ultimate capacity of the [12] Mascia NT, Soriano J. Benefits of timber-concrete composite action in rural
bridges. Mater Struct 2004;37(266):122–8.
bonded assembly with the stiffness of the used adhesive; the dis- [13] Natterer J, Hoeft M. Zum Tragverhalten von Holz-Beton-
persion of the test in terms of maximum capacity seems to increase Verbundkonstruktionen, Forschungsbericht CERS Nr. 1345, IBOIS, EPFL; 1987.
with the stiffness of the used adhesive. A reduction about 30% is [14] Pham HS. Optimisation et comportement en fatigue de la connexion bois-BFUP
pour de nouveaux ponts mixtes. In: Ph.D thesis, Ecole Nationale des Ponts et
obtained for adhesion properties after ageing. Chaussées; 2007.
Theoretical approaches are also being developed to assess the [15] Le Roy R, Pham HS, Foret G. New wood composite bridges. Eur J Environ Civil
level of stress suffered by the assembly under these conditions. Eng 2009;13(9):1125–39.
[16] Hamelin P. Collage dalle sur acier, Résultats du projet national MIKTI, Journée
This should help to further inform the designer on the importance
technique de l’AFGC: construction métallique et mixte: nouveaux matériaux,
of the choice of the adhesive. A method for measuring strains by nouvelles conceptions; 2010.
image correlation was also tested. The made attempts for recording [17] Jurkiewiez B, Morvan M, Si Larbi A, Ferrier E. Bonding connection for steel–
image and applying image correlation proved to be quite satisfac- concrete flexural member. CCC2008—challenges for civil construction. Porto;
1985. p. 8.
tory. Additional work must be done to compare the obtained strain [18] Bouazaoui L, Perrenot G, Delmas Y, Li A. Experimental study of bonded steel
fields with finite element modelling. This could be a promising concrete composite structures. J Constr Steel Res 2007;63:1268–78.
 G. Youssef et al. / Construction and Building Materials 53 (2014) 381–391 391

[19] Bouazaoui L, Jurkiewiez B, Delmas Y, Li A. Static behaviour of a full-scale steel– [43] Chataigner S. Conception et dimensionnement d’un ancrage de hauban plat en
concrete beam with epoxy bonding connection. Eng Struct 2008;30:1981–90. matériaux composites. In: Ph.D thesis Ecole des Ponts; 2008.
[20] Si Larbi A, Ferrier E, Jurkiewiez B, Hamelin P. Static behaviour of a steel– [44] Weaver CA. Behavior of FRP-reinforced glulam–concrete composite bridge
concrete beam connected by bonding. Eng Struct 2007;29:1034–42. girders. In: Ph.D thesis, University of Maine; 1999.
[21] SiLarbi A, Ferrier E, Hamelin P. Concrete to steel lap joint failure criteria under [45] Dias A. Mechanical behavior of timber–concrete joints. In: Ph.D thesis; ISBN
combined shear and peeling stress. J Constr Steel Res 2009;65:386–94. 90-9019214-X; 2005.
[22] Jurkiewiez B, Meaud C, Michel L. Non linear behaviour of steel–concrete epoxy [46] Faust T, Selle R. Properties of composite joints of timber. LWAC composite
bonded composite beams. J Constr Steel Res 2011;67:389–97. structures. In: Lacer no. 4 –Leipzig annual civil engineering report. University
[23] Loulou L, Durabilité de l’assemblage mixte bois béton collé sous chargement of Leipzig, Federal Republic of Germany; 1999. p. 231–40.
hydrique. In: Ph.D thesis, Université Paris Est; 2013. [47] Benítez MF. Development and testing of timber/concrete shear connectors. In:
[24] Chataigner S, Caron JF, Duong VA, Diaz A. Experimental and numerical 6th World conference on timber engineering WCTE, paper no. 832, Whistler
investigation of shear strain along an elasto-plastic bonded lap joint. Constr Resort, British Columbia, Canada, 31 July–3 August; 2000.
Build Mater 2011;25(2):432–41. [48] Mascia NT, Soriano J. Benefits of timber–concrete composite action in rural
[25] Blonski A, Etude de la rigidité de systèmes de connexions pour un pont type bridges. Mater Struct 2004;37:122–8.
bois-béton. In: Master report, Ecole des Ponts ParisTech; 2009. [49] Branco J et al. Experimental analysis of laterally loaded nailed timber-
[26] Piazza M, Ballerini M. Experimental and numerical results on timber-concrete toconcrete connections. Constr Build Mater 2009;23:400–10.
composite floors with different connection systems. In: ‘6th WCTE’, [50] Mungwa MS et al. Experimental study of a composite wood–concrete beam
proceedings of the world conference on timber engineering, Whistler Resort, with the INSA-Hilti new flexible shear connector. Constr Build Mater
British Columbia, Canada, CD, 2000. 1999;13:371–82.
[27] Steinberg E, Selle R, Faust T. Connectors for timberlightweight concrete [51] Holschemacher K, Klotz S, Weibe D. Application of steel fibre reinforced
composite structures. J Struct Eng 2003;129(11):1538–45. concrete for timber–concrete composite constructions. In: Lacer no. 7 – Leipzig
[28] Chataigner S. Modélisation aux éléments finis de poutres bois béton annual civil engineering report. University of Leipzig, Federal Republic of
assemblées par collage avec différents adhésifs et sous flexion 4 Germany; 2002. p. 161–70.
points. Département Laboratoire d’Autun, Rapport de Recherche; 2010. [52] Branco J. Behaviour of dowel-type fasteners in timber–concrete connections.
[29] Bocquet JF, Barthram C, Girardon S, Tavakoli I, Le Roy R. Prise en compte des In: Master thesis, Engineering School, University of Minho, Portugal; 2003. [in
caractéristiques d’interface: connexions bois béton, Journées « Conception et Portuguese]
gestion des ouvrages en bois » de l’AFGC, Epinal; 2010. [53] Heiduschke A, Kasal B. Composite cross sections with high performance fiber
[30] Loulou L, Caré S, Le Roy R, Bornert M. Damage of wood-concrete composite reinforced concrete and timber. Forest Prod J 2003;53(10):74–8.
subjected to variable hygrometric conditions. In: 14th International [54] Clouston P, Civjan S, Bathon L. Experimental behavior of a continuous metal
conference on experimental mechanics, ICEM14, Poitiers, France, July 4–9; connector for a wood–concrete composite system. Forest Prod J
2010. http://dx.doi.org/10.1051/epjconf/20100628002. 2004;54(6):76–84.
[31] Benkhdija Y. Etude des contraintes induites par un gradient thermique ou [55] Pommier R, Muszynski L. Evaluation du comportement d’un assemblage bois
hydrique sur un assemblage collé bois/béton. In: Master report, Université collé par méthodes de caractérisation optiques.25eme rencontres de l’AUGC,
Blaise Pascal, 2010. mai 23–25; 2007. [Bordeaux].
[32] EN 26891. Timber structures – joints made with mechanical fasteners – [56] Bruck HA, McNeill SR, Sutton MA, Peters WH. Digital image correlation using
general principles for the determination of strength and deformation Newton–Raphson method of partial differential correction. Exp Mech
characteristics 1991; ISO 6891–1983 (E). 1989;29:261–7.
[33] American Society for Testing and Materials (ASTM). Standard test methods for [57] Pop O, Meite M, Dubois F, Absi J. Identification algorithm for fracture
bolted connections in wood and wood-base products. D 5652–95. Annual Book parameters by combining DIC and FEM approaches. Int J Fract
of ASTM Standards; 2000. 2011;170:101–14 [3].
[34] EN 1994–1-1. Eurocode 4: design of composite steel and concrete structures: [58] Meite M, Pop O, Dubois F, Absi J. Characterization of mixed-mode fracture
general rules and rules for buildings; 2004. based on a complementary analysis by means of full-field optical and finite
[35] Johnson RP, Anderson D. Designeŕs handbook to Eurocode 4 – Part 1. 1: design element approaches. Int J Fract 2013;180:41–52.
of composite steel and concrete structures. London; 1993. [59] Caré S, Nguyen QT, Beddiar K, Berthaud Y. Times to cracking in reinforced
[36] Pires Carvalho E, Mantilla Carrasco E. Influence of test specimen on mortar beams subjected to accelerated corrosion tests. Mater Struct
experimental characterization of timber–concrete composite joints. Constr 2010;43(1–2):107–24.
Build Mater 2010;24:1313–22. [60] Care S, Nguyen QT, L’Hostis V, Berthaud Y. Mechanical properties of the rust
[37] Sutton MA, Wolters WJ, Peters WH, Ranson WF, McNeil SR. Determination of layer induced by impressed current method in reinforced mortar. Cem Concr
displacements using an improved digital correlation method. Image Vis Res 2008;38(8-9):1079–91.
Comput 1983;1:133–9. [61] Nguyen QT, Caré S, Berthaud Y, Millard A, Ragueneau F. Experimental and
[38] Sutton MA, Cheng MQ, Peters WH, Chao YJ, McNeill SR. Application of an numerical behaviour of reinforced mortar plates subjected to acceleretad
optimized digital correlation method to planar deformation analysis. Image corrosion. Int J Num Anal Methods Geomech 2011;35(10):1141–59.
Vis Comput 1986;4:143–51. [62] Bornert M, Vales F, Gharbi H, Nguyen Minh D. Multiscale full-field strain
[39] Guyot A. Etude de l’essai Push-Out sur collage bois-BFUP. In: Internship report, measurements for micromechanical investigations of the hydromechanical
IUT Chalon sur Saone; 2009. behaviour of clayey rocks. Strain 2010;46:33–46.
[40] Gouny F, Fouchal O, Pop O, Maillard P, Rossignol S. Mechanical behaviour of an [63] Ceccoti A, Fragiacomo M, Giordano S. Long-term and collapse tests on a
assembly of wood-geopolymer-earth bricks. Constr Build Mater timber–concrete composite beam with glued-in connection. Mater Struct
2013;38:110–8. 2006;40:15–25.
[41] Volkersen O. Die Nietkraftverteilung in zugbeanspruchten mit konstanten [64] Benzarti K, Chataigner S, Quiertant M, Marty C, Aubagnac C. Accelerated ageing
laschenquerschritten. Luftfahrtforschung 1938;15:41–7. behavior of the adhesive bond between concrete specimens and CFRP
[42] Hart-Smith LJ. Adhesive-bonded double-lap joints. In: NASA Technical Report overlays. Constr Build Mater 2011;25:523–38.
California 90846; 1973.