Professional Documents
Culture Documents
Abstract: This investigation presents results of the statistical and probabilistic analyses of 3,269 normal weight concrete cylinder compression
tests used for recently constructed highway bridges in California. A new model for in-place strength of concrete structures is proposed as a func-
tion of the specified compressive strength, the normalized 28-day cylinder strength, and age of the structure based on a realistic strength–age
relation for hardened concrete. The model prediction indicates that concretes in cast-in-place bridge structures designed with specified
compressive strengths of 25 MPa (3.6 ksi), 28 MPa (4.0 ksi), and 35 MPa (5.0 ksi) reach their maximum strengths at about 40 years, with
approximately 98% of the maximum strength occurring during the first 10 years. Also, through significance testing on the 28-day cylinder
strengths, it was established that the California Department of Transportation practice of using an expected concrete strength instead of the
specified strength for seismic design of bridge components is justified. An expression for predicting the 28-day strength of concrete cylinders as
a function of the strength of companion cylinders, also proposed herein, could prove a useful tool for quality control of concrete during
construction. DOI: 10.1061/(ASCE)CF.1943-5509.0000404. © 2014 American Society of Civil Engineers.
Author keywords: Concrete; Compressive strength; Concrete tests; Quality control; Statistics; Bridges; Probability; Concrete construction.
h i fc,cyl (4)
t
fc,cyl ¼ 1,743:5 ln ðtÞ 2 44:81 (2)
Eq. (4) is most useful as a tool for predicting the 28-day cylinder
35 MPa ½5,000 psi concrete: fc,cyl
t
¼ 13:0333 ln ðtÞ þ 3:248 strength based on the strength of companion cylinders tested at an
h i earlier age. For example, Caltrans Standard Specifications (Caltrans
t
fc,cyl ¼ 1,890:3 ln ðtÞ þ 471:15 (3) 2010b) require that if the compressive strength at 7 days indicates to
the engineer that the concrete will not attain the strength specified at 28
t
In Eqs. (1)–(3), fc,cyl 5 compressive strength of concrete cylinder days, the mix design or concrete fabrication must be corrected before
(in MPa or psi) at age t measured in days, where 3 # t # 56. additional concrete is placed. In this case, Eq. (4) may be written as
Because of the large scatter in the data points of Figs. 1–3,
1
Eqs. (1)–(3) are more relevant as predictors of cylinder compressive 28
fc,cyl ¼ ft (5)
strength gain with respect to early age (t # 56 days) rather than ½0:2752 lnðtÞ þ 0:083 c,cyl
Fig. 1. Variation of concrete strength with age for fc9 5 25 MPa (3,600 psi) (note: 1 psi 5 6.894757 kPa)
Fig. 2. Variation of concrete strength with age for fc9 5 28 MPa (4,000 psi) (note: 1 psi 5 6.894757 kPa)
Fig. 3. Variation of concrete strength with age for fc9 5 35 MPa (5,000 psi) (note: 1 psi 5 6.894757 kPa)
Construction engineers often rely on historical data or some rule of Twenty-Eight–Day Cylinder Strength Results
thumb to obtain an indication of the likely 28-day compressive
strength, given the strength of companion cylinders tested at an earlier Because the 28-day strength has historically been used as the basis
age. One such rule of thumb for 100% portland cement concrete for concrete compressive strengths under 41 MPa (6 ksi) (AASHTO
mixes with 335–390 kg/m3 (564–658 lb/yd3) of portland cement and 2010), subsequent analyses of the cylinder data will be based on the
a water/cement ratio of 0.5 was 65% of 28-day strength at 7 days and 28-day strength. Accordingly, subsequent references to mean cyl-
80% of 28-day strength at 14 days (Caltrans 2011). This rule of thumb inder strength would be taken as the 28-day mean cylinder strength.
is unarguably highly restrictive, especially with the increasing use Table 1 also shows summary statistics for the 28-day cylinder strength
of supplementary cementitious materials in today’s projects. Also, data normalized by the respective specified strengths, fc9.
previous experience or historical data may not always be available The dispersion of cylinder strengths about the mean values, as
to the field engineer, especially when new mix designs are specified indicated by the standard deviation (or coefficient of variation), is
for a project. Therefore, Eq. (5) would be a useful addition to the attributable to batch-to-batch as well as within-batch variations (ACI
tools available to a construction engineer for quality control of Committee 214 2011). Batch-to-batch and within-batch variations in
concrete during construction. concrete cylinder results are the result of differing characteristics,
9
H0 : f c,cyl # fce (6)
9
H1 : f c,cyl . fce (7)
where fce9 5 1:39 fc9 for 25 MPa (3.6 ksi) concrete; and fce9 5 1:3 fc9
for 28 and 35 MPa (4.0 and 5.0 ksi) concretes.
Downloaded from ascelibrary.org by KANSAS STATE UNIV LIBRARIES on 07/18/14. Copyright ASCE. For personal use only; all rights reserved.
Fig. 7. Frequency distribution of cylinder strength data for fc9 5 5,000 psi
and corresponding fitted normal distribution curve (note: 1 psi
5 6.894757 kPa)
8 2 3
9 of differences in concrete placement procedure, consolidation
< 9c 2 ln f c,cyl þ 0:5 ln 1 þ V 2 =
Downloaded from ascelibrary.org by KANSAS STATE UNIV LIBRARIES on 07/18/14. Copyright ASCE. For personal use only; all rights reserved.
Parameter f3 , the normalized mean strength of hardened con- l ¼ 0:3 1 2 28=t (18)
crete at age t, describes the rate of strength gain of in-place concrete
with respect to time. The long-term strength–age relationship of
concrete was obtained by analyzing and plotting the long-time data Note that the portion of the curve from 34 to 40 years was based on
from four studies initiated by the Portland Cement Association as the predictions of ACI Committee 209 (1992) and Gardner (2004).
compiled by Wood (1991) (Fig. 8). The strength–age data were Substituting Eq. (17) and the value of f2 into Eq. (13), the mean
based on mean values of compressive strengths of concrete cylinders in-place compressive strength of concrete at any age t, is
and cubes stored outdoors and tested at ages ranging from 3 days to
f c,in-place ¼ 0:81 e l f1 fc9
t
34 years. Although concretes from the long-term studies reported by (19)
Wood (1991) were manufactured with pre-1957 cements, data from
more recent concretes (Lange 1994; Wood 1991) showed that both Eq. (19) describes the mean in-place strength of a cast-in-place
concretes exhibited similar long-term trends in strength development. concrete structure as a function of time, the specified compressive
strength used in the design, and the applicable normalized mean
28-day cylinder compressive strength. It should be noted that the
form of in-place strength model in Eq. (19) is quite robust, enabling
Table 4. Values of Parameter f2 from the Literature it to be used by different owners with values of normalized mean
Author f2 5 ð f c,hard, 28-day Þ Remarks 28-day cylinder strengths applicable to its practice or specific con-
crete mix design.
Campbell and Tobin (1967) 0.88 Slabs
Using Eq. (19) and data for f1 from Table 1, the mean in-place
Bloem (1968) 0.67–0.82 Poorly to well
concrete strength at various ages for cast-in-place California bridge
field-cured drilled cores
structures were calculated and shown in Table 5 and Fig. 9 (labeled
Petersons (1968) 0.804–0.84–0.88 Columns–slabs–walls
as Author). Table 5 shows that the predicted mean 28-day in-place
Davis (1976) 0.65 Columns and walls
strength of the bridge structures varied from 1:08 fc9 for 35 MPa
ACI Committee 228 (2003) 0.89–0.94 Horizontal elements
(5 ksi) concrete to 1:17 fc9 for 25 MPa (3.6 ksi) and 28 MPa (4 ksi)
and Bickley (1984)
concretes. Table 5 (column 2) shows that the maximum strength of
prediction for 35 MPa (5.0 ksi) concrete. Also, because the predi-
ctions of the present model for all three concrete strengths are in As stated previously, values of Vf1 as given in Table 1 were
better agreement with that for shallow elements than tall elements, affected by batch-to-batch as well as within-batch variations, and
it is indicative that the effect of member height on in-place strength, therefore, are more representative of the upper bound values. Batch-
represented by the indicator variable in the model by Bartlett and to-batch variations in the standard deviation estimates are those
MacGregor (1996), may have been overestimated. attributable to constituent material suppliers and the concrete pro-
The prediction of the present model was also compared with in- ducer, whereas those resulting from sampling of the batch sample,
place strength data obtained using drilled cores for the seismic ret- specimen preparation, curing, and testing constitute within-batch
rofit of San Diego-Coronado Bay Bridge in California (Table 7). The variation (ACI Committee 214 2011). For a given structure, and
approximate age of the bridge was 28 years. The core strengths were considering the data used in the present investigation, a better
converted to the equivalent in-place strengths using modification representation of the variability may be estimated using the lower
factors for the effects of length-to-diameter ratio (Fl=d ), diameter limit of the confidence interval for the standard deviation. Hindo
(Fdia ), moisture condition (Fmc ), and damage sustained during and Bergstrom (1985) suggested values of confidence levels of 75%
drilling (Fd ), in accordance with ACI Committee 214 (2010), as for ordinary structures, 90% for very important buildings, and 95%
follows: Fl=d (dried specimen) (Table 7), Fdia 5 1:0 [100-mm (4-in.) for crucial parts of nuclear power plants. However, in the present
diameter cores used], Fmc 5 0:96 (air-dried), and Fd 5 1:0 (un- investigation for bridges, the 95% confidence limit is used to maintain
damaged cores used). Table 7 lists the data. The average in-place consistency with the definition of other variables used in this study.
strength was calculated as 1:57 fc9 (Table 7). The lower limit of the 95% confidence interval on the standard
Assuming a value of normalized mean 28-day cylinder strength deviation, SL was calculated using the following equation
equal to 1.45, the predicted average in-place strength using Eq. (19) qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
is SL ¼ ðn 2 1ÞS2 =x2a=2 (21)
28-yr
f c,in-place ¼ 0:81 el f1 fc9 ¼ 0:81 1:329 1:45 fc9 ¼ 1:56 fc9 where x2a=2 5 appropriate point on the x2 distribution with (n 2 1)
degrees of freedom. x2a=2 may be read off tables of cumulative x 2
This value is close to 1:57fc9 obtained by taking cores on the bridge distribution or determined using computer-based statistical tools.
in-place. Although this result is promising, a larger number of in-place The resulting values of Vf1 are shown in Table 8. The small decrease
test results will be required to further verify the model predictions. in the Vf1 values was the result of the large sample size used in this
investigation. This should, however, not detract from the general
applicability of the proposed procedure. For example, for the
Variability of In-place Concrete Strengths 25 MPa (3.6 ksi) data and a sample size of 50, Vf1 decreases from 0.19
to 0.05 instead of 0.18 obtained for the current sample size of 673.
The variability associated with in-place strength of concrete, The mean value of the variability in the ratio of 28-day hardened
Vc,in-place may be calculated by considering the variability in the strength to cylinder strength, Vf2 was obtained from the experi-
variables f1 , f2 and f3 , denoted by their coefficients of variation, Vf1 , mental test data of Bickley (1984), Davis (1976), Drysdale (1973),
Vf2 and Vf3 . Following the general principles for calculating vari-
ability (Ang 1973; Ang and Cornell 1974), Vc,in-place is given by
Table 8. Lower Limits of 95% Confidence Interval on Variability Vf1
Lower limit of Coefficient of variation
Table 7. in-place Strength of San Diego-Coronado Bridge at 28 Years Standard 95% confidence based on lower limit of
Using Drilled Cores deviation, interval on 95% confidence interval
As-built Equivalent fc9 [MPa (ksi)] (S=fc9) standard deviation on standard deviation
specified Core in-place In-place 25 (3.6) 0.28 0.27 0.18
Pier/core strength, strength, strength, strength 28 (4.0) 0.27 0.26 0.17
location MPa (psi) MPa (psi) L/D Fl=d MPa (psi) /fc9 35 (5.0) 0.17 0.15 0.11
2–3 22.5 (3,250) 42.1 (6,100) 1.8 0.995 40.2 (5,826) 1.79
20–1 28.0 (4,000) 43.4 (6,300) 1.9 0.999 41.7 (6,040) 1.51 Table 9. Values of Coefficient of Variation Vf2
20–2 28.0 (4,000) 35.2 (5,100) 1.4 0.954 32.2 (4,669) 1.17
References Vf2 Remarks
36–2 22.5 (3,250) 35.2 (5,100) 2.0 1.0 33.8 (4,896) 1.51
36–3 22.5 (3,250) 36.5 (5,300) 2.0 1.0 35.1 (5,088) 1.57 Bickley (1984) 0.09 Horizontal elements
28–1 22.5 (3,250) 43.4 (6,300) 2.0 1.0 41.7 (6,048) 1.86 Drysdale (1973) 0.15 Columns
28–2 22.5 (3,250) 37.9 (5,500) 1.4 0.954 34.8 (5,037) 1.55 Tso & Zelman (1970) 0.11 Columns
Mean in-place strength 1.57 Davis (1976) 0.11 Columns
Acknowledgments
The writers wish to thank the following coworkers for their assis-
tance in collecting concrete cylinder and core test data: Rosme Aguilar,
Branch Chief , Concrete Materials Testing Laboratory; Jaro Simek,
Office of Earthquake Engineering (OEE), and Jeremiah Jackson,
Volunteer at OEE. Thanks are also due to Dr. Kyoung Lee for his as-
sistance in reviewing the final paper. The contents of this paper reflect
Downloaded from ascelibrary.org by KANSAS STATE UNIV LIBRARIES on 07/18/14. Copyright ASCE. For personal use only; all rights reserved.
the views of the authors and not necessarily the official views of
the California Department of Transportation.
Bartlett, F. M., and MacGregor, J. G. (1995). “Equivalent specified concrete Davis, S. G. (1976). “Further investigations into the strength of concrete in
strength from core data.” Concr. Int., 17(3), 52–58. structures.” Technical Report 42.514, Cement and Concrete Associa-
Bartlett, F. M., and MacGregor, J. G. (1996). “Statistical analysis of the tion, Slough, U.K.
compressive strength of concrete in structures.” Mater. J., 93(2), 158– Drysdale, R. C. (1973). “Variation of concrete strength in existing build-
168. ings.” Mag. Concr. Res., 25(85), 201–207.
Bickley, J. A. (1984). “The evaluation and acceptance of concrete quality Gardner, N. J. (2004). “Comparison of prediction provisions for drying
by in-place testing.” In-situ/nondestructive testing of concrete, SP-82, shrinkage and creep of normal strength concretes.” Can. J. Civ. Eng.,
V. M. Malhotra, ed., American Concrete Institute, Farmington Hills, 31(5), 767–775.
MI, 95–109. Hindo, K. R., and Bergstrom, W. R. (1985). “Statistical evaluation of the in-
Bloem, D. (1968). “Concrete strength in structures.” J. Am. Concr. Inst., place compressive strength of concrete.” Concr. Int., 7(2), 44–48.
65(3), 176–187. Lange, D. A. (1994). “Long-term strength development of pavement con-
California Department of Transportation (Caltrans). (2001a). “California cretes.” J. Mater. Civ. Eng., 10.1061/(ASCE)0899-1561(1994)6:1(78),
test method 521: Method of test for compressive strength of molded 78–87.
concrete cylinders.” Transportation Laboratory, Division of Engineering Mirza, S. A., Hatzinikolas, M., and MacGregor, J. G. (1979). “Statistical
Services, California Department of Transportation, Sacramento, CA descriptions of strength of concrete.” J. Struct. Div., 105(ST6), 1021–1037.
Æhttp://www.dot.ca.gov/hq/esc/ctms/CT_ChooseVersion.htmlæ (Jan. 24, Petersons, N. (1968). “Should standard cube test specimens be replaced
2012). by test specimens taken from structures?” Materiaux et Constructions,
California Department of Transportation (Caltrans). (2001b). “California 1(5), 425–435.
test method 539: Method of sampling fresh concrete.” Transportation Stewart, M. G. (1995). “Workmanship and its influence on probabilistic
Laboratory, Division of Engineering Services, California Department models of concrete compressive strength.” Mater. J., 92(4), 361–372.
of Transportation, Sacramento, CA Æhttp://www.dot.ca.gov/hq/esc/ Tabsh, S. W., and Aswad, A. (1997). “Statistics of high-strength concrete
ctms/CT_ChooseVersion.htmlæ (Jan. 24, 2012). cylinders.” Mater. J., 94(5), 361–364.
California Department of Transportation (Caltrans). (2005). “California Tso, W. K., and Zelman, I. M. (1970). “Concrete strength variation in actual
test method 540: Method for making, handling, and storing con- structures.” J. Proc. 67(12), 981–988.
crete compressive test specimens in the field.” Transportation Lab- Wood, S. L. (1991). “Evaluation of the long-term properties of concrete.”
oratory, Division of Engineering Services, California Department of Mater. J., 88(6), 630–643.