Вы находитесь на странице: 1из 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/237190073

Strength evaluation of an existing concrete bridge based on core and non-


destructive test data

Article  in  Canadian Journal of Civil Engineering · February 2011


DOI: 10.1139/l87-026

CITATION READS

1 153

2 authors, including:

Andrew Scanlon
Pennsylvania State University
86 PUBLICATIONS   576 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Andrew Scanlon on 21 April 2016.

The user has requested enhancement of the downloaded file.


Strength evaluation of an existing concrete bridge based on core and nondestructive
test data

ANDREWSCANLON
Department of Civil Engineering, UniversiQ of Alberta, Edmonton, Alta., Canada T6G 2G7
AND

LEONIDMIKHAILOVSKY
Alberta Tran.sportation, Edmonton, Alta., Canada T6G 2x3
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by Dalian Nationalities University on 06/06/13

Received April 9, 1986


Revised manuscript accepted October 24, 1986

The use of concrete cores and nondestructive testing for strength evaluation of existing structures is illustrated through an
evaluation of an existing concrete bridge. The results confirm that rebound hammer number should not be used alone as an
indicator of in situ compressive strength. Reasonably good correlation is demonstrated between pulse velocity and compressive
strength, with a slight improvement when pulse velocity and rebound number are combined.
A probabilistic approach to structural evaluation of existing structures is proposed in which the variability of in situ concrete
compression strength and uncertainty associated with indirect nondestructive testing are incorporated. An example of shear
strength evaluation is given, which indicates a range of increase in shear strength compared with initial design parameters of
17-37% depending on the amount and type of data available.

L'usage de prClkvements de bCton par carottage et d'essais non destructifs dans I'Cvaluation de structures existantes est
illustri pour le cas d'un pont en beton arme. Les rCsultats confirment que les essais de rebondissement du marteau ne devraient
pas &treutilisCs seuls comme indicateurs de resistance en compression in situ. Une bonne correlation est obtenue entre la vitesse
de propagation d'onde et la resistance en compression, avec une legbre amClioration quand la vitesse de propagation d'onde
et les rCsultats des essais de rebondissement du marteau sont combinks.
Les auteurs proposent une approache probabiliste pour I'Cvaluation des structures existantes qui tient compte de la variabilite
For personal use only.

de la rksistance en compression in situ de bCton et de I'incertitude associCe aux mesures faites par essais non destructifs.
L'approche est illustrie par un exemple d'evaluation de la rCsistance en cisaillement du beton. L'exemple indique une
rCsistance rCelle supCrieure a la valeur utiliste pour la conception initiale de I'ouvrage, I'Ccart variant de 17 a 37% selon la
quantitC et la nature des donnees disponibles.
[Traduit par la revue]
Can. J. Civ. Eng. 14. 145- 154 (1987)

Introduction
In 1980 the Canadian Standards Association Technical Com- The uncertainty associated with the evaluation of safety of an
mittee on Highway Bridges introduced, through clause 1 2 of existing structure can be significantly less than that for an
CAN3-S6, a probabilistic or statistical approach to the evalua- unbuilt new structure, since the existing structure has already
tion of existing bridge structures. As stated in the 1980 Supple- given evidence of its level of performance over its lifetime to
ment no. 1 to CAN3-S6 (Canadian Standards Association date, and because it is possible to obtain information on its
1980), the approach "provides a method whereby the strength current in situ condition. The uncertainty presumably decreases
of a bridge is determined on a statistical basis and compared to with an increase in the quantity and quality of data obtained
the effects of any desired load." from the structure.
The basic concepts underlying the probabilistic approach to This paper describes an investigation carried out to assess the
safety evaluation are outlined by MacGregor (1976), with par- variability of concrete strength and member dimensions in an
ticular reference to the determination of load and resistance existing bridge structure located near Lethbridge, Alberta. Di-
factors for limit states design of reinforced concrete structures. mensional variability was assessed from a series of measure-
Development of load and resistance factors for design of new ments of the structural members. The concrete strength vari-
structures requires estimates of variability of parameters af- ability was assessed based on a combination of compressive
fecting strength. These estimates are obtained primarily from strength tests on cores, and two series of nondestructive tests.
surveys of published data. Progressive refinement of the load Application of probability-based concepts to strength evalua-
and resistance factors can be expected as more statistical data tion is illustrated through an example using the data obtained.
are gathered and made available.
It has been suggested (Scanlon 1984) that a similar approach Data acquisition
is possible for evaluation of existing structures by considering The investigation was conducted on a 34 year old two-lane
the variability of strength parameters in the actual structure bridge over a Lethbridge Northern Irrigation District (LNID)
under consideration. In this case, the necessary data for the canal. The bridge was scheduled for demolition as part of a
evaluation can be obtained by taking appropriate measurements highway upgrading programme.
and by performing various in situ and laboratory tests. The superstructure consisted of three continuous concrete
spans 12 200, 18 300, and 12 200 m m (40, 60, and 4 0 ft) long
NOTE: Written discussion of this paper is welcomed and will be as shown in Fig. 1. Dimensions of the slab and the five
received by the Editor until July 3 I, 1987 (address inside front cover). T-girders are shown in Fig. 2. The girders varied in depth from
146 CAN. J . CIV. ENG. VOL. 14. 1987

FIG. I . Elevation of bridge.


Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by Dalian Nationalities University on 06/06/13

A 8 C D E

FIG.2. Cross section of bridge.


For personal use only.

FIG.4. Rigid caliper for stem thickness measurements.

sisting of member dimensions, pulse velocities, and rebound


numbers were obtained. In addition, a series of cores was taken
to provide direct measurements of concrete strength and to
provide a basis for correlation of the pulse velocity and rebound
readings with compressive strength.
Dimensions
Dimensions of the deck slab and girders were measured. The
slab dimensions were measured by coring through the slab at
seven locations. The vertical stem dimensions were obtained
using a device consisting of a steel square edge, clamps, levels,
and rigid scale as shown in Fig. 3. Vertical dimensions h , and
h2were measured by steel rule and the stem depth was obtained
by subtracting h2 from h , . Stem thickness measurements were
FIG.3. Measurement of girder stem height. taken using a caliper device as shown in Fig. 4. The reference
dimension, 1, was measured using a steel tape and checked
1448 mm (4 ft 9 in.) over the piers to 686 mm (2 ft 3 in.) over every 20 min to ensure that the dimension remained constant
the abutments and at centre span. for measurements in each girder zone. The dimension d was
The minimum 28-day compressive strength of concrete was measured using a sliding caliper. Stem thickness measurements
specified as 20.7 MPa (3000 psi) and reinforcement shown on were made to the nearest 0.1 mm. Accurate measurements of
the design drawings consisted of deformed and nondeformed stem thickness were required for the determination of pulse
square and round steel bars. The deck was covered with 80 mm velocities for concrete in the girders. The length of each girder
(3 118 in.) asphalt pavement. of the centre span was measured to the nearest millimetre using
Testing was restricted to the interior span. In situ data con- a steel tape.
SCANLON AND MIKHAILOVSKY

Legend
N = North Zone
M = Middle Zone
S = South Zone
t t
PlER PlER
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by Dalian Nationalities University on 06/06/13

I
k0
z Grid Cell Location Numbers 350

FIG.5. Grid cells on girder stems.

North South Coring


(a) End End
Twenty-one locations were selected at random in the girder
stems. Cores were identified according to their location on the
efors tCell
ocafion*
in Test
for Cells in
Locat~on
stem grid layout. A magnetic inductance meier was used to
avoid, where possible, reinforcement embedded in the con-
Zone S Zones N and M crete. One core was discarded b e c a w of a steel bar inclusion.
For personal use only.

Cores were drilled with a 100 mm (4 in.) diameter diamond


drill. Cores taken from the stems were placed in two groups.
The first group consisted of 10 single cores, each having a
length of approximately half the girder width. The second
group consisted of 11 pairs of cores, each pair obtained by
drilling through the stem and breaking the core in two pieces,
each approximately half the girder stem width.
Compressive strength tests and pulse velocity tests were
conducted on all of these cores. To assess the effect of surface
roughness, pulse velocity readings were taken with and without
the natural rough ends. Smooth ends were obtained by sawing
FIG. 6. Test locations within grid cell for pulse velocity and re- with a diamond blade.
bound hammer readings: ( a )Test locations for pulse velocity readings; Rebound readings were not taken directly on the cores be-
( b ) test locations for rebound hammer readings. cause of difficulties associated with applying a proper hammer
Pulse velocities blow to the curved surface of the core. Instead, rebound
The development of ultrasonic testing, in which pulse veloc- readings were taken around the core hole in situ.
ity is determined from the time taken for a stress wave to Correlation between compressive strength and nondestruc-
traverse a known path length, is described by Malhotra (1976). tive test parameters was performed on the cores taken from the
The test is covered by ASTM Standard (2597. In this investiga- stems. Cores were also taken from 1 1 locations in the slab.
tion testing was conducted with a C-4902 "V-meter" manu- Seven full-depth and 4 half-depth cores were taken. The
factured by James Electronics Inc. The digital display showed 7 full-depth cores were used to measure slab thickness.
the pulse propagation time in microseconds with an accuracy of
0. I p s . Readings were taken on the girder stems at grid loca- Test results
tions illustrated in Figs. 5 and 6. Results of the dimensional survey, compressive strength
To define grid cell locations within the bridge, the girders tests, and nondestructive tests are summarized below. Further
were identified by letters A-E as shown in Fig. 2. Each girder details are presented in a report by Mikhailovsky and Scanlon
was then divided into three zones, N (north zone), M (middle (1985).
zone), and S (south zone) as shown in Fig. 5. Within each
zone, a grid was marked on the girder stem and each cell was Dimensions
numbered according to the sequence shown in Fig. 5. Measured and specified slab thickness values are compared
in Table 1 for the seven slab core locations. The results indicate
Rebound numbers that instead of the specified slab thickness of 178 mm (7 in.),
Rebound testing, also described by Malhotra, was conducted a thickness of 254 mm (10 in.) was provided. The reason for
in accordance with ASTM Standard C805 with an "H-meter" the large discrepancy was not identified.
model C-7311 manufactured by James Electronics Inc. Nine The comparisons between measured and specified girder
readings were taken in each cell of the grid layout in the height, girder thickness, and centre span length are summar-
sequence shown in Fig. 6. ized in Tables 2-4. For these cases the agreement between
148 CAN. J. CIV. ENG. VOL. 14. 1987

TABLEI. Measured and specified slab thicknesses TABLE


3. Measured and specified girdcr thicknesses

Slab thickness (mm) Stem thickness (mm)

Core Measured Specified Mean


NO. no. LM LS LMILS measured Specified
value value
Zone TM Ts T MI T S
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by Dalian Nationalities University on 06/06/13

Mean
Standard deviation
Coefficient of variation

TABLE
2. Measured and specified girder heights

Girder hcight (mm)


Mean* of
Specified measured Mean
Section dimension dimensions Standard deviation
no. hs hM h~lhs Coefficient of variation
- -
For personal use only.

TABLE
4. Girder lengths of the middle span

Girder length (mm)


Specified Measured
Girder span span
NO. no. Ls LM LMILS
Mean
Standard deviation
Coefficient of variation
*Mean of 10 measurements: 2 symmetrical locations in
5 girders.

mean measured values and specified values is generally very Mean


good. Standard deviation
Coefficient of variation
Pulse velocities
Pulse velocity readings at 451 cell locations defined as values for each cell, the mean was calculated as 40.6 with a
shown in Fig. 5 were recorded for the five T-girders. Six coefficient of variation of 9.8%.
additional readings were discarded because of extremely rough
Cores
concrete surfaces preventing proper contact between the trans-
Results of compressive strength tests, pulse velocity mea-
ducer and the concrete. The results are summarized in the
surements, and rebound readings for the 31 cores taken from
histogram shown in Fig. 7. The mean pulse velocity is
the girder stems are listed in Table 5. The histograms shown in
3.97 mm/ps with a coefficient of variation of 4.1%. Consid-
Fig. 8 provide an indication of the variability of compressive
ering each of the 15 girder zones separately, the mean pulse
strength, pulse velocity, and rebound number.
velocity ranges from 3.89 to 4.07 mm/ps while the coeffi-
The coefficient of variation of the core compressive strength
cients of variation range from 2.2 to 5.4%. These results sug-
values is more than five times that of the pulse velocities,
gest that the concrete quality is quite uniform over all five
indicating that the variability of pulse velocity is not a direct
girders.
measure of the variability of compressive strength.
Rebound readings Comparing mean and coefficient of variation of pulse veloc-
Rebound readings were taken at 457 cell locations. At each ity in situ at the core locations with pulse velocity directly on
location, nine readings were taken and the mean and minimum the cores in the laboratory the following values were obtained:
values of the nine readings were recorded.
The mean and coefficient of variation of cell mean values are Mean CV
47.7 and 6.7% respectively. It should be noted that the coeffi-
cient of variation of all rebound readings as opposed to all mean In situ 4.049 3.7%
values would be considerably higher. Taking the minimum Cores 4.199 3.1%
SCANLON AND MIKHAILOVSKY

FREQUENCY

I3Or
FREQUENCY
I
f - 36.83

n = 31
f = 36.83 MPa
$ = 6.583 MPa
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by Dalian Nationalities University on 06/06/13

*
o r-
n 2 ~ vc ( \ ? f i r ~ ~ ~
o
N N N l O l O l O * $ : ;
COMPRESSION STRENGTH
OF CORES ( M P a l
PULSE VELOCITY (=)
PS
FREQUENCY
FIG.7. Histogram of pulse velocities in girder stems.

Correlation between compressive strength of cores


and nondestructive test parameters
A number of researchers have proposed expressions relating
For personal use only.

compressive strength with pulse velocity and (or) rebound


number (e.g. Facaoaru 1969; Chung and Law 1983; Samarin
and Dhir 1984; S t u m p et al. 1984). In some cases the expres-
sions include parameters related to the concrete mix propor-
tions, such as coarse and fine aggregatelcement ratios. For
existing structures these parameters are generally unknown.
Even when a reasonable estimate can be obtained from either
job records or samples taken from the structure, significant
variations can be expected depending on quality control of
batching and placement of concrete as well as possible effects
of aging. FREQUENCY

To evaluate concrete strength in existing structures, a direct


correlation between measured strengths and nondestructive test n
parameters appears to be a more practical approach. This
means that relationships must be developed for each individual
structure.
In this investigation a number of relationships were devel-
oped between compressive strength, pulse velocity, and re-
bound number.
To provide information on the effect of the number of data
points, regression analyses were carried out using two data
groups: a group of 20 cores and a group of 6 cores. The 20 cores
were selected randomly from the 31 cores taken from girder
stems. The group of 6 cores was selected from the group of NUMBERS Rmi,
20 cores to provide a set of core strengths covering the full
range of values. FIG. 8. Histograms of data for cores: (a) Histogram of compressive
strength of 31 cores; (b) histogram of pulse velocities of 20 cores;
Pulse velocity and compressive strength (c) histogram of minimum rebound numbers around 20 cores.
Regression analyses were performed on the data for 20 cores
using both a power and an exponential curve. An analysis was
aIso performed on the data of 6 cores using a power curve. The [3] f: = 0 . 0 0 1 7 2 ~ ~ . ' ~ (6
~ cores)
relationships obtained are as follows: (r = 0.944, 6, = 4.40MPa)

[l] f: = 0.020V5.227 (20 cores) where f: = estimated concrete strength (MPa), V = measured
(r = 0.832, 6, = 4.26 MPa) pulse velocity (mm/p,s), r = correlation coefficient, and 6, =
standard error of correlation (MPa).
[2] ff = 0.1 88e'.25V (20 cores) Figure 9 shows the scatter diagram and regression function
(r = 0.829, 6, = 4.02 MPa) based on the power curve for 20 cores.
150 CAN J. CIV. ENG VOL. 14. 1987

TABLE5. Results of tests on girder stem cores CONCRETE STRENGTH


(MPal

Compressive Pulse Rebound no. $


Core strength* velocityt
no. (MPa) (mm/p,s) Mean Min. Max.

I 36.77 4.293 44.9 32 51


2 43.33 4.256 44.9 35 52
3 44.09 4.337 47.1 38 51
4 44.74 4.286 45.9 39 51
5 33.73 3.993 38.1 35 43
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by Dalian Nationalities University on 06/06/13

6 38.04 4.274 37.1 28 47


7 33.49 4.195 42.2 34 50
8 47.95 4.326 49.9 43 54
9 35.62 4.136 43.5 37 49
10 37.28 4.084 46.9 43 51
II 37.25 4.242 41.4 36 47
12 29.79 4.044 43.2 35 52
13 32.56 4.256 43.9 3.9 4.0 4.1 4.2 4.3 4.4 4.5 4.6
31 49
14 38.93 4.204 41.8 33 47 PULSE VELOCITY v (=)PS
15 21.07 3.917 41.9 31 48
16 20.33 3.920 44.0 40 50 FIG.9. Scatter diagram and the best-fitting curve,f, = 0.20~'."'
17 37.31 4.147 40.9 38 48 for the data on 20 cores.
18 35.28 4.268 39.8 32 47
19 38.44 4.304 40.1 30 45 CONCRETE STRENGTH
20 41.65 4.164 40.6 36 44 (MPa)
21 36.07 4.120 44.6 39 48
22 43.33 4.326 40.2 33 50
23 41.82 4.129 44.8 43 51
For personal use only.

24 37.34 4.299 42.9 37 49


25 29.42 3.972 38.9 33 45
26 32.56 4.148 39.2 34 43
27 45.76 4.381 47.0 42 50
28 38.93 4.204 39.6 33 48
29 35.16 4.134 45.1 32 54
30 44.74 4.388 44.6 36 52
31 27.51 4.035 40.9 31 53
I I . ' ' ' I ' I ~ f ~ ~ s
Mean 36.8 4.19 42.6 35.4 48.8 2 8 29 X) 31 32 33 34 35 36 37 38 39 40 41 42 43
Standard MINIMUM REBOUND NUMBER Rmi,
deviation 6.6 0.13 3.0 4.0 3.0
Coefficient of FIG. 10. Scatter diagram and the best-fitting curve, f, =
variation 0.18 0.3 1 0.07 0.11 0.06 1 . 4 1 0 for
~ ~the~data
~ on 20 cores.
*Compressive strengths shown are corrected for L I D ratio less than 2.
tPulse velocity readings shown are for both ends smooth.
$Values shown are based on eight in sir~rreadings taken around each core scatter diagram and best-fitting curve based on Rmi,.
hole. The values for correlation coefficient, r , obtained in this
study are low compared with values normally obtained for new
Rebound number and compressive strength concrete. Although the causes were not investigated in detail,
Eight rebound numbers were taken around each core hole. it is possible that carbonation of the surface has masked the
Regression analysis using a power curve was performed to effect of the underlying strength. Since surface carbonation
detemine the correlation between compressive strength and normally produces an increase in rebound number, the mini-
each of mean, minimum, and maximum rebound numbers, mum values may be more representative of the compressive
RnIcan,
R m i n and
, R,,,,. The regression equations and corre- strength in the member.
sponding correlation coefficients obtained for the 20-core
group are as follows: Pulse velocity, rebound number, and compressive strength
Because pulse velocity and rebound number measure differ-
[4] R,,: f: = 1.410~!:;~ ent parameters associated with concrete quality, it has been
(r = 0.504, 6, = 5.76 MPa) suggested (e.g. Facaoaru 1969) that a better estimate of
strength can be obtained using a combination of the two non-
[5] R,,,,: f,' = 9.351~:::~~ destructive test methods. Since the correlation between
( r = 0.127, 6, = 6.66 MPa) strength and Rnli,was found to be better than that using R,,,,,
[6] R,,,: f: = 22.510~!::' or R,,,, only R,, was considered in the combined approach.
( r = 0.040, 6, = 6.86 MPa) Using a power curve, the following relationships were
obtained:
These results indicate a weak correlation between compres-
sive strength and Rmi,, and little or no correlation between [7] f: = 0 . 0 0 4 5 ~ ~ . ~ ' ~ ~ : ? (20 cores)
compressive strength and R,,,, or R,,,. Figure 10 shows the (r = 0.896, 6, = 3.28 MPa)
SCANLON A N D MlKHAlLOVSKY 151
ACTUAL CONCRETE TABLE
6. Data sets used to estimate mean and standard deviation of
concrete strength
Case Data used
A Compressive strength tests on 31 cores
B Pulse velocities for 3 1 cores, regression curve from 20 cores
C Pulse velocities for 31 cores, regression curve from 6 cores
D Pulse velocities and minimum rebound numbers for 3 1 cores,
regression curve for 6 cores
E Pulse velocities and minimum rebound numbers from 69 ran-
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by Dalian Nationalities University on 06/06/13

dom locations in the bridge, regression curve from 6 cores

[lo] f: = -25.6 + 0.655R ,,,, + 0.128V4


( r = 0.892, 6, = 3.08 MPa)
ESTIMATED CONCRETE
STRENGTH fe (MPa) The comparison between actual and estimated concrete
strengths for three of the relationships described above is
ACTUAL CONCRETE shown in Fig. I I. The results indicate significantly better cor-
STRENGTH ;1
(MPa) relation between strength and pulse velocity than between

I ibi
/+IS% strength and minimum rebound number. The best correlation
was obtained when pulse velocity and minimum rebound
readings were combined, although the improvement, compared
with pulse velocity alone, is quite small and the use of com-
bined testing does not appear to be justified in this instance.
The difference between expressions for 6 and 20 cores is also
quite small although it should be noted that the 6 cores were
selected from the group of 20 to cover the full range of com-
For personal use only.

pressive strength.

Evaluation of the mean and standard deviation of


concrete strength
20 30 40 50
ESTIMATED CONCRETE Various combinations of data obtained from the bridge were
STRENGTH f, (MPa) used to estimate the mean and standard deviation of concrete
strength. These quantities are required to evaluate the strength
ACTUAL CONCRETE of concrete in a probabilistic sense as well as the strength
STRENGTH fd
(MPa)
(resistance) of structural members.
I Since the estimates are based on limited samples of data, it
is of interest to know the levels of confidence associated with
these estimates. In each case the 95% confidence limits on the
estimate of the mean were calculated using the relationship
6
[I I] p = X + t,,,,, -
G
where p = population mean, 2 = sample mean, 6 = sample
standard deviation, n = number of data points in the sample,
and t,,,,, = parameter of the t-distribution at confidence level a,
with m = n - 1 degrees of freedom.
For the case where the sample standard deviation is based on
ESTIMATED CONCRETE indirect measurements the correlation error must be included,
STRENGTH 1,
(MPa)
and the sample standard deviation is calculated from (Samarin
FIG. 1 I. Comparison between the actual and the estimated concrete and Dhir 1984)
strength: (a)f, = 1 . 4 1 0 ~ ~ :(rebound);
;~" (b)f, = 0.020V~~~''(pulse
velocity); (c)f, = 0 . 0 0 4 5 ~ ~ (combined).
~ ~ ~ " ~ ~ ; ~ ~
where 6, = standard deviation of the values estimated from the
correlation curve and 6, = standard error of correlation.
[8] ff = 0 . 0 0 2 0 3 ~ ~ ' ~ (6~ cores)
~ ~ ; ~ ~ To illustrate the effect of different sets of data on the esti-
(r = 0.948, 6, = 4.16 MPa) mated mean and standard deviation of the concrete strength, the
cases summarized in Table 6 are considered.
Samarin and Dhir (1984) proposed an expression of the form Cases A-D involve various combinations of core data while
case E involves indirect (nondestructive) it1 situ data and a
[9] ff=Ao+AIRm,,,+A2V3
regression curve based on 6 cores. Case E represents a typical
This form, with R ,,,, replaced by R ,,,,, gave the following application of nondestructive test methods with limited core
relationship: data.
CAN. J . CIV. ENG. VOL. 14, 1987

TABLE
7. Mean and standard deviation of concrete cornprcssive strcngth and calculated shcar strength based on
various sets of data
- - - - -

Confidcncc limits
Total on rncan (95%) Factored
Sample Standard Standard standard shear
mean
-
deviation error deviation Coefficient Lowcr Upper strength
xc 6c 6, 6 of limit l~rnit +R % increase'"
Case (MPa) (MPa) (MPa) (MPa) variation (MPa) (MPa) (kN) in strength
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by Dalian Nationalities University on 06/06/13

'Relative to strength based on original specified compressive strength of 20.7 MPa (3000 psi)

Results for each case are summarized in Table 7. The greater within 6%. The other cases, B-D, are intermediate between
uncertainty associated with the nondestructive test results com- the two extremes.
pared with the direct measurement of core strengths is evident The differences are discussed further in the evaluation of
from a comparison of coefficients of variation and lower limit shear strength discussed in the following section.
values of the mean. It is expected that an optimum sampling
strategy would involve a combination of direct and indirect Probabilistic evaluation of strength
measurements. Work is currently underway at the University of The basic relationship between load and strength (resistance)
Alberta to develop such a strategy using a Bayesian statistics using a limit states design approach is given by
approach.
For case A, involving direct strength tests only, the interval [ I 31 R(e-PaV~]3 Cl[ePavu]
for p (at a = 95%) was calculated as (34.42, 39.24 MPa), i.e. or
For personal use only.

there is a 95% probability that the true mean lies within the
range 34.42-39.24 MPa. The standard deviation was calcu- [ 141 R y, [e-@vH] 3 U y, [esUVu]
lated as u = 6.58 MPa.
or
For cases B and C, the mean and standard deviation were
estimated on the basis of the pulse velocities obtained for the 1151 R b 3 UA
3 1 cores using regression equaiions based on both 20 cores and
6 cores. where R , 0 = mean values of strength and load, V,, V, =
Using the power curve obtained for 20 cores, the mean value coefficients of variation of strength and load, a = separation
function = 0.75, P = safety index, R , U = nominal values of
of strength was calculated as = 35.94 with a standard devi-
ation 6, = 6.04 MPa. Because the strength is being measured strength and load, y, = RIR, y, = U / U , and +,A = resis-
indirectly, the correlation error must be included. The standard tance and load factor respectively.
error associated with the correlation curve was calculated as The derivation and assumptions for these relationships are
6, = 4.26 MPa. The overall standard deviation is calculated given by MacGregor (1976). MacGregor has shown how the
from [12]: relationship can be used as a basis for determining load fac-
tors (A) and resistance factors (+) for design using available
data on material and dimensional variability. Application of the
procedure to strength evaluation of existing structures will now
be illustrated by considering the shear strength at a critical
6 = 7.39 MPa section of the bridge-for which field data were obtained.
The variability associated with dimensions and steel proper-
The 95% confidence limits on the mean can now be calcu- ties could also be included using the procedures outlined by
lated using [I I] as (33.23, 38.65 MPa). Using the correlation MacGregor (1976) from data obtained from the structure or
relationship obtained from 6 cores, the mean value of concrete based on published information on variability of these quan-
strength based on pulse velocities for the 31 cores was cal- tities. For simplicity it will be assumed that the only variability
culated as % = 37.35 MPa with a standard deviation of is that associated with concrete strength since this is the domi-
7.74 MPa. The standard error for the correlation curve was nant factor. For other design parameters, such as dimensions
calculated as u,= 4.40 MPa, giving a total standard deviation and steel yield strength, it will be assumed that the nominal and
of 6 = 8.90 MPa and the 95% confidence limits on the mean mean values are identical with zero standard deviation.
are (34.09, 40.6 MPa). First, the shear strength is calculated on the basis of the
Using the procedure outlined above, the calculations were specified 28-day strength (20.7 MPa (3000 psi)) and published
repeated for the remaining cases to give the results summarized data on concrete strength variability. The calculations are then
in Table 7. repeated using data on mean and standard deviation of concrete
The greater uncertainty associated with indirect (nondestruc- strength for various combinations of core strengths, pulse ve-
tive test) data compared with direct (core) data is illustrated locity data, and rebound data.
through a comparison of case A with E. The total standard At the design stage the in situ properties are not known and
deviation for case E (indirect data) is 38% greater than for the factored (design) strength is based on published data on
case A (direct data). The sample means on the other hand are variability. To illustrate the calculation procedure, the fol-
SCANLON AND MIKHAILOVSKY 153

lowing data given by MacGregor (1976) are used:


Coefficient of variation of fi,
Calculation of coefficietit of variation of shear strength
The standard deviation for the prediction of V , is
Mean concrete compressive strength in structure,
f, = 0.675fL + 7.6
= 21.56 MPa
Therefore, the coefficient of variation for the shear strength is
Correction factors for errors in design equation for shear,
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by Dalian Nationalities University on 06/06/13

7, = l.lV, + 1.2VS
Contribution of design equation error to coefficient of variation
of shear strength, The contribution of inaccuracy in the design equation to
uncertainty in predicting strength is included through the
V, = 0.15
relationship
Safety index for shear,
[21] v, = = \ / 0 . ~ ~ 8+
' 0 . 1 5 ~= 0.203
Substituting into [13], the factored shear strength is
The design equation for shear strength given in CSA
C A N 3 3 6 is
These calculations are repeated using the sample mean and
coefficient of variation of concrete strength values listed in
Table 7. The computed factored shear strength values range
For the section considered, the appropriate design data are from 295 to 347 kN, i.e. 17-37% increase compared with the
f: = 28-day compressive strength = 20.7 MPa strength based on initial specified concrete strength as indicated
For personal use only.

(3000 psi) in Table 7. The range in values reflects the varying degrees of
uncertainty associated with the types of data used to obtain the
b , = web thickness = 432 mm (17 in.) estimates of the mean and coefficient of variation of the in sit11
d = effective depth = 1067 mm (42 in.) concrete strength.
It should be emphasized that the procedure outlined above is
A, = area of stirrups = 258 mm' (0.40 in.') valid only if the concrete in the bridge girders can be consid-
fy = reinforcement yield strength = 248 MPa (36 ksi) ered as homogeneous, as was the case for the bridge under
consideration. An initial condition survey should be undertaken
s = stirrup spacing = 635 mm (25 in.) to identify any localized areas of distress or deterioration in the
structure. This can be done by visual inspection and non-
Calculation of mean shear strength destructive testing. Such conditions would require special at-
The nominal design strength is calculated from [I61 as tention in the evaluation of strength, perhaps including repairs
of the affected areas.
For the bridge considered in the study, the in situ strength
was found to be considerably higher than implied by the design
drawings. If the original design had been found to be mar-
ginally unsafe in terms of shear strength, an evaluation of this
type could provide a rational basis for finding the structure to
have adequate strength. Of course, other cases involving in situ
concrete strength less than the specified strength would be
Following the procedure outlined by MacGregor (1976) and treated in the same manner and the evaluation procedure would
considering only variability in concrete strength, the mean be effective in identifying an unsafe condition.
shear strength is obtained from Whether or not a detailed investigation of concrete strength
is required for evaluation of an existing bridge depends on the
critical design condition for the structure. For example, if the
strength is governed by flexure at an underreinforced section,
the concrete strength will have little influence on the calculated
strength. On the other hand, if shear strength is critical and the
amount of shear reinforcement is low, the concete strength can
have a significant effect on the calculated shear resistance, as
illustrated in the example.
The shear strength example illustrated how in sit11 data can
+
where f, = 0.675ff 7.6 = 21.56 MPa (MacGregor 1976) = be used to evaluate the strength of an existing structure. Further
mean concrete strength in the structure. work is required to calibrate the calculated strength against the
Including the effects of errors in the design equation, the load and resistance factors incorporated in the governing design
corrected mean shear strength is given by code. For example, MacGregor (1976) has shown that the
154 CAN. I. CIV. ENG. VOL. 14, 1987

relationship given by [13] leads to a +-factor of 0.67 for shear Sciences and Engineering Council of Canada through Oper-
strength and load factors of 1.25 for dead load and of 1.5 for ating Grant A5 153 is also gratefully acknowledged.
live load. CANADIAN STANDARDS ASSOCIATION Technical Committee on
If the governing design code is based on factors significantly Highway Bridges. 1980. Supplement no. 1 to CSA Standard
different than these, an appropriate adjustment would have to CAN3-S6-M78: Design of highway bridges.
be made to ensure that the calculated strength is compatible CHUNG, H. W., and LAW,K . A. 1983. Diagnosing in situ concrete by
with the code-specified factored loads. ultrasonic pulse technique. Concrete International Design & Con-
struction, S(10): 42-49.
Summary and conclusion FACAOARU, 1. 1969. Non-destructive testing of concrete in Romania.
Proceedings, Symposium on Non-Destructive Testing of Concrete
An example of an evaluation of an existing bridge structure
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by Dalian Nationalities University on 06/06/13

and Timber, Institution of Civil Engineers, London, pp. 39-49.


has been used to illustrate how strength data from cores and MACGREGOR, J. G. 1976. Safety and limit states design for reinforced
nondestructive test results can be used to evaluate the safety of concrete. Canadian Journal of Civil Engineering, 3(4): 484-5 13.
an existing structure. Methods for correlating direct strength MALHOTRA, V. M. 1976. Testing hardened concrete: Nondestructive
tests and nondestructive tests were presented as well as a meth- methods. American Concrete Institute, Monograph 9.
odology for including material variability and sampling error in MIKHAILOVSKY, L., and SCANLON, A. 1985. Evaluation of existing
strength evaluation using a probabilistic approach. The greater bridge structure by nondestructive test methods. Structural Engin-
the uncertainty in the data, for example using indirect non- eering Report no. 128, University of Alberta, Edmonton, Alta.
SAMARIN, A., and DHIR,R . K. 1984. In Determination of in situ
destructive testing instead of direct tests on cores, the lower the strength rapidly and confidently by nondestructive testing. Edited
estimate of strength of the structure. For the structure consid- by V. M. Malhotra. Publication SP-American Concrete Institute,
ered, the estimated shear strength based on several types of no. 82: 77-94.
field data was in all cases greater than the strength calculated SCANLON, A. 1984. Reliability index of existing concrete structures.
on the basis of the original design compressive strength. Symposium on Evaluation of Existing Concrete Structures: Safety,
American Concrete Institute Convention, Phoenix, Symposium
Acknowledgements Volume.
STURRUP, V. R., VECCHIO,F. J., and CARATIN, H. 1984. In Pulse
Technical assistance and financial support for this study pro- velocity as a measure of concrete compressive strength. Edited by
vided by the Bridge Engineering Branch of Alberta Transporta- V. M. Malhotra. Publication of SP-American Concrete Institute,
For personal use only.

tion is gratefully acknowledged. The support of the Natural no. 82: 201 -227.

View publication stats

Вам также может понравиться