Академический Документы
Профессиональный Документы
Культура Документы
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/266735401
CITATIONS READS
6 120
4 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Jovan Tatar on 18 September 2015.
ABSTRACT
Efforts are underway to develop standardised tests to assess the bond capacity of FRP bonded
to a concrete substrate. While several bond test configurations have been proposed and some
standard tests have been promulgated, many use difficult-to-handle specimens or require
relatively complex test apparatuses. This paper focuses on a simple beam-bending bond test
based on concrete modulus of rupture (MOR) tests, which are simple to conduct and familiar
to most in the anticipated community of users. This study addresses the following parameters
that must be properly accounted for in the development of such a standard test: a) specimen
size; b) ratio of FRP width-to-beam width; and c) ratio FRP-to-concrete modulus.
Furthermore, the study contrasts simplified and rigorous methods of calculating internal
specimen forces with measured in situ stresses in order to validate assumptions necessary in
developing a standard test method.
Keywords: bond, concrete, FRP, standard test
1. Introduction
It is well established that debonding of adhesive-applied fibre-reinforced polymer composite
(FRP) retrofit materials from their concrete substrate often is a critical limit state. A variety of
test methods intended to quantify this behaviour have been proposed and tested. Proposed
methods generally fall into variations of two fundamental forms shown schematically in
Figure 1: pull-off tests [1,2] and beam tests [3-8]. Pull-off tests are generally relatively
small specimens requiring specialized test apparatus. Being small, pull-off tests are well
suited to accelerated aging or environmental conditioning. On the other hand, the smaller size
of the specimen may not capture certain effects of scale and/or limit the geometry of the FRP
and substrate investigated. Beam specimens overcome many of the drawbacks of pull-off
specimens; significantly, they better represent actual in situ behaviour. Some researchers
have used very small test specimens (Karbhari and Engineer [3] used specimens only 25 mm
square loaded over a 200 mm span) which may be limited in their applicability, while others
have used large specimens (Harries et al. [7] used specimens having a 250 x 150 mm section
tested over a 4.5 m span) which are not practical for standardized test applications. Larger
beam tests require internal steel reinforcing and their behaviour is affected by both internal
steel and concrete compression block response. Hinged beam tests [9], shown at the bottom
of Figure 1b, overcome variation associated with concrete compression block behaviour at
the expense of more complex specimen preparation.
The objective of the present work is to promote the development of a practical standardised
test method for assessing FRP-to-concrete bond characteristics. For a test method to be
practical, and therefore universally adopted, it must a) provide a reliable and repeatable
engineering property that describes bond behaviour; b) be sufficiently compact to be easily
handled and allow repetition; and c) have a form that is both suitable for environmental
exposure and accelerated aging tests while also capturing the general geometry of in situ
applications. A test method for direct tension pull-off testing of FRP-concrete interfaces has
been recently standardised [10]. This method is primarily intended as a quality control and/or
acceptance test and does not yield values of an engineering property suitable for design.
Therefore it remains that a method for assessing the more realistic in situ bond capacity of an
FRP-concrete interface is required.
Gartner et al. [4] and Evaslage et al. [5] independently focused on the notched three point
bend test as being suitable for standardisation. Together, the authors, under the auspices of
ASTM Committee D30 Plastics, are drafting a new ASTM standard for the method
described in this paper.
applied load
FRP concrete three point P
applied bending FRP
load
FRP
concrete
four point
bending notch h
h/2
notched
beam S/2 S/2 w
L/2 L/2 b=h
hinged L = 3h
beam
a) pull-off style tests b) beam style tests Figure 2. Test specimen configuration.
Figure 1. Schematic representation
of bond test methods.
S/2 S/2
L/2 L/2
a) simplified calculation (Eq. 3) b) rigorous calculation (Eq. 4)
Figure 3 Internal stresses in beam specimen.
a 2 2 0.5 (5)
Where is the ratio of axial stiffnesses of the FRP and concrete:
= K*w/Ecbh (6)
Where Ec is the modulus of elasticity of the concrete. Bond stress is then calculated from Eq.
2. While accounting for the more realistic behaviour of the beam specimen, this formulation
requires the moduli of both the FRP and concrete to be known. In some instances, this may
not be possible in so far as environmental conditioning may affect one or both of these
values. In both formulations (Eqs 3 and 4) concrete is assumed to remain linear in
compression. In general, this assumption is considered valid for concrete stresses below about
0.6fc. Perfect bond between FRP reinforcement and concrete is also assumed for both
methods, which generally does not hold true as the ultimate capacity of the specimen is
reached.
4. Experimental Program
An experimental program was undertaken to assess the performance of the proposed
standardised test over a range of anticipated parameters. Parameters considered include a) the
use of 10-series (b=h=102mm) and 15-series (b=h=152mm) and cross-sections; b) FRP type;
c) FRP width; and d) concrete compressive strength. Varying these parameters fully evaluate
the terms in the equations for the rigorous method (Eq. 4, 5, and 6) and will provide the range
of conditions that will be encountered in bond testing. Varying both the FRP type and width
results in a significant range of values of FRP stiffness, K*w to be considered. Table 1
provides specimen identification, details and relevant material properties for all specimens
tested. The notches were provided by saw cutting resulting in a kerf of approximately 2 mm
in all cases. Three repetitions of each 15-series case were tested. The adhesive used had
manufacturers specified tensile and shear capacities of 24.8 MPa.
Table 1. Specimen identification, details and relevant material properties.
ID h = b = L/3 f c Ec 4725 fc ' FRP K* w S
mm MPa GPa N/mm mm mm
10-C1 102 28.0 25.0 1.02 mm CFRP fabric 71,680 25 203
10-C2 102 79.0 42.0 1.02 mm CFRP fabric 71,680 25 203
10-G1 102 28.0 25.0 1.02 mm GFRP fabric 26,540 25 203
10-G2 102 79.0 42.0 1.02 mm GFRP fabric 26,540 25 203
15-C1 152 32.0 26.1 1.65 mm CFRP strip 256,000 25 380
15-C2 152 32.0 26.1 1.65 mm CFRP strip 256,000 50 380
15-G1 152 32.0 26.1 1.90 mm GFRP strip 78,600 25 380
15-G2 152 32.0 26.1 1.90 mm GFRP strip 78,600 50 380
15-G3 152 32.0 26.1 1.90 mm GFRP strip 78,600 75 380
It is noted that there is a practical limit to the amount of FRP that may be applied to the
otherwise unreinforced beam specimens. If too much FRP is applied to the beams, the
behaviour changes from FRP debonding to concrete shear, which is not of interest in this test.
Thus the reinforcement quantity should be limited so that the ultimate capacity of the
specimen is lower than that required to cause a shear failure, which may be estimated as
0.33 f c 'bh (MPa and mm).
5. Test Procedure
All beams dimensions were measured in order to permit actual dimensions, rather than the
nominal dimensions reported in Table 1, to be used for subsequent calculations. All beams
were tested in three-point flexure over a span of 3h as shown in Figure 2. Load is applied at a
constant rate until failure. The load rate is established to result in an increase in FRP bond
stress of between 0.4 and 0.8 MPa/min; in general, this results in failure in less than three
minutes. Strain gages were applied to the FRP to definitively establish the FRP stress (Eq. 1).
Additional gages were arrayed vertically above the notch to establish the shape of the
concrete strain profile, depth of the neutral axis, and to verify that the concrete compressive
strains remain in the elastic range.
6. Test Results and Discussion
Table 2 summarises the applied load (P) and FRP strain (FRP) at ultimate capacity obtained
from the tests. The ultimate FRP strain is used in Eq. 1 to calculate the experimental average
bond strength (exp). This bond stress can only be considered average since a single strain
gage was used to measure FRP strain at the notch. Table 2 also reports the peak recorded
concrete strain at a location 12.5 mm below the extreme compression fibre (c) and the
location of the neutral axis (, see Figure 3b) determined from strain data. Finally,
experimental bond strength is compared to the calculated values produced by both the simple
and rigorous formulations for determining the average bond strength () presented previously.
The experimental bond strength was also determined using an alternate method. sp was
calculated using the measured depth of the compression stress block in Eq 4. Theoretically
these two methods of measurement will produce the same result. In nearly all cases,
however, exp is less than sp, and in a few cases they are comparable. This difference may be
due to the effect that the progressive debonding has on the distribution of the strain across the
width of the reinforcement (see Section 6.1). exp is based on measurement of strain at a
single point on the FRP reinforcement while sp is based on the total force in the FRP
reinforcement calculated from the applied load using equilibrium. Given that exp is a more
direct measure of force in the FRP reinforcement, it is compared to both the simplified and
rigorous methods in the form of the ratios shown in Table 2.
Both the simple and rigorous methods described by Eqs 3 and 4, respectively, overestimate
the experimentally observed average bond stress calculated using strain data (Eq. 1). For the
152-mm deep specimens, the average ratio of experimental-to-predicted debonding stress is
0.67 for the simple method (Eq. 3) and 0.75 for the rigorous method (Eq. 4). Similar ratios for
the 102-mm deep specimens are 0.80 and 0.94, respectively. The test results indicate that
both the simplified and rigorous methods provide unconservative estimates of the average
bond stress. If the rigorous method is compared the alternative bond strength measurement,
however, the results are more encouraging (Table 3). For the 10-series, the average ratio is
0.99 while for the 15-series the average ratio is 1.03, indicating excellent agreement for both
series.
The application of FRP to the notched beams results in a strengthening effect proportional to
the FRP stiffness. The average load carrying capacity (P in Table 3) of notched 152 mm
beams having no FRP was 5930 N. As is typically observed, as additional FRP is added, the
incremental strengthening effect is reduced. While not approached in this test program, this
effect is limited by the shear capacity of the beam specimens. Bond stress results obtained
from the 152-mm deep specimens are generally lower than anticipated for a good FRP
application. The failure of most of these specimens was characterised as an adhesive failure
between the FRP and adhesive. This failure suggests improper preparation of the FRP strips
prior to installation. In addition, 10-C1-A and 10-G1-A had lower concrete strength but had
significantly larger ultimate bond capacities than companion specimens having higher
strength concrete. Following testing it was determined that the surface preparation of the
higher strength concrete specimens resulted in insufficient roughness, which led to adhesive
failure modes at lower loads. Nonetheless, poor bond performance does not affect the intent
or calculations presented here; indeed, if the test method were used to assess environmental
conditioning, poor bond results may very well be expected.
Table 2. Test results and calculated values of FRP stress.
experimental simplified rigorous
exp sp sim rig t exp t sp
ID P FRP c t exp Eq. 5
and 4 t rig t rig
Eq. 1 Eq.4 Eq. 3
t sim Eq. 5
N MPa MPa MPa MPa
10-C1-A 20515 9290 6.55 - - - 7.27 0.90 0.111 6.29 1.04 -
10-C2-A 12846 5280 3.73 -480 0.07 3.89 4.55 0.82 0.089 3.91 0.95 0.99
10-G1-A 14359 15300 4.00 -1700 0.04 4.30 5.09 0.79 0.069 4.34 0.92 0.99
10-G2-A 8140 7830 2.05 -530 0.05 2.45 2.88 0.71 0.055 2.45 0.84 1.00
15-C1-A 18326 - - - - - 3.35 - 0.133 2.97 - -
15-C1-B 14803 960 1.29 -142 0.17 2.48 2.72 0.47 0.133 2.40 0.54 1.03
15-C1-C 17681 - - -791 0.25 3.05 3.18 - 0.132 2.87 - 1.06
15-C2-A 32555 1650 2.22 -439 0.18 2.73 3.00 0.74 0.183 2.69 0.83 1.01
15-C2-B 28138 1499 2.02 -459 0.14 2.33 2.57 0.79 0.183 2.32 0.87 1.00
15-C2-C 24139 1094 1.47 -408 0.14 2.00 2.20 0.67 0.183 1.99 0.74 1.00
15-G1-A 9238 1956 0.81 - - - 1.65 0.49 0.075 1.47 0.55 -
15-G1-B 10795 2161 0.89 -530 0.22 1.84 1.96 0.45 0.076 1.72 0.52 1.07
15-G1-C 10253 2545 1.05 - - - 1.88 0.56 0.076 1.63 0.64 -
15-G2-A 17983 3410 1.41 -759 0.15 1.49 1.65 0.85 0.107 1.45 0.97 1.03
15-G2-B 19460 2424 1.00 -695 0.12 1.60 1.76 0.57 0.105 1.56 0.64 1.03
15-G2-C 16671 2019 0.84 - - - 1.55 0.54 0.107 1.34 0.63 -
15-G3-A 27462 3925 1.62 -968 0.21 1.55 1.68 0.96 0.129 1.48 1.09 1.05
15-G3-B 25727 2991 1.24 - - - 1.60 0.78 0.130 1.39 0.89 -
15-G3-C 27324 3069 1.27 -314 0.11 1.49 1.65 0.77 0.127 1.47 0.86 1.01
concrete
-0.2 15-G2-A 15-G2-B -0.2
proportion of beam height, a
-0.6 -0.6
h/2 sawcut
h/2 sawcut
-0.7 -0.7
-0.8 -0.8
0.4P
-0.9 -0.9
FRP FRP
-1 -1
-2000 -1000 0 1000 2000 3000 4000 5000 -1000 -500 0 500 1000 1500 2000 2500
measured strain (me) measured strain (me)