ACI MATERIALS JOURNAL
Title no. 102-M25
Three-Day Prediction of Concrete Compressive
Strength Evolution
by Marco Viviani, Branko Glisic, and Ian F. C. Smith
Although there are several procedures predicting concrete
compressive strength, reliable methodologies involve either extensive
testing or voluminous databases. This paper presents a simple and
efficient procedure to evaluate the activation energy and the rate
constant of concrete. These two parameters can be used for a rapid
prediction of the mechanical properties of concrete and particularly
the evolution of compressive strength. They also allow separation
of effects due to physical phenomena such as humidity loss. The
procedure uses an experimentally-determined parameter called
“hardening time” as an indicator of equivalent maturity when
comparing two hardening profiles. Test results from specimens of
six concrete types validate the approach.
Keywords: hydration; maturity; strength.
INTRODUCTION
At early age, the mechanical properties of cement-based
materials are time-dependent and involve hydration. The
hydration process is a thermally-activated reaction that may
be described by the Arrhenius equation. This equation
establishes the progression of a chemical reaction in terms of
rate of reaction k.1 The integral over time of the rate of
reaction gives the degree of reaction. Two independent and
parallel research areas have been generated through applying
degree of reaction indexes in this research. For the purposes
of this paper, they are called “predictions” and “separation of
effects.” Predictions of mechanical properties of concrete are
possible based on the empirical relationship between the
degree of reaction (hydration) and physical properties such
as compressive strength, tensile strength, and elastic
modulus.2-6 Separation of effects involves decoupling the
contributions to the total deformation of a physical and
chemical phenomenon during hardening.7 Unfortunately, the
separation of an effect cannot be done by direct comparison of
deformation time-histories, measured in concrete pours that
are hardening in different environments. The effects of the
temperature after similar elapsed times of hydration change
with the thermal expansion coefficient (TEC), and this
coefficient depends on the degree of hydration.8 To perform
predictions and separate effects, knowledge of maturity
indexes is required. Maturity indexes need to be determined
experimentally for each concrete type. This article describes
a new methodology to determine two common maturity
indexes. These indexes lead to the prediction of the evolution
of compressive strength in six different concretes.
RESEARCH SIGNIFICANCE
A maturity method is used to predict the compressive
strength evolution of concrete. Values for the activation
energy and the rate of reaction are necessary to implement
this approach. Determination of these values usually requires
either extensive tests or large databases. This has resulted in
ACI Materials Journal/July-August 2005
TECHNICAL PAPER
limited use of maturity methods. A simple and fast method-
ology to determine these values and consequently predict
compressive strength evolution is presented. More timely
knowledge of compressive strength evolution will lead to
savings during construction and improve safety.
BACKGROUND
The Arrhenius equation states that the rate of a chemical
reaction k increases exponentially with absolute temperature,
regardless of the degree of reaction already obtained (refer to
Eq. (1))
–E
k = A exp --------a- (1)
RT
where
A = frequency factor, s–1;
Ea = activation energy (KJ/mole);
k = reaction rate;
R = gas constant (KJ × mole–1 × K–1); and
T = absolute temperature, K.
The degree of reaction is calculated by integrating Eq. (1)
over time. The rate of reaction k is constant when the tempera-
ture of the hydration process is constant (T = Tr = constant
imply k = kr = constant). Equation (2) uses kr to predict the
compressive strength. This empirical equation is widely used.9
kr ( t – t0 )
S ( k r, t ) = S u ------------------------------ (2)
1 + kr ( t – t0 )
where
kr = rate of reaction at reference temperature Tr;
S = compressive strength at age t;
Su = ultimate compressive strength;
t0 = age at start of strength development, h; and
t = time, h.
With the exception of controlled laboratory conditions, the
temperature of the hydration process changes during the
reaction and then Eq. (2) becomes inapplicable. To overcome
this difficulty, it is sufficient to change the time-history into
a degree of reaction history. This can be done using the
equation of Freisleben-Hansen and Pedersen.10 Observing
that hydration of cement is a chemical reaction, the Arrhenius
ACI Materials Journal, V. 102, No. 4, July-August 2005.
MS No. 04-069 received April 26, 2004, and reviewed under Institute publication poli-
cies. Copyright © 2005, American Concrete Institute. All rights reserved, including the
making of copies unless permission is obtained from the copyright proprietors. Perti-
nent discussion including authors’ closure, if any, will be published in the May-June 2006
ACI Materials Journal if the discussion is received by February 1, 2006.
231
Marco Viviani is an assistant at the Applied Computing and Mechanics Laboratory
(IMAC), Swiss Federal Institute of Technology in Lausanne (EPFL), Switzerland. His
research interests include predictive modeling of hardening materials and the
monitoring of structures.

Branko Glisic is Head of the Solution and Service Department of Smartec SA, Manno,
Switzerland. His research interests include the development of smart sensors for
structural health monitoring and industrial engineering applications.

Ian F. C. Smith is a professor and Director of the Applied Computing and Mechanics
Laboratory (IMAC-EPFL). His research interests include measurement systems,
active structures, and applied computing in civil engineering.

law is integrated to describe cement hydration through a new

index called equivalent age (Et)

t
exp Q  --- – ----- dt
1 1
Et ( t ) =
∫  T T r
(3) Fig. 1—Concept of equivalent age.
t0

where
Et = equivalent age, h;
Q = activation energy divided by gas constant, Ea/R;
t = time, h;
t0 = time at hydration start, h;
T = temperature of concrete, K; and
Tr = reference temperature, K.
Et is the integral in time of the ratio between the rates of
reaction k1 and kr of two specimens of the same concrete
types. One is a fictitious specimen and is assumed to be kept
at a constant temperature Tr (generally 20 °C in Europe and
23 °C in the U.S.).
The other specimen is real and has a temperature profile T1
= T1(t). At every time t*, the real specimen has an equivalent
age Et,1(t*). This means that at the time t*, it has the same
degree of reaction that the reference process will have after
a total time Et,r(t*), being cured at T = Tr. Where time is
converted in equivalent age, the temperature of the process
assumes the value T = Tr. Thus, if T = Tr = constant, Eq. (2)
Fig. 2—Standard and stiff FO sensors.
is applicable (refer to Eq. (4)) for cases when temperature
varies during hydration.
Hardening time
k r ( Et – Et 0 ) A long gauge fiber-optic (FO) deformation sensor has
S ( k r, Et ) = S u -------------------------------------- (4)
1 + k r ( Et – Et 0 ) recently been developed.14 The FO sensor is particularly suit-
able for concrete because of its robustness, temperature
compensation, insensibility to magnetic fields, and a
where precision of 2 µm. Moreover, FO sensors follow the defor-
S = compressive strength at age t; mation of fresh concrete without disturbing the strain field of
Su = ultimate compressive strength; the host material.15 The stiffness and the thermal expansion
kr = rate of reaction at reference temperature Tr; coefficient (TEC) of the FO sensor are influenced mainly by
Et = equivalent age at time t; and the characteristics of the protective tube. For instance, the
axial stiffness of standard FO is very low because it is housed
Et0 = equivalent age at start of strength development. in a plastic protective tube. Glisic proposed a new sensor
The equivalent age is of great interest for predictions and called FO “setting” sensor with a higher axial stiffness using
for separation of effects because it allows direct comparisons a protective tube made of stainless steel15,16 (refer to Fig. 2).
of concrete pours (or specimens) that are hydrating at The setting sensor, once embedded in concrete together with
different speeds (Fig. 1). Moreover, when used in predictions, a standard sensor of the same gauge length, leads to determi-
it takes into account the so-called cross over effect of nation of the hardening time (refer to the following). When
concrete,9 which affects predictions made with other degree concrete is placed, the standard sensor measures the swelling
of reaction indexes.9,11-13 of concrete while the stiff sensor is not initially influenced by
The procedure explained in the following allows the the deformations of the concrete matrix; therefore, the differ-
calculation of the activation energy and can be used to ence between deformations measured by the two sensors
determine the datum temperature without modification. increases.

232 ACI Materials Journal/July-August 2005

Fig. 3—Hardening time.

After concrete hardens, both sensors measure only the

deformation of the concrete matrix and the difference Fig. 4—Determination of activation energy Ea.
between the deformations measured by the two sensors
remains constant (Fig. 3). The hardening time is defined as
the time when the derivative of the difference between the
deformations measured by setting and standard sensors
becomes zero.

EXPERIMENTAL INVESTIGATIONS
Determination of the activation energy Ea
Originally, hardening time was intended to be an equivalent
of setting time. Studies of the mechanism of force transmission
between sensors and concrete-matrix indicate that hardening
time depends on the degree of concrete hydration. This
degree is unknown and will be denoted as α = α*. Values for
hardening time depend on the following factors
• Degree of reaction (types of concrete, temperature of
hydration, and time); and
• Sensors features (thermal expansion properties and
stiffness).
The strategy adopted for determining the activation energy
uses two specimens of the same type of concrete. Both
specimens have the same dimensions. They are monitored
with a stiff and a soft sensor. Aside from their stiffness, each
pair of sensors has the same features. One specimen is
wrapped with glass wool. The glass wool acts as insulation Fig. 5—Determination of rate of reaction kr , frequency
and keeps the temperature of this specimen at a higher level factor A, and ultimate compressive strength Su.
than the temperature of the other specimen. This induces a
higher rate of reaction in the insulated cylinder. The
temperature is measured in both specimens (Fig. 4). The (with exception of Test 1 where test are made at 24 and 72 h).
degree of reaction, in terms of equivalent time Et, is expressed The 24-h test has not been found to be representative for
through Eq. (3). For both specimens, at the hardening time, the slowly hydrating concrete.
degree of reaction index Et has the same value. Temperature
profiles are inserted in Eq. (3) for each specimen and the COMPARISON OF PREDICTIONS
integral is calculated to the hardening time. As a result, two AND EXPERIMENTAL RESULTS
equations with two unknown values (Et and Ea) are obtained. Hardening time, activation energy, and rate of reaction were
Resolution leads to determination of the activation energy Ea evaluated and applied to six different types of concrete (refer
(refer to Fig. 4). to Table 1 to 6) using the procedure presented previously. Five
were concrete types commonly used in civil engineering.
Predictions of compressive strength They were made with different types of aggregate. Air
The activation energy is necessary but not sufficient for entrainers, high-range water-reducing admixtures, and
determining the rate constant kr (Eq. (1)). The value of kr is different types of cements were used (refer to Table 1 to 6).
needed to predict mechanical properties (Eq. (4)). The value The results shown in Fig. 6 to 11 have been obtained within
of kr can be determined if two compressive tests using stan- the first 72 h. All predictions obtained were realistic and
dard specimens of the same composition, humidity, acceptable without any correction according to the criteria
boundary conditions and temperature histories, are given in the code TEX-426-A (refer to Table 7 and 8). The
performed at different equivalent ages Et. This allows deter- quality of the prediction was verified after 7, 21, and 28 days.
mination of kr through the application of Eq. (4) (Fig. 5). The maximum deviation between predicted and tested values
Compressive tests have been carried out after 48 and 72 h of each test is presented in Table 8. Zero equivalent age in

ACI Materials Journal/July-August 2005 233

Table 1—Mixture proportions Test 1 Table 6—Mixture proportions Test 6
w/c 0.45 w/c 0.18
Cement CEM II/A-LL 42.5 R 325 kg/m3 Cement CEM I 52.5 N 1051.1 kg/m3
High-range water-reducing admixture 0.9% High-range water-reducing admixture 35.1 kg/m3
Air-entraining admixture 0.1% Steel fiber Not available
Aggregate 0-32 Hüttwangen Air-entraining admixture None
Maximum temperature difference 5 °C Silica fume 273.3 kg/m3
Aggregate 0-4 sand of Fontainebleau
Table 2—Mixture proportions Test 2 Maximum temperature difference 14.5 °C
w/c 0.45
Cement CEM I 42.5 R 350 kg/m3 Table 7—Verification criteria for maturity
High-range water-reducing admixture 0.8% prediction*
Air-entraining admixture None Verification criteria Adjusting procedure
Aggregate 0-32 Sergey s* ≤ 0.90s
s* ≥ 1.10s Develop new S-M relationship
Maximum temperature difference 15 °C
Three consecutive within Evaluate batching and placement,
0.90s ≤ s* ≤ 0.95s adjust s-M* relationship if needed
Table 3—Mixture proportions Test 3 1.05s ≤ s* ≤ 1.10s
Better correlations S-M relationship accepted
w/c 0.48
*Code: TEX-426-A; s = predicted strength; and s* = independent test results.
Cement CEM I 42.5 N HS 360 kg/m3
High-range water-reducing admixture 0.8%
Table 8—Maximum error between predicted
Air-entraining admixture None
strength and independent test results
Aggregate 0-32 Sergey
Maximum errors
Maximum temperature difference 20.2 °C
Test Day Maximum error, %
1 7 +4.5
Table 4—Mixture proportions Test 4
2 28 –5.1
w/c 0.48 3 28 +5.1
Cement CEM III/A 32.5 N 360 kg/m3 4 21 –7.4
High-range water-reducing admixture 0.8% 5 28 –6.4
Air-entraining admixture None 6 13 +3.7
Aggregate 0-32 Sergey
Maximum temperature difference 14.5 °C

Table 5—Mixture proportions Test 5

w/c 0.48
Cement CEM II/A-LL 32.5 R 360 kg/m3
High-range water-reducing admixture 0.8%
Air-entraining admixture None
Aggregate 0-32 Sergey
Maximum temperature difference 21.6 °C

Fig. 6 to 11 does not always refer to the pouring time. Because

poured concrete temperature is influenced by ambient
temperature in the initial phases, the zero equivalent age is
taken to be the point where cooling (if it occurs) slows to a
variable rate. If no cooling occurs, the zero time is taken to be
the batching time.
DISCUSSION OF RESULTS
The methodology presented assumes that the hardening
time is an indicator of the degree of reaction. Tests support
this assumption for the concrete that was studied. More
mixtures will be tested to clarify the limits of applicability.
Constraints on the testing procedure (such as minimum
difference in temperature profiles) could be added for a
better definition of hardening time when necessary. The
relationship between the hardening curve and the degree of
reaction is an important issue for the extension of the
Fig. 6—Compressive strength versus equivalent age for Test
Series 1. Calibration strengths of young concrete are used
to predict strength evolution and, this prediction is verified
by independent test results using cylinders containing more
mature concrete.
methodology to the general field of hardening materials, and
this will be the subject of further study. The basis of the
proposed methodology involves passing from mechanical
properties of concrete (hardening time) to thermodynamic-
chemical properties (activation energy and rate constant) and
back again to mechanical properties (compressive strength).
Some codified methods use similar concepts by inserting the
final setting time into maturity-strength equations and

234 ACI Materials Journal/July-August 2005

Fig. 7—Compressive strength versus equivalent age for Test
Series 2. Calibration strengths of young concrete are used
to predict strength evolution, and this prediction is verified
by independent test results using cylinders containing more
mature concrete.
Fig. 9—Compressive strength versus equivalent age for Test
Series 4. Calibration strengths of young concrete are used
to predict strength evolution, and this prediction is verified
by independent test results using cylinders containing more
mature concrete.

Fig. 8—Compressive strength versus equivalent age for Test
Series 3. Calibration strengths of young concrete are used
to predict strength evolution, and this prediction is verified
by independent test results using cylinders containing more
mature concrete.
Fig. 10—Compressive strength versus equivalent age for
Test Series 5. Calibration strengths of young concrete are
used to predict strength evolution, and this prediction is
verified by independent test results using cylinders containing
more mature concrete.

performing regression analyses. A recently developed

method17 uses a variant of the setting time to determine the
equivalent age and thus helps to determine strength-maturity
relationships.
Maturity methods are still rarely used in practice. This lack
of acceptance is partially related to limited experience with
these approaches. Confidence in the methodology presented
herein would be increased through performing more
compressive tests during the early age of concrete. For
example, using a given pair of compressive-strength values,
the value of kr and Su are obtained, and a predictive curve can
be calculated. Using other pairs, an envelope of curves is
obtained. A standard apparatus for the application of this
methodology is under development. Due to reusability and
robustness of equipment, an inexpensive and in-place appli-
cation of the methodology is feasible.
Fig. 11—Compressive strength versus equivalent age for
SUMMARY AND CONCLUSIONS Test Series 6. Calibration strengths of young concrete are
Compressive strengths of several widely used concrete used to predict strength evolution, and this prediction is
mixtures have been successfully predicted using a procedure verified by independent test results using cylinders containing
that involves early age deformation monitoring. The same more mature concrete.

ACI Materials Journal/July-August 2005 235

procedure has been applied to a special concrete to study the
applicability of the methodology to other types of hardening
materials. This methodology allows a fast and accurate
prediction of compressive strength on site. Seventy-two
hours are sufficient to gather the necessary data and provide
accuracy of less than 8% error. It is also an attractive procedure
for the determination of the activation energy and the rate
constant. Separation of various contributions to deformation
(autogenous, thermal, and humidity loss) is thus possible
in-place and in real time. More timely knowledge of
compressive strength evolution will lead to savings during
construction and improve safety.
ACKNOWLEDGMENTS
This project was supported in its early stages through a project funded by
the Swiss Commission for Technology and Innovation (CTI) and Cemsuisse
(Swiss Cement Fabricators Association). The authors are grateful to K.
Scrivener, EPFL, for valuable advice and for providing testing support. We
also express special thanks to P. Gallay who has helped design and build
testing apparatus.
REFERENCES
1. Arrhenius, S., “On the Reaction Velocity of the Inversion of Cane Sugar
by Acids,” Zeitschrift für Physikalische Chemie, No. 4, 1889, pp. 226-232 (as
translated and published in Margaret H. Back and Keith J. Laidler, 1967,
“Selected Readings in Chemical Kinetics,” Pergamon, Oxford, 1967).
2. De Shutter, G., and Taerwe, L., “Degree of Hydration-Based Description
of Mechanical Properties of Early Age Concrete,” Material and Structure,
V. 29, 1996, pp. 335-344.
3. Plowman, J. M., “Maturity and the Strength of Concrete,” Magazine
of Concrete Research, V. 8, No. 22, 1956, pp. 13-22.
4. Carino, N. J.; Lew, H. S.; and Volz, C. K., “Early Age Temperature
Effects on Concrete Strength Prediction by the Maturity Method,” ACI
236
JOURNAL, Proceedings V. 80, No. 2, Mar.-Apr. 1983, pp. 93-101.
5. Kee, C. F., “Relation between Strength and Maturity of Concrete,”
ACI JOURNAL , Proceedings V. 68, No. 3, Mar. 1971, pp. 196-203.
6. Knudsen, T., “On Particle Size Distribution in Cement Hydration,”
Proceedings, 7th International Congress on the Chemistry of Cement
Editions Septima, V. II, 1980, pp. I-170–I-175.
7. Turcry, P.; Loukil, A.; Barcelo, L.; and Casabonne, J. M., “Can the
Maturity Concept be Used to Separate the Autogenous Shrinkage and
Thermal Deformation of a Cement Paste at Early Age?” Cement and
Concrete Research, V. 32, 2002, pp. 1443-1450.
8. Laplante, P., and Boulay, C., “Evolution du Coefficient de Dilatation
Thermique du Béton en Fonction de sa Maturité aux tout Premiers Âges,”
Materials and Structures, V. 27, 1994, pp. 596-605.
9. Carino, N. J., and Lew, H. S., “The Maturity Method: From Theory to
Application,” Proceedings of the 2001 Structures Congress & Exposition,
P. C. Chang, ed., ASCE, Reston, Va., 2001, 19 pp.
10. Freiesleben Hansen, P., and Pedersen, J., “Maturity Computer for
Controlled Curing and Hardening of Concrete,” Nordisk Betong, V. 1,
1977, pp. 19-34.
11. Klieger, P., “Effects of Mixing and Curing Temperatures on Concrete
Strength,” ACI JOURNAL, Proceedings V. 54, No. 12, 1956, pp. 1063-1082.
12. Nurse, R. W., “Steam Curing of Concrete,” Magazine of Concrete
Research, V. I, No. 2, 1949, pp. 79-88.
13. Saul, A. G. A., “Principles Underlying the Steam Curing of Concrete
at Atmospheric Pressure,” Magazine of Concrete Research, V. 2, No. 6,
1951, pp. 127-140.
14. Inaudi, D., “Fiber Optic Sensor Network for the Monitoring of Civil
Engineering Structures,” PhD thesis, No. 1612, EPFL, Lausanne, Switzerland,
1997, 182 pp.
15. Glisic, B., “Fibre Optic Sensors and Behaviour in Concrete at Early
Age,” PhD thesis, No. 2186, EPFL, Lausanne, Switzerland, 2000, 195 pp.
16. Glisic, B., and Simon, N., “Monitoring of Concrete at Very Early
Age Using Stiff SOFO® Sensor,” Cement and Concrete Composites, V. 22,
2000, pp. 115-119.
17. Pinto, R. C. A., and Hover, K. C., “Application of Maturity Approach
to Setting Times,” ACI Materials Journal, V. 96, No. 6, Nov.-Dec. 1999,
pp. 686-691.
ACI Materials Journal/July-August 2005

Disc. 102-M24/From the July-Aug. 2005 ACI Materials Journal, p. 224
Sulfate Attack of Mortars Containing Recycled Fine Aggregates. Paper by S. T. Lee, H. Y. Moon, R. N. Swamy,
S. S. Kim, and J. P. Kim
Discussion by William G. Hime
FACI, Principal Emeritus, Wiss, Janney, Elstner Associates, Inc., Northbrook, Ill.
The paper written by the authors is important in what it
says, but what was not said may have even greater significance.
First, the title and paper discuss “recycled fine aggregates,”
but what was investigated was recycled concrete. Unfortu-
nately, there was no petrographic study of the “fine aggregate”
and, thus, no information as to the proportions of cement paste
and true aggregates is available. This could be very important.
Cement fines may consist largely of calcium carbonate, and
would certainly promote thaumasite formation.
Second, the authors used a low SO3/C3A and high C3S/
C2S cement, both of which would promote ettringite and
gypsum formation.
Finally, the authors used a 34,000 ppm sulfate solution to
promote early distress. Certainly this may not relate at all to
concrete that is occasionally exposed to even a 2000 ppm
sulfate solution, which is considered to be severe exposure
by ACI 201.2R.20
AUTHORS’ CLOSURE
The authors would like to thank the discusser for his
interest in our paper and for his valuable comments.
The “recycled fine aggregates” used in the paper were
obtained from a local supplier. The data on the physical
properties of them (Table 2 and 3 from the original paper)
were also provided from the supplier.
The commercial Type I portland cement produced in
South Korea was examined. The authors agree with the
discusser that the low SO3/C3A and high C3S/C2S cement
may be attributable to produce a great amount of ettringite
and/or gypsum, especially in a high concentration of sulfate
Disc. 102-M25/From the July-Aug. 2005 ACI Materials Journal, p. 231
Three-Day Prediction of Concrete Compressive Strength Evolution. Paper by Marco Viviani, Branko Glisic,
and Ian F. C. Smith
Discussion by Ken W. Day
FACI, Australia
The paper reports an original, elegant, and evidently well-
conducted research into the problem of predicting concrete
strength evolution at an early age. The discusser’s only
concern is that the problem was satisfactorily overcome
several years ago and the solution proposed, while more
elegant and theoretically sound, is not likely to be competitive
in effort, expense, or earliness.
The discusser’s technique involves measuring Arrhenius’
early age without preliminary determination of the activation
energy and using a feedback correction automatically
218
DISCUSSION
solution, during sulfate attack process. It should be noted,
however, that this paper focused on the effects of the recycled
fine aggregates on the sulfate resistance of mortars with
different replacement levels and types.
Both cement and recycled fine aggregates are, indeed, the
materials being used in real construction sites in Korea.
Thus, the ultimate purposes of the paper were to estimate
sulfate resistance of mortars containing the recycled fine
aggregates and to suggest the proper replacement levels of
the aggregates for environmental and economical benefits.
To achieve these purposes, a standard test (ASTM C 1012)
was carried out to assess the durability of the mortars,
although some details (for example, water-cement ratio [w/c]
and fines-cement ratio) do not exactly match with the standard.
While the sulfate solution used in the study represents an
extremely severe exposure condition, the authors adopted a
highly concentrated solution (5% Na2SO4) to evaluate the
resistance within a short period (approximately 1 year)
according to ASTM C 1012.
Finally, further research, as stated by the discusser, is
certainly needed to investigate the relationship between the
content of paste adhered to aggregate and sulfate attack.
REFERENCES
20. ACI Committee 201, “Guide to Durable Concrete (ACI 201.2R-01),”
American Concrete Institute, Farmington Hills, Mich., 2001, 41 pp.
21. ASTM C 1012, “Standard Test Method for Length Change of
Hydraulic-Cement Mortars Exposed to Sulfate Solution,” ASTM International,
West Conshohocken, Pa., 6 pp.
derived by a computer from a strength versus log equivalent
age relationship. A prediction can be obtained as soon as a
test cylinder (whose equivalent age is also monitored by a single
channel meter attached to it by a rubber band) has attained a
minimum strength of approximately 2 MPa (300 psi).
This technique was first reported in the U.S. at the 1991
ACI Spring Convention in Boston, Mass., and a small
modification was noted in the Second Edition of Concrete
Mix Design, Quality Control and Specification18 in 1999.
An excerpt of the relevant section of the book in question
ACI Materials Journal/May-June 2006
appears at www.kenday.id.au. It is not correct to say that
maturity methods are rarely used in practice, although this
may possibly be the case in Switzerland or the U.S.
The technique has been used to control stressing of large
precast units at ages as young as 7 hours and of in-place
floors at ages of approximately 2 days.

REFERENCES
18. Day, K. W., Concrete Mix Design, Quality Control and Specification,
2nd Edition, Spon Press, London, 1999, 416 pp.

AUTHORS’ CLOSURE
The authors thank the discusser for his interest in our
work. In his discussion, the discusser directly and indirectly
claims that his method has three advantages compared with
the approach presented in our article: Fig. A—Strength to log equivalent age data, Series 4.
1. Earliness in predicting strength;
2. Sufficient scientific soundness; and
3. Reduced cost and effort.
Furthermore he declares that his method can “predict
strength” as soon as young concrete strength reaches 2 MPa.
Finally, the discusser declares that it is incorrect to think
that maturity is scarcely used in the market. The authors will
explain why we disagree with these claims.

Earliness in predicting strength

Let us imagine that an engineer comes to us to ask for a
strength-maturity curve of a mixture. His mixture has never
been tested by the discusser, the authors, or others. After 3 days
of measurements on the mixture, the discusser and the Fig. B—Strength to log equivalent age data, Series 5.
authors have the following data:
• Strength at 2 and 3 days;
• Measurements of concrete temperature; and
• Measurements of concrete deformations.
The authors will provide the following data to the engineer:
1. Strength equivalent age prediction for any age; and
2. Activation energy and rate of reaction.
The discusser, however, has to wait at least for 28 days to
find a parameter called “control age” and establish a correla-
tion between the 28-day strength and the 7-day (or control
age) strength. These entries are necessary in his method.
Furthermore, the discusser needs more measurements to
correct his initial guesses on the activation energy value.

Sufficiently scientific sound methodology

The method proposed by the discusser is based on the
existence for every concrete of a “knee point” in the graph
strength-log of equivalent age. In Fig. A, the strength to log
equivalent age graph for Test Series 4 is provided. No knee
point is identifiable. Figure B shows the strength to log
equivalent age graph for Test Series 5. In this figure, several
knee points appear (which is the right one?). In Figure C, an
example given by the discusser on his website is shown. The
knee point is identified using 222 specimens coming from 27
different samples. The authors fear that, although this “knee
point” might appear for many concrete types, for others, it
might be difficult to identify even using an enormous
number of strength test results.
The discusser’s method uses only estimates for the value
of the activation energy. Such estimates are not scientifically
acceptable values and can lead to unexpected errors. No
explanation has been found in the discusser’s documentation
on the methods used for the “correction automatically
derived by a computer” of the activation energy.
Fig. C—Strength versus log equivalent age, from
www.kenday.id.au/ConadIndex.html.
Furthermore, no explanation is given on how the initial
time t0 is determined by the software. Initial time t0 is an
important value to determine when the concrete starts to gain
strength. Errors may lead to incorrect predictions. Finally,
elimination of early temperature data as shown on the
discusser’s website is arbitrary (decisions regarding data
relevancy is left to users).
Reduced costs and efforts
In the discussion, we could not find an estimation of the
cost of the discusser’s method, so it is difficult to make a

ACI Materials Journal/May-June 2006 219

direct comparison. The authors would appreciate to know
how much the initial development of a maturity curve for a
mixture costs the discusser. We hope that the minimum
necessary number of strength data is less than the 222 points
that are plotted in Fig. C. If this is not the case, we have no
doubt that our methodology is less expensive.
Our method is not on the market, thus, a commercial price
is not available. Indeed, we developed a system that allows
the reuse of sensors, thus reducing equipment expenses.
Reading units for optical sensors (that can monitor thermo-
couples at the same time) are available for rent worldwide
for the same price of a common multipurpose data logger.
The authors are confident that the method proposed in the
paper can provide quicker and more scientifically sound
predictions at competitive costs. The authors add, nevertheless,
that in the research, advertising and selling a maturity device
was neither a priority for the authors nor for the authors’
industrial partners.
Worldwide use of maturity methods
The authors have no idea if maturity methods are
frequently used in Australia. Indeed we could not understand
the meaning of the description “frequently used.” The
world’s largest cement and concrete manufacturers are in
Europe and America (Lafarge, Cemex, Italcementi Holcim,
Heildelberg cement, etc.) as well as many of the world’s
larger construction companies. In these countries, few
engineers use maturity methods.
Disc. 102-M32/From the July-Aug. 2005 ACI Materials Journal, p. 279
Influence of Cation on Corrosion Behavior of Reinforcing Steel in High-pH Sulfate Solutions. Paper by Laura
Mammoliti and Carolyn M. Hansson
Discussion by William G. Hime
FACI, Principal Emeritus, Wiss, Janney, Elstner Associates, Inc., Northbrook, Ill.
The paper presents an interesting study showing that a
magnesium ion can promote corrosion of steel in
concrete. The authors, however, fail to discuss the
complete logic of this.
Contrary to many ACI and federal documents that show
that there is a “corrosion threshold” of some weight level of
chloride (for example, 2 lb per cubic yard of concrete), the
corrosion threshold depends on the hydroxyl ion concentration
or, for a chloride-containing environment, on the ratio of
Cl–/OH– ionic concentrations.
This ratio explains chloride environments, but the more
obvious relationship simply between the inverse of the
hydroxyl ion concentration and corrosion is often
ignored.
220
Prediction of concrete strength
The authors acknowledge that the discusser has developed
a tool for estimating the strength of concrete using maturity
and statistical analysis of data. This system could be classified
as a statistical tool that uses a database of temperature and
strength data that have been collected unsystematically.
Under the conditions of training the database with a “sufficient”
number of data, such software might give good estimations
of strength for common concretes that are cured in traditional
ways. The value for a “sufficient number” of tests necessary
to correctly use this approach could not be determined by the
authors in the available documentation.
True predictions of strength development are not possible
with the discusser’s system. This is because “prediction” for
a hardening material such as concrete means that one makes
a statement that says what the strength will be at a certain age
before the concrete ever reaches that age. The discusser
cannot estimate the 28-day strength of a mixture without
having a priori set of data on 28-day strength of such a
mixture. In this sense, the method presented in our paper is
one of the few systems available that truly predicts the 28-
day (or any other age) strength. For these reasons, the
authors affirm that the discusser’s method estimates rather
than predicts strength.
CONCLUSIONS
Maturity methods are rarely-used methods for deter-
mining strength. Problems underlying the use of maturity
techniques have not been solved by the method proposed by
the discusser, as he claims. For example, more scientific
work is necessary to study the limitations of maturity.
The magnesium ion is a special cation, having a property
hardly any other common cations have: it takes the hydroxyl
ion out of solution.
Much less obvious is the likelihood that sulfate anions
may increase hydroxyl ion concentrations. As an example,
Na2SO4 + Ca(OH)2 → 2 NaOH + CaSO4·2H2O
That reaction won’t really occur in pure solutions because
calcium hydroxide is relatively insoluble. In concrete,
however, if gypsum is used up by a reaction with, for example,
monosulfate, the reaction goes from left to right. Because
most (but not all) portland cements now form monosulfate as
a normal hydration reaction, the equation moves from left to
right, and hydroxyl ion concentration increases.
ACI Materials Journal/May-June 2006