4 GUIDE TO EVALUATION OF STRENGTH TEST RESULTS OF CONCRETE (ACI 214R-11)

Table 3.1—Principal sources of strength variation results that are biased but uniform. Laboratory equipment and
Batch-to-batch variations Within-batch variations procedures should be calibrated and checked periodically.
Variations in characteristics and Improper sampling from the batch Testing personnel should be trained ACI Certified Technicians
proportions of ingredients: sample. or the equivalent who are evaluated routinely.
• Aggregates;
• Cementitious materials, including Variations due to fabrication
pozzolans; and techniques: CHAPTER 4—ANALYSIS OF STRENGTH DATA
• Admixtures. • Substandard conditions;
• Incorrect tools; 4.1—General
Changes in w/cm caused by: • Poor quality, damaged, or A sufficient number of tests is needed to accurately indicate
• Poor control of water; distorted molds;
• Variation of aggregate stockpile • Nonstandard molding and the variation of the concrete strength and permit application
moisture conditions; consolidation; and of appropriate statistical procedures for interpreting the test
• Variable aggregate moisture • Incorrect handling of fresh test
measurements; and samples. results. Statistical procedures provide a sound basis for
• Retempering. determining the potential quality and strength of the concrete
Differences in curing:
Variations in mixing, transporting, • Delays in beginning initial curing; and for expressing results in the most useful form.
and sampling: • Temperature variation;
• Mixing time and speed; • Variable moisture control;
• Distance between plant and • Nonstandard initial curing; 4.2—Statistical functions
placement; • Delays in bringing cylinders to A strength test result is the average strength of all specimens
• Road conditions; and the laboratory;
• Failure to obtain a representative • Rough handling of cylinders in of the same age, fabricated from a sample taken from a single
sample from the batch. transport; and batch of concrete. A strength test cannot be based on only
• Improper final curing.
Variations in placing, and one cylinder. ACI 318 states that a “strength test shall be the
consolidation:* Variations in sample testing: average strength of at least two 6 x 12 in. (150 x 300 mm)
• Chute, pump, or buggy; • Uncertified tester;
• Specimen surface preparation;
• Internal or external vibration; and cylinders, or three 4 x 8 in. (100 x 200 mm) cylinders of the
• Different operators. • Inadequate or uncalibrated testing same concrete batch tested at the same age…”
equipment;
Variations in concrete temperature • Nonstandard loading rate; and In this guide, strength test results are assumed to follow a
*
and curing: • Poor record keeping. normal distribution. Figure 4.1 shows the “bell-shaped
• Season;
• Ambient humidity; and curve” characteristic of the normal distribution. The normal
• Wind speed. distribution is mathematically defined completely by two
*
Applies to in-place strength of the structure. statistical parameters: the population mean μ and standard
deviation σ. A mathematical characteristic of the normal
distribution is that 68.27% of the data lies within 1 standard
at early ages. These differences will not be reflected in
deviation from the mean, and that 95.45% of the data is
specimens fabricated and stored under standard laboratory
within 2 standard deviations. On the chart are notes indicating
conditions (ASTM C31/C31M) but will be reflected in
the number of samples n; the sample standard deviation s,
strength assessments using nondestructive testing methods
which is an estimate of σ; the coefficient of variation V; and
or strength testing of cores.
the sample mean X, which estimates μ. Also on the chart is a
histogram of the measured test results sorted into strength
3.3—Influence of within-batch variations
on concrete strength ranges. Each dot represents a test, which indicates that this
Testing to determine compliance with contract specifications data set is normally distributed. Plotting histograms is one of
should be conducted according to the methods specified in the easiest ways to check the data for normality.
the contract documents, such as ASTM C31/C31M and When evaluating concrete strength tests, a normal
C39/C39M. Acceptance tests assess the compressive distribution cannot always be assumed. A normal distribution
strength of the concrete when prepared and cured under is appropriate in most cases when the concrete strength does
standard conditions, not the in-place strength. It is not the not exceed 10,000 psi (70 MPa) (Cook 1982). However,
purpose of such tests to determine the in-place strength of the Cook (1989) further reported that a skewed distribution may
structure, but rather to assure that the concrete delivered and result for high-strength concrete when the limiting factor is
placed was the concrete specified. Deviations in field the aggregate strength. When data are not symmetrical about
sampling, specimen preparation, curing, and testing procedures the mean, but concentrated to the right or left, the distribution is
may cause lower strength test results. Field sampling, said to be skewed. When a distribution curve is either too
making, curing, handling, and testing of specimens should peaked or too flat, kurtosis is said to exist. Data exhibiting
be performed by ACI Certified Technicians or equivalently skewness, or kurtosis, may not be normally distributed; and
trained and certified personnel. Provisions for maintaining any analysis presuming a normal distribution may be very
specified curing conditions should be made. Specimens from misleading. Skewness and kurtosis should be considered for
concrete that are hardening and gaining strength should not statistical evaluation of high-strength concretes. Cook (1989)
be disturbed until sufficient strength is achieved to prevent provides simplified equations that calculate relative skewness
cylinder damage (ASTM C31/C31M). and kurtosis for a particular data set. Further discussion of
Using accurate, properly calibrated testing equipment and these topics is beyond the scope of this document; interested
proper sample preparation procedures is essential. Test results readers should consult statistical references.
with low variability do not necessarily indicate accurate test When there is satisfactory control of a concrete mixture,
results. A routinely applied, systematic error can provide strength test values tend to cluster near the average value