Anda di halaman 1dari 20

Non-destructive Testing and Evaluation, 2015

http://dx.doi.org/10.1080/10589759.2015.1046872

I-SonReb: an improved NDT method to evaluate the in situ


strength of carbonated concrete
Marco Breccolotti* and Massimo F. Bonfigli

Department of Civil and Environmental Engineering, University of Perugia, PerugiaI-06125, Italy


(Received 5 December 2014; accepted 22 April 2015)
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

Concrete strength evaluated in situ by means of the conventional SonReb method can
be highly overestimated in presence of carbonation. This latter, in fact, is responsible
for the physical and chemical alteration of the outer layer of concrete. As most of the
existing concrete structures are subjected to carbonation, it is of high importance to
overcome this problem. In this paper, an Improved SonReb method (I-SonReb) for
carbonated concretes is proposed. It relies on the definition of a correction coefficient
of the measured Rebound index as a function of the carbonated concrete cover
thickness, an additional parameter to be measured during in situ testing campaigns. The
usefulness of the method has been validated showing the improvement in the accuracy
of concrete strength estimation from two sets of NDT experimental data collected from
investigations on real structures.
Keywords: concrete strength; existing structures; non-destructive techniques; SonReb
method; carbonation

1. Introduction
It is well known [1] that the phenomena of concrete carbonation significantly reduce
the accuracy of the concrete strength estimation by means of methods based on the
measurement of the surface hardness such as the rebound hammer test. In fact, carbonation
is responsible for a possible over-estimation of the concrete strength as it could return
rebound numbers up to 50% higher than those that would be obtained in non-carbonated
concretes.[2]
Carbonation is a chemical process where CO2 reacts with H2 O and with CaOH2 or
CZSZH to form Calcium Carbonate CaCO3 . This reduces concrete porosity, resulting in
an increase in both concrete strength and stiffness and in a reduction of concrete pH. The
physical and mechanical characteristics of carbonated concretes were experimentally
evaluated, among others, by Jerga [3] and Chang and Chen [4].
The presence of carbonation is generally not negligible in old reinforced concrete (RC)
structures, while it is just for these buildings that the evaluation of the load bearing
capacity to imposed and seismic loads is most frequently required. For instance, De Luca
et al. [5], investigating a 70-year-old building, found an average carbonation depth of
50 mm. Similar values have been found by Huang et al. [6] in several interior and exterior
columns and in some beams of a 35-year-old RC building. While there are currently
research efforts for increasing the reliability of concrete strength estimation through NDT
methods,[7 10] it is the Authors opinion that the influence of carbonation in the
assessment of existing RC buildings is still not taken into due consideration. For instance,

*Corresponding author. Email: brec@unipg.it

q 2015 Taylor & Francis


2 M. Breccolotti and M.F. Bonfigli

Kim [11] proposed a correction coefficient based on the age of the concrete irrespectiveof
the effective depth of carbonation, while other researchers used only the surface rebound
method [12] to assess the concrete compressive strength neglecting the presence of
carbonation.
According to Breysse [13], a possible improvement on the accuracy of NDT estimates
must be searched by combining different NDT techniques. The method proposed by the
Authors is a step towards this direction since the traditional SonReb method is combined
with the carbonation assessment method to take into account quantitatively the effect of
concrete surface carbonation.
In their previous work [14], Authors investigated the effect of carbonation on the
SonReb results modelling in a FEM environment the impact of the rebound hammer
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

plunger on the carbonated concrete cover. That analysis allowed the definition of a
correction coefficient of the measured rebound index Qxc as a function of the carbonated
concrete layer thickness xc :
Q0
kc xc ; 1
Qxc
where the rebound index Q was defined as the ratio between the post- and pre-impact
velocities of the mass on the plunger of the rebound hammer and Q0 is the expected
value of the rebound number for non-carbonated concrete.
The coefficient kc xc was expressed in the following general form:
xc =b3
kc xc b1 eb2 e : 2
In that paper, the numerical values of coefficients b1 ; b2 ; b3 were obtained analysing
the results of FE analysis and comparing these latter with experimental data gathered from
several literature references.
In this paper, Authors will provide a better definition of the corrective coefficient kc xc
and will show the benefits in terms of accuracy improvement in the concrete strength
estimation by means of the Improved SonReb (I-SonReb) method achieved taking into
account the carbonated concrete thickness together with the classical Rebound index R and
ultrasonic pulse velocity UPV values V.
As will be shown, carbonation mainly affects the rebound hammer measurements, but
it also has, on a lesser extent, an influence on the measured ultrasonic pulse velocities.
Thus, in the first part of the paper, Authors will evaluate the influence of
carbonation on the measured UPV values to decide if a proper correction has to be
applied to them.
In the second part of the paper, the Improved SonReb method (I-SonReb) is presented.
The application of the proposed method to 2 NDT data-sets from tests on real structures
allowed to highlight the increased accuracy achieved on the concrete strength evaluation.
Finally, in the last part, the steps necessary to carry out I-SonReb tests are summarised
and a review of possible methods to assess the in situ carbonated concrete thickness is
mentioned.

2. The SonReb method


The idea of correlating the concrete strength f c simultaneously to the Rebound number R
and to the ultrasonic pulse velocity V, thus originating the SonReb method, dates back to
the early 1970s, thanks to Facaoaru.[15] Up to now, different correlation formulae,
ranging from linear, polynomial, exponential, etc., have been proposed. Among them, one
Non-destructive Testing and Evaluation 3

of the most widely accepted correlation surfaces for the SonReb method available in the
literature has the following general form:

f c V; R a1 V a2 R a3 3

Several values for the parameters a1 ; a2 ; a3 can be found in the bibliographical


references (e.g. [16] [18]). These relations, referred to the cube compressive strength Rc ,
are summarised in Table 1. The existence of several different correlation equations is due
to the fact that there are many factors influencing the relation among V, R and Rc , so that a
generally valid expression cannot be obtained. It is thus mandatory, as suggested by many
Authors and Technical Standards (e.g. [19]), the calibration of the parameters of Equation
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

(3) for the specific concrete being investigated by a suitable number of destructive testings.

3. Influence of carbonation on SonReb estimation


While the effect of carbonation on the rebound number has been already addressed by the
Authors,[14] its influence on UPV values still needs to be properly investigated. As known
well, a compression wave travels inside a homogeneous and isotropic material, like in this
work concrete is considered to be, with the velocity V:
s
1 2 nd Ed
V ; 4
1 nd 1 2 2nd r

where nd and Ed are, respectively, the dynamic Poissons ratio and the dynamic elastic
modulus of the material.
According to Equation (4), concrete carbonation affects the UPV measurements, since
it produces an increase in concrete elastic modulus Ed ; that is, nevertheless, partially
counterbalanced by the increase in the material density r. Besides, according to Pham and
Prince [20], carbonation does not induce any changes on the concrete Poissons ratio.
However, there is no consolidated literature on the influence of concrete carbonation on
this parameter, thus a sensitivity study has been carried out.
In order to investigate the effects of carbonation on the measured UPV values, the
following assumptions have been made:
(1) The ultrasonic emitter and receiver probes are located in front of each other,
placed on the opposite faces of the concrete specimen (direct transmission layout),
thus at a distance S equal to the specimen width.
(2) Under the fully carbonated layer, there is a further layer of partial carbonation
whose thickness is equal, if physically possible, to that of the fully carbonated layer.
(3) The faces of the specimen are parallel and so are the surfaces that separate the fully
carbonated layers from the partially carbonated ones and those that separate the
latter from the inner, non-carbonated, concrete.

Table 1. Examples of SonReb correlation equations found in the literature.


Authors Correlation surface
Di Leo and Pascale [16] Rc V; R 1:201029 V 2:446 R 1:058
Gasparik [17] Rc V; R 8:061028 V 1:85 R 1:246
RILEM [18] Rc V; R 9:2710211 V 2:6 R 1:4
4 M. Breccolotti and M.F. Bonfigli

Under these assumptions, the problem is simplified but, at the same time, these
hypotheses resemble the direct transmission configuration, which is the most frequent and
reliable configuration to carry out in situ UPV measurements. As a consequence, the
results that have been obtained could be applied to the majority of real world situations.
In details, hypotheses 1 and 2 allow to know, a priori, which is the fastest path the
ultrasonic waves will move along to reach the receiver. Without hypothesis 1, the path
could be different from the straight one connecting the emitter and the receiver, depending
on the travelling speeds in the different materials. Moreover, without hypotheses 1 and 3,
the effects of wave diffraction that depends on the incident angle of the waves on the
interfaces between the different layers should also be considered. Hypothesis 2 is also used
to take into account the experimental results obtained by Chang and Chen [21] regarding
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

the presence in the concrete of a zone of partial carbonation that cannot be measured by
means of phenolphthalein solutions.
The aforementioned hypotheses lead to a problem, the geometry of which is shown in
Figure 1. In this figure, S is the thickness of the specimen, xc;tot xc;a xc;b the total
thickness of the carbonated layers and xpc;tot xpc;a xpc;b # xc the total thickness of the
partially carbonated layer.

Figure 1. Geometry of the UP direct transmission layout.


Non-destructive Testing and Evaluation 5

We define as v0 ; vpc ; vfc , the speeds of ultrasonic pulses (UP) into non-carbonated,
partially carbonated and fully carbonated concretes, respectively.
Obviously from the geometry of the problem and hypothesis 3, only the following two
conditions can be met:
(a) xc;tot # S=2 and xpc;tot xc;tot :
(b) S=2 , xc;tot # S and xpc;tot S 2 xc;tot :
The condition (a) corresponds to the situation in which is still present a non-carbonated
concrete core into the RC element, while condition (b) corresponds to the situation where
the entire specimen is fully or partially carbonated.
The physical and mechanical properties of the three materials (fully carbonated,
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

partially carbonated and non-carbonated concretes) have been taken, for a specific
concrete strength, from the works by Jerga [3] and Chang and Chen [4]. In Table 2, the
values of densities r, static elastic modulus Es and static Poissons ratio n are reported.
From Equation (4), the ratio between V fc and V 0 can be expressed as:

s s
V fc 1 2 nd;fc 1 nd;0 1 2 2nd;0 Ed;fc r0
 ; 5
V0 1 nd;fc 1 2 2nd;fc 1 2 nd;0 Ed;0 rfc

where the values with the subscript 0 are referred to the non-carbonated concrete and those
with the subscript fc refer to the carbonated concrete. Assuming that nd;0 0:15, the first
factor at the second member of Equation (5) takes values ranging between 0.984 and 1.026
if nd;fc [ 0:1; 0:2. It can be concluded that reasonable changes of the value of nd for a
carbonated concrete with respect to the value of the non-carbonated one do not affect in a
significant manner the mutual ratio between the travelling speeds in the various layers.
This observation allows to consider nd as a constant value regardless of the carbonation
development in the concrete.
Several relations between the static and dynamic elastic modulus of concrete can be
found in the literature. In this paper, the provision of the British Standard BS 8110-2 [22]
has been used:

Es 19
Ed 6
1:25

with Es and Ed measured in GPa.


With the aforementioned assumptions, the UPV into the three types of concrete can be
calculated using Equation (4). The obtained values are given in Table 3.

Table 2. Physical and mechanical properties of a concrete with different degrees of carbonation
obtained from data available in [3] and [4].

Static El. modulus


Concrete type Density rkg=m3 Es MPa Static Poissons ratio n
Non-carbonated concrete 2300 17,967 0.15
Partially carbonated concrete 2325 20,617
Fully carbonated concrete 2400 26,946
6 M. Breccolotti and M.F. Bonfigli

Table 3. Estimated UPV through the non-carbonated, and partially and fully carbonated concretes.

Ultrasonic pulse velocity (m/s)


V0 3685
V pc 3794
V fc 4021

These values refer to a specific concrete having a compression strength equal to


24.2 MPa, but it can be reasonably assumed that this trend will be similar for concretes of
different strengths.
The correction coefficient kv S; xc;tot that should be applied to the measured UPV
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

value V m to determine the ultrasonic pulse velocity V 0 that would have been measured in
absence of carbonation can be expressed as:
V0
kv S; xc;tot : 7
Vm
In order to find the expression of kv S; xc;tot , the aforementioned cases (a) and (b) are
addressed.
Case (a): xc;tot # S=2 In this situation, the ultrasonic wave travels through a thickness
of fully carbonated and partially carbonated layers equal to xc;tot for both of them, and it
travels as well through a layer of non-carbonated concrete of a thickness equal to
S 2 2xc;tot . The average travelling speed V m is thus given by the following equation:

1
Vm x   : 8
c;tot
S
1
V fc V1pc 2 V20 V10

The correction coefficient kv S; xc;tot is given in this situation by:


 
V 0 V 0 xc;tot 1 1 2
kv S; xc;tot 2 1: 9
Vm S V fc V pc V 0

Case (b): S=2 , xc;tot # S In this case, the ultrasonic wave travels through layers of
completely carbonated concrete of thickness xc;tot and through a layer of partially
carbonated concrete of thickness S 2 xc;tot . The average travelling speed V m is given by
the following equation:
1
Vm x   10
c;tot
S
1
V fc 2 V pc V pc
1 1

and the correction coefficient kv S; xc;tot is equal to:


 
V 0 V 0 xc;tot 1 1 V0
kv S; xc;tot 2 11
Vm S V fc V pc V pc

Obviously, the lower limit for kv S; xc;tot is obtained when xc;tot S:


 V0
kv;min kS; xc;tot xc;tot S 12
V fc
Non-destructive Testing and Evaluation 7
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

Figure 2. Variation of kv values depending on the carbonated concrete thickness xc;tot and on the
total depth of the element S.

In Figure 2 is plotted a chart obtained inverting Equations (9) and (11), whose lines
provide the value of the coefficient of correction kv for a couple of known values S and
xc;tot . The lower line is the one associated with kv;min .
It can be observed that the maximum error induced by carbonation on ultrasonic pulse
velocity is expected to be approximately 8% if the whole concrete depth is carbonated, a
condition that occurs only in very old and very exposed concrete elements such as, for
instance, old RC bridge girders. For most structures, the total carbonated thickness xc;tot is
expected to be less than 100 mm while the depth of the element is expected to be greater
than 300 mm (red shaded area in Figure 2). In these situations, the maximum error
produced by the different UP velocities in carbonated concrete is small, roughly equal to
3%. Although this observation has been drawn for a specific concrete, it is the Authors
opinion that similar results would be obtained for every other concrete. For these reasons,
the correction on the UP velocity will be ignored in the Improved SonReb method. The
values of the corrective coefficient shown in Figure 2, if necessary, could be used to tackle
the influence of carbonation on UPV values.

4. Definition and validation of the I-SonReb method


The aim of this section is to investigate, by means of experimental data, the usefulness of
Equation (2) in increasing the accuracy of the in situ concrete strength evaluation taking
into account the carbonated concrete thickness as additional input data in the conventional
SonReb correlation formula.
While the Q value used in Equation (1) is defined as the ratio between the velocity just
after and just before the impact, in this work rebound hammer measurements refer to the
traditional R index, which is a function of the distances travelled by the impacting mass.
8 M. Breccolotti and M.F. Bonfigli

In detail, this value is defined as the ratio between the distances travelled by the mass
during its motion along the rod after and before the impact with the plunger in contact with
the concrete. This choice is due to the fact that, up to now, almost all the available data
come from testings on existing structures carried out with the traditional rebound
hammers. Theoretically, the Q and R indexes should coincide, but in real-world situations,
they slightly differ for the action of friction between the moving mass and the rod and for
the action of gravity that tends to brake or to accelerate the mass depending on the
inclination of the rebound hammer. It is thus assumed that Equations (1) and (2) are also
valid for the R index.
The expression of the correction coefficient kc given by Equation (2) has been further
improved by taking into consideration that kc should be equal to 1 if xc 0. With this
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

condition, the number of coefficients to be determined is reduced and the following


expression is obtained:
xc =b2
kc xc 1 2 eb1 eb1 e 13

Combining Equation (13) with Equations (3) and (1) (written in terms of R value), the
following relation that define the I-SonReb method is obtained:
xc =b2
f c R; V; xc a1 R1 2 eb1 eb1 e a2 V a3 14

Equation (14) is a function of five unknown parameters that have to be fitted through
non-linear regression analysis on experimental data coming from tests on homogeneous
concrete.
To validate the proposed method, two data-sets collected in situ on existing structures
have been analysed.

4.1 Experimental investigations


The proposed correlation formula has been tested with experimental data taken from real
structures. A total of 131 records were available from experimental investigations carried
out on different buildings and bridges.
Each data record is composed by the measured ultrasonic pulse velocity V, rebound
index value R, effective concrete compressive strength f c evaluated by compression
tests on drilled cores and the carbonated layer thickness xc measured by means of a
phenolphthalein solution on the extracted cores. As stated in [2] the internal moisture
condition of the concrete can significantly alter the measured rebound index values. It is
however recognised that the analysis and correlations carried out in the next paragraphs
have been performed on concretes having similar moisture conditions because they are
exposed to the same ambient conditions. The same applies for what it concerns the
influence of moisture condition on the UPV measurements. If different moisture
conditions are encountered in the investigated points, suitable countermeasures should be
taken to correct this effect.
The theoretical weak influence of concrete carbonation on the UPV stated in par. 3 can
be inferred observing Figures 3 and 4, where the unspecific experimental data mentioned
above are represented.
In Figure 3, R-values are plotted against f c values for carbonated (blue diamonds) and
non-carbonated (red points) concretes. The two plotted curves have the form f c aR b
and they interpolate the data distribution for carbonated and non-carbonated concretes
Non-destructive Testing and Evaluation 9

70 Noncarbonated concrete
Noncarb. concr. fitting curve
Carboneted concrete
60 Carb. concr. fitting curve

50
fc(MPa)

40
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

30

20

30 35 40 45 50 55 60
Rebound Index (R)

Figure 3. Measured R values related to f c values obtained from compression tests on concrete
cores.

separately. It can be noticed that carbonated concretes tend to show R values greater than
that of non-carbonated concretes having the same concrete strength f c . From this image, it
is also clear that the correlation between compressive strength and rebound index is
generally very weak. New technologies have been recently introduced to improve the

70
Noncarbonated concrete
Noncarb. concr. fitting curve
Carboneted concrete
60 Carb. concr. fitting curve

50
fc(MPa)

40

30

20

2500 3000 3500 4000 4500


UPV(m/s)

Figure 4. Measured V values related to f c values obtained from compression tests on concrete
cores.
10 M. Breccolotti and M.F. Bonfigli

accuracy of rebound hammer measurements, such as instruments that can evaluate more
precisely the rebound index Q through the measurement of velocities rather than distances.
Nevertheless, this improvement is not sufficient to eliminate the great dispersion of the
rebound hammer data. For this reason, rebound index values should always be used
together with UPV values since some of the weakness of rebound hammer test do not
affect UPV measurements and vice versa, so that an overall accuracy improvement is
realised by the combination of these two NDT methods.
Ultrasonic pulse velocities V measured on the same concretes are plotted against f c
values in Figure 4. The red and blue markers have the same meaning of those in the
previous figure. In this case, the red and blue curves have the form f c cV d and
interpolate data coming from non-carbonated and carbonated concretes, respectively. As
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

expected, the data distribution confirms that the influence of carbonation on UPV values is
smaller than the one observed on R values with interpolating curves for the carbonated and
non-carbonated concretes very close to each other.
The whole database available to Authors, shown in Figures 3 and 4, was a collection of
values coming from different structures and different concretes. After a first general
analysis of the whole data, in order to perform further relevant investigations, the
extraction of homogeneous sets from it was necessary. Thus, the database was divided into
subsets, where each of them was associated with a given structure. In order to ensure the
greatest statistical significance, the following analyses were performed on the two largest
subsets.
The numbers of available records are 34 and 10 for the first and the second data-sets,
respectively. They include cases coming from both carbonated and non-carbonated
concretes, with the former ones showing values of xc up to 80 mm. According to the
findings reported in par. 2, the weak influence of carbonation on the UPV values is
neglected. The application of the proposed correlation formula to these data-sets is
described in the next paragraphs.

4.2 Data-set 1
The NDT measurements of the Data-set 1 come from a ten-floor RC building located in
Rome investigated in 2013. The tests were carried out only on the RC columns. One of
these elements is shown in Figure 5 with the location of the extracted core and the grid to
carry out the Rebound and the UPV tests. The numerical values of the ND parameters are
reported in Table 4.
Some values may seem slightly anomalous, for example, samples 3 and 12, but a
certain scatter is typical of the rebound hammer and UPV tests, hence no filtering was
applied to the input data.
The following four different correlation Equations (15a 15d) have been fitted to the
experimental data:

f c a1 R a2 15a

f c a1 V a3 15b

f c a1 R a2 V a3 15c

xc =b2
f c a1 R1 2 eb1 eb1 e
a2 V a3 15d
Non-destructive Testing and Evaluation 11
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

Figure 5. RC column investigated in Data-set 1: the location of the extracted core and the grid to
carry out the Rebound and the UPV tests are visible.

which namely represent the rebound hammer test, the ultrasonic pulse velocity test, the
classical SonReb method and the I-SonReb method. The Levenberg Marquardt nonlinear
least squares algorithm [23] implemented in the MatLab environment has been used as
fitting algorithm.
To evaluate the goodness of fit, it is a common practice the use of synthetic parameters
such as the coefficient of determination value R 2 :
Pn  2
yi 2 y^ i
R 2 1 2 Pi1
n  2 ; 16
i1 yi 2 y

where n is the sample size, yi the observed data, y their mean value and y^ i the predicted
values.
The coefficients in Equation (15) and the R 2 values obtained from fitting on the
available data are summed up in Table 5.
From these results, some conclusions can be drawn. First of all, it can be observed, as
expected, that increasing coefficients of determination have been obtained switching from
the rebound method to the UPV technique to the SonReb method. A further increase (18%)
in the R 2 value has been achieved using the I-SonReb method. The increased goodness of
fitting can also be argued from Figure 6 where the experimental compressive strengths are
compared with the predicted values obtained using the classical SonReb method and the
12 M. Breccolotti and M.F. Bonfigli

Table 4. Data-set n.1.

Sample no. V (m/s) R xc (mm) fc (MPa)


1 4053.5 39.3 0 45.2
2 4202.7 42.9 0 45.7
3 4416.1 45.6 11 30.7
4 4070.5 45.2 0 46.3
5 3628.6 41.2 10 29.5
6 4070.8 48.6 10 45.3
7 4339.9 48.8 0 58.6
8 4313.9 49.2 0 60.5
9 3916.7 45.8 0 37.7
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

10 3997.9 43.8 10 32.5


11 3768.4 47.3 10 36.7
12 3426.7 53.0 10 45.6
13 4280.1 50.9 10 54.1
14 4065.1 44.0 10 43.7
15 3662.0 44.4 0 40.1
16 3640.1 44.0 0 36.2
17 4129.6 49.0 0 59.8
18 4020.2 39.3 0 42.6
19 4376.2 46.3 5 46.8
20 4026.5 36.9 0 44.2
21 4310.8 47.7 0 52.4
22 3257.9 42.8 10 29.7
23 3985.6 44.9 11 51.2
24 3737.3 44.7 8 39.1
25 3558.6 42.8 0 35.8
26 4058.5 46.1 5 49.8
27 3826.7 41.2 0 42.3
28 3666.5 39.1 0 37.1
29 3168.2 39.3 20 26.5
30 3923.4 41.2 0 36.4
31 3623.1 35.8 0 28.4
32 3610.0 46.9 10 38.9
33 4105.3 48.2 7 50.7
34 4195.4 46.8 0 56.6

proposed method. Are, in fact, evident the beneficial corrections applied to most of
the estimated strengths of carbonated concretes taking into account the effect of the
carbonation depth. Just as an example, test 3 shows a reduction in the prediction error from
20% to 10% and test 21 goes from 10% error down to approximately 1%. The increased

Table 5. Fitting data-set n.1.

Correlation formula a1 a2 a3 b1 b2 R2 sres


f c a1 R a2 0.162 1.469 0.356 7.565
f c a1 V a3 1.123 1025 1.831 0.457 6.945
f c a1 R a2 V a3 9.043 1026 1.036 1.381 0.598 6.068
f c ax1=bR1 2 eb1 3.367 1025 1.360 1.082 21.794 20.818 0.713 5.309
b1 e c 2 a2
e V a3
Non-destructive Testing and Evaluation 13

65
Strength of extracted cores 0
0
60 Predicted strength with SonReb method 0
Predicted strength with ISonReb method 0
0 10
55
11 5
7 11
5
10
50 0
0
0
0 10
fc(MPa)

10 0
45 10 10
0 0 0
0 0
10 8
40 0
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

0
10
35
0 10

30 20

25

5 10 15 20 25 30
Test no.
2
Predicted strength measured

1.8 Normalized strength


Predicted strength using SonReb method
1.6 Predicted strength using ISonReb method
strength ratio

1.4
1.2
1
0.8

0 5 10 15 20 25 30 35
Test no.

Figure 6. Data-set 1: experimental and predicted compressive strengths of concrete samples using
the classical SonReb method and the I-SonReb method. The numbers indicate the carbonation depths
in millimetres.

goodness of fitting achieved using the I-SonReb method is also confirmed observing the
values of the standard deviation of the residuals sres , also shown in Table 5, defined as:
sP
Y 2 Y est 2
sres ; 17
n2p
where Y are the observed values, Y est the predicted values, n the number observations and
p the number of the parameters of the model.

4.3 Data-set 2
The NDT measurements of the Data-set 2 come from an experimental campaign carried
out in 2011 on several piers and cap beams of a viaduct. The numerical values of the ND
parameters are reported in Table 6, while the fitted coefficients in Equation (15) and the
14 M. Breccolotti and M.F. Bonfigli

Table 6. Data-set n.2.

Sample no. V (m/s) R xc (mm) fc (MPa)


1 3637.8 46.0 55 26.6
2 4388.5 45.1 35 34.2
3 3773.0 45.9 30 39.0
4 3999.7 41.3 20 40.3
5 4914.9 39.6 40 45.0
6 4315.0 43.0 25 45.7
7 4257.1 45.6 15 47.2
8 4414.0 45.4 20 57.0
9 4190.9 49.7 25 60.6
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

10 4103.8 35.7 80 22.9

corresponding R 2 and sres indexes are listed in Table 7. In this data-set, the measured
carbonation thickness, in the range of 15 80 mm, is higher than that observed in the
previous set. As it can be observed, for instance, in Figure 7, the measured carbonated
thickness is uniformly affected by carbonation.
In this case, the increased accuracy obtained using the proposed correlation formula is
also evident. A 28% increase in the R 2 value has been obtained using the I-SonReb method
in comparison to the classical SonReb correlation formula.
Data-set 2 has very few samples, so that it may be inferred that the increase in R 2 is
spurious since it could be due to the increase in the number of regressors. Thus, for this set
2
of data, the values of the adjusted coefficient of correlation R were also computed, which
is defined as:

2 n21
R 1 2 1 2 R 2 18
n2p21

where n is the number of samples and p the number of regressors.


2
For Equation (15c), R 0:64 was obtained, while for the improved SonReb method
2
given by Equation (15d), a value of R 0:717 has been obtained. As a non-negligible
increase has been achieved also for this parameter, it means that even if the sample is small
in size, the better fitting is unlikely to be due by chance, but it is related to the influence of
carbonation being kept into consideration (Figure 8).

4.4 General considerations


The improvements achieved on the R 2 and sres values show that taking into consideration
the carbonated layer thickness in the in situ concrete strength evaluation through Equation
(14) can effectively help increasing the accuracy of the assessment carried out using the
SonReb method.

Table 7. Fitting data-set n.2.

Correlation formula a1 a2 a3 b1 b2 R2 sres


f c a1 R a2 0.047 1.793 0.304 10.615
f c a1 V a3 5.754 1024 1.342 0.178 11.537
f c a1 R a2 V a3 7.873 1028 2.037 1.486 0.635 8.178
f c ax1=bR1

2 eb1 5.063 1026 1.294 1.363 20.835 47.174 0.813 7.105
eb1 e 2 a2 V a3
c
Non-destructive Testing and Evaluation 15
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

Figure 7. Cap beam investigated in Data-set 2 (up) and the phenolphthalein test on the extracted
core (bottom).

Further consideration can be made regarding the coefficients b1 and b2 of Equation (13).
The two couples of values reported in Tables 5 and 7 produce corrective coefficients kc
whose trends are shown in Figure 9. The corrections applied by means of these coefficients
are very reasonable. For the first data-set, with a maximum carbonation thickness of 20 mm,
the reduction in the rebound index is small, approximately equal to 18%, for carbonation
depth grater than 22 mm. In the second data-set, relating to very exposed RC elements, the
carbonation depths reached values of 80 mm. In this case, a more severe reduction in the
rebound index has been identified with a corrective coefficient kc equal to 0.57 for
carbonation depths of 80 mm. It must be observed that the first data-set has values of
carbonation depths ranging from zero up to 20 mm, so the correction coefficients above this
value are just an extrapolation not supported by experimental data and no comparison can be
made with the one predicted using the second data-set.

5. Practical application of I-SonReb method


In this section are summarised the steps involved in the I-SonReb method. Several
techniques to measure the concrete carbonation thickness that minimise the damage to the
structure are briefly described, too. Assuming that the influence of carbonation on the UPV
16 M. Breccolotti and M.F. Bonfigli

65
Strength of extracted cores 25
60 Predicted strength with SonReb method
20
Predicted strength with ISonReb method
55 15

50 35
25
40
45
fc(MPa)

20

30
40
55
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

35

30
80

25

20

1 2 3 4 5 6 7 8 9 10
Test no.
Predicted strength measured

1.5
1.4 Normalized strength
Predicted strength using SonReb method
1.3
strength ratio

Predicted strength using ISonReb method


1.2
1.1
1
0.9
0.8
1 2 3 4 5 6 7 8 9 10
Test no.

Figure 8. Data-set 2: experimental and predicted compressive strengths of concrete samples using
the classical SonReb method and the I-SonReb method. The numbers indicate the carbonation depths
in millimetres.

0.8

0.6
kc

0.4
Dataset 1
0.2 Dataset 2

0
0 20 40 60 80 100
xc (mm)

Figure 9. Relationship between kc and xc using Equation (13) and the fitted coefficients b1 and b2.
Non-destructive Testing and Evaluation 17

can be considered negligible, the I-SonReb method could be carried out performing the
following steps:
(1) Choose a suitable number of places in the structure, where extract cores to
calibrate the coefficients of the correlation Equation (14). If possible, it is
advisable to choose these sampling locations among elements with different
carbonation conditions (exposed and non-exposed concretes). The number of
samples can be inferred from international standards. For example, EN 13791,[24]
suggests using data coming from at least 18 cores, or 9 if a basic curve of reference
is available.
(2) Measure the rebound index R and the ultrasonic pulse velocity V in the position
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

identified at point 1) following the rules given by appropriate international


standards, e.g. [19,25 27]. In particular, attention must be paid to avoid testing
locations located nearby steel reinforcements. If it is not possible to avoid the
measurement in proximity of reinforcing steel proper correction, as that indicated
by Bungey et al. [28], must be used.
(3) Extract cores and measure the thickness of the carbonated layer on these cores by
means of a phenolphthalein solution.
(4) Test cores to evaluate the compressive strength f c of concrete.
(5) Find the appropriate fitting coefficients of Equation (14) by non-linear fitting of
data-set.
(6) Perform an extended test campaign on the structure measuring for each sampling
point, the rebound index R, the ultrasonic pulse velocity V, the carbonation
thickness xc and the distance S between the emitter and the receiver ultrasonic
probe. Perform transparency-only UPV measurements and follow recognised
standards for Rebound Index and UPV measures on concrete.
(7) Use the correlation expression calibrated during step 5 to evaluate the concrete
strength based on NDT data coming from step 6.
It is useful to remark that if testing a good quality concrete batch cured in
homogeneous conditions, there is a possibility that the dispersion of the measured
strengths may become very low. Traditional nonlinear NDT correlation laws in these
particular cases do not perform better than a linear regression in these particular cases,
the same being true also for the I-SonReb method. However, it still gives a positive
contribution to the evaluation by taking into consideration the influence of carbonation
that may be very different from point to point within the structure. Yet such conditions are
quite infrequent in old existing structures, where these NDTs are most frequently used.
In such structures, concrete comes from casts performed in different moments, with
different curing conditions and from different production batches, so that such a low
dispersion is unlikely to be observed.

5.1 Carbonation depth evaluation


To determine the carbonated concrete thickness, it is advisable to use suitable techniques
that minimise the damage inflicted to the structure. The traditional way of performing this
assessment is to extract micro-cores from the structure and then apply 1% phenolphthalein
in 70% ethyl alcohol solution on the lateral surface of the specimen. As known well, this
solution is colourless when p H,9 as in the case of a carbonated concrete, while it assumes
a magenta colour in the presence of a more basic environment, like the one realised by a
non-carbonated concrete. Nevertheless, even if core drill bits can have a diameter as low as
18 M. Breccolotti and M.F. Bonfigli

approximately 12 mm, it is generally hard to extract intact micro-cores from low-strength


concretes.
An alternative way is the Carbontest. This promising method has been patented and
is ideal for a large-scale assessment of carbonation thickness. Its working principle,
described by Felicetti [29], consists of drilling a hole in the concrete with a traditional drill
to which is attached an appropriately sized device that neatly collects the produced powder
into a cylindrical tube. The carbonation thickness evaluation is thus carried out applying a
phenolphthalein solution to this powder and then measuring the height of the colourless
zone, as in the other techniques. Knowing the depth of the perforation and the total height
of powder column inside the tube, by means of a simple proportion, it is easy to estimate
the carbonated layer thickness. This methodology allows to estimate xc faster than the
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

extraction of micro-cores and with an almost negligible damage.


Another promising way to the non-destructive carbonation evaluation on real
structures is the one described by Bouchaala et al. [30]. The Authors, using the following
phenomenological stress strain relationship to describe the nonlinear behaviour of the
concrete:

s K 0 11 b1 d1 2 . . .  a1; sign1;
_ 19

where K 0 is the elastic modulus, 1 the strain, b and d the classical Landau and Lifchitz type
of nonlinearity and a the non-classical nonlinear parameter, recognised that carbonation
significantly alters the non-linear parameter a, which have been assessed using Non-linear
Resonant Ultrasonic Spectroscopy. However, further studies are still needed to discover
ways to easily evaluate the a parameter in real-world structures and to set up a proper
testing methodology.
Yet exist even other ways to assess concrete carbonation depth, such as Thermo-
Gravimetric Analysis (TGA), X-Ray Diffraction Analysis and Fourier Transform Infrared
Spectroscopy, but all these methods are not suitable for a fast assessment for in situ
investigations.

6. Conclusions
In this paper, the Improved SonReb method (I-SonReb) for the estimation of in situ
compressive strength in carbonated concrete has been presented. It makes use, together
with the classical rebound number R and the ultrasonic pulse velocity V, of additional
input data, the carbonation thickness xc , for the indirect estimation of the in situ concrete
compressive strength.
The method is, thus, particularly well adapted for the evaluation of concrete strength in
RC buildings and structures built in the last decades since it takes properly into account the
effect of carbonation always present in these old constructions.
In this paper, it has been first showed that the carbonation alters not only the rebound
index values but, to a lesser extent, also the ultrasonic pulse velocities. Parametric studies
allowed for judging as negligible this effect, since counterbalancing effects (increase of
concrete stiffness and mass) are produced by the carbonation.
The Improved SonReb method has been subsequently applied to two sets of
experimental data collected from existing concrete structures. The results obtained from
these experimental data show an increase in the R 2 values of 18% and 28%, respectively,
if compared with the coefficients of determination obtained with the traditional SonReb
method.
Non-destructive Testing and Evaluation 19

Finally, indications have been provided on some current available techniques to


evaluate by means of NDT methods the in situ carbonation depths.

Acknowledgements
Authors would like to thank MOST S.r.l. for providing the experimental data used in this research
project.

Disclosure statement
No potential conflict of interest was reported by the authors.
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

References
[1] Amasaki S. Estimation of strength of concrete structures by the rebound hammer. CAJ Proc
Cem Concr. 1991;45:345 351.
[2] Malhotra V, Carino NJ. Handbook on nondestructive testing of concrete. 2nd ed. Boca Raton,
FL: CRC Press; 2003.
[3] Jerga J. Physico-mechanical properties of carbonated concrete. Constr Build Mater.
2004;18:645 652.
[4] Chang C-F, Chen J-W. Strength and elastic modulus of carbonated concrete. Mater J.
2005;102:315 321.
[5] Luca FD, Verderame GM, Manfredi G. Eurocode-based seismic assessment of modern heritage
RC structures: the case of the tower of the nations in naples (Italy). Eng Struct.
2014;74:96 110.
[6] Huang N, Chang J, Liang M. Effect of plastering on the carbonation of a 35-year-old reinforced
concrete building. Constr Build Mater. 2012;29:206 214.
[7] Giannini R, Sguerri L, Paolacci F, Alessandri S. Assessment of concrete strength combining
direct and NDT measures via bayesian inference. Eng Struct. 2014;64:68 77.
[8] Nguyen NT, Sbarta Z-M, Lataste J-F, Breysse D, Bos F. Assessing the spatial variability of
concrete structures using NDT techniques laboratory tests and case study. Constr Build Mater.
2013;49:240 250.
[9] Bilgehan M, Turgut P. Artificial neural network approach to predict compressive strength of
concrete through ultrasonic pulse velocity. Res Nondestruct Eval. 2010;21:1.
[10] Brozovsky J. High-strength concrete NDT with rebound hammer: influence of aggregate on test
results. Nondestr Test Eval. 2014;29(3):255 268.
[11] Kim J-K, Kim C-Y, Yi S-T, Lee Y. Effect of carbonation on the rebound number and
compressive strength of concrete. Cem Concr Compos. 2009;31(2):139 144.
[12] Szilagyi K, Borosnyoi A, Zsigovics I. Rebound surface hardness of concrete: Introduction of an
empirical constitutive model. Constr Build Mater. 2011;25(5):2480 2487.
[13] Breysse D. Nondestructive evaluation of concrete strength: An historical review and a new
perspective by combining {NDT} methods. Constr Build Mater. 2012;33(0):139 163.
[14] Breccolotti M, Bonfigli MF, Materazzi AL. Influence of carbonation depth on concrete strength
evaluation carried out using the Sonreb method. NDT&E Int. 2013;59:96 104.
[15] Facaoaru I. Non-destructive testing of concrete in Romania. Proceedings symposium on non-
destructive testing of concrete and timber, London. June 1969. London: Institute of Civil
Engineers; 1970.
[16] Di Leo A, Pascale G. Prove non distruttive sulle costruzioni in cemento armato, Il giornale
delle prove non distruttive. [Non-destructive testing on concrete buildings].
1994;25:2480 2487.
[17] Gasparik J. Prove non distruttive nelledilizia. [Non-destructive testing on buildings]. Brescia;
1992.
[18] RILEM. NDT 4 Recommendations for in situ concrete strength determination by combined
non-destructive methods. London; 2009.
[19] ASTM C805. Standard test method for rebound number of hardened concrete. West
Conshohocken, PA: ASTM International. 2013.
20 M. Breccolotti and M.F. Bonfigli

[20] Pham ST, Prince W. Non destructive characterisation of cement mortar during carbonation.
World Acad Sci Eng Technol. 2013;7:1738 1742.
[21] Chang C-F, Chen J-W. The experimental investigation of concrete carbonation depth. Cem
Concr Res. 2006;36:1760 1767.
[22] BS 8110-2: structural use of concrete. Code of practice for special circumstances. International
Science Index. World Academy of Science, Engineering and Technology. 1985.
[23] Seber G, Wild C. Nonlinear regression. Hoboken, NJ: John Wiley & Sons; 2003.
[24] EN 13791. Assessment of in situ compressive strength in structures and precast concrete
components. Brussels: CEN; 2007.
[25] ASTM C597. Standard test method for pulse velocity through concrete. West Conshohocken,
PA: ASTM International; 2009.
[26] EN 12504-2. Testing concrete in structures part 2: non-destructive testing determination of
rebound number. 2012.
Downloaded by [University of Perugia], [Marco Breccolotti] at 00:06 21 July 2015

[27] EN 12504-4. Testing concrete part 4: determination of ultrasonic pulse velocity. Brussels:
CEN; 2004.
[28] Bungey J, Millard S, Grantham M. Testing of concrete in structures. 4th ed. New York, NY:
Taylor & Francis; 2006.
[29] Felicetti R. Assessment of deteriorated concrete cover by combined while-drilling techniques.
J Infrastruct Syst. 2011;18:25 33.
[30] Bouchaala F, Paya C, Garnier V, Balayssac J. Carbonation assessment in concrete by nonlinear
ultrasound. Cem Concr Res. 2011;41:557 559.

Anda mungkin juga menyukai