SP-284 — 5

Using Historical Cylinder Strength Data for

Structural Evaluation

F. Michael Bartlett

Synopsis: Section 20.2 of ACI 318-08 gives provisions for determining required dimensions and material
properties for the strength evaluation of an existing concrete structure. Subsection 20.2.3 states “If required,
concrete strength shall be based on results of cylinder tests from the original construction or tests of cores
removed from the part of the structure where the strength is in question. For strength evaluation of an existing
structure, cylinder or core test data shall be used to estimate an equivalent fc’.” The commentary cites two
methods presented in ACI 214.4R for determining an equivalent-to-specified fc’ using cores from an existing
structure. There are no criteria provided or references cited, however, suggesting how to compute an equivalent
fc’ based on cylinder strength data. The paper first identifies necessary conditions for using original cylinder test
data to determine a concrete compressive strength for structural evaluation. To investigate strength compliance
during initial construction, methods to compute an equivalent-to-specified fc’ value are presented that are based
on inverting the conventional acceptance criteria given in ACI 318-08 Section 5.6.3.3. To evaluate an older
structure, a method to determine an equivalent-to-specified strength based on the lower-bound fractile of the
concrete strength represented by fc’ is presented.

Keywords: concrete cylinder tests, equivalent specified concrete strength, specified concrete
compressive strength, strength evaluation, structural concrete

5-1
Bartlett

F. Michael Bartlett, FACI, is Professor and Associate Chair (Undergraduate) in the Department of Civil and
Environmental Engineering, University of Western Ontario, London, Canada. He is a member of ACI
Committees 214 and 562. As a member of ACI 214, he has led the development of the 2003 and 2010 editions of
the ACI 214.4R “Guide for Obtaining Cores and Interpreting Compressive Strength Results”.

INTRODUCTION
Section 20.2 of ACI 318-08 gives provisions for determining required dimensions and material properties for
the strength evaluation of an existing concrete structure. Subsection 20.2.3 states “If required, concrete strength
shall be based on results of cylinder tests from the original construction or tests of cores removed from the part of
the structure where the strength is in question. For strength evaluation of an existing structure, cylinder or core
test data shall be used to estimate an equivalent fc’. The method for obtaining and testing the cores shall be in
accordance with ASTM C42.” The commentary to this subsection draws the user’s attention to two methods
developed by ACI Committee 214 for determining fc’ from cores taken from an existing structure. There are no
criteria provided or references cited, however, concerning the computation of an equivalent fc’ based on cylinder
strength data.

This paper is intended to rectify this deficiency. It starts with a brief discussion of potential pitfalls that can
occur when using original cylinder test data for strength evaluation of an existing structure, and identifies
conditions necessary to follow this procedure. Then, paralleling the structure of ACI 214.4R-10 (ACI 214,
2010), it offers two procedures: one applicable for investigating low cylinder strength test results in new
construction, based on the provisions of ACI 318-08 (ACI 318, 2008) Section 5.6.3.3, and the other applicable
for the strength evaluation of an existing structure based on determining a lower-bound estimate of the fractile of
the concrete strength represented by fc’ as derived from the cylinder strength data. A brief comparison of the
results obtained using these two methods is then presented and discussed.

RESEARCH SIGNIFICANCE
The provisions in Section 20.2 of ACI 318-08 permit the determination of the compressive strength of concrete
in an existing structure using the original cylinder strength test results. Although this procedure may be flawed
because it does not directly measure the in-place strength of concrete in a structure, it provides a low-cost
approach that is clearly welcome in practice. Current design criteria, including particularly the calibration of
strength reduction factors used to verify structural safety under ultimate loading conditions, recognize that the
specified compressive strength of concrete is a lower bound on the cylinder strength. This paper therefore
proposes two procedures for the computation of an equivalent fc’ based on cylinder strength data for two
extremely common circumstances: (1) when low strength test results occur during construction of a structure;
and, (2) when an equivalent-to-specified strength is sought for the strength evaluation of an existing structure
after it has been occupied.

NECESSARY CONDITIONS TO USE CYLINDER STRENGTH DATA

The use of cylinder data to assess the strength of concrete in an existing structure is potentially deceiving
because cylinders give no direct indication of actual in-place strengths. If the concrete was improperly cured or
has deteriorated over time, the actual in-place strength could be markedly less than the strengths obtained by
testing the original cylinders. Conversely the actual in-place strengths could be markedly greater than the original
strengths, as cement continues to hydrate over time, particularly for coarse-ground cements produced before the
1970s. For example, 40-year old cores from the main foundations of Lions’ Gate Bridge in Vancouver indicated
compressive strengths between 39 and 86 MPa, even though the original specified strength was believed to be 21
MPa (Buckland, 1981).

5-2
 Using Historical Cylinder Strength Data for Structural Evaluation

Thus a decision to determine the concrete strength in an existing structure from the original cylinder test results
would be appropriate only if the following conditions are satisfied:
1. Complete cylinder test records from a reputable testing company are available. The reports should
include the test specification adopted, typically ASTM C39/C39M (ASTM, 2010 or earlier), and details
including the type of fracture, if different from the typical cone failure, and the presence of any defects in
the specimen or caps.
2. Visual inspection indicates no evidence that the concrete has deteriorated from its original as-placed
condition.
3. If there is more than one concrete strength class in the structure – footings, columns and slabs in a
building often have unique specified strengths, for example – the reported cylinder test results can readily
be classified according to the strength classes.

It is further assumed in this paper that the strength evaluation being conducted is to assess the safety of the
structure when subjected to load combinations that do not involve earthquake loads, so it is appropriate to obtain
a lower-bound estimate of the concrete compressive strength. If developing a model for analysis of the structure,
or checking serviceability limit states, it is usually more appropriate to use the estimated mean concrete strength
instead of a lower bound value. It is often unsafe to use a lower-bound estimate of concrete strength when
assessing the probable seismic performance of an existing structure because this causes the capacity of concrete
components to be underestimated, and so may underestimate the demands on the energy-dissipating elements of
the structure (e.g., Sezen et al, 2011)

METHOD 1: ASSESSMENT OF STRENGTH AT TIME OF CONSTRUCTION

Section 5.6.3.3 of ACI 318-08 provides the following acceptance criteria for concrete, based on the results of
compressive cylinder tests:
1. The average of any set of three consecutive strength tests must exceed fc’, and
2. For fc’ < 5000 psi (35 MPa), no individual strength test may fall by more than 500 psi (3.5 MPa) below fc’,
or
3. For fc’ > 5000 psi (35 MPa), no individual strength test may fall by more than 0.10 fc’ below fc’.
A strength “test”, in the context of these criteria, is the average of two or more cylinder breaks observed when the
concrete has a specified age, typically 28 days.

These acceptance criteria have remained unchanged since the 1971 edition of ACI 318 (ACI 318, 1971). They
are intended to balance the risks of the consumer, who doesn’t want to accept inferior quality concrete in the
structure, and the producer, who doesn’t want to see concrete of acceptable quality rejected. Both the consumer’s
risk and the producer’s risk can be reduced by testing larger numbers of specimens, so these acceptance criteria
also implicitly represent a further optimization balancing the expected costs of a faulty outcome with the costs of
testing additional specimens. The method based on these criteria may therefore not be appropriate for assessing
older structures, when the producer is no longer a major stakeholder and the consequences of overestimating the
strength on the structural safety may be more severe, but are potentially useful when assessing new construction,
where the producer is a major stakeholder.

If low-strength concrete test results occur during construction, Section 5.6.5.2 of ACI 318-08 makes provision
for investigating the in-place concrete strength by tests of cores drilled from the area in question if “calculations
indicate that load-carrying capacity is significantly reduced”. ACI 318-08 does not provide guidance on how to
compute a value of specified compressive strength to be used in such calculations, but in this circumstance it is
credible simply to invert the traditional concrete-strength-acceptance criteria. Thus the equivalent-to-specified fc’
is taken as the lesser of:
1. Every arithmetic average of any three consecutive strength tests, and
2. For all single strength test strengths less than 4500 psi (31.5 MPa), the test strength plus 500 psi (3.5
MPa), or
3. For all single test strengths greater than 4500 psi (31.5 MPa), the test strength multiplied by (1/0.9).
This approach can be used to determine the equivalent-to-specified strength for a class of concrete used in a
specific project. If the construction records are detailed, it may be possible to assign unique equivalent-to-

5-3
Bartlett

specified strengths to specific components in a structure to assess their structural adequacy, as illustrated in the
example below.

Example Application of Method 1

Table 1 summarizes cylinder strength data from a series of consecutive concrete placements during the
construction of a concrete high-rise building. The data have been generated by numerical simulation but are
based on a real project where the specified 28-day concrete compressive strength was 12000 psi (82.7 MPa).
None of the 28-day test results, shown in the fourth column, satisfy this requirement.

Table 1 — Data for Example Illustrating Method 1

Placement 28-day Min. Avg. Test/0.9 Equiv. fc’

Test Date Location Test (psi) of 3 (psi) (psi) (psi)
1 24-Oct-10 Cols B1 &B2 10540 10490 11711 10490
2 30-Oct-10 Cols C1 & C2 11160 10367 12400 10367
3 12-Nov-10 Bases D8 & E7 9770 10073 10856 10073
4 14-Nov-10 Cols C6 & C7 10170 9993 11300 9993
5 19-Nov-10 Cols B6 & B7 10280 9913 11422 9913
6 20-Nov-10 Cols E3 & E4 9530 9763 10589 9763
7 25-Nov-10 Cols C3 & C4 9930 9723 11033 9723
8 26-Nov-10 Cols B3 & B4 9830 9523 10922 9523
9 8-Dec-10 Cols B5 & E5 9410 9107 10456 9107
10 9-Dec-10 Cols D4 & D5 9330 9107 10367 9107
11 10-Dec-10 Cols F1 & F2 8580 9107 9533 9107
12 12-Dec-10 Cols D1 & D2 9600 9170 10667 9170
13 15-Dec-10 Cols E6 & F6 9450 9210 10500 9210
14 16-Dec-10 Cols C5 & F5 10020 9537 11133 9537
15 17-Dec-10 Bases G8 & H7 9140 9360 10156 9360
16 18-Dec-10 Cols F3 & F4 9940 9360 11044 9360
17 21-Dec-10 Cols D8 & E7 9000 9000 10000 9000

The analysis using Method 1 is also presented in Table 1. The computed values are not rounded to the nearest
10 psi (0.05 MPa), as is usual practice, to expose the calculation procedures. The “Minimum Average of 3” is the
minimum value obtained by inverting the first acceptance criterion assuming the test result is the first, the second,
or the third value in the set of three used to compute the average. For example, the three averages involving the
Test 3 result are for: Tests 1, 2 and 3; 2, 3 and 4; and 3, 4, and 5. They equal 10490, 10367 and 10073 psi (72.3,
71.5 and 69.4 MPa), respectively, so the minimum value of 10073 psi is shown. Values shown for the first two
tests and last two tests involve the averaging of three or fewer specimens in the set: for example, for Test 1, the
minimum average is the minimum of: the Test 1 result; the Test 1 and 2 results averaged; or, the Test 1, 2 and 3
results averaged. The “Test/0.9” value is obtained by inverting the second acceptance criterion for the case
where the test strengths exceed 4500 psi (31.5 MPa). The equivalent fc’ value shown in the right column is the
lesser of the “Minimum Average of 3” and the “Test/0.9” values.

The range of equivalent fc’ values obtained using the method may be practically useful in resolving the strength
compliance issue. For example, elements cast on or before November 14th have equivalent fc’ values above
10000 psi (69. MPa) and so may be sufficient given the factored demands. The elements cast between December
8th and 12th with equivalent fc’ values less than 9200 psi (63.4 MPa) are more likely to be insufficient, however,
and so more likely to require remedial measures.

5-4
 Using Historical Cylinder Strength Data for Structural Evaluation

METHOD 2: POST-CONSTRUCTION STRENGTH ASSESSMENT

While Method 1 has potential for assessing new construction, it may not be appropriate for assessing an older
structure, when the producer is no longer a major stakeholder and the consequences of overestimating the
strength on the structural safety may be more severe. In this instance it is appropriate to develop a method to
compute a suitable lower-bound exclusion fractile as ACI 214.R4-10 (ACI 214, 2010) does for cores. This
requires two steps: determining what fractile of the concrete cylinder strength corresponds to fc’, and then,
knowing that fractile, determining how to compute the equivalent specified strength that corresponds to a lower-
bound (i.e., safe) estimate of that fractile.

Fractile of Cylinder Strength Represented by fc’

The definition of a fractile is illustrated using the probability density function of cylinder test strengths shown
in Figure 1. Cylinder strengths are typically assumed to be normally distributed, as shown (e.g., ACI 214, 2002).
The “fractile represented by fc’” is the area under the probability density function to the left of fc’, shown shaded
and denoted A in the figure. It represents the probability that a test cylinder strength will be less than or equal to
fc’. Typically the average cylinder strength is much greater than the specified strength fc’, because Section 5.3 of
ACI 318 requires the concrete producer to achieve a required average strength so that the likelihood of not
passing the strength acceptance criteria, as shown above, is in the order of 1%. This overstrength is reflected in
the reliability analyses used to compute the strength reduction factors in Section 9.3.2 of ACI 318-08 and so must
be accounted for in any calculation of an equivalent-to-specified strength that will be used with these resistance
factors.
Probability Density

f ’c

Cylinder Test Strength

Figure 1 — Fractile of cylinder strength represented by fc’.

Two approaches to determine this fractile are as follows:

1. Compute fractile values that are consistent with the bias coefficients and coefficients of variation of
concrete cylinder strength data for ordinary ready-mix concretes reported by Nowak et al. (2005), which
were “analysed” to determine the statistical parameters used for calibration of ACI 318-08.

2. Compute fractile values that are consistent with the required average compressive strengths, fcr’, specified
in Clause 5.3.2.1 of ACI 318-08.

Fractile Based on Statistical Parameters for Cylinder Strength Data Reported by Nowak et al. (2005)

Nowak et al. (2005) collected a significant quantity of concrete cylinder strength data to derive the strength
reduction factors now presented in Section 9.3.2 of ACI 318-08. The actual calibration was based on parameters
obtained by “analysis” of the raw strength data. The bias coefficient, representing the ratio of the mean value to

5-5
Bartlett

the nominal value, was assumed to vary from 1.31 to 1.08 as the specified strength increases from 3000 psi to
12000 psi (20.7 MPa to 82.7 MPa) as shown in Figure 2(a). The coefficient of variation, which is the ratio of the
standard deviation to the mean, was deemed to reduce from 0.15 to 0.11 over this range as shown in Figure 2(b).
The fractile represented by fc’ therefore increases as the specified strength increases: assuming concrete strengths
to be normally distributed, the probability of the actual cylinder strength being less than a specified value of 3000
psi (20.7 MPa) is 8.2% while the probability of the actual cylinder strength being less that a specified value of
12000 psi (82.7 MPa) is 25%, as shown in Figure 2(b).

1.4 0.30

1.2
0.25

Coefficient of Variation, Fractile

1.0 Fractile
0.20
Bias Coefficient

0.8
0.15
0.6 Coefficient of Variation

0.10
0.4

0.05
0.2

0.0 0.00
3,000 4,500 6,000 7,500 9,000 10,500 12,000 3,000 4,500 6,000 7,500 9,000 10,500 12,000
Specified Compressive Strength (psi) Specified Compressive Strength (psi)

Figure 2 — Statistical Parameters for cylinder strength assumed by Nowak et al (2005) for resistance factor
derivation: (a) Bias coefficient; (b) Coefficient of variation and associated fractile value.

It is more realistic to determine fractiles based on the actual data, which don’t particularly follow the derived
parameters shown in Figure 2. The statistical parameters for ordinary strength concretes shown in Table 2, taken
from Table 5-2 of Nowak et al., are based on over 11,000 standard cylinder tests. The bias coefficients are
relatively constant for specified strengths between 3500 and 5000 psi, and there is no clear relationship between
the specified strength and the coefficient of variation or standard deviation. The fractile represented by fc’ ranges
from 4.4% to 15.9%. For specified strengths between 3000 and 5000 psi (20.7 and 35 MPa), the average fractile
is 9.1%.

Table 2 — Fractiles Consistent with Ordinary Strength Concrete Data Reported by Nowak et al (2005)

fc’(psi) n Bias CoV Std dev fractile

3000 4016 1.33 0.145 0.193 0.044
3500 527 1.24 0.115 0.143 0.046
4000 2784 1.21 0.155 0.188 0.131
4500 1919 1.19 0.160 0.190 0.159
5000 1722 1.22 0.125 0.153 0.075
6000 130 1.22 0.075 0.092 0.008
Total 11098
1000 psi = 6.985 MPa

The statistical parameters shown in Table 3, taken from Table 5-4 of Nowak et al. (2005), are for high-strength
concretes. The numbers of tests for each specified strength, n, are typically smaller than those for the ordinary
concrete strength data shown in Table 2. The bias coefficients do not display a consistent trend with increasing
fc’ and the value shown for the specified strength of 12000 psi (84.7 MPa) is quite low. The standard deviations
and coefficients of variation are quite consistent, irrespective of the specified concrete strength. The fractiles
represented by fc’ range from 8.4% to 36.2%, and the average for specified strengths between 7000 and 10000 psi
(48.3 and 69 MPa) is 12.5%.

5-6
 Using Historical Cylinder Strength Data for Structural Evaluation

Table 3 — Fractiles Consistent with High Strength Concrete Data Reported by Nowak et al (2005)

fc’ (psi) n Bias CoV Std dev fractile

7000 210 1.19 0.116 0.138 0.084
8000 753 1.09 0.088 0.096 0.174
9000 73 1.16 0.100 0.116 0.084
10000 635 1.13 0.115 0.130 0.159
12000 381 1.04 0.109 0.113 0.362
Total 2052
1000 psi = 6.985 MPa

The statistical parameters shown in Table 4, taken from Table 5-3 of Nowak et al. (2005), are for plant-cast
concretes used for precast concrete construction. The number of test results for a specified strength of 5500 psi
(37.9 MPa) is very small, which may explain the low coefficient of variation and very low fractile for that
strength category. The average fractile represented by fc’ for the other three strength categories is 6.6%.

Table 4 — Fractiles Consistent with Plant-Cast Concrete Data Reported by Nowak et al (2005)

fc’ (psi) n Bias CoV Std dev fractile

5000 330 1.32 0.105 0.139 0.010
5500 26 1.20 0.045 0.054 0.000
6000 493 1.16 0.080 0.093 0.042
6500 325 1.08 0.070 0.076 0.145
Total 1174
1000 psi = 6.985 MPa

Thus the fractile of the compressive cylinder strengths represented by fc’ typically varies between 4% and 17%
according to the data collected by Nowak et al (2005).

Fractile Consistent with the Required Average Compressive Strength Specified in ACI 318-08

Clause 5.3.2.1 of ACI 318-08 specifies a required average compressive strength fcr’ to be used as the basis for
proportioning the concrete mixture that is intended to ensure that the likelihood of failing each of the concrete
acceptance criteria, described in the previous section, is in the order of 1%. For fc’ ≤ 5000 psi (35 MPa),
equations presented in Table 5.3.2.1 require that the required average compressive strength be the larger of:

fcr’ = fc’ + 1.34 ss (ACI 318 Eq. 5-1)

or

fcr’ = fc’ + 2.33 ss – 500 psi (ACI 318 Eq. 5-2)

where ss is the sample standard deviation of cylinder strengths obtained from test batches. ACI 318 Eq. 5-1
governs for ss <505 psi, and implies that the specified strength fc’ represents the 9% fractile of the cylinder
strength distribution, irrespective of the magnitude of fc’. ACI 318 Eq. 5-2 governs for larger sample standard
deviations and implies the specified strength represents a fractile that is constant for a given standard deviation,
irrespective of fc’, but reduces from 9% as ss increases.

For fc’ > 5000 psi (35 MPa), equations presented in Table 5.3.2.1 require that the required average compressive
strength be the larger of:

fcr’ = fc’ + 1.34 ss (ACI 318 Eq. 5-1)

5-7
Bartlett

or

fcr’ = 0.9 fc’ + 2.33 ss (ACI 318 Eq. 5-3)

ACI 318 Eq. 5-1 governs for ss < 0.101 fc’ and again implies that the specified strength fc’ represents the 9%
fractile of the cylinder strength distribution, irrespective of the magnitude of fc’. ACI 318 Eq. 5-3 governs for
larger sample standard deviations and implies the specified strength represents a fractile that is constant for a
coefficient of variation of the cylinder strength, irrespective of fc’, but reduces from 9% as ss increases.

It has been shown (Bartlett and MacGregor, 1996) that the actual mean strength of concrete mixes with a given
standard deviation does not correspond particularly well to the required strengths computed using ACI 318 Eqs.
5-1 to 5-3. This is also apparent from the values shown in Tables 3, 4 and 5: while the overall average fractile
based on 14,324 cylinder tests is approximately 9%, the individual average fractiles can be either markedly lower
or markedly higher than 9%.

Practically, the procedure to determine an equivalent specified strength should be based on a single value for
the fractile representing the percentage of cylinder tests less than fc’. Based primarily on the data presented in
Tables 3, 4 and 5, the most suitable single value is the 9-10% fractile. This is consistent with the fractile implicit
in ACI 318 Eq. 5-1, which often governs the required average compressive strength for a given specified
strength.

Procedure for Computing an Equivalent-to-specified Compressive Strength

With the target fractile defined as 9-10%, it remains only to develop a procedure to determine the associated
equivalent specified strength. The conventional approach to estimate a fractile value (e.g., ACI 214.4-R10) is the
Tolerance Factor Approach involving a tolerance factor, K, that accommodates the uncertainties of both the
sample mean and the sample standard deviation caused by smaller sample sizes:

fc’ = f c – K sc (1)

where f c is the mean and sc is the sample standard deviation of the cylinder strengths. Table 5 presents suitable
values of K for a known sample size n that correspond to a 75% confidence level and have been normalized to
eliminate the sample size effect for more than 30 specimens. The values shown correspond to a fractile of 10%,
which is sufficiently close to the target 9-10% fractile.

Table 5 — Tolerance Factors K for Use in Eq. (1)

n 3 4 5 6 8 10 12 15 18 21 24  30
K 2.16 1.84 1.70 1.61 1.51 1.45 1.40 1.37 1.33 1.32 1.30 1.28

ACI 214.4-R10 (2010) provides a procedure to modify Equation (1) to account for the uncertainty of strength
correction factors, which does not diminish as the sample size increases. There are no strength correction factors
involved with the interpretation of concrete cylinder strength data, so this modification is unnecessary.

ACI 214.4-R10 (2010) also provides an alternate procedure to determine the equivalent specified strength that
generally gives greater values than the Tolerance Factor Approach because core strengths tend to be more
variable than the actual in-place strength. While it would be possible to devise a similar alternative procedure,
based on the precision statements for cylinder testing given in ASTM C39, this seems unwarranted.

5-8
 Using Historical Cylinder Strength Data for Structural Evaluation

Example Application of Method 2

Original construction records give the following eight cylinder strength test results for a class of concrete in a
structure: 6030, 5460, 6500, 6190, 6570, 5030, 5260, and 5720 psi (note 1000 psi = 6.895 MPa). The original
records are complete and indicate that the testing was done in accordance with ASTM C39. Visual inspection
indicates no evidence that the concrete has deteriorated from its original as-placed condition.

To determine an equivalent specified strength value from these data to use for strength evaluation of the
structure using Method 2:
 The mean and standard deviation of the cylinder strengths are computed to be 5845 psi and 570 psi,
respectively
 From Table 5, for n = 8, K = 1.51
 The equivalent specified strength, from Equation (1), is (5485 – 1.55x570 =) 4984 psi, say 4980 psi.
If these data are analysed using Method 1, the minimum equivalent specified strength is 5337 psi, say 5340 psi.
In this case, the additional safety margin provided using Method 2 corresponds to an equivalent specified strength
value that is 6.7% less than the minimum specified strength value obtained using Method 1.

COMPARISON OF RESULTS OBTAINED USING METHODS 1 AND 2

Figure 3 shows a comparison of results obtained using Methods 1 and 2. The data shown are derived by
simulation: cylinder strengths were generated assuming the cylinder strength population has a mean value of
5870 psi (40.5 MPa) and a standard deviation of 587 psi (4.05 MPa). For the assumed standard deviation, the
mean strength is equal to the required target strength of a mixture with fc’ specified to be 5000 psi (35 MPa) in
accordance with ACI 318 Eq. 5-1. The mean strength predicted using Method 1 reduces as the sample size
increases as shown in Figure 3(a) because the equivalent-to-specified strength is based on the minimum average
of three consecutive tests, or the minimum test strength, and lower realizations of these minima are likely to be
realized as the sample size increases. This may not in practice be a serious issue because, as demonstrated in the
example involving Method 1, the equivalent-to-specified strength can be computed uniquely for each individual
cylinder strength test result, so the minimum value shown in Figure 3(a) need not be applied to the entire data set.
The mean strength predicted using Method 2 increases as the sample size increases, particularly for relatively
small sample sizes, because the Tolerance Factors, K, shown in Table 5 reduce as the sample size increases. The
standard deviation of the predicted strength, shown in Figure 3(b), is a measure of the accuracy of the predicted
strength and reduces as the sample size increases for both Methods 1 and 2. This reduction is particularly
apparent when Method 2 is used because the larger sample sizes decrease the variability of the sample mean and
sample standard deviation of the cylinder strengths, f c and sc, respectively.

5500 400
Std. Deviation of Predicted Strength (psi)

Method 1
Mean Predicted Strength (psi)

5250 300
Method 1

5000 200
Method 2 Method 2

4750 100

4500 0
0 5 10 15 20 25 0 5 10 15 20 25
Number of Tests Number of Tests

Figure 3 — Comparison of results obtained using Methods 1 and 2: (a) Mean predicted strength; (b) Standard
deviation of predicted strength.

5-9
Bartlett

SUMMARY
This paper has presented two approaches for determining an equivalent-to-specified strength from original
cylinder strength test data for the strength evaluation of a concrete structure. Although permitted by Section 20.2
of ACI 318-08, this procedure creates potential pitfalls because cylinders give no direct indication of actual in-
place strengths. It is not recommended that either approach be considered unless: (1) complete cylinder test
records are available; (2) visual inspection indicates no evidence that the concrete has deteriorated from its
original as-placed condition; and, (3) the available cylinder strength data can be readily classified according to
the concrete strength classes present in the structure.

The first method proposed simply inverts the strength acceptance criteria given in Section 5.6.3.3 of ACI 318-
08 to compute an equivalent specified strength. It is recommended for evaluating components of an existing
structure when these strength acceptance criteria are not met. It has the potential advantage of allowing unique
equivalent specified strength values to be computed for each available cylinder strength test result, and so may
facilitate distinguishing regions of the structure where the concrete strength, although less than specified, is
sufficient to resist the design loadings.

The second method proposed involves computation of a suitable lower-bound of the fractile of the cylinder
strength probability distribution represented by the specified cylinder strength fc’. Cylinder test data reported by
Nowak et al (2005), gathered as part of the calibration that derived the strength reduction factors in ACI 318-08,
suggest that the specified strength represents approximately the 9-10% fractile of the cylinder strength
distribution. A tolerance factor approach is proposed to determine the 75% confidence limit on the 10% fractile
value. This approach gives equivalent specified strengths that are less than those obtained using the first method,
and so is recommended for the strength evaluation of an existing structure after construction and occupancy,
when the original ready-mix concrete producer is no longer a significant stakeholder.

ACKNOWLEDGMENTS
The author gratefully acknowledges financial support from the Natural Sciences and Engineering Research
Council of Canada (NSERC) in the form of a Discovery Grant.

The author also gratefully acknowledges the support and friendship provided by Dr. Andrew Scanlon over the
past two decades. Andy has cheerfully and effectively advised and mentored a number of academics, particularly
in their early careers as Assistant Professors. I am sincerely grateful for Andy’s genuine interest in, and support
for, my path through academia.

REFERENCES
ACI Committee 214, 2010, “Guide for Obtaining Cores and Interpreting Compressive Strength Results (ACI 214.4R-10)”,
American Concrete Institute, Farmington Hills, Michigan, 17 pp.

ACI Committee 214, 2002, “Evaluation of Strength Test Results of Concrete (ACI 214R-02)”, American Concrete Institute,
Farmington Hills, Michigan, 20 pp.

ACI Committee 318, 2008, “Building Code Requirements for Structural Concrete (ACI 318-08) and Commentary”, American
Concrete Institute, Farmington Hills, Michigan, 465 pp.

ACI Committee 318, 1971, “Building Code Requirements for Reinforced Concrete (ACI 318-71), American Concrete
Institute, Detroit, Michigan, 78 pp.

American Society for Testing and Materials (ASTM) International, 2010, “Standard Test Method for Compressive Strength of
Cylindrical Concrete Specimens”, American Society for Testing and Materials (ASTM C39/C39M)”, ASTM International,
West Conshohocken, PA, 7 pp.

5-10
 Using Historical Cylinder Strength Data for Structural Evaluation

Bartlett, F.M. and MacGregor, J.G., 1996, “Statistical Analysis of the Compressive Strength of Concrete in Structures”, ACI
Materials Journal, V. 93, No. 2, pp. 158-168.

Buckland, P.G., 1981, “The Lions’ Gate Bridge – Investigation”, Canadian Journal of Civil Engineering, Vol. 8, No. 2, pp.
241-256.

Nowak, A. S.; Szerszen, M. M.; Szeliga, E. K.; Szwed, A.; and Podhorecki, P. J., 2005, “Reliability-Based Calibration for
Structural Concrete,” Report No. UNLCE 05-03, University of Nebraska, Lincoln, NE, Oct. 2005.

Philleo, R. E., 1981, “Increasing the Usefulness of ACI 214: Use of Standard Deviation and a Technique for Small Sample
Sizes,” Concrete International, Vol. 3, No. 9, pp. 71-74.

Sezen, H., Hookham, C., Elwood, K., Moore, M. and Bartlett, M., 2011, “Core Testing Requirements for Seismic Evaluation
of Existing Reinforced Concrete Structures’’, submitted to Concrete International, 28 December 2010.

5-11
Bartlett

5-12