Cement & Concrete Composites 30 (2008) 515–523

Contents lists available at ScienceDirect

Cement & Concrete Composites

journal homepage: [Link]/locate/cemconcomp

A humidity-adjusted maturity function for the early age strength prediction

of concrete
Wei-Chong Liao *, B.J. Lee, C.W. Kang
Department of Civil Engineering, Feng Chia University, Taichung 407, Taiwan

a r t i c l e i n f o a b s t r a c t

Article history: The early age strength development of concrete is determined by a proposed maturity method, which
Received 9 August 2006 considers not only the curing temperature history at the core of the concrete specimens, but also the rel-
Received in revised form 22 February 2008 ative humidity of the environment. The humidity factor is incorporated with the original rate constant
Accepted 25 February 2008
model to form a new maturity function for the prediction of concrete strength development. In order
Available online 2 March 2008
to calibrate this humidity-adjusted rate constant, compression tests were conducted on concrete cylin-
ders cured at different conditions of temperature and humidity. By comparing the ratio of rate constants
Keywords:
of concrete cylinders cured at the same temperature but different humidities, the humidity factor could
Relative humidity
Rate constant
be quantified. Verification programs (including concrete cylinders cured at the programmed temperature
Maturity and relative humidity, submerged in water baths and cured outdoors, and air-dried outdoors) were used
Humidity-adjusted equivalent age to determine the applicability of the prediction model. Experimental results show that this new maturity
function is able to predict the in-place strength development of ordinary concrete cylinders at early age
with a maximum difference of 10%.
Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction The main purpose of the maturity method is to predict the

strength gain of a concrete structure after a certain curing period,
The strength development of concrete is a function of curing so that the construction schedules could be accelerated and cata-
temperature and age, described by the so-called maturity function strophic failure during formwork removal could be avoided. Usu-
[1,2]. Saul proposed a summation of temperature over the curing ally, the strength gains of concrete and mortar specimens as a
period as the maturity function, in which the datum temperature function of maturity index should be established first [9–11]. The
of 10.5 °C was recommended [3]. Saul also presented the matu- measured temperature history of the concrete specimens in the
rity concept as: concrete of the same mix at the same maturity field can be transformed into a maturity index (e.g., rate constant
has approximately the same strength whatever combination of or equivalent age), and the in-place strength can be estimated. Sev-
temperature and time make up the maturity. However, several eral laboratory and field tests have been performed to verify the
researchers [4,5] found that this temperature–time factor (or applicability of various maturity functions [11–13]. For example,
Nurse–Saul function) underestimated the influence of temperature Lachemi et al. have successfully applied the maturity method to
on the strength development of concrete at early age and overesti- predict the setting time in slipforming operations [14]. Detailed re-
mated it at later maturity. This drawback comes from the assump- view of procedures to calibrate the maturity index of a given con-
tion that temperature has a linear effect on the strength crete mix and methods to estimate the in-place strength of
development [6]. Based upon the Arrhenius equation, Hansen concrete using the laboratory built strength–maturity relation
and Pedersen [7] proposed an equivalent age to convert the age can be found in Malhotra and Carino [6].
of concrete cured at different temperature to an age at a reference To date, most of the maturity models are based upon mortar
temperature. This equivalent age was related to the activation en- cubes or concrete cylinders cured at 100% relative humidity
ergy of the concrete mix. The apparent activation energy is not a (100% RH). However, insufficient water or moisture supply often
constant for a curing temperature below 20 °C [7]. Recent study occurs during the curing period of the site-cast concrete. These
shows that the apparent activation energy depends on tempera- existing maturity models usually overestimate the strength gain
ture as well as curing age [8]. of the concrete cured outdoors. Since the relative humidity of the
in-place concrete structure is not always 100% RH, it is necessary
to consider the influence of the relative humidity on the strength
* Corresponding author. Tel.: +886 424521037. development of the concrete. Although the internal moisture state
E-mail address: wcliao@[Link] (W.-C. Liao). and surface to volume ratio will affect the hydration rate of a

0958-9465/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/[Link].2008.02.006
516 W.-C. Liao et al. / Cement & Concrete Composites 30 (2008) 515–523

concrete structure, this paper only considers the relative humidity authors [11,13]. From Eqs. (1)–(3), the humidity-adjusted rate con-
of the environment and the temperature at the concrete core as the stant (or modified rate constant) can be expressed as
dominant factors for the strength development of concrete, since  
J u kh ðh  h0 Þ J kh ðh  h0 Þ
both quantities are feasible to measure during the curing period RT c ;h ¼ RT c ;100   ¼ A100 eBT c  u  ð5Þ
1 þ kh ðh  h0 Þ 1 þ kh ðh  h0 Þ
of the site-cast concrete. Moreover, humidity of the environment
can be obtained from the local weather record. This study proposes where A100 is the same A as shown in [11], which is the regression
a humidity-adjusted equivalent age of concrete and a procedure to constant obtained from concrete specimens cured at 100% RH.
calibrate the rate constant as a function of the curing temperature The humidity-adjusted affinity ratio c0 , which is the ratio of the
and the environmental relative humidity. This humidity-adjusted rate constant of concrete cured at arbitrary temperature and rela-
rate constant model reduces to a conventional rate constant model tive humidity during time interval Dt to the rate constant for cur-
under 100% RH. Some verification programs are performed to esti- ing at the reference temperature and reference relative humidity,
mate the strength gain of concrete cured under variable tempera- can be defined as
ture and relative humidity.  
RT;h J kh ðh  h0 Þ 1 þ khr ðhr  h0 Þ
c0 ¼ ¼ eBðTT r Þ  u   
RT r ;hr J ur 1 þ kh ðh  h0 Þ khr ðhr  h0 Þ
2. Humidity-adjusted rate constant model (HARCM) 
h0  h
¼ eBðTT r Þ   ð6Þ
Based upon previous arguments, a proper maturity function h0  h =J u
should consider the influence of the relative humidity on the where hr denotes the reference relative humidity, which is 100% RH
strength development of the concrete. Therefore a humidity-ad- in this study, khr represents the humidity rate constant at reference

justed rate constant RT,h is proposed as temperature Tr, h* is 100  h, while hr ¼ 100  hr ¼ 0, and Jur is the
RT;h ¼ AðT; hÞ  eðBTÞ ð1Þ regression humidity constant at Tr. In the derivation of Eq. (6), the
relation in Eq. (4) is applied for simplification. It is evident that
where A is a regression constant obtained from the compression when h equals hr, c0 will reduce to the affinity ratio as that defined
data of concrete cylinders or mortar cubes cured at different humid- by TC method [11]. The humidity-adjusted equivalent age t 0e can be
ities h and curing temperature T, and B is a temperature sensitivity obtained as
factor. It is noted that A depends on both humidity and tempera- X
ture, which is different from the form used by Tank and Carino t0e ¼ c0 Dt ð7Þ
[11]. When h is 100% RH, A will reduce to the form as in [11]. For The strength gain for concrete cylinders cured at arbitrary temper-
simplicity, the maturity function proposed by Tank and Carino ature and humidity history can be calculated from the humidity-ad-
[11] will hereafter be denoted as the TC method. justed rate constant model (HARCM) as
In this study, the humidity factor Jh is defined as
RT r ;hr ðt 0e  t 00r Þ
RT ;h S ¼ Su  ð8Þ
Jh ¼ c ð2Þ 1 þ RT r ;hr ðt0e  t 00r Þ
RT c ;hr 
J ur khr ðhr
 h0 Þ
RT r ;hr ¼ A100  eðBT r Þ   ¼ A100  eðBT r Þ ¼ kr ð9Þ
where RT c ;h is the rate constant for concrete cured at a fixed temper- 1 þ khr ðhr  h0 Þ
ature Tc, with relative humidity h%, and RT c ;hr is the rate constant for
where Su is the regression limiting strength of concrete of the same
concrete cured at the same temperature Tc, with reference curing
mix, t0e is the humidity-adjusted equivalent age following the trans-
relative humidity hr. A calibration test to obtain the value of Jh is
formation law given by Eqs. (6) and (7), t 00r is the initial strength
necessary. In order to calibrate the humidity factor, it is suggested
development age (days) at the reference temperature Tr and refer-
that concrete cubes or cylinders be cured at three different temper-
ence humidity hr. From Eq. (9), RT r ;hr will equal to kr, and t 00r equal
atures and four relative humidities for each temperature condition.
to t0r as those defined by TC method [11]. However, the humidity-
Concrete specimens cured in water baths are considered as the
adjusted equivalent age t 0e considers the humidity effect on the
100% RH curing condition, which is the same as the conventional
strength development of concrete. The importance of this humidity
maturity method. The rate constant RT,h is obtained from the regres-
factor on the strength gain of concrete cylinders will be demon-
sion data of the concrete compressive strengths at different ages,
strated in the experimental part of this study. It is noted that Su is
and the humidity factor Jh as defined by Eq. (2) can be obtained.
a function of curing temperature and relative humidity. Curve fit-
Experimental data reveal that the relation between the humidity
ting to obtain Su(T, h) is necessary. By taking the strength difference
factor Jh and relative humidity h can be accurately fitted by a
between two consecutive time steps using Eq. (8), the strength gain
reverse hyperbolic curve as
of a concrete specimen can be evaluated from the following incre-

J u kh ðh  h0 Þ mental equation
Jh ¼  ð3Þ
1 þ kh ðh  h0 Þ X X Su RT r ;hr Dt0e
S¼ dS ¼ ð10Þ
½1 þ RT r ;hr ðt 0e þ Dt0e  t00r Þ½1 þ RT r ;hr ðt0e  t 00r Þ
where h* = 100  h, Ju is the nondimensional regression limiting
humidity factor, kh is the humidity rate constant from the data where Dt 0e is the humidity-adjusted equivalent age increment,
regression with dimension (% RH)1, and h0 is the regression initial which is the age difference between two consecutive humidity-ad-
humidity constant (% RH). Experimental data (which will be dem- justed equivalent ages calculated from Eqs. (6) and (7).
onstrated later) show that h0 is almost constant irrespective of
the curing temperature. By imposing that Jh equals one when h 3. Experimental procedures
equals hr(=100), we have
3.1. Calibration test
1
kh ¼ ð4Þ
h0 ð1  J u Þ
In order to apply the humidity-adjusted rate constant model to
The constant Ju varies linearly with the curing temperature. It is estimate the strength gain of concrete cylinders, the constants
interesting that Eq. (3) has the same reverse hyperbolic form as A100, B, Ju, khr , h0, t00r and Su(T, h) in Eqs. (8) and (9) have to be deter-
the strength gain versus maturity relation proposed by other mined from a standard calibration test. For a specified concrete mix,
 W.-C. Liao et al. / Cement & Concrete Composites 30 (2008) 515–523 517

three constant curing temperatures of 12 °C, 27 °C and 40 °C, were Table 2

adopted in this study. The relative humidities were set at 100% RH Mix proportions of concrete

(cured in water baths), 90% RH, 80% RH and 70% RH for each tem- Specimen Unit weight (kg/m3) Air content (%)
perature, respectively. A thermocouple was inserted into the core fc0 ðMPaÞ w/c (%) Water Cement Sand Gravel
of the concrete cylinder during the casting of the concrete speci-
21 0.81 240 289 924 734 2
men. This thermocouple was connected to the data acquisition sys- 35 0.63 240 375 853 734 2
tem to record the temperature of the concrete cylinder every 5 min. 56 0.44 240 539 718 734 2
By measuring the penetration resistance on mortar sieved from the
fresh concrete as specified by ASTM C403/C403M, the setting time
of concrete could be determined. The 1, 2, 4, 8, 16 and 32
final set time were scheduled as the compression test dates for the used to record the humidity histories of the curing cabinet and
concrete cylinders as shown in Table 1. concrete specimens.
Three concrete mixtures were used in the experimental pro- Concrete cylinders submerged in an outdoor water bath were
gram. Type I OPC, river sand, and gravel were used as the basic the second program to simulate the ideal in-field curing condition.
materials. The gravel has a SSD relative density of 2.60, with water The relative humidity was 100% in this case. Only the temperature
absorption of 1.68%, and moisture content of 1.24%. The maxi- histories of the concrete cores were recorded through the thermo-
mum diameter of the gravel is 12.7 [Link] river sand has a SSD couples. In the last program, concrete specimens were air-dried
relative density of 2.50, with absorption of 2.04%, and a moisture outdoors to simulate the real field curing condition. Both temper-
content of 1.79%. The fineness modulus of the fine aggregate ature history at the concrete core and the environmental humidity
was 2.46. The cement has a relative density of 3.15. There was were recorded through an AD/DA card. Specimens were removed
no admixture used in this study. Concrete cylinders were cured on the same scheduled dates as the first program to measure their
according to the ASTM standard C31-83. Table 2 shows the mix strengths. All the compressive strength data were compared with
proportioning of the three mixes. The fc0 represents the 28 day de- the strength gains predicted from the humidity-adjusted rate con-
sign strength. The slumps of the 21 MPa and 35 MPa concrete stant model. Since the concrete mixture used in the verification
batches were 20.5 cm and 16 cm, respectively. The dimensions of programs should be the same as the calibration test, only the dif-
the concrete cylinders used in this study were /7.5 cm  15 cm. ferent curing conditions for the verification programs are demon-
After casting the hardened concrete specimens were kept inside strated in Fig. 1.
a temperature and humidity controlled cabinet or water baths
according to the designed curing conditions. Before the final set 4. Results and discussion
time, the concrete specimens were kept within the steel mould
and inside the temperature/humidity controlled cabinet. Care 4.1. Humidity and temperature dependence of the rate constant
was taken for the low temperature and low humidity specimens.
Each set of concrete compression data represents three cylinders. For simplicity, only the results of the 21 MPa concrete cylinders
Since there are 12 kinds of temperature and humidity variations are reported here. The strength–age data for concrete cured at dif-
with six scheduled testing dates and three specimens as a test ferent temperatures and humidities are given in Table 3. These
set, and two additional specimens needed to record the curing data were used to calculate the constants RT,h in Eq. (1) and t0
temperature of each curing condition, the total amount of concrete and Su defined by other authors [11,13] (see Table 4). It is noted
cylinders used in the calibration test for a specified mixture was that RT,h is similar to kT as defined in TC method, therefore a least
240. If the degree of hydration of a concrete specimen as a function square regression analysis can be used to obtain these three con-
of environmental curing humidity/temperature can be derived stants. In this paper, RT,h is adopted as the rate constant instead
from some theoretical approach, then the amount of concrete cyl- of kT, since RT,h is both temperature and humidity dependent. Figs.
inders needed in the calibration test could be reduced. Fig. 1 shows 2–4 show the strength–age data for the 21 MPa concrete cylinders
the procedure of the calibration test. cured at 40 °C, 27 °C and 12 °C with different humidities, respec-
tively. Each strength data represents an average from three con-
3.2. Verification program crete cylinders with a standard deviation of less than 6% in most
cases. Fig. 5 summarizes the humidity-adjusted rate constant RT,h
Three kinds of verification programs were performed to exam- as a function of curing temperature using Eq. (1). It is seen that
ine the applicability of the proposed humidity-adjusted rate con- the humidity-adjusted rate constants vary exponentially with the
stant model. In the first program, concrete cylinders of the same temperature (Fig. 5). As expected, the limiting Su decreased with
mixture were cured within a chamber according to the pre-pro- increasing temperature (see Table 4), which exhibits the same
grammed temperature and humidity variations. The humidity var- trend as that reported by other researchers [11,13]. Furthermore,
ied between 75% and 85% RH, and the curing temperature varied for the same curing temperature the limiting Su increased with
from 27 °C to 36 °C. Compressive strengths of the concrete speci- increasing relative humidity. However, for the same curing tem-
mens were measured at designated ages of 1, 2, 3, 5, 7 and 14 days, perature, the humidity-adjusted rate constant RT,h decreased with
respectively. Pico RH02 temperature/humidity transmitters were increasing humidity (see Fig. 5). The larger difference of the
humidity between the concrete core and the outer environment
might accelerate the water migration and mass transport phenom-
Table 1 enon inside the concrete cylinder, which will experience a higher
Scheduled dates for compression test (1 = setting time at 100% RH) RT,h at a lower humidity at early ages.
Temperature (°C) 1 (h) 2 (h) 4 (h) 8 (h) 16 (h) 32 (h) The values of A and B at different humidities are listed in Table
12.0 13 26 52 104 208 416
5. It is interesting that the constant B is almost invariant with the
(17.33 days) relative humidity. An average value of B = 0.05511 (1/°C) is
27.0 7.25 14.5 29 58 116 232 adopted as the temperature sensitivity factor in this study. Table
(9.67 days) 5 reveals that the coefficient A in RT,h decreases with curing humid-
40.0 5 10 20 40 80 160
ity; therefore it is necessary to quantify the influence of the humid-
(6.67 days)
ity on the magnitude of RT,h.
518 W.-C. Liao et al. / Cement & Concrete Composites 30 (2008) 515–523

Material selection and preparation

Mix design

Trial batch

Mortar final setting test

Calibration program
Casting of concrete cylinders

12 oC cured Verification programs

Water bath cured (100% RH),
temperature/humidity chambers
(90% RH, 80% RH, 70% RH)

27 oC cured Controlled temperature/

Water bath cured (100% RH), humidity chambers,
temperature/humidity chambers Outdoor water bath
,
(90% RH, 80% RH, 70% RH) Air-dried outdoors

40 oC cured
Water bath cured (100% RH),
temperature/humidity chambers
(90% RH, 80% RH, 70% RH)

Concrete compression test

Data analysis

Fig. 1. Experimental procedures for calibration test of the humidity factor and maturity verification program.

Table 3
Compressive strengths for 21 MPa concrete cured at different temperature and Table 4
relative humidity (Specimen: /75  150 mm) The rate constants RT,h for 21 MPa (3000 psi) concrete cylinders cured at different
temperature and relative humidity
Curing temperature Age (days) Strengths (MPa) at different humidities
Tc (°C) Humidity Tc (°C) Su (MPa) RT,h (days1) t0 (days)
70% RH 80% RH 90% RH 100% RH
100% RH 40 22.97 0.5435 0.1421
12 0.54 0.41 0.48 0.70 0.76
27 27.34 0.2697 0.2150
1.08 1.64 1.74 2.08 2.36
12 27.83 0.1199 0.3075
2.17 3.82 4.51 4.77 4.96
4.33 7.32 7.73 8.44 9.36 90% RH 40 17.55 0.7156 0.1446
8.67 9.46 10.80 12.47 14.38 27 19.82 0.3450 0.2025
17.33 12.19 13.28 14.97 18.47 12 21.20 0.1497 0.3163

27 0.30 0.50 0.64 0.73 0.63 80% RH 40 15.86 0.7420 0.1483

0.60 2.14 2.44 3.05 2.76 27 18.44 0.3646 0.2033
1.24 4.64 4.74 4.99 5.36 12 18.50 0.1590 0.3729
2.42 7.46 7.87 8.49 9.66 70% RH 40 13.71 0.7753 0.1454
4.83 11.41 11.90 12.40 15.31 27 16.96 0.3753 0.2217
9.67 12.86 14.11 14.98 19.58 12 16.65 0.1636 0.3858
40 0.21 0.64 0.68 0.76 0.80
0.42 2.23 2.58 2.65 2.85
0.83 5.13 5.46 6.00 6.55 4.2. Calibration of humidity factor Jh
1.75 7.94 9.05 10.05 10.88
3.33 10.45 11.68 12.64 15.27
In order to calibrate the humidity factor Jh in Eq. (2), the ratios
6.67 10.99 12.84 13.91 17.41
of RT c ;h to RT c ;100 are evaluated and listed in Table 6. The distribution
 W.-C. Liao et al. / Cement & Concrete Composites 30 (2008) 515–523 519

20
21 MPa concrete
cured at 40ºC 21 MPa concrete
15 cured at 12ºC
+
Compressive strength (MPa)

S = R T,h(t-t 0 )

Compressive strength (MPa)

15 +
+ Su 1+ R T,h(t-t 0 )

+
10 +
10
70% RH experiment 70% RH experiment
+
70% RH curve fit 70% RH curve fit
+ 80% RH experiment 80% RH experiment
5 80% RH curve fit 80% RH curve fit
+ 90% RH experiment 5 + + 90% RH experiment
S = R T,h(t-t 0 ) 90% RH curve fit 90% RH curve fit
+ Su 1+ R T,h(t-t 0 ) 100% RH experiment 100%RH experiment
100% RH curve fit + 100% RH curve fit
+
0 +
0 2 4 6 8 0
0 5 10 15
Age (days) Age (days)

Fig. 2. Compressive strength data for concrete cylinders cured at 40 °C and different Fig. 4. Compressive strength data for concrete cylinders cured at 12 °C and differ-
humidities. ent humidities.

21 MPa concrete 100% RH exp.

20 cured at 27ºC 0.8 100% RH curve fit
90% RH exp.
+
S = R T,h(t-t 0 ) 90% RH curve fit
Compressive strength (MPa)

+
Rate constant RT,h (days-1)

Su 1+R T,h(t-t 0 ) 80% RH exp.

15 + 0.6 80% RH curve fit
70% RH exp.
70% RH curve fit
+

10 21 MPa concrete
0.4
+ 70% RH experiment
+
70% RH curve fit
80% RH experiment
80% RH curve fit
5 + + 90% RH experiment
90% RH curve fit
0.2
+ 100% RH experiment + RT,h =A (T,h )e
BT

100% RH curve fit

+
0
0 5 10 0
Age (days) 10 20 30 40
Temperature (ºC)
Fig. 3. Compressive strength data for concrete cylinders cured at 27 °C and differ-
ent humidities. Fig. 5. The rate constants RT,h for concrete cured at different humidities.

of Jh as a function of the relative humidity is accurately fitted by a reference temperature and 100% RH applying the regression con-
reverse hyperbolic curve as shown in Fig. 6. The regression limiting stants listed in Table 5.
Ju, humidity rate constant kh, and initial humidity constant h0 are The humidity-adjusted equivalent age t 0e can be obtained by
given in Table 7. The average value of h0 is 5.12% RH for this employing Eqs. (6) and (7). It is noted that Ju in Eq. (6) is a linear
21 MPa concrete mixture. For other concrete mixtures, 35 MPa function of curing temperature as indicated in Eq. (11). Therefore
and 56 MPa, which are not shown in this paper, the variations of this humidity-adjusted affinity ratio considers not only the curing
h0 with temperatures were also small. The limiting Ju is a linear temperature effect but also the influence of relative humidity on
function of curing temperature and can be expressed as the strength development of concrete cylinders.
Ju ¼ C  T þ D ð11Þ
2
with C = 0.0009586 1/°C, D = 1.4624 and R = 0.998. If Tr is set as Table 5
27 °C, then the limiting humidity factor at reference temperature The regression constants A and B for 21 MPa concrete cylinders RT,h = A(T,h)  e(BT)
is Jur = 1.4883, and the corresponding humidity rate constant
Humidity 100% RH 90% RH 80% RH 70% RH Average
khr ¼ 0:400ð%RHÞ1 using Eq. (4). The humidity-adjusted rate con-
A (days1) 0.06277 0.07649 0.08228 0.08388 –
stant at Tr and hr can be calculated from Eq. (9) as
B (1/°C) 0.05397 0.05587 0.05502 0.05557 0.05511
RT r ;hr ¼ 0:27801= C which is same as the rate constant kr at the same
520 W.-C. Liao et al. / Cement & Concrete Composites 30 (2008) 515–523

Table 6
Humidity factors for 21 MPa concrete cylinders
1
Tc (°C) Humidity (% RH) RT,h (days1) Jh
40 100 0.5435 1 21 MPa concrete
90 0.7156 1.3167
0.8

Relative strength S/Su

80 0.7420 1.3653 +
+
70 0.7753 1.4266
27 100 0.2697 1
90 0.3450 1.2791 0.6
+ 100% RH Tr= 27ºC HARCM
80 0.3646 1.3519
70 0.3753 1.3916 70% RH 12ºC
70% RH 27ºC
12 100 0.1199 1 0.4
+ + 70% RH 40ºC
90 0.1497 1.2483
80 0.1590 1.3259 80% RH 12ºC
70 0.1636 1.3639 80% RH 27ºC
0.2 80% RH 40ºC
+

The relative strength gains for concrete cylinders cured at 70% +

0
and 80% RH, and 90% and 100% RH in the calibration test are shown
in Figs. 7 and 8, respectively. The symbols represent the experi- 0 5 10 15
20
mental data from the calibration tests, while the solid line employs Modified equivalent age te' (days)
Eq. (8) with RT r ;hr ¼ 0:27801= C for this 21 MPa concrete mixture.
Typical values of the humidity-adjusted equivalent ages t0e and Fig. 7. The relative strength gains for concrete specimens cured at 70% and 80% RH.
the predicted relative strength S/Su for the 80% RH case are listed
in Table 8. It is evident that the relative strength gain between
the experimental data and theoretical prediction is less than 8% perature history at the center of a concrete cylinder was used, i.e.,
in most cases except for specimens with ages less than 2 final set- the humidity was assumed as 100% RH, then the strength gain for
ting time (see the percent error column in Table 8). If only the tem- the 80% cured case can be yielded from the TC method as shown in
the last column of Table 8. Compared to the experimental strength
1.5
data listed in Table 3, it is evident that the TC method will overes-
timate the strength gain of a concrete cylinder cured under insuf-
ficient moisture supply in most cases. It is noted that the concrete
21 MPa concrete
specimens were cured at room temperature/humidity first, and
1.4 then moved to a temperature/humidity chamber after the setting
+ time. Difference between the room temperature/humidity and de-
signed curing temperature/humidity might cause larger compres-
Humidity factor Jh

+
1.3
sive strength error in the first two days in some cases.
In the calibration test the limiting Su for each curing condition is
40ºC exp. + well defined as listed in Table 4, since the curing temperature and
40ºC curve fit humidity were kept as constant as possible. In converting the
1.2 27ºC exp. humidity-adjusted equivalent age t0e , the curing temperature and
27ºC curve fit
+ 12ºC exp.
12ºC curve fit
1.1
Ju kh(1 00 -h-h 0 )
Jh =
1+ kh(100 -h-h 0 ) 0.8
+
1 +
70 80 90 100
+
Humidity (% RH)
Relative strength S/Su

0.6
Fig. 6. The humidity factors Jh for concrete cured at different humidities. 21 MPa concrete

Table 7 0.4
Humidity rate constants for concrete specimens 100% RH Tr =27ºC HARCM
90% RH 12ºC
Type Temperature Ju kh h0 Average h0 +
90% RH 27ºC
(MPa) (°C) (% RH1) (% RH) (% RH)
90% RH 40ºC
21 40 1.5004 0.4877 4.1 5.12 0.2
100% RH 12ºC
27 1.4890 0.4011 5.1
+ 100% RH 27ºC
12 1.4736 0.3433 6.15
+ 100% RH 40ºC
35 40 1.9899 1.6838 0.60 0.50
27 1.9592 1.8956 0.55 +
0
12 1.9313 3.0680 0.35 0 5 10 15 20
56 40 1.8579 0.3885 3.00 2.67 Modified equivalent age te' (days)
27 1.8365 0.4598 2.60
12 1.7047 0.5915 2.40
Fig. 8. The relative strength gains for concrete cylinders cured at 90% and 100% RH.
 W.-C. Liao et al. / Cement & Concrete Composites 30 (2008) 515–523 521

Table 8
The predicted strengths for 21 MPa concrete cylinders cured at h = 80% RH

Curing temperature Age t t 0e (days) S (MPa) S/Su (1) S/Su (2) Absolute difference Percent error S (MPa)
ð1Þð2Þ
(°C) (days) HARCM HARCM (expected) (prediction) (1)(2) ð1Þ  100% TC
method
12 0.542 0.378 0.48 0.026 0.043 0.0171 65.38 0.52
1.083 0.667 1.75 0.094 0.112 0.0173 19.15 2.02
2.167 1.267 4.51 0.244 0.226 0.0176 7.38 4.65
4.333 2.572 7.73 0.418 0.396 0.0218 5.26 8.88
8.667 5.271 10.80 0.584 0.584 0.0005 0.09 14.06
17.333 10.402 13.28 0.718 0.739 0.0212 2.92 18.74
27 0.302 0.370 0.72 0.035 0.039 0.0040 11.29 0.42
0.604 0.717 2.22 0.132 0.122 0.0100 7.58 2.19
1.208 1.430 4.74 0.257 0.252 0.0045 1.95 5.19
2.417 2.881 7.87 0.427 0.426 0.0011 0.23 9.51
4.833 6.101 11.90 0.645 0.621 0.0247 3.72 14.84
9.667 12.632 14.11 0.765 0.775 0.0101 1.31 19.51
40 0.208 0.401 0.68 0.043 0.049 0.0065 13.95 0.47
0.417 0.893 2.58 0.162 0.159 0.0161 1.85 2.58
0.875 2.016 5.46 0.344 0.334 0.0171 2.91 6.17
1.667 4.209 9.05 0.571 0.526 0.0560 7.88 10.36
3.333 9.001 11.68 0.736 0.709 0.0248 3.67 14.78
6.667 19.531 12.84 0.810 0.843 0.0332 4.07 18.13

relative humidity records during each time interval are necessary. and 100% RH), and are tested at six scheduled dates for the
Since the humidities in the calibration tests are programmed in ad- compressive strength data.
vance, only a constant humidity value is substituted into Eq. (6). (2) From the strength–age data at a fixed temperature and
Fig. 9 depicts the temperature histories for concrete cured at 70% humidity, find the Su, RT,h, and t0 following the TC method.
and 80% RH during the calibration test. It is noted that if the con- (3) Find the constant A(T, h) and B from Eq. (1) for a fixed
crete cylinders are cured under variable temperature and humid- humidity as listed in Table 5, and the average B can be
ity, the relative strength gain S/Su will lose its physical meaning obtained.
because the denominator Su listed in Table 4 was obtained under (4) For a fixed curing temperature, calculate the humidity factor
a constant relative humidity but variable temperature condition. Jh as defined in Eq. (2), and tabulated in Table 6.
This Su is not a constant in the variable temperature/humidity (5) Fit Jh with an inverse hyperbolic curve using Eq. (3) and
environment. However, the absolute strength gain of the concrete obtain the Ju, kh and h0. The average h0 can be obtained.
can be yielded through Eq. (10), and will be explained in the next (6) For the Ju at different temperature, find the linear regression
section. constant C and D from Eq. (11).
As a summary, the data reduction for the calibration procedure
can be described as follows: 4.3. Verification of humidity-adjusted rate constant model (HARCM)

(1) For concrete of the same mixture, cast concrete cylinders or After the data reduction procedure described above, the neces-
mortar cubes and cure at three different temperatures sary constants A100, C, D, average B, RT r ;hr , average h0, t00r and Su to
(12 °C, 27 °C, 40 °C). For each temperature, the concrete cyl- form a humidity-adjusted rate constant model are known. To
inders are cured at four different humidities (70%, 80%, 90% examine the applicability of this model, three kinds of verification
programs were conducted. The corresponding temperature and
45 humidity records of these three programs are shown in Fig. 10.
For concrete cured in outdoor water baths, the peak temperatures
40 were due to the solar radiation at noon. For the outdoor air-dried
specimens, the temperature shows variations similar to those of
35 70% RH the second program, since both sets of specimens were cured out-
doors at the same time. It is evident that during rain the tempera-
80% RH ture dropped while the relative humidity reached a plateau. Due to
30
Temperature (ºC)

the variations of temperatures and humidities in these programs,

the limiting strength Su also varies with temperature and humidity.
25
It is assumed that Su is a bi-quadratic function of curing tempera-
ture T and humidity h as
20
2
Su ðT; hÞ ¼ a0 þ a1 T þ a2 h þ a3 T 2 þ a4 T  h þ a5 h þ a6 T 2  h
15 2 2
þ a7 T  h þ a8 T 2  h ð12Þ

10 By using the 12 regression values Su (three temperatures  four

humidity combinations) listed in Table 4, and a least square
5 scheme, these coefficients were obtained as

0
a0 ¼ 8:045; a1 ¼ 9:303; a2 ¼ 0:2466; a3 ¼ 0:1957;
0 5 10 15 3
Age (days) a4 ¼ 0:2136; a5 ¼ 0:8623  10 ; a6 ¼ 0:450  102 ;

Fig. 9. Temperature histories for concrete cylinders cured at 70% and 80% RH.
a7 ¼ 0:1241  102 ; and a8 ¼ 0:2645  104 :
522 W.-C. Liao et al. / Cement & Concrete Composites 30 (2008) 515–523

Concrete cured at controlled T and h that for the outdoor air-dried concrete specimens, if the humidity
Relative humidity (%)

100 40
factors were not taken into account, there would be an erroneous

Temperature (ºC)
temperature
90 humidity strength estimate as shown by the dotted line which almost coin-
35 cides with outdoor water cured case. The maximum error is 49%
80 of the TC approach compared with the experimental data. This sug-
30 gests that if the concrete cylinder is cured under insufficient mois-
70
ture or water supply environment, the rate constant model should
60 25 consider the humidity factor to obtain a reasonable strength gain.
0 5 10 15 In these three verification programs, only concrete cylinders of
Age (days)
the same size as the calibration test were adopted, therefore the
size effect of the maturity index due to different surface to volume
50
Concrete cured at outdoor water bath ratio was not considered. This aspect of the maturity method needs
Temperature (ºC)

further study. Due to the capacity of the temperature/humidity

40 chamber, only concrete cylinders of /7.5 cm  15 cm were used
in this study. For concrete members in which the member sizes
30 are considerably larger than the specimens used here, the self-cur-
ing effect of the interior of the member and the internal tempera-
ture rise due to hydration should be considered as well.
20
0 5 10 15
Age (days)

humidity 5. Conclusions
Relative humidity (%)

100 temperature 50
Temperature (ºC)

In order to predict the early age compressive strength gain of

80 concrete cylinders cured under variant temperature and humidity
40
conditions, a humidity-adjusted rate constant model and humid-
60 ity-adjusted equivalent age were developed in this study. In the
30
calibration test, concrete specimens were cured under three differ-
40 ent temperatures and four different humidities. From the compres-
Concrete air-dried outdoors
20 sive strengths at different ages the humidity influencing factor was
0 5 10 15
Age (days) quantified. Several verification programs for concrete cylinders
cured under programmed temperature/humidity condition; cured
Fig. 10. Temperature and humidity history for the verification tests. outdoors with air dry conditions and cured in outdoor water bath
were also performed to validate the applicability of this humidity-
adjusted rate constant model. From this study, the following con-
From this the Su (in MPa) at arbitrary temperature and humidity
clusions can be drawn:
can be interpolated. Therefore, the relative strength gains S/Su will
also exhibit some peak-trough behavior. To overcome this, an incre-
1. For a specified concrete mixture, a calibration test to quantify
mental form of Eq. (10) in estimating the strength gain induced by
the influence of the environmental relative humidity is needed.
the temperature/humidity variations is adopted. Fig. 11 shows the
The procedure follows the specification as stated in ASTM
predicted strength gains of these three verification programs,
C1074, however, the calibrated concrete specimens should be
where good correlation can be seen. The difference between the
cured under at least four different humidities (e.g., 70%, 80%,
experimental data and the prediction from the humidity-adjusted
90% and 100% RH) for each curing temperature (e.g., 12 °C,
rate constant model is within 10% for most cases. It is interesting
27 °C, and 40 °C, respectively).
2. A humidity factor Jh is introduced to evaluate the influence of
humidity on the magnitude of the rate constant. Under the
21 MPa concrete same curing temperature, the humidity factor Jh increased with
20
decreasing relative humidity. This might result from the higher
RH gradient between the concrete core and the ambient envi-
ronment at early age.
15 3. The relation of humidity factor as a function of RH was accu-
+ rately fitted by a reverse hyperbolic curve in this study. Since
Strength (MPa)

there are three parameters to be determined in a reverse hyper-

+
bolic curve, it is suggested that the concrete specimens be cured
+ under at least four humidities if the least squares method is
10
+ employed.
prediction for controlled T and h
+ 4. In order to estimate the in-place strength of concrete cylinders
controlled T and h
cured at arbitrary temperature and relative humidity, it is nec-
5 predction for outdoor water cured
+ essary to interpolate the limiting Su obtained from the calibra-
cured at outdoor water bath
tion tests as a function of temperature and relative humidity.
prediction for air-dried outdoors
An accumulative strength gain equation has been applied
+ air-dried outdoors
successfully to estimate the concrete cylinder strength develop-
0 air dried outdoors TC method
ment under variable temperature and relative humidity history.
Verification programs show that this humidity-adjusted equiv-
5 10 15
Age (days) alent age and humidity-adjusted rate constant model can
predict the in-place strength growth of concrete cylinders with
Fig. 11. The strength prediction for the verification programs. a maximum error of 10%.
 W.-C. Liao et al. / Cement & Concrete Composites 30 (2008) 515–523 523

5. Application of this proposed maturity function to other con- Copenhagen, Session BII. Copenhagen: Danish Institute for Building
Research; 1956.
crete structures needs further study, since size effects on the
[5] Alexander KM, Taplin JH. Concrete strength, cement hydration and maturity
internal moisture distribution and self-curing of the concrete rule. Aust J Appl Sci 1962;13:277.
member were not investigated. [6] Malhotra VM, Carino NJ. CRC Handbook on nondestructive testing of
concrete. CRC Press Inc.; 1991.
[7] Hansen P Freiesleben, Pedersen EJ. Maturity computer for controlled curing
and hardening of concrete. Nord Betong 1977;1:19–34.
Acknowledgements [8] Kim JK, Han SH, Lee KM. Estimation of compressive strength by a new apparent
activation energy function. Cement Concrete Res 2001;31(2):217–25.
[9] Carino NJ. Maturity method; theory and application. J Cement Aggr, ASTM
Partial supports from the National Science Council, Taiwan, un- 1984;6(2):61–73.
der Grants NSC91-2211-E035-026 and NSC912745-P-035-003 are [10] Knudsen T. On particle size distribution in cement hydration. In: Proceedings,
acknowledged. 7th International congress on the chemistry of cement, vol 2. Paris, 1980. p.
170–5
[11] Tank RC, Carino NJ. Rate constant functions for strength development of
References concrete. ACI Mater J 1991;88(1):74–83.
[12] Carino NJ, Lew HS, Volz CK. Early age temperature effects on concrete strength
[1] Nurse RW. Steam curing of concrete. Mag Concrete Res 1949;1(2):79–88. prediction by the maturity method. ACI J 1982;80(2):92–101.
[2] McIntosh JD. Electrical curing of concrete. Mag Concrete Res 1949;1(1):21–8. [13] Carino NJ, Tank RC. Maturity functions for concrete made with various
[3] Saul AGA. Principles underlying the steam curing of concrete at atmospheric cements and admixtures. ACI Mater J 1992;89(2):188–96.
pressure. Mag Concrete Res 1951;2(6):127–40. [14] Lachemi M, Hossain KMA, Anagnostopoulos C, Sabouni AR. Application of
[4] McIntosh JD. The effect of low temperature curing on the compressive strength maturity method to slipforming operations: performance validation. Cement
of concrete. In: Proceedings RILEM symposium on winter concreting Concrete Comp 2007;29:290–9.