You are on page 1of 15

Composite Structures 192 (2018) 52–66

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Design method and verification of steel plate anchorages for FRP-to-concrete T


bonded interfaces

Hugo C. Biscaiaa, , Carlos Chastreb
a
FSE, UNIDEMI, Faculdade de Ciências e Tecnologia/Universidade Nova de Lisboa, 2829-516 Caparica, Portugal
b
CERIS, ICIST, Faculdade de Ciências e Tecnologia/Universidade Nova de Lisboa, 2829-516 Caparica, Portugal

A R T I C L E I N F O A B S T R A C T

Keywords: Concrete structures Externally Bonded Reinforced (EBR) with Fibre Reinforced Polymers (FRP) have been stu-
Design methodology died and used since the end of the last century. However, several issues need to be better studied in order to
FRP improve performance. The influence of size of anchorage plates used on Reinforced Concrete (RC) structures
Concrete strengthened with EBR FRP composites, the external compressive stress to be applied on the anchorage plate and
Mechanical anchorage
the numerical simulation of this region are some of the topics that need to be more carefully studied in order to
Bond
clarify the performance of the FRP-to-concrete interface within the anchorage plate region. This study proposes a
Interfaces
design methodology to estimate the amount of external compressive stress necessary to be applied on the an-
chorage plate of EBR systems with FRP composites, in order to avoid premature debonding. The external
compressive stress imposed on the FRP composite is intended to simulate the effect produced by a mechanical
anchorage system tightened to the EBR system. The results from the design proposal, when compared with the
numerical ones, were efficient enough on the prediction of the bond strength improvement of FRP-to-concrete
interfaces.

1. Introduction have been used, such as FRP spike anchors [23–26], chemical or me-
tallic screw fastening with or without washers [27,28], metallic cy-
The use of Fibre Reinforced Polymers (FRP) in the rehabilitation of linder or steel plates installed at the FRP end [29], increasing the width
concrete structures has become increasingly popular since the end of of the composite at its ends [29–31], or using the innovative gradient
the last century. Due to the advantages in comparison to other materials anchorage proposed by Michels et al. [32] in which the use of con-
such as steel, FRP composites have attracted the attention of several ventional steel mechanical fasteners are not needed. The installation of
researchers for their use mostly on the strengthening of reinforced steel plates as mechanical anchorages of EBR FRP composites is, per-
concrete (RC) structures by developing studies based on experimental haps, the most obvious method for the prevention of the premature
work (e.g. [1–6]), analytical theories (e.g. [7,8]) or based on numerical debonding phenomenon. However, the design of those mechanical
simulations (e.g. [9–12]) using the Finite Element Method (FEM). anchorage systems needs to be better studied, as the compression
However, researchers have faced several challenging issues in the stresses are applied often without knowing their real influence on the
search to take full advantage of them. One of the most challenging is- strengthening system.
sues is the understanding of the debonding phenomenon of FRP com- The addition of a steel plate to mechanically anchor the FRP com-
posites from the substrate when its maximum strain is quite far away posite consists of tightening the steel plate against the FRP composite
from the rupture value. This is known as premature debonding and with fasteners (chemical or metallic). Thus, it is expected that the FRP
several researchers have identified it in the literature, mainly for con- composite may not easily debond when a field of compression stresses
crete structures Externally Bonded Reinforced (EBR) with FRP compo- perpendicular to the bonded area is induced to the FRP-to-concrete
sites, independently of the model used to describe the interfacial law interface. On RC beams, this method is usually applied at the end of the
corresponds to a linear function (e.g. [13–15]), a power (e.g. [16–18]) FRP composite. However, this methodology raises another important
or an exponential function (e.g. [19,20]). question: how much stress must be applied to the steel plate anchorage
Therefore, alternative bonding methods have been proposed in order to eliminate the premature debonding of the FRP composite?
[21,22] or, in addition to the EBR method, other devices or strategies This is a question that, to the best of the authors’ knowledge, has not yet


Corresponding author.
E-mail address: hb@fct.unl.pt (H.C. Biscaia).

https://doi.org/10.1016/j.compstruct.2018.02.062
Received 16 December 2017; Received in revised form 12 February 2018; Accepted 20 February 2018
Available online 23 February 2018
0263-8223/ © 2018 Elsevier Ltd. All rights reserved.
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

Fig. 1. Relation between the Mohr-Coulomb failure criterion and nonlinear bond-slip models.

been properly answered and it is clear that when this method is used, 2. Theoretical formulation
the researchers have often opted to omit this information, e.g. [33–37].
The main focus of this study is on the flexural strengthening of RC 2.1. Fundamental assumptions
beams with FRP composites. Within this context, the present work is
focused on giving an appropriate answer to that question by proposing The proposed design of the mechanical anchorage is based on the
a method to estimate the external compression stress that needs to be following assumptions:
applied to the FRP composite in order to avoid premature debonding
and, thus, to reach its tensile rupture. The proposal is based on a bond- (i) the FRP-to-concrete interface is subjected only to shear deforma-
slip relationship that dependents on the external stress level applied to tions leading to a rupture mode that is consistent with the Mode II
the FRP-to-concrete interface. The authors have already proposed a interfacial fracture;
bond-slip model under these external stress conditions, in which the (ii) across the width of the FRP composite, the distribution of the bond
failure debonding process was discussed in depth [38]. The Mohr- stresses is assumed to be uniform;
Coulomb failure criterion was used to explain the increase of the in- (iii) the FRP-to-concrete interface has a behaviour that can be ap-
terface strength due to external compressive stresses. proximated to the Mohr-Coulomb failure criterion. This criterion
Furthermore, from other tests carried out by the authors [39] and in is only based on stresses which the rupture envelope is defined
Silva's work [40], it was found that when the steel plate is tightened according to:
against the FRP composite an increase of the strains in Carbon (C) FRP
τmax = c + σ·tan(ϕ) (1)
laminates was noticed, which decreased the rupture load that could be
sustained by the conjunct FRP-adhesive-substrate. Another issue found where c is the cohesion of the interface; σ is the external stress
in Silva’s work [40] was that when the steel plates were tightened too perpendicular to the bonded area; and ϕ is the internal friction
much against the composite, the rupture of some fibres due to shear was angle of the interface;
observed. Of course, without knowing the proper answer to the ques- (iv) a rigid steel plate and a uniform pressure distribution along the
tion previously raised, other researchers may have been facing these anchorage are assumed;
same problems and this proves the urgent necessity to understand, first, (v) from the previous assumption, an exponential function is used to
this topic and, secondly, to propose models that can predict those ef- represent the local bond-slip relationship of the FRP-to-concrete
fects. interface according to [38]:
Therefore, the proposed design method begins with the definition of
τ (s ) = A2 ·B·(e−B·s−e−2B·s )·Ef ·t f + τres·(1−e−B·s ) (2)
the stress beyond which the composite will break due to shear and,
afterwards, the maximum external compressive stress that needs to be where Ef and tf are, respectively, the Young’s modulus and the
applied to the steel plate to avoid the premature debonding of the FRP thickness of the FRP composite; A and B are constants that are
composite is calculated and, thus, to achieve its tensile rupture. To defined from the experimental strain-slip curve and where a
clearly show how the proposed design method can be used, practical lengthy bond length is considered; τres is the residual shear stress
examples are presented and solved. Comparisons with several numer- developed within the interface due to the influence of the external
ical results obtained from a commercial Finite Element (FE) code are compressive stresses;
also reported. Different double-shear tests and 4-point bending tests of (vi) the FRP composite has a Winkler’s beam, where the elastic foun-
Reinforced Concrete (RC) T-beams flexurally-strengthened with Glass dation is the bonding agent, usually an epoxy resin with a linear
(G) FRP composites were modelled. The results obtained from the constitutive behaviour;
proposed design method showed a good approximation to the numer- (vii) the elastic foundation has a bidirectional behaviour, i.e. it has the
ical results, with a relative error lower than 10%. same behaviour under external tension or compression stress.

2.2. Influence of the external stresses on the resulting bond-slip law

The effectiveness of the external compression stresses applied to the

53
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

To assume a zero friction angle between the moving material surfaces,


when lateral compressive stresses exist, implies neglecting the con-
tribution of compressive stresses and the load transmitted to the FRP-to-
concrete is underestimated [38]. Dry friction can also be defined by Eq.
(1) assuming that c = 0 MPa.
Unlike the cases with external compression stresses, the bond
stresses developed in the interface with external tensile stresses, will
tend to zero with the increase of the slips within the interface. The
newer maximum bond stress is reduced accordingly to the Mohr-
Coulomb failure envelope and an exponential bond-slip function that
might represent these cases is:

τ (s ) = 4·(e−B·s−e−2B·s )·[τmax,0−σt ·tan(ϕ)] (4)

where τmax,0 is the maximum bond stress in an interface free of any


external stresses; and σt is the tensile external stress applied to the in-
terface that, in the case of concrete substrates, could be limited by the
mean tensile strength of the concrete (fctm).
From all the considerations made so far, it is also important to
highlight here that the local bond-slip behaviour of FRP-to-concrete
interfaces free of any external stresses should be known or, at least, the
values of the maximum bond stress (τmax,0), the corresponding slip
(smax,0) and the Mode II fracture energy (GF,0) should be known and
quantified.
In terms of the Mohr-Coulomb rupture criterion, the values of co-
hesion (c) and internal friction angle (ϕ) of the interface should be
known as well. Notwithstanding, cohesion and maximum bond stress
can be considered the same as shown in Fig. 1. The definition of the
internal friction angle is quite important because increasing it will re-
sult in a more efficient anchorage but ignoring it will mean that the
external stresses won’t affect the interfacial behaviour of the bonded
interface [38]. Among other aspects that are beyond the scope of this
study, the internal friction angle (and all other parameters) may vary,
essentially due to the surface treatment, type of substrate (e.g. concrete,
timber, steel, etc.), type or thickness of adhesives, FRP composites,
humidity of the substrate at time of the application of the FRP com-
posite and elevated temperatures, e.g. [15,38,41–44]. Of course, dur-
ability issues also affect the local bond-slip relationship and for a rig-
Fig. 2. Winkler foundation model.
orous design of bonded joints they should be considered as well.

FRP composite in order to provide an efficient mechanical anchorage 2.3. Definitions of the FRP Winkler’s beam
can be determined through a bond-slip relationship that assumes such
influence. For that, the proposal made by the authors [38] and pre- As will be shown in subsequent sections, to carry out some needed
viously defined in Eq. (2) is used. Fig. 1 shows the influence of the verifications and make the design of the mechanical anchorage, the
external stresses on the nonlinear bond-slip relationship. Here, it is Winkler’s beam is assumed where the elastic foundation is the bonding
assumed that when the rupture surface is reached, the interface begins agent and the FRP composite the “beam”. In such case, the equilibrium
to debond, reaching a residual line which corresponds to dry friction as equation of the FRP beam on an elastic foundation (adhesive) is:
shown by the dotted line in Fig. 1. d 4v
From the point of view of the external compression stresses, the Ef If + Kv = q
dx 4 (5)
definition of the bond-slip relationship will be the result of considering
the superposition of two functions corresponding to two different stress where q is the load applied to the FRP composite (see Fig. 2); If is the
fields [38]. One corresponds to a free external stress situation where moment of inertia of the FRP beam; and K is defined as:
perpendicular stresses are not considered or are at least sufficiently K = bf K r (6)
small to be ignored and the other is characterized by the presence of
stresses (perpendicular to the bonded area), induced by external pres- The complete solution of Eq. (6) is mathematically well-known and
sure, which usual bond-slip models fail to adequately represent. further developments can be found in Appendix A. By assuming the
Ordinary bond-slip models do not consider the component material Winkler's beam theory, the FRP beam is on an elastic foundation
friction that is accompanied by shear stresses at the interface. Indeed, (bonding agent plus the substrate) that when externally pressured will
bond stresses cannot tend to zero for large slip when maximum bond deform. The values of the maximum shear and maximum bending
stress should approach to a constant value as represented in Fig. 1, moment installed in this bonded system are important, as it will be seen
referred to as residual stress and defined as follows: later, for establishing two distinct rupture criteria: one, that limits the
external pressure to prevent the FRP shear rupture and, the other one
τres = σ·tan(ϕ). (3)
that takes into account the increase of the strains in the FRP composite
From the equilibrium of the FRP-to-concrete interface, the con- during the installation of the steel mechanical anchorage onto the FRP/
tribution of these shear stresses due to friction is the main factor re- substrate bonded system. Therefore, at x = x = 0 (see Fig. 2), the
sponsible for the increase in the load transmitted to the FRP composite. maximum shear in the FRP Winkler’s beam is calculated according to:

54
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

q
Vmax =
4β (7)
where β is defined as:

K
β= 4
4Ef If (8)

and the maximum bending moment can be found at (see Appendix A):
π
x=
4β (9)

where its value is determined according to:

2 q q
Mmax = π · ≈ 0.08· 2 .
8e 4 β2 β (10)

3. Proposed methodology

Mechanical anchorages might be efficient if the correct external


compressive stress is applied to the system, otherwise the strength of
the system may be compromised. The steps needed to design such an
anchorage can be synthetically enumerated as follows:

1. Define the maximum external compressive stress limited by the


shear strength of the FRP composite;
2. Estimate the strain in the FRP composite due to the application of
the external compressive stress;
3. Assume a local bond-slip model adjustable to the external stress
applied to the anchorage and that accurately represents the real
debonding behaviour of the FRP-to-concrete interface;
4. Assume that the rupture is based on the Mohr-Coulomb criterion;
5. From the previous points, the Mode II fracture energy is determined
Fig. 3. Flowchart of the proposed methodology to estimate the compression stress needed
and, consequently, the interface strength is estimated; to have an efficient anchorage and the strain induced to the FRP composite due to this
6. The comparison between the strength and the service load trans- external compression stress.
mitted to the FRP composite is made according to:
FEd ⩽ FRd (11) The FRP composite to be used may be chosen from CFRP, GFRP,
Aramid (A) FRP to Basalt (B) FRP or other material within the family of
where FEd and FRd are the applied and the strength loads, respectively.
these polymers. Also, the use of these FRPs as laminates or sheets can be
Fig. 3 shows the flowchart of the proposed methodology and each step
considered as well. However, it is worth bearing in mind that there are
will be fully described in the following sections.
some parameters to be considered in the proposed design method that
may change with the type of FRP composite chosen for the EBR of the
4. Design implementation
concrete beam.
The epoxy resins usually used to promote the bonding between the
4.1. Dimensions of the steel plate anchorage
FRP composite and the substrate do not have an ideally linear beha-
viour, but close to the rupture vicinity they do not show significant
Based on the dimensions of the concrete beam to be flexurally-
nonlinearities at room temperatures as shown in some experimental
strengthened with the FRP composite, the thickness or the width of the
works found in the literature, e.g. [47–52]. Therefore, a linear beha-
composite was defined previously. In this calculation, it is assumed in
viour is assumed where the displacements (v) are proportional to the
the equilibrium of the critical section of the beam (see Fig. 2) that the
stresses (σ) applied to the resin. The stiffness coefficient of the resin, Kr,
rupture of the FRP composite and the yielding of the steel reinforce-
is defined as:
ments can be reached. Thus, based on the width of the FRP composite
obtained, the anchor steel plate may have the same width plus σ
Kr = .
v (12)
70–100 mm, which will permit the holes for the metallic bolts to be
drilled without damaging the steel reinforcement or the main FRP where σ is the stress developed in the perpendicular direction of the
composite. Moreover, the thickness of the anchorage steel plate should bonded area, i.e. consistent with pure Mode I; and v is the relative
not be less than 8–10 mm in order to prevent excessive transversal displacement produced in the same direction consistent with pure Mode
deformations of the plate. I. Usually, instead of Eq. (12), Dimitrov’s formula [53] is used:
λ·Er
4.2. Properties of materials Kr =
br (1−υ2) (13)
The concrete of the beam to be strengthened is assumed to be so stiff where br is the width of the adhesive agent (e.g. resin), which may be
that the elastic foundation of the FRP Winkler’s beam will be defined considered equal to the width of the FRP composite bf; υ is the Poisson's
only by the adhesive properties. Furthermore, the concrete compressive ratio that can be assumed a value from 0.20 to 0.30 in the case of GFRP
strength of the beam should not be lower than 17.5 MPa [46,47] be- composites [54,55] and from 0.25 to 0.34 in the case of CFRP com-
cause otherwise, the efficiency of the bonding will be seriously com- posites [55,56]; Er is the Young modulus of the resin (see Table 1 for
promised due to low bond stress transfer between materials. several examples of epoxy resins); and λ is a coefficient that depends on

55
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

Table 1
Young modulus of several commercial epoxy resins.

Supplier Resin Young modulus, Er (GPa) Observations

Sika Sikadur-30 9.6 (compression); 11.2 (tension) Mortar based on epoxy resins
Sika Sikadur-300 2.8 (bending); 3.5 (tension) Tested after 7 days at 23 °C
Sika Sikadur-330 3.8 (bending); 4.5 (tension) Tested after 7 days at 23 °C
S&P S&P resin 50 2.6 Tested after 14 days at 15–20 °C
S&P S&P resin 55 2.6 Tested after 14 days at 15–20 °C
S&P S&P resin 220 > 7.1 –
BASF Mbrace Resin 55 2.8 (compression) Tested after at 23 °C and 65% RH
BASF Mbrace Resin 220 6.8 (compression); 7.9 (bending) Tested after at 23 °C and 65% RH
MAPEI MapeWrap 31 1.4 (compression); 3.0 (bending) Tested after 7 days
Fyfe Tyfo S 3.2 (compression or tension); 3.1 (bending) Tested after 3 days at 60 °C
Fyfe Tyfo WS 3.2 (tension); 3.1 (bending) Tested after 3 days at 60 °C
Fyfe Tyfo MB-3, Tyfo TC 1.2 (tension or bending) Tested after 3 days at 60 °C
Fyfe Tyfo 103, Tyfo 111 2.8 (compression) Tested after 7 days at 23 °C
Fyfe Tyfo 111 3.2 (tension); 3.1 (bending) Tested after 3 days at 60 °C
Fyfe Tyfo WP 2.1 (tension); 2.6 (bending) Tested after 3 days at 60 °C

Table 2
Values for the coefficient λ [53].

Lanchor/br 1 2 3 5 10 20 30 50

λ 1.05 0.78 0.66 0.54 0.45 0.39 0.33 0.30

the anchored length (Lanchor) and width of the resin as shown in Table 2.

4.3. FRP induced strain due to the external compression stress

The strains on the outer surface of the Winkler’s FRP beam can be
determined accordingly to:
tf
εf = − · χ
2 (14)
Fig. 4. Typical load-slip curve of an interface subjected to an external compression stress.
where tf is the thickness of the FRP composite which, if applied with the
wet lay-up technique, it may be considered as the thickness of the
number of layers used. When Eqs. (A.12) and (A.13) are introduced into FRP-to-concrete interface, which is designated by residual load (Fres).
Eq. (14) the strain developed at the outer surface of the FRP beam is However, the advantage of the latter alternative is the easier way that
calculated according to: the design and calculus of anchorage are presented, since the Mode II
fracture energy released during the failure debonding process of the
tf M
εf = · . interface does not need to be quantified, as it will be explained later.
2 Ef If (15)
(i) Using the bond-slip model (considering fracture energy)
Therefore, replacing in Eq. (15) the bending moment M developed
in the FRP Winkler’s beam by its maximum value Mmax, as defined
The steel plate anchorage can be designed by finding the energy
previously in Eq. (10), yields:
released in the full debonding of the FRP-to-concrete interface sub-
2 σ·Af β 2 σ·Af β 2 jected to the compressive external stress. So, the equivalent Mode II
ε fσ,max = π · ≈ 0.16·
4e 4 K K (16) fracture energy in such a situation is calculated according to:
s0,res
where σ is the external compression stress applied to the FRP composite GF ,c = ∫0 τ (s ) ds (17)
(see Fig. 2); and Af is the cross sectional area of the FRP composite.
where s0,res is the slip corresponding to the initiation of the constant
4.4. Strength verifications of the interface bond stress distribution along the interface, i.e. when the constant stage
starts to influence the interface, and is calculated according to [38]:
4.4.1. Strength of the interface under the steel plate anchorage 2
σ tan(ϕ) Lanchor
Two alternatives can be followed to estimate the strength of a me- s0,res = .
chanical anchorage: 2Ef t f (18)
where Lanchor is the anchor length adopted for the anchorage steel plate.
(i) considering the fracture energy released during the failure de- Introducing Eqs. (2) and (18) into Eq. (17), leads to:
bonding process of the interface; and
2
σ tan(ϕ) Lanchor 2
σ tan(ϕ) Lanchor
(ii) assuming a uniform bond stress distribution along the interface ⎛ −B· −B· ⎞
GF ,c = GF ,0·⎜1 + e Ef t f −2e 2Ef t f
equal to the residual bond stress. ⎟
⎝ ⎠
2
The first alternative will provide more accurate results because it ⎛ 2 −B·
σ tan(ϕ) Lanchor

σ tan(ϕ) Lanchor e 2Ef t f 1
intends to accurately estimate the maximum load transmitted to the + σ tan(ϕ)·⎜ + − ⎟.
FRP-to-concrete interface (Fmax), as shown in Fig. 4 while the other way ⎜ 2Ef t f B B⎟
⎝ ⎠ (19)
is more conservative corresponding to a lower load transmitted to the

56
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

Remembering that parameters A and B are constants that are de- In this case, the full anchored length is assumed to be subjected to a
fined by the strain-slip curve experimentally obtained from a sufficient uniform residual bond stress and the load transmitted to the FRP
long interface, both can be determined, respectively [38]: composite is calculated according to the following equation [38]:
2GF ,0 Fres ≡ FRd = τres bf Lanchor (29)
A=
Ef t f (20) where Lanchor is the anchored length. Replacing in Eq. (28) the value of
and the residual bond stress as defined in Eq. (3), leads to:

ln(2) Fres ≡ FRd = σ tan(ϕ) bf Lanchor . (30)


B=
smax,0 (21) A final note to the common parameters that are needed in both
or, alternatively to Eq. (21) [38]: approaches: the external stress (σ), the internal friction angle (ϕ) of the
interface (for which, unless stated otherwise, the model proposed by
2τmax,0 Biscaia et al. [15,62] can be assumed) and the anchored length (Lanchor),
B= .
GF ,0 (22) which, until now, has not yet been determined and, with its definition,
The fracture energy of the interface subjected to the compression completes the design of a mechanical anchorage that should be applied
stress still depends on the anchored length, which is, in fact, one to the FRP composite in order to improve its final strength. Further-
parameter to be considered in the design. Therefore, regarding Eq. (11), more, this case may not need to consider the strain in the FRP com-
the strength of the interface should be calculated as follows: posite induced by the compression stress because, as shown in Fig. 4,
two distinct load points can be found with the same value of Fres and, in
FEd ⩽ bf · 2GF ,c Ef t f (23) both cases, the values of Fres is always lower than FRd and the influence
of the initial strain in the FRP induced by the external compression
which according to some studies (e.g. [16,43,57–60]), the second term
stress is already taken into account in that difference.
of Eq. (23) can be used independently of the shape of the bond-slip
relationship. The FRP composite has a linear constitutive behaviour
4.4.2. Anchored length
until rupture and therefore, from Eq. (23), the maximum strain in the
The anchored length is determined from the condition stated in (10)
FRP composite is:
through the knowledge of the service load that the FRP composite will
2GF ,c be subjected to. The anchored length can be now determined from Eq.
εf ,max = .
Ef t f (24) (23) or, conservatively, as follows:

However, as mention earlier, an initial strain is induced into the FRP FEd ⩽ σ tan(ϕ) bf Lanchor . (31)
composite when a compression stress is applied to the mechanical an- However, considering that the external compression stress increases
chorage device. So, during the debonding process of the FRP-to-con- the strains in the FRP composite and recognizing that the FRP has an
crete interface, as the maximum strains move towards the FRP an- elastic constitutive behaviour, Eq. (31) can be rewritten accordingly:
chored end, the design maximum strain in the FRP composite due to the
action of a pull load should be determined by the following expression: ⎡ t f2 β 2·bf br ·Ef ⎤
FEd ⩽ σ ⎢tan(ϕ) bf Lanchor −0.16· ·(1−υ2)⎥.
⎣ λ · Er ⎦ (32)
2GF ,c σ
(εf ,max )Ed = −ε f ,max
Ef t f (25) The minimum external compressive stress that needs to be applied
onto the GFRP composite to prevent the debonding of the GFRP-to-
where εσf ,max
is the strain developed in the FRP composite due to the
concrete interface is obtained by isolating σ in Eq. (32) and redefining σ
application of the external compression stress as defined in Eq. (16).
equal to σ2 (for the second alternative), which leads to:
The maximum load FEd transmitted to the FRP composite can be found
by multiplying the Young’s modulus and the cross sectional area of the FEd
σ⩾
t 2f β2·bf br ·Ef
FRP composite in both terms of Eq. (25), which yields: tan(ϕ) bf Lanchor −0.16· ·(1−υ2)
λ·Er (33)
σ1·Ef ·bf ·br ·(t f β )2
FEd = bf · 2GF ,c Ef t f −0.16· ·(1−υ2). where FEd is calculated from the design value of the FRP accordingly to
λ·Er (26)
Eq. (28). The minimum stress to be applied to the FRP composite is then
It should be further noted that the fracture energy due to the in- obtained using Eq. (33), which must be lower than that obtained from
fluence of the external compression stress GF,c depends on the amount the shear verification.
of external stress applied to the FRP composite as shown from Eq. (19).
Therefore, the maximum external stress σ1 in Eq. (26) cannot be iso- 4.5. Shear verification
lated and after equating the service and the strength loads, its solution
can be obtained by trial and error in order to satisfy the following To verify the shear condition of the FRP composite when submitted
equation: to the external compression stress, the following condition should be
σ1·Ef ·bf ·br ·(t f β )2 satisfied:
FEd−bf · 2GF ,c Ef t f + 0.16· ·(1−υ2) = 0.
λ·Er (27) VEd ⩽ VRd (34)

where FEd is determined from the design value of the FRP strain, εfd, where VEd is the shear design calculated according to Eq. (7); and VRd is
according to: the shear capacity of the FRP composite calculated as explained next.
εfk The “rule of mixtures” simplification is assumed and the shear strength
εfd = of the FRP composite (τf) is determined as:
γf (28)
τf ≈ τm vm + τfib vfib (35)
where εfk is the characteristic value of the FRP strain; and γf is a safety
coefficient as recommended in Fib bulletin 14 [61]. where τm and τfib are the shear strength of the matrix and fibres, re-
spectively; and vm and vfib are the volume fraction of the matrix and
(ii) Adopting a simplified approach fibres where vm + vfib = 1. According to the FIB bulletin 14 [61], ty-
pical values for the volume fraction of fibres in prefabricated strips vary

57
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

from 0.50 to 0.65 and the shear strength of matrices (τm) may vary which corresponds to a higher value than that obtained from the pre-
between 10 and 30 MPa [61]. Furthermore, if the fibres have negligible vious verification from Eq. (27) and, therefore, it may be conservatively
shear strength, Eq. (35) can be reduced to: considered to be the external compressive stress needed to obtain its
τf ≈ τm vm. rupture. Hence, this value cannot exceed the maximum compression
(36)
stress that can be applied to the GFRP composite without causing its
For a uniform shear stress distribution along the FRP thickness, Eq. rupture due to shear and this needs to be confirmed, as will be checked
(36) allows the maximum external compression stress that can be ap- in the subsequent subsection. It should be noted also that if theses va-
plied to the composite to be found. Making the maximum shear in Eq. lues are both higher than the maximum stress limited due to shear
(7) equal to the shear design in Eq. (36) and considering the safety (σshear), the materials and/or its dimensions previously used should be
coefficient γf, the maximum external compressive stress is limited by: redefined and stresses σ1 and σ2 should be recalculated, once again,
until one of their values becomes lower than σshear. Otherwise, the an-
4βτm vm t f
σshear ⩽ . chorage plate will not be properly designed.
γf (37)
5.2. Shear verification
5. Application example
The maximum external stress that can be applied to the GFRP
To show how the proposed design method is applied, an example of composite is determined according to Eq. (37). The shear strength and
the volume fraction of the matrix were assumed as equal to 10 MPa and
a GFRP-to-concrete interface based on a previous work of the authors
[62] is presented. Double-lap shear tests were performed with GFRP 0.50 [61], respectively, leading to
composites externally bonded on concrete with and without external 4βτm vm t f
compressive stresses externally applied to the composite. The results σshear ⩽ ⇒ σmax = 4.48 MPa.
γf (40)
obtained there [62] are considered in this example and, therefore, some
reference parameters used in this example are assumed unless other- Therefore, the compression stress limit obtained for this example is
wise specifically stated. Thus, it is assumed that the thickness of the approximately 4.5 MPa, which means that for higher external com-
GFRP composite is tf = 2.54 mm (1.27 mm per layer), the width is pression stresses perpendicularly applied to the GFRP composite, the
bf = 80 mm, the Young’s modulus is Ef = 20.39 GPa and the char- rupture of the GFRP due to shear will occur.
acteristic rupture strain is εfk = 2.10% [49,62]. Moreover, the average
concrete tensile strength is fctm = 2.69 MPa and an epoxy resin was used 5.3. Induced strain in the FRP composite due to the external compression
with, according to the supplier [63], a Young’s modulus of stress
Er = 3.2 GPa. The bond-slip relationship free of any external pressure is
defined according to an exponential bond-slip model in which the The maximum strain in the FRP due to the application of the ex-
maximum bond stress is τmax,0 = 4.24 MPa, smax,0 = 0.101 mm, ternal compression stress calculated in (39) is determined with the help
B = 9.0 mm−1 and GF,0 = 1.026 N/mm. The internal friction of the of Eq. (16), leading to:
GFRP-to-concrete interface was found to be equal to ϕ = 1.064 rad. An
anchorage steel plate with dimensions 150 × 150 × 10 mm σ·br ·β 2
ε fσ,nax = 0.16· ·(1−υ2) = 0.059%
(length × width × thickness) will be initially considered in this ex- λ·Er (41)
ample to anchor the GFRP free end. Also, the value of the safety coef- where
ficient, γf, adopted is, according to the recommendation in [61], equal
to 1.50, i.e. application of wet lay-up GFRP systems under normal σ = max{σ1;σ2}. (42)
quality control conditions. Therefore, using σ2 = 2.80 MPa, the strain obtained in (41) re-
presents a load of 2.43 kN and, under these conditions, this means that
5.1. Verifying the strength of the interface the design value of the GFRP strain is 1.342% instead of 1.400%, as
determined from Eq. (28).
5.1.1. Using the bond-slip model (considering fracture energy)
Based on the fracture energy released during the failure debonding 5.4. Final drawings, preparation and strengthening of the RC beam
process, the strength load Frd can be found from Eq. (26), where

3λEr Despite being outside the scope of this work, to get the mechanical
β= 4 = 0.132 mm−1 anchorage finally designed, it should be mentioned that the determi-
br t f3 Ef ·(1−υ2) (38) nation of the number of metallic anchors should follow, e.g., Annex C of
and, in the current example, for a relation Lanchor/br equal to 1.0 (i.e. ETAG 001 [64], which, for the current example, the use of 6 steel an-
150/80), by performing a linear regression of the values presented in chors, Graded 8.8, with a diameter of 8 mm can be considered (e.g. HST
Table 2, λ = 0.81. Also, the Poisson’s ratio was assumed at υ = 0.22. R M8/50 from Hilti supplier). This should be enough to sustain the
Therefore, the maximum external stress σ1 in Eq. (26) is obtained by necessary load to apply the external compression stress of 2.80 MPa at
trial and error in order to satisfy Eq. (27), which, for the current ex- both GFRP anchored ends.
ample, returned σ1 = 2.66 MPa. For this solution, Fig. 5 shows the main procedures for installing the
mechanical steel plate on the left and the final drawings of the
strengthening procedure are depicted on the right (Fig. 5b.2 to d.2).
5.1.2. Simplified approach
Regarding the installation process, it should be mentioned that after the
The anchored length considered in this example is 150 mm and the
corresponding surface preparation, the adhesive should first be spread
minimum external compressive stress necessary to prevent the de-
along the bonded area of the unstrengthened concrete beam (key 1 in
bonding of the GFRP-to-concrete interface is obtained by solving Eq.
Fig. 5a). The adhesive should also be spread on the same area of the
(33), which leads to
mechanical steel plate (2 in Fig. 5b.1). The FRP composite is then ap-
εfd Ef bf t f plied on the adhesive (3 in Fig. 5c.1) and the remaining width at the
σ⩾ ⇒ σ2 ⩾ 2.80 MPa
bf br ·Ef ·t 2f β2 FRP free end, i.e. where the mechanical anchorage will be positioned,
tan(ϕ) bf Lanchor −0.16· ·(1−υ2) (39)
λ·Er should be also filled with some FRP composite (4 in Fig. 5c.1), in order

58
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

Fig. 5. Main procedures for the installation of the mechanical anchorage with the corresponding final drawings.

to accommodate and distribute the stresses uniformly along the an- 6. Finite element analysis
chorage area and ensure that these stresses may be uniformly trans-
ferred to the bonded interface region. After the curing period of the FRP To show the accuracy of the proposed design methodology, nu-
composite, the holes are drilled without compromising the fibres and/ merical simulations of the example previously illustrated and other
or the integrity of main FRP strip, i.e. along the additional FRP com- closed cases using the Finite Element Method (FEM) are presented next.
posite (5 in Fig. 5c.1). After cleaning the dust inside the holes and the Two cases are shown: one corresponds to the simulation of double-
surrounding area, the steel plate anchor (6 in Fig. 5d.1) can finally be shear tests, where several external compression stresses were applied to
attached by applying to each metallic anchor (7 in Fig. 5d.1) the cor- the GFRP composite to simulate the effect of the mechanical anchorage;
responding load that needs to be transmitted to the bonded interface the other corresponds to the simulation of flexurally-strengthened RC T-
(Fig. 5e). beams with GFRP composites tested in a 4-point bending test until its
Table 3 summarizes all the calculations developed in the application failure. In the latter case, the influence of the mechanical anchorage on
example that allowed the design of the anchor steel plate. The reduced the final load-bearing capacity of the RC T-beams was tested.
number of steps and simple equations needed attest to the simplicity of
the proposed method for the design of these anchor devices often used
to avoid the premature debonding phenomenon of the FRP composite 6.1. Simulation of double-shear tests
from the concrete beam.
For the FE simulations of the double-shear tests a 2D version of the
commercial FE code ATENA [65] was used. The properties of the
concrete were modelled with an available model suitable for concrete
designated in the FE package by “SBeta Material” which is the German

Table 3
Summary of the steps and results obtained from the application example.

Steps Equation Equation Results Observations


number

#1 3λEr 8 0.132 mm−1 λ is determined from Lanchor/br ratio as defined in Table 2


β= 4
br t 3f Ef ·(1 − υ2)

#2 σ 1·br bf Ef ·(t f β )2 27 σ1 = 2.66 MPa Equation solved by tentative and error due to the dependency of fracture
εfd Ef bf t f −bf · 2GF ,c Ef t f + 0.16· ·(1−υ2) = 0 energy with σ as it can be seen from Eq. (19). The safety coefficient γf to
λ·Er
be considered can be found in the Fib bulletin 14 [61] in its Table 3.1
#3 ε fd Ef bf t f 33 σ2 = 2.80 MPa The value of the internal friction angle of the interface (ϕ) is needed
σ2 ⩾
bf br Ef ·t 2f β2
tan(ϕ) bf Lanchor − 0.16· ·(1 − υ2)
κ·Er
#4 4βτm vm t f 37 σshear = 4.48 MPa The shear strength of the matrix (τm) and the volume fraction of the
σshear ⩽
γf matrix (vm) are needed
#5 σshear > max{σ1; σ2} – – If the condition is not verified, then new materials and/or geometrical
dimensions should be defined and the 1st step should retaken
#6 σ·t f β2 16 0.059% Where σ = max{σ1; σ2} and σ ≤ σshear
ε fσ,max ≈ 0.16· ·br ·(1−υ2)
λ·Er

59
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

abbreviation for Stahlbeton Analyse (Reinforced Concrete Analysis) Section 5.2 (σsheat = 4.48 MPa). In the perspective of the design, due to
[65]. The GFRP composite constitutive behaviour was considered to be the safety coefficient herein adopted, the external compression stress of
linear with fragile rupture. The contact between the GFRP and the 2.80 MPa is sufficient enough to prevent the debonding of the GFRP
concrete was modelled with dimensionless interface elements, which composite from the anchorage since the values used for the design of
are governed by the Mohr-Coulomb rupture criterion. Thus, before the the strengthened concrete beam are all based on design values as well
rupture line is reached, a linear relationship between the bond stress (see Fig. 2) being, in the particular case of the GFRP composite, the
and the relative displacement consistent with Mode II is assumed with corresponding force installed in the GFRP limited by 1.342% (see
stiffness Ktt = 4.2 × 104 MN/m3, i.e. the ratio between the maximum Section 5.3).
bond stress and the corresponding slip: Here, it should also be mentioned that the snapback phenomenon
τmax described in the literature (e.g. [20,57,66–68]) cannot be captured by
Ktt = . the FE analysis because the displacements imposed to the GFRP com-
smax (43)
posite are always applied at its loaded end. Nevertheless, this is not
After the rupture line is reached, an exponential softening behaviour relevant for the present discussion since this is a post-peak phenomenon
of the interface was assumed by introducing the same exponential that has low importance for the anchorage design. Moreover, the
softening stage of the pure Mode II bond-slip relationship used along snapback phenomenon would only occur in the case with no external
the examples herein presented in Section 5. stresses applied to the interface (σ = 0.0 MPa) because all the other
On the other hand, perpendicular to the bonded area, the stiffness cases have a stage with a constant residual bond stress that prevents the
Knn was almost neglected since it plays an irrelevant role in the failure snapback phenomenon from occurring.
debonding process of the interface. Other relevant parameters to com- If the predicted service and the residual loads obtained from the
pletely define the elements of interface, such as the cohesion and the proposed design method with the numerical results are now compared,
internal friction of the interface, were defined in accordance with the it can be seen in Fig. 7 that the relative differences between both are
experimental results obtained during the experimental campaign car- less than 10%, which shows the accuracy of the proposed design
ried out by the authors fully described elsewhere [44]. Thus, for a method with the FEM. It should be noted also that each point in Fig. 6
GFRP-to-concrete interface, the friction angle is 1.064 rad, whereas the represents a situation with a different external compression stress, the
cohesion was set equal to the maximum bond stress obtained from the lowest point being σ = 0.0 MPa and the highest σ = 5.0 MPa.
local bond-slip relationship, c = 4.24 MPa.
Both elements were modelled with quadrilateral finite elements
with different sizes. In the case of the concrete, the brick elements used 6.2. Simulation of RC T-beams
had 5.0 mm of side, whereas the size of the brick elements used for the
GFRP element were smaller, with 0.5 mm of side. The mesh along the In the numerical simulation of the flexurally-strengthened RC T-
interface was also refined and it was ensured that the nodes along the beams with GFRP composites, the 3D version of ATENA was used. The
interface were spaced 0.5 mm apart. Hence, the model had a total of concrete and the GFRP were modelled with brick elements. In this
2281 elements and 3399 nodes. A constant displacement increase of ATENA version, the properties of the concrete were modelled with an
0.02 mm/step was applied to the GFRP composite to simulate the loads available model designated as “CC3DNonLinCementitious2”, which is a
transmitted to the composite and several steps were considered. model that combines constitutive models for tensile (fracturing) and for
To easily compare with the example given in Section 5, 11 different compression (plastic) behaviour of concrete. For more detailed in-
situations were modelled, where the 150 mm length of the mechanical formation, the ATENA theory manual [65] should be consulted.
anchorage was replaced by external compression stresses from 0.5 to Due to the symmetry of the RC T-beam, only one-fourth of the RC T-
5.0 MPa with intervals of 0.5 MPa between each situation. Naturally, beam was modelled as can be seen from the model shown in Fig. 8. A
the situation free of the mechanical anchorage had no external com- monotonic displacement history was chosen to simulate the loads ap-
pression stresses applied on the GFRP composite. The load-slip results plied to the beam at 1.0 m from each pin rolled support. The increment
obtained from the 11 simulations are shown in Fig. 6. Also, the rupture adopted was equal to 0.4 mm per step of the simulation and several
load of the GFRP is shown in Fig. 6 to easily identify the modelled steps were computationally analysed. The free span of the T-beam is
situations beyond which the rupture of the GFRP composite is reached. 3.0 m. Details of the steel reinforcements and remaining configurations
Therefore, it can be clearly seen that when a compression stress of can be found elsewhere [12].
4.50 MPa is applied on the GFRP composite, the rupture of the GFRP-to-
concrete interface is reached, which, in this particular example, almost
coincides with the limit imposed by the shear verification performed in

Fig. 6. Load-slip responses of the GFRP-to-concrete interface under different external Fig. 7. Comparisons between maximum load and residual load, obtained from the pro-
compression stresses. posed design model and from the FEA under different external compression stresses.

60
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

3, 4 in Fig. 9 and curve 3 in Fig. 10, respectively), a classical rupture


was reached, i.e. the T-beam failed due to the crushing of the concrete
at a compression ultimate strain of εcu = −0.35%. However, it should
be mentioned that for the RC T-beam flexurally-strengthened with an
80 mm wide GFRP composite, the rupture of the GFRP composite is
reached after the concrete was crushed. Nevertheless, in both cases, the
rupture criteria in the concrete (crushing of the concrete at
εcu = −0.35%) was reached and this was the failure mode considered
for these RC T-beams.
In terms of the load bearing capacity improvement regarding the
corresponding unstrengthened RC T-beam which reached 65.0 kN, an
increase of 68.3% (109.4 kN) and 66.3% (108.1 kN) was found from the
T-beams with 80 mm and 120 mm GFRP, respectively. However, when
the bearing capacity of the RC T-beams without the anchorage system is
compared with the unstrengthened T-beam, increases of 41.5%
(92.0 kN) and 42.9% (92.9 kN) are determined for the T-beams with
80 mm and 120 mm GFRP, respectively.
From the point of view of the design, the vertical dashed line in
Fig. 10 represents the strain limit of the GFRP composite obtained from
the proposed model. This result shows that the strain limit of 1.342%
(see Section 5.3) drops between the GFRP debonding (Point d) and the
concrete crushing (Point c) at a mid-span displacement of 58.5 mm.
With a safety factor γf of 1.50 [61], the results obtained for the current
example provide an estimation for the strength of the T-beam of
93.4 kN, which represents 85.4% of the real load bearing capacity of the
flexurally-strengthened RC T-beam.
Denoting μΔ,ref as the ductility coefficient in displacement of the
strengthened RC T-beams defined by the ratio between the mid-span
displacement at the rupture (Δr) and at the yielding (Δy) of the steel
reinforcements of the RC T-beam:
Δr
μΔ,ref = ,
Δy (44)

Fig. 8. Finite element models: (a) mesh; (b) RC T-beam free of any mechanical anchorage the ductility coefficient of the T-beam with the 80 mm and 120 mm
device; and (c) RC T-beam with a mechanical anchorage device installed. wide anchored GFRP composite reached 4.7 and 4.4, respectively.
However, this ductility coefficient of the T-beams without the steel
The external compression stresses were induced to the steel plate plate anchorage installed is 3.5 and 2.6 for the cases with 80 mm and
anchorage located at both ends of the RC T-beam by means of simu- 120 mm GFRP composites, respectively.
lation of three steel discrete reinforcements subjected to a temperature Fig. 9 also shows the strains in the EBR T-beams and in the steel
change. This was, in fact, the first step of the numerical analysis, which reinforcements in the xx-direction at different stages of the simulations
had the purpose of simulating the application of an external compres- of the T-beams with 120 mm GFRP: (i) at the yielding of the steel re-
sion stress of 4.48 MPa to the steel plate anchor, which, according to the inforcements (Point y); when the GFRP debonds from the T-beam in the
previous examples, should be sufficient to get the tensile rupture of the unanchored solution (Point d); and at the concrete crushing (Point c).
GFRP composite. Only after this initial step is finished all the following Fig. 10 also shows the strains in the xx-direction of the T-beams with
steps had a regular displacement increase applied to the T-beam in 80 mm wide GFRP and beyond the same three points identified in
accordance to what was already mentioned earlier. It should be noted Fig. 9, one last point corresponding to the rupture of the GFRP com-
that the width of the GFRP composite used to flexurally-strengthened posite is also depicted. These results prove that the negative strains
the RC T-beams in [12] was 120 mm and for this reason two GFRP (compression) in the concrete of the T-beam with the mechanical an-
widths of 80 mm (the same as the examples presented earlier) and chorage installed reached 0.35% at the top of the beam and close to the
120 mm are considered here. Thus, the results obtained for the latter T- applied load. On the other hand, the debonding of the GFRP composite
beams could be compared with the experimental one allowing the va- can be identified by the visible separation of the GFRP element from the
lidation of the modelling to be carried out in this particular numerical beam at the nearest region to the support of the T-beam without the
study (see Fig. 9). steel plate anchorage (see Point d in Figs. 8 and 9). The steel re-
At the end of the analyses, the load vs. mid-span displacement were inforcements have yielded in both cases, which can be attested to by the
all gathered and grouped into two graphs for comparisons between the decrease in stiffness of all the simulated T-beams shown in Figs. 8 and 9
solutions considered here, as shown in Figs. 9 and 10. The experimental after approximately 14.0 mm.
and numerical load vs. mid-span displacement of one unstrengthened As an example, Fig. 11 shows the amount of slip developed within
RC T-beam is also considered in these graphs. From these figures, it can the GFRP-to-concrete interface for the unanchored (Fig. 11a) and an-
be seen that the use of the mechanical anchorage has a significant effect chored (Fig. 11b) cases, with the 80 mm wide GFRP composite at GFRP
on the load bearing capacity of the RC T-beams. The use of the steel rupture (see Fig. 10, Point r). In this particular slip analysis it can be
plate as an anchorage for the GFRP composite allowed us to change the stated that a major bonded area of the interface of the T-beam without
rupture mode of these RC T-beams (curves 5 and 4 in Figs. 8 and 9, the mechanical anchorage has slipped 23 mm, which, regarding the
respectively). Hence, in the case of the RC T-beams flexurally- bond-slip relationship, means that no bond stresses are being trans-
strengthened with the GFRP composite with the steel plate anchorage, ferred between the GFRP and the concrete, i.e. the GFRP debonded
instead of the premature debonding of the GFRP from the beam (curve from the concrete. On the other hand, the highest slips in the other T-
beam are localized between the support and the applied load, where the

61
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

Fig. 9. Strains in the xx-direction at the yielding of the steel reinforcements (Point y); GFRP debonding (Point d); and concrete crushing (Point c, εcu = 0.35%): (a) EBR T-beam with
120 mm wide GFRP without the mechanical anchorage device; and (b) EBR T-beam with 120 mm wide GFRP with the mechanical anchorage device.

slips have reached 0.331 mm. These slip distributions shows how the joints are mechanically anchored, for example, using steel fasteners
use of a steel anchorage plate installed at both GFRP free ends may that, when tightened, introduce compressive stresses into the bonded
improve the stress transfer between materials and delaying or even area, therefore improving the strength of the bonded joint. The design
preventing its premature debonding from the concrete substrate. proposal presented here aims to assist engineers, practitioners and
academics in designing these steel mechanical anchorages. From the
7. Conclusions results obtained and presented in this study the following conclusions
can be enumerated:
In this work, the knowledge about the performance of FRP-to-con-
crete interfaces externally subjected to compressive stresses was dis- • external compression stresses improve the performance of FRP-to-
cussed in depth. This corresponds to the same cases where bonded concrete interfaces allowing premature debonding of the composite

62
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

Fig. 10. Strains in the xx-direction at the yielding of the steel reinforcements (Point y); GFRP debonding (Point d); concrete crushing (Point c, εcu = 0.35%); and GFRP rupture (Point r,
εfu = 2.20% [49,62]): (a) EBR T-beam with 80 mm wide GFRP without the mechanical anchorage device; and (b) EBR T-beam with 80 mm wide GFRP with the mechanical anchorage
device.

63
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

Fig. 11. Slip distribution throughout the bonded length at GFRP rupture (Point r): (a) EBR T-beam with 80 mm wide GFRP without the mechanical anchorage device; and (b) EBR T-beam
with 80 mm wide GFRP with the mechanical anchorage device.

from the concrete surface to be avoided in an early stage, i.e. far composite can be eliminated;
away from its rupture strain value; • the steel plate anchorages do not represent an ideal solution for
• the design of the steel mechanical anchorages is based on bond-slip anchoring FRP composites to the substrate since, when the com-
relationships that consider the influence of external stresses. Such pression stresses are transversely applied, the longitudinal strains in
dependency is governed by the internal friction angle of the inter- the FRP composite increase, which reduces the maximum loads that
face and by its cohesion, according to the Mohr-Coulomb rupture can be applied to the composite afterwards;
criterion. The cohesion of the interface assumes the maximum bond • the expression proposed to estimate the value of the strain induced
stress of the bond-slip law and the friction angle is mainly re- to the FRP due to the external compression stress has demonstrated
sponsible for the changes in the bond-slip when external stresses are a fair accuracy with the final results. Also, it can be concluded from
considered; the example reported here, that the compression stress applied to
• the proposed design method is easy to implement in an international the mechanical anchorage increased the strain in the GFRP com-
code, once some parameters for the bonded joint such as cohesion, posite (reaching 0.059%), which will reduce the maximum strain to
internal friction angle, mechanical properties of the adhesion agent be considered for design.
and FRP composite, are entirely identified and known;
• the numerical study based on a commercial FE code confirms the
results obtained from the design proposal. Furthermore, RC T-beams Acknowledgments
flexurally-strengthened with GFRP composites were modelled and
the influence of using mechanical anchorages on the load bearing The first author is grateful to Fundação para a Ciência e Tecnologia
capacity of the RC T-beams were presented. The results attested well for partial financing of the work under the UNIDEMI Strategic Project
to the importance of considering a design method that could be PEst-OE/EME/UI0667/2014 and for the Post-Doctoral grant SFRH/
easily followed because the premature debonding of the FRP BPD/111787/2015.

Appendix A

Assuming the anchorage zone as a Winkler’s beam where the elastic foundation is the bonding agent and the FRP composite the “beam”, the
equilibrium equation of the FRP beam on an elastic foundation (adhesive) can be represented by Eq. (5). In this appendix, the mathematical solutions
obtained from the differential equation in (5) are presented. Thus, introducing parameter β defined in Eq. (8) into Eq. (5) the homogeneous solution
of Eq. (5) is determined:
vh = e β·x [C1cos(βx ) + C2sin(βx )] + e−βx [C3cos(βx ) + C4sin(βx )] (A.1)
where C1, C2, C3 and C4 are all constants that are obtained from the boundary conditions of the problem. Despite well defined (finite values), for the
sake of the simplicity, the dimensions of the anchored and the bonded lengths are assumed to be so large that they are infinite. In those cases,
according to Timoshenko [45] when βL > 5, the beam is considered to have a long length with a response similar to infinite beams. Naturally, as βL
increases the response of a finite and an infinite beam are more alike.
The bonded length, i.e. the right side of the FRP beam, the homogeneous solution in Eq. (A.1) is defined by assuming that when x → ∞ the
relative displacement v must be zero (v = vh = 0) leading to C1 = C2 = 0. Therefore, the solution for the right side of the bonded length (see Fig. 2)
is:
vr = e−β·x [C3cos(βx ) + C4sin(βx )]. (A.2)
The anchored length, i.e. the left side of the FRP beam (see Fig. 2), the solution of Eq. (5) is the sum of the homogeneous solution stated in Eq.
(A.1) plus the particular solution, which gives:
q
vl = e β·x [C5cos(βx ) + C6sin(βx )] + e−βx [C7cos(βx ) + C8sin(βx )] +
K (A.3)
where C5, C6, C7 and C8 are constants to be found from the boundary conditions of the problem. So, when x → ∞ the FRP beam should have a finite
relative displacement, which is only obtained by ensuring that the coefficients in the term eβx in Eq. (A.1) are zero. Therefore, the solution of Eq.
(A.3) stays:

64
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

q
vl = e−β·x [C7cos(βx ) + C8sin(βx )] + .
K (A.4)
The boundary conditions are all defined by imposing compatibility and equilibrium conditions at x = x = 0 as follows:
vl = vr , (A.5)

dvl dv
+ r = 0,
dx dx (A.6)

d 2vl d 2vr
= ,
dx 2 dx 2 (A.7)
and
d3vl d3vr
+ = 0.
dx 3 dx 3 (A.8)
From the boundary conditions stated in Eqs. (A.5) to (A-8), constants C3, C4, C7 and C8 are obtained:
q
C3 = −C7 = and C4 = C8 = 0
2K (A.9)
leading to:
q −βx
vr = e ·cos(βx )
2K (A.10)
and

q ⎡ e−βx
vl = − cos(βx ) + 1⎤.
K⎢⎣ 2 ⎥
⎦ (A.11)
The maximum shear in the FRP Winkler’s beam occurs at x = x = 0 and it is calculated according to Eq. (7). It is also important to notice that
when an external compression stress is applied to the FRP composite, the strains in the FRP composite are affected which has direct implications on
the rupture mode of the FRP-to-concrete bonded joint as documented in [15,39]. For instance, the authors [39] observed in a previous work that the
strains in a CFRP laminate externally bonded to a steel profile increased when the steel anchorage was pressured against the CFRP/substrate bonded
system which had implications on the subsequent maximum strain reached in the CFRP at the rupture instant.
Therefore, to determine the relation between the external compression stress applied to the FRP composite and the strains in it, the maximum
bending moment developed in the FRP Winkler's beam is established. To that end, the curvatures of the right side of the FRP beam are calculated
according to:
d 2vr
χ=−
dx 2 (A.12)
and the bending moment is:
d 2vr
M = Ef If · .
dx 2 (A.13)
Introducing the second derivative of Eq. (A.10) with respect to x into Eq. (A.13), yields:
q·β 2 −βx
M = Ef If · ·e ·sin(βx ).
K (A.14)
The maximum bending moment can be now determined by solving the first derivative of Eq. (A.14) with respect to x and equating to 0, i.e.
finding the point of zero shear stress, leads to Eq. (9). Introducing Eq. (9) into Eq. (A.14), the maximum bending moment developed in the FRP
Winkler’s beam is determined as stated in Eq. (10).

References [10] Al-Mahaidi R, Pham HB, Saouma V. Discrete-smeared crack finite element model-
ling of CFRP debonding mechanisms in RC members. FRPRCS-8, July 16–18,
University of Patras, Patras, Greece, 2007.
[1] Arduini M, Nanni A. Behavior of precracked RC beams strengthened with carbon [11] Obaidat YT, Heyden S, Dahlblom O. The effect of CFRP and CFRP/concrete inter-
FRP sheets. J Compos Constr 1997;1(2):63–70. face models when modelling retrofitted RC beams with FEM. Compos Struct
[2] Cosenza E, Manfredi G, Realfonzo R. Behavior and modeling of bond of FRP rebars 2010;92:1391–8.
to concrete. J Compos Constr 1997;1(2):40–51. [12] Biscaia HC, Chastre C, Silva MAG. A smeared crack analysis of reinforced concrete
[3] Buyukozturk O, Hearing B. Failure behavior of precracked concrete beams retro- T-beams strengthened with GFRP composites. Eng Struct 2013;56:1346–61.
fitted with FRP. J Compos Constr 1998;2(3):138–44. [13] Biscaia H, Chastre C, Silva MAG. Linear and nonlinear analysis of bond-slip models
[4] Bizindavyi L, Neale KW. Transfer lengths and bond strengths for composites bonded for interfaces between FRP composites and concrete. Compos Part B J
to concrete. J Compos Constr 1999;3(4):153–60. 2013;45(1):1554–68.
[5] Rahimi H, Hutchinson A. Concrete beams strengthened with externally bonded FRP [14] Chen JF, Yuan H, Teng JG. Debonding failure along a softening FRP-to-concrete
plates. J Compos Constr 2001;5(1):44–56. interface between two adjacent cracks in concrete members. Eng Struct
[6] Sebastian WM. Significance of midspan debonding failure in FRP-plated concrete 2007;29:259–70.
beams. J Struct Eng 2001;127:792–8. [15] Biscaia HC, Chastre CC, Silva MAG. Double shear tests to evaluate the bond strength
[7] Malek AM, Saadatmanesh H, Ehsani MR. Prediction of failure load of R/C beams between GFRP/concrete elements. Compos Struct 2012;94(2):681–94.
strengthened with FRP plate due to stress concentration at the plate end. ACI Struct [16] Nakaba K, Kanakubo T, Furuta T, Yoshizawa H. Bond behavior between fiber-re-
J 1998;95(1):142–52. inforced polymer laminates and concrete. ACI Struct 2001;98(3):359–67.
[8] Saadatmanesh H, Malek AM. Design guidelines for flexural strengthening of RC [17] Ferracuti B, Savoia M, Mazzotti C. Interface law for FRP-concrete delamination.
beams with FRP plates. J Compos Constr 1998;2(4):158–64. Compos Struct 2007;10:523–31.
[9] Camata G, Spacone E, Zarnic R. Experimental and nonlinear finite element studies [18] Mazzotti C, Savoia AM, Ferracuti B. A new single-shear set-up for stable debonding
of RC beams strengthened with FRP plates. Compos B Eng 2007;38:277–88. of FRP-concrete joints. Constr Build Mater 2009;23(4):1529–37.

65
H.C. Biscaia, C. Chastre Composite Structures 192 (2018) 52–66

[19] Dai J, Ueda T, Sato Y. Development of the nonlinear bond stress-slip model of fiber 2014;48(9):2961–81.
reinforced plastics sheet-concrete interfaces with a simple method. J Compos Constr [44] Biscaia HC, Silva MAG, Chastre C. Influence of external compressive stresses on the
2005;9(1):52–62. performance of GFRP-to-concrete interfaces subjected to aggressive environments:
[20] Biscaia HC, Chastre C, Silva MAG. Nonlinear numerical analysis of the debonding an experimental analysis. J Compos Constr 2015;20(2):1–19.
failure process of FRP-to-concrete interfaces. Compos B Eng 2013;50:210–23. [45] Timoshenko SP. Strength of materials: Part II, Advanced theory and problems. 3rd
[21] De Lorenzis L, Teng JG. Near-surface mounted FRP reinforcement: an emerging ed. Princeton, New Jersey: D. Van Nostrand Company Inc.; 1956.
technique for strengthening structures. Compos B Eng 2007;38(2):119–43. [46] Cánovas M. Refuerzo de elementos estructurales de hormigón armado mediante
[22] Mostofinejad D, Shameli SM, Hosseini A. EBROG and EBRIG methods for encolado de bandas de acero con adhesivos epoxídicos. Informes de la Construcción
strengthening of RC beams by FRP sheets. Eur J Environ Civ Eng 1985;37(373):27–38. [in Spanish].
2014;18(6):652–68. [47] Chastre C. Comportamento da Ligação Aço-Resina-Betão em Elementos Estruturais.
[23] Zhang HW, Smith ST. FRP-to-concrete joint assemblies anchored with multiple FRP MSc Dissertation in Structural Engineering, Instituto Superior Técnico, 1993. [in
anchors: experimental investigation. In: Ye L, Feng P, Yue Q, editors. Advances in Portuguese].
FRP composites in civil engineering. Berlin, Heidelberg: Springer; 2011. p. 553–6. [48] Rodrigues C, Silva M. Cyclic compression behaviour of polymer concrete. J Polym
[24] Breña SF, McGuirk GN. Advances on the behavior characterization of FRP-anchored Eng 2007;27(6–7):525–45.
carbon fiber-reinforced polymer (CFRP) sheets used to strengthen concrete ele- [49] Rodrigues CC. Comportamento às Acções Cíclicas de Pilares de Betão Armado
ments. Int J Concr Struct Mater 2013;7(1):3–16. Reforçados com Materiais Compósitos PhD Thesis Faculdade de Ciências e
[25] Bournas DA, Pavese A, Tizani W. Tensile capacity of FRP anchors in connecting FRP Tecnologia, Universidade Nova de Lisboa; 2005 [in Portuguese].
and TRM sheets to concrete. Eng Struct 2015;82:72–81. [50] Hufenbach W, Gude M, Böhm R, Zscheyge M. The effect of temperature on me-
[26] Llaurad PV, Ibell T, Gomez JF, Ramos FJG. Pull-out and shear-strength models for chanical properties and failure behaviour of hybrid yarn textile-reinforced ther-
FRP spike anchors. Compos B Eng 2017;116:239–52. moplastics. Mater Des 2011;32:4278–88.
[27] Martinelli E, Napoli A, Nunziata B, Realfonzo R. Inverse identification of a bearing- [51] Reis JML. Effect of temperature on the mechanical properties of polymer mortars.
stress-interface-slip relationship in mechanically fastened FRP laminates. Compos Mater Res 2012;15(4):645–9.
Struct 2012;94(8):2548–60. [52] Pires, JMC. Mechanical behaviour at elevated temperatures of GFRP pultruded
[28] Realfonzo R, Martinelli E, Napoli A, Nunziata B. Experimental investigation of the composite profiles. MSc Dissertation in civil Engineering, Instituto Superior
mechanical connection between FRP laminates and concrete. Compos B Eng Técnico, 2012.
2013;45(1):341–55. [53] Dimitrov, D. Der Balken un die platte als gründungskörper. PhD Thesis, Karlsruhe
[29] Grelle SV, Sneed LH. Review of anchorage systems of externally bonded FRP la- University, 1955. [in German].
minates. Int J Concr Struct Mater 2013;7(1):17–33. [54] Yilmaz C, Akalin C, Kocaman ES, Suleman A, Yildiz M. Monitoring Poisson’s ratio of
[30] Travassos N. Caracterização do comportamento da ligação CFRP-betão. MSc Thesis, glass fiber reinforced composites as damage index using biaxial Fiber Bragg Grating
Instituto Superior Técnico, 2004. [in Portuguese]. sensors. Polym Test 2016;53:98–107.
[31] Biscaia HC, Micaelo R, Teixeira J, Chastre C. Numerical analysis of FRP anchorage [55] Gay D, Hoa SV. Composite materials: design and applications. 2nd ed., CRC Press,
zones with variable width. Compos B Eng 2014;67:410–26. Taylor & Francis Group, ISBN: 1-4200-4519-9, 2007.
[32] Michels J, Martinelli E, Czaderski C, Motavalli M. Prestressed CFRP strips with [56] Tavakkolizadeh M, Saadatmanesh H. Repair of steel bridges with CFRP plates. In:
gradient anchorage for structural concrete retrofitting: experiments and numerical Proceedings of the first international conference of advanced polymer composites
modeling. Polymers 2014;6(1):114–31. for structural applications in construction, ACIC 2002, Thomas Telford, London,
[33] Yoshitake I, Kim YJ, Yumikura K, Mimura Y. Composite strips with various anchor 2002, pp. 211–218. ISBN: 0 7277 3122 X.
systems for retrofitting concrete beams. Int J Concr Struct Mater 2011;5(1):43–8. [57] Mazzotti C, Savoia M, Ferracuti B. An experimental study on delamination of FRP
[34] Jung WY, Kwon MH, Ju BS. The flexural improvement of RC beams using an in- plates bonded to concrete. Constr Build Mater 2008;22(7):1409–21.
serted plate between concrete and FRP bonding surface. Int J Civ Environ Struct [58] Biscaia HC, Borba IS, Silva C, Chastre C. A nonlinear analytical model to predict the
Constr Archit Eng 2014;8(6):757–61. full-range debonding process of FRP-to-parent material interfaces free of any me-
[35] Wu YF, Liu K. Characterization of mechanically enhanced FRP bonding system. J chanical anchorage devices. Compos Struct 2016;138:52–63.
Compos Constr 2013;17(1):34–49. [59] Yu T, Fernando D, Teng JG, Zhao XL. Experimental study on CFRP-to-steel bonded
[36] Kalfat R. The strengthening of post-tensioned slabs using CFRP composites at White interfaces. Compos B Eng 2012;43(5):2279–89.
City, London. In: Proc., Structural Faults and Repair, 12th Int. Congress and [60] Dehghani E, Daneshjoo F, Aghakouchak AA, Khaji N. A new bond-slip model for
Exhibition (CD-ROM), Engineering Technics Press, Edinburgh, UK, 2008. adhesive in CFRP–steel composite systems. Eng Struct 2012;34:447–54.
[37] Deifalla A, Ghobarah A. Strengthening RC T-beams subjected to combined torsion [61] Fédération Internationale du Béton (FIB): Bulletin d'information n. 14. Externally
and shear using FRP fabrics: experimental study. J Compos Constr bonded FRP reinforcement for RC structures. July, 2001.
2010;14(3):301–11. [62] Biscaia H. Behaviour and modelling of GFRP-to-concrete interfaces of reinforced
[38] Biscaia HC, Chastre C, Silva MAG. Bond-slip model for FRP-to-concrete bonded concrete elements exposed to aggressive environments PhD Thesis Faculdade de
joints under external compression. Compos B Eng 2015;80:246–59. Ciências e Tecnologia, Universidade Nova de Lisboa; 2012 [in Portuguese].
[39] Biscaia HC, Chastre C, Cruz D, Viegas A. Prediction of the interfacial performance of [63] FYFE Co. LLC. Tyfo SEH-51 composite using Tyfo S Epoxy; May, 2002.
CFRP laminates and old timber bonded joints with different strengthening techni- [64] ETAG 001, Guideline for European technical approval of metal anchors for use in
ques. Compos B Eng 2017;108:1–17. concrete. Part one: Anchors in general. Edition 1997, 1st Amended November 2006,
[40] Silva, CP. Comportamento de ligações adesivas entre compósitos de FRP e ele- 2nd Amended April 2013, EOTA, 2013.
mentos estruturais de aço. MSc Dissertation in Civil Engineering, Faculty of Science [65] Cervenka V, Jendele L, Cervenka J. ATENA Program Documentation – Part 1 –
and Technology, Nova University of Lisbon, 2015. [In Portuguese]. Theory. Cervenka Consulting, Prague, May, 2016.
[41] Biscaia HC, Silva MAG, Chastre C. An experimental study of GFRP-to-concrete in- [66] Carrara P, Ferretti D, Freddi F, Rosati G. Shear tests of carbon fiber plates bonded to
terfaces submitted to humidity cycles. Compos Struct 2014;110:354–68. concrete with control of snap-back. Eng Fracture Mech 2011;78:2663–78.
[42] Biscaia HC, Chastre C, Silva MAG. Modelling GFRP-to-concrete joints with interface [67] Cornetti P, Carpinteri A. Modelling the FRP-concrete delamination by means of an
finite elements with rupture based on the Mohr-Coulomb criterion. Constr Build exponential softening law. Eng Struct 2011;33(6):1988–2001.
Mater 2013;7:261–73. [68] Wu YF, Xu XS, Sun JB, Jiang C. Analytical solution for the bond strength of ex-
[43] Biscaia HC, Silva MAG, Chastre C. Factors influencing the performance of externally ternally bonded reinforcement. Compos Struct 2012;94:3232–9.
bonded reinforcement systems of GFRP-to-concrete interfaces. Mater Struct

66

You might also like