Construction and Building Materials 409 (2023) 133643

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Determination of coefficient “s” and apparent activation energy for fib

model code’s maturity-based strength function when applied to fly
ash concrete
Y. Sun a, b, *, H.S. Lee a, b, *
a
Department of Architectural Engineering, Hanyang University, 1271 Sa 3-dong, Sangrok-gu, Ansan 15588, Republic of Korea
b
Department of Smart City Engineering, Hanyang University, 1271 Sa 3-dong, Sangrok-gu, Ansan 15588, Republic of Korea

A R T I C L E I N F O A B S T R A C T

Keywords: The maturity-based strength function recommended by fib model code 2010 can be easily utilized to estimate the
fib model code compressive strength of normal-strength or high-strength cement concrete. However, the lack of calculation
Maturity-based strength function approaches and guidelines for determining the coefficient “s” and the apparent activation energy Ea in the fib
Compressive strength
model code’s maturity method have limited its further application to fly ash (FA) concrete. In this study, with
Fly ash concrete
Apparent activation energy
reference to ASTM C1074, this limitation is addressed by adopting the equivalent mortar method to determine
Equivalent mortar method the coefficient “s” for the concrete under investigation. The feasibility and superiority of the equivalent mortar
Arrhenius equation method were validated through two cases. Moreover, the original strength function in fib model code is modified
Iterative searching method to incorporate the reaction rate constant, allowing the use of Arrhenius equation to determine the mixture-
specific Ea value. The iterative searching method is also supplemented to find the best-fit Ea value. The results
demonstrated that the Ea value, obtained either through the linear Arrhenius plot or the iterative searching
method, does not match the default Ea value (33.33 kJ/mol) provided by the fib model code, and can achieve
relatively lower standard error for strength estimation of FA concrete.

1. Introduction concrete with or without FA. However, the discrete measurement of CS

at specific time intervals fails to provide real-time quality information
Concrete is one of the most widely used construction materials, and it regarding concrete structures cast in a complex environment or assist in
plays a major role in economic development. However, the high demand optimizing the time of formwork removal [11]. fib model code 2010
for concrete has caused the concrete industry to be a major consumer of provides a simple time-strength relationship that describes the CS pro­
natural resources and a significant contributor to anthropogenic carbon gression in concrete [12], as shown in Eq. (1).
dioxide emissions [1]. To address this problem, coal fly ash (FA), a by-
fcm (t) = βcc (t) • fcm (1)
product from coal combustion in thermal power plants [2], is used as
a partial replacement for cement in concrete. Substituting cement with { [ ( )0.5 ] }
FA not only helps dispose of industrial waste, but also reduces green­ 28
βcc (t) = exp s × 1 − (2)
house gas emissions during the cement production process. t
Extensive research has been conducted to study the performance of
FA cement concrete worldwide [3–9]. Compressive strength (CS), as a Where fcm (t) represents the CS of concrete at time t in days (MPa); the
crucial property of concrete, has been widely tested and used as a key function βcc (t), defined by Eq. (2), describes the strength development
indicator for assessing concrete quality [10] by construction companies, over time; fcm corresponds to the 28d CS of concrete (MPa); and “s” is the
structural designers, and researchers. It is necessary for them to acquire coefficient related to the cement strength class and fcm value.
the time-dependent information on the CS development of cement The default “s” values for the different strength classes of cement and

Abbreviations: FA, fly ash; CS, compressive strength; w/b, water to binder ratio; FA/b, fly ash to binder ratio; EMM, equivalent mortar method; EM, equivalent
mortar; LAP, linear Arrhenius plot; ISM, iterative searching method; SE, standard error; AE, absolute error; RE, relative error.
* Corresponding authors at: Department of Architectural Engineering, Hanyang University, 1271 Sa 3-dong, Sangrok-gu, Ansan 15588, Republic of Korea (Y. Sun).
E-mail addresses: [email protected] (Y. Sun), [email protected] (H.S. Lee).

https://doi.org/10.1016/j.conbuildmat.2023.133643
Received 8 November 2022; Received in revised form 19 September 2023; Accepted 1 October 2023
Available online 24 October 2023
0950-0618/© 2023 Elsevier Ltd. All rights reserved.
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

28d CS value are given in fib model code as below [12]. w FA

s = c1 × + c2 × (6)
b b
fcm ≤ 60 MPa, CEM 32.5 N: s = 0.38,
( )n1
fcm ≤ 60 MPa, CEM 32.5R & 42.5 N: s = 0.25, m1 • w ( )n2
fcm ≤ 60 MPa, CEM 42.5R, 52.5 N, & 52.5R: s = 0.20, s=
b
n1 × m2 •
FA
(7)
fcm > 60 MPa, all strength class: s = 0.2. m1 • (0.5) b

where CaO, SiO2, and Al2O3 in Eq. (4) represent the mass percentages of
This strength function, which involves only two unknown values
the corresponding oxides in the composite powder; s0 in Eq. (5) is the
(coefficient “s” and fcm ), allows for easy estimation of CS. However, it
default coefficient “s” given in fib model code; c1 and c2 in Eq. (6), are
may not be directly applicable to FA cement concrete due to the distinct
constants depending on strength class of cement, for 32.5 N, 32.5R &
hydration mechanism of FA cement system and different strength evo­
42.5 N: c1 = 0.528, c2 = 0.527, and for 42.5R, 52.5 N & 52.5R: c1 =
lution characteristics. To adapt the fib model code’s strength function for
0.481, c2 = 0.441 [16]; m1 , n1 , m2 , and n2 in Eq. (7) are constants, and
FA concrete, various “s”-related formulas have been proposed. Bamforth
they are 0.537, 0.39, 0.695, and 0.097, respectively.
et al. [13] suggested assuming cement class N (CEM 32.5R & 42.5 N)
Considering concrete is generally cast in a sophisticated environ­
when FA content exceeds 20 %, and Class S (CEM 32.5 N) when it ex­
ment, temperature is a critical factor affecting its CS evolution. The
ceeds 35 %, as described by Eq. (3). Chen et al. [14] considered the
maturity method, recommended by ASTM C1074 [19], includes the
chemical composition of FA and cement, and related the coefficient “s”
Nurse-Saul (NS) approach [20,21] and the Friesleben Hansen-Pedersen
to CaO/(SiO2 + Al2O3) of the composite binder, as shown in Eq. (4).
(FHP) approach [22], both widely employed to assess the temperature
Bhaskara et al. [11] introduced an exponential function (Eq. (5)) to
effect on strength development. The NS technique assumes a linear
determine the coefficient “s” considering FA content (FA/b) and curing
relationship between curing temperature and rate of strength gain for
time. They compared this approach with the shrinking-core hydration
concrete, requiring a mixture-specific value for datum temperature T0 to
strength model proposed by Wang [15], and concluded that the fib
determine the temperature-affected CS. The FHP approach, on the other
model code’s strength function provided superior strength prediction for
hand, assumes an exponential function, as described in Eq. (8), to ac­
FA concrete. Additionally, with the strength class of cement taken into
count for temperature effect and requires determining the apparent
account, Vollpracht et al. [16] proposed a simple linear relationship (Eq.
activation energy (Ea) for the given concrete mixture [22]. According to
(6)) between coefficient “s” and w/b plus FA/b based on a collection of
the calculation procedures provided by ASTM C1074, both approaches
CS data. Sun et al. [17] assumed the exponential function cannot well
can be independently applied to cement concrete based on their specific
reflect the variation of coefficient “s” and utilized a power-type function,
requirements. In fib model code 2010, FHP equation is also combined
as shown in Eq. (7), to calculate “s”, considering the effect of w/b and
with Eq. (1) to address the temperature effect on CS estimation [12].
FA/b. Among all mentioned “s”-related formulas, Eq. (6) appears to be
However, due to the limitation of Eq. (1) to cement concrete without FA
the most comprehensive one, for it takes into account as multiple factors
addition, only a fixed Ea value of 33.33 kJ/mol is provided. As a result,
as possible. However, Eq. (6) neglects the curing time effect on “s” value.
Eq. (8) is transformed into Eq. (9) when a reference curing temperature
Eqs. (4) and (5) imply that coefficient “s” is related to the curing time,
of 20℃ is set [23].
and the “s” value for FA concrete when curing time exceeds 28d is higher
than the one before 28d. Vollpracht’s validation results also reveal that ∑n [ ( )]
Ea 1 1
late-age CS for FA concrete is underestimated by Eq. (1) with their te = Δti • exp − − (8)
R Ti Tr
proposed “s” values [16]. Therefore, no listed “s”-related formula has
i=1

considered all potential influencing factors for FA concrete. On the other ∑

n [
4000
]
hand, despite different formulas to determine coefficient “s”, a te = Δti • exp 13.65 − (9)
Ti
comparative analysis of these formulas under the same condition is yet i=1

to be conducted, and guidelines for selecting an appropriate coefficient

Where Δti is the curing time (d) when the absolute temperature is Ti (K),
“s” for FA concrete are still lacking [18]. Furthermore, considering FA or
Tr is the reference curing temperature (K); Ea is the apparent activation
cement from different sources might have different physical properties,
energy (J/mol); R is the ideal gas constant, and equals to 8.314 J/
chemical composition and mineralogical phases, it might bring about
(mol⋅K); and te is the equivalent age (d).
different strength development characteristics for FA concrete even with
According to existing research [24,25], it is generally believed that
the same mixture proportion, causing the variability of “s” values.
incorporation of low-calcium FA leads to a reduction in the Ea value of
Consequently, more reliable approach needs to be proposed to deter­
concrete. Conversely, for blast furnace slag (BFS) concrete, an increase
mine raw materials and mixture-related coefficient “s” for FA concrete.
in BFS content results in an increased Ea [25,26]. The fib model code
⎧
⎪ FA itself also acknowledges that the Ea value depends on various factors,
⎪ 0.2 ≤ 20%
⎪
⎪
⎪
⎪ b including “the type and strength class of cement, water/binder ratio, addi­
⎪
⎨
FA tions and admixtures” [12]. However, there is currently no available data
s = 0.25 20% < < 35% (3) to modify the Ea value concerning the use of additions and admixtures in
⎪
⎪ b
⎪
⎪
⎪
⎪ FA normal-strength or high-strength concrete. Moreover, fib model code
⎪
⎩ 0.38
b
≥ 35% does not provide step-by-step calculation procedures to determine the
mixture-specific Ea value, in contrast to ASTM C1074. Sun et al. [17]
⎧( )
validated Eq. (7) with the default Ea value, suggesting that the addition
⎪
⎪ CaO
⎪ − 0.38
⎨ + 2.12 × s0 t < 28d of FA or BFS does not alter the temperature sensitivity of concrete.
SiO2 + Al2 O3
s= ( ) (4) However, the suitability of this default value has not been reflected in
⎪
⎪ CaO
⎪
⎩ − 1.15 + 3.70 × s0 t > 28d their predicted results. Soutsos et al. [23] applied the Ea determined
SiO2 + Al2 O3
according to the calculation procedures in ASTM C1074 to fib model
⎧ ( ) code’s maturity-based strength function, and found no improvement in
FA
⎪
⎪
⎪ 0.268 × exp 0.0132
⎨ strength estimation for slag concrete under adiabatic curing conditions.
b
s= (
t < 28d
) t > 28d (5) This only implies that the Ea value obtained from ASTM C1074 is not
⎪
⎪
⎪
⎩ 0.315 × exp 0.0135
FA applicable to the maturity-based strength function recommended by fib
b model code. The best-fit Ea value for concrete is still unclear.

2
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

Consequently, it is necessary to develop effective methods to determine { [ ( )0.5 ] }

the specific Ea values for the fib model code’s maturity-based strength fcm (t) = fcm • exp s × 1 −
28
(10)
function. k•t
The inability to accurately determine “s” and Ea greatly limits the
In analogy to ASTM C1074, similar calculation procedures to
application of fib model code’s maturity-based strength function to FA
determine Ea is proposed for fib model code’s maturity method: (i)
concrete. This study aims to fill this gap by proposing simple and reliable
Prepare the investigated concrete and cure them at a minimum of three
approaches to determine these two values, thereby enabling accurate
different temperatures; (ii) Test the CS of concrete at various curing
strength estimation for different FA concrete mixtures. Furthermore, this
times and temperatures; (iii) Fit the CS of concrete cured at the reference
study also explores the variability of the “s” and Ea values resulting from
temperature using Eq. (1) to obtain fcm and s (Note: the 28d CS of con­
different calculation methods. The findings of this research can serve as
crete can also be used as the value of fcm). Maintain these two values,
a valuable supplement to the existing maturity method recommended by
and fit the CS of concrete cured at non-reference temperatures using Eq.
fib model code, and improve the strength prediction performance for FA
(10) to determine the k value; (iv) Use LAP with the x-axis as 1/T and the
concrete structures.
y-axis as ln(k) to determine the Ea value for the specific concrete
mixture.
2. Methodologies
These procedures allow for Ea calculation and CS estimation in the
application of the fib model code’s maturity method. Importantly, the
2.1. Determination of coefficient “s”
proposed procedures for Ea determination can also be seamlessly inte­
grated with EMM proposed in Section 2.1. By employing EM, it be­
Considering diverse physicochemical properties of cement and FA,
comes feasible to simultaneously determine both the coefficient “s” and
the formula-based approach for determining the “s” value may result in
the Ea value that closely align with the investigated concrete.
its variability. Besides, the strength class of cement used in concrete
In addition to LAP, we also incorporate an alternative approach,
cannot be always acquired [13], which might also affect the applica­
known as iterative searching method (ISM) proposed by Lee et al. [27],
bility of certain listed formulas mentioned in the Introduction. There­
to determine the best-fit Ea value for the fib model code’s maturity
fore, a practical approach, called the equivalent mortar method (EMM)
function. This approach involves varying Ea to minimize the prediction
is utilized in this study to determine the coefficient “s”. The EMM,
error, resulting in the determination of the best-fit Ea value. The
originally recommended in ASTM C1074 to determine the Ea value of
searching range of Ea generally needs to be pre-defined, and it is then
concrete [19], involves preparing an equivalent mortar (EM) that
discretized with an increment of 0.01 kJ/mol. Each Ea value from this
mimics the composition of the tested concrete. The EM is formulated to
range is taken into Eq. (8) to determine the equivalent age for strength
have the same fine aggregate to binder ratio (by mass) as the coarse
estimation. The standard error (SE) between the predicted and experi­
aggregate to binder ratio in the concrete mixture being investigated.
mental CS data are calculated by Eq. (11). By plotting the relationship
Notably, none of the “s”-related formulas or fib model code considers the
between SE and the corresponding Ea, the best-fit Ea can be identified
influence of aggregates on the coefficient “s” [11–14,17,23]. As the “s”
with the smallest SE value. Note that the procedure related to ISM in this
value reflects the rate of strength gain, which is primarily controlled by
study considers CS data tested at different curing temperatures. ISM can
cement hydration of cementitious materials, it is assumed that the EM
also be applied to CS data at only one single curing temperature to
shares a similar coefficient “s” with its corresponding concrete. Sun’s
explore the temperature effect on Ea value.
formula, given in Eq. (6), is proposed based on mortar test results, and
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
subsequently validated using CS data of mortar and concrete collected √
√1 ∑ N
( )2
from literatures [17], indirectly supporting our assumption. On the SE = √ fp − ft (11)
other hand, considering the curing time might affect the “s” value and N i=1
maturity method is usually applied to early-age CS estimation, EMM is
mainly restricted within 28d in this study. where SE is the standard error (MPa), N represents the number of CS
EMM follows these basic procedures: (a) Prepare an EM that matches data, fp is the predicted CS (MPa), and ft is the tested CS (MPa).
the concrete in terms of w/b and FA/b; (b) Test the early-age CS of the Unlike LAP, ISM does not rely on the modified strength function.
EM, cured at 20℃ and not exceeding 28d of curing; (c) Fit Eq. (1) to Instead, it systematically explores different Ea values one by one to find
determine the coefficient “s” using the tested CS values of EM; (d) Use out which can best predict CS. Because ISM is conducted based on the
the obtained “s” value from EM along with the 28d CS of the concrete to FHP equation, it requires that the reaction rate constant for the strength
predict its early-age strength development by Eq. (1). development should follow the Arrhenius equation. Therefore, it is
Note that in Step (b), it is recommended to acquire at least three generally advisable to perform LAP beforehand to verify the presence of
average CS values of EM at different testing times to achieve a better fit a linear relationship for concrete within the investigated temperature
of Eq. (1). Including 28d as one of the testing times is not necessary, as range. In our study, both LAP and ISM are employed to determine the Ea
the 28d CS (fcm in Eq. (1)) can also be obtained through nonlinear values for the fib model code’s maturity method when applied to FA
regression of the strength function. concrete. The Ea obtained from these two methods will also be compared
with the default value (33.33 kJ/mol) provided by the fib model code.
2.2. Determination of Ea value
3. Experimental information for case studies
Both fib model code and ASTM C1074 utilize the maturity concept
via FHP approach to evaluate the CS development of concrete over time. Four experimental setups regarding the CS test of FA concrete were
However, the fib model code’s strength function does not incorporate a collected from literature. These experimental results serve as four cases
parameter equivalent to the reaction rate constant (k) used in the to validate the proposed approaches as described in Section 2. Case 1
strength function recommended by ASTM C1074. Therefore, the direct and Case 2 are used for analysis of coefficient “s”, and Case 3 and Case 4
application of the linear Arrhenius plot (LAP) from ASTM C1074 to the for determination of Ea values. The essential experimental information
fib model code’s maturity method is not feasible. To leverage the for the four cases is provided below.
Arrhenius equation, we assume a reaction rate constant to be 1 in Eq. (1) Case 1 corresponds to the experiments conducted by Soutsous et al.
for concrete cured at the reference curing temperature. For concrete [25]. The plain Portland cement concrete and low-calcium FA cement
cured at non-reference curing temperatures, Eq. (1) is modified to concrete, with the average 28d CS values of 30 and 50 MPa, were pre­
incorporate the reaction rate constant k as shown in Eq. (10). pared and cured at 20℃. For simplification, four concrete mixtures are

3
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

Fig. 1. Comparison of different “s” related-formulas: (a) Bamforth’s (Eq. (3)) and Bhaskara’s (Eq. (5)) formulas; (b) Vollpracht’s formula with cement strength class
of 32.5 N, 32.5R, or 42.5 N (Eq. (6)); (c) Vollpracht’s formula with cement strength class of 42.5R, 52.5 N, or 52.5R (Eq. (6)); (d) Sun’s formula (Eq. (7)).

denoted as “PC30”, “PC50”, “FA30” and “FA50”, respectively, where the temperatures (5, 20 and 30 ℃) are selected, and the CS values tested at
numerical value in the label represents the average 28d CS. For example, 3, 7, and 28d are utilized for analysis, in order to avoid crossover effect.
“PC50” refers to the cement concrete with the average 28d CS of 50 MPa. Case 4 corresponds to the experimental results from Ragan [30].
CEM I 52.5 N cement with the standard strength of 57 MPa was used for Unlike the previous cases, FA mortar specimens, rather than concrete,
all four concrete mixtures. The free w/b for PC50 and PC30 were 0.46 were prepared. Four cementitious mixtures were used, labelled as
and 0.66, respectively. For FA50 and FA30, approximately 30 % cement “FA25-0.5”, “FA25-0.6”, “FA35-0.6”, and “FA35-0.7”, respectively.
was replaced by FA (by mass), and the free w/b were 0.35 and 0.52, These labels provide the information on FA/b and w/b. For example,
respectively. The CS tests were performed on all concrete specimens at 1, “FA35-0.6” signifies a mortar specimen with FA/b of 35 % and a w/b of
2, 3, 5, 7, 14, and 28d. Additionally, the EM specimens were prepared 0.6. No cement mortar without FA addition was made in Ref. [30]. The
accordingly in Ref. [25]. Only the CS values tested within 32d for EM mortar specimens with different cementitious mixtures were cured
were selected for validation (note the 28d CS value were not given in under three different temperatures (4.4, 22.8 and 29.4℃), and their
Ref. [25]), considering that longer curing time might affect the coeffi­ corresponding CS values were tested at 2, 3, 4, 7, 10, 14, 21, and 28d,
cient “s”. All specimens in Case 1 have been labeled to indicate whether respectively.
they are mortar or concrete.
The experiment conducted by Qiu [28] is selected as Case 2. In 4. Result and discussion
Ref. [28], the cement concrete with or without 6 % Ca(NO3)2 (by mass of
binder) as frost resistant admixture and low-calcium FA concrete, with 4.1. Analysis of coefficient “s”
the average 28d CS values of 40 MPa, were prepared and cured under
the standard conditions (T:20 ℃, R.H.:≥95 %). CEM I 42.5R cement was 4.1.1. Variability of formula-determined coefficient “s”
used, and w/b for all three mixtures was 0.4. For FA concrete, FA/b was Various “s”-related formulas mentioned in the Introduction are
20 %. In Case 2, three mixtures, denoted as “C40” for cement concrete, examined to assess the variability of the coefficient “s” under the same
“C40-FA” for FA concrete, and “C40-C” for cement concrete with Ca condition. Considering use of Eq. (4) from Chen et al. [14] requires prior
(NO3)2 addition were prepared. Their corresponding EM are denoted as knowledge of the chemical composition of FA and cement, it is not
“M40”, “M40-FA”, and “M40-C”, respectively. The CS test were con­ analyzed in this part. Instead, it will be applied in the actual cases in
ducted on both EM and concrete at 1, 2, 3, 7, and 28d. Section 4.1.3, and compared with other methods. Eq. (3) proposed by
Dhir’s experimental results [29] are employed as Case 3. Ten con­ Bamforth et al. [13], and Eq. (5) by Bhaskara et al. [11], are analyzed
crete mixtures were prepared, divided into five groups labelled as “M1”, together since they only consider the FA/b effect on “s”. In Fig. 1(a), it
“M2”, “M3”, “M4” and “M5”. Each group consisted of two mixtures, one can be observed that Eq. (3), characterized by a piecewise function,
for cement concrete denoted as “M × -OPC” and the other for FA con­ consistently yields lower “s” values compared to Eq. (5) with the same
crete denoted as “M × -FA”, where “×” represents the group number. For FA/b value. Furthermore, as the FA content increases, the difference in
cement concrete from M1-OPC to M5-OPC, w/b were 0.9, 0.745, 0.613, the “s” value between these two formulas becomes more pronounced.
0.507, and 0.442, respectively. For FA concrete from M1-FA to M5-FA, For instance, when FA/b reaches 75 %, the “s” value calculated by Eq.
w/b were 0.492, 0.477, 0.448, 0.426, and 0.403, respectively, and (5) is 0.721, whereas Eq. (3) only estimates a value of 0.38, significantly
FA/b were 63.5 %, 52.9 %, 40.1 %, 28.0 % and 15.5 %, respectively. The lower than the expected value from Eq. (5).
investigated concrete specimens cured under three different curing Both Eq. (6) proposed by Vollpracht et al. [16] and Eq. (7) by Sun

4
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

Table 1 values are presented in Fig. 2. βcc(t) with Eq. (3)-derived “s” values
The calculated coefficient “s” at specific conditions. exhibit the most rapid strength gain, while βcc (t) with “s” values from Eq.
w/ FA/ Eq. Eq. Eq. (6) Eq. (7) demonstrate the slowest rate of strength development. The vari­
c b (3) (5) (7) ability of “s” values might cause significant deviations in CS estimation.
32.5 N, 32.5R, 42.5R, 52.5 N,
42.5 N 52.5R Although there is minimal difference between the results derived from
Eqs. (3) and (5), it cannot be guaranteed that these two formulas
0.5 0 0.2 0.268 0.264 0.241 0.370
10 0.2 0.306 0.317 0.285 0.556
consistently yield similar results for all FA concrete mixtures. As shown
% in Fig. 1, with increasing FA/b and w/b, the increase rate of “s” value
50 0.38 0.519 0.528 0.461 0.650 determined by Eq. (7) gradually slows down, and the derived values
% become similar to those from Eq. (6). However, the “s” value from Eq.
(5) increases significantly with increasing FA/b, leading to a notable
difference compared to Eq. (6). Besides, Sun et al. [17] demonstrated the
et al. [17] consider the combined effect of w/b and FA/b on the “s”
linear formula proposed by Vollpracht et al. [11] cannot perfectly fit the
value, making them suitable for comparison. Fig. 1 (b)-(d) illustrates the
CS evolution of FA mortar from their own experiments. The coefficient
variation of “s” when FA content ranges from 0 to 80 % and w/b varies
“s” may differ from case to case, and even for the same concrete mixture
from 0.3 to 0.5. It is evident that Eq. (6) yields a calculated “s” range of
with FA and cement from different sources. Therefore, a more reliable
0.158–0.686 and 0.144–0.593 for different strength classes of cement.
method is needed to determine the raw material and mixture-related
Conversely, Eq. (7) results in a calculated “s” range of 0.281–0.680. This
coefficient “s”.
implies that when FA content and w/b are high, the calculated “s” values
from these two formulas align closely. These maximum “s” values are
4.1.2. Applicability of EMM
slightly lower than the result from Eq. (5) (0.77), but significantly higher
Case 1 and 2 are utilized to validate the applicability of our proposed
than the result from Eq. (3) (0.38). When w/b and FA/b are low, the “s”
EMM regarding “s” calculation. Following the procedures outlined in
values derived from Eq. (7) are evidently higher than those from Eq. (6).
Section 2.1, the CS evolution of the EM from Case 1 is first fitted using
Based on Fig. 1, different “s” related formulas can yield significantly
Eq. (1), as depicted in Fig. 3. The determined values of fcm and coeffi­
different or closely aligned values. To illustrate the potential differences
cient “s” for EM are presented in Table 2. With the fixation of “s” values
between the proposed formulas more intuitively, a specific condition is
from EM, the CS development of the investigated FA concrete can also be
considered where CEM I cement is used for concrete preparation and w/
estimated based on Eq. (1), as shown in Fig. 3. Considering the tested
b is set at 0.5. Three different FA contents, namely 0 %, 10 %, and 50 %
28d CS values of concrete in Case 1 exhibit relatively large fluctuation,
are examined. The calculated “s” values from different formulas are
the maximum and minimum 28d CS values within the fluctuation range
presented in Table 1. It can be observed that the “s” values determined
are also employed in Eq. (1), respectively, to represent the potential
by Eq. (5) and Eq. (6) with cement strength class of 32.5 N, 32.5R or
strength development range over time. They are plotted as the upper and
42.5 N show minimal differences. However, the “s” values from Eq. (3)
lower lines of predicted CS, respectively. It can be observed that in
are notably lower than those from Eqs. (5) and (6). On the other hand,
Fig. 3, the majority of the tested CS values fall within the predicted
Eq. (7) yields significantly higher “s” values for FA concrete compared to
fluctuation range, indicating that the coefficient “s” derived from EMM
the results from Eq. (6).
can be effectively applied for FA concrete strength prediction. Similarly,
Additionally, the time-dependent function βcc(t) with different “s”
EMM is also applied to experimental results of Case 2. CS fitting of EM

Fig. 2. Comparison of βcc(t) at specific conditions: (a) 0% FA; (b) 10% FA; (c) 50% FA.

5
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

Fig. 3. CS fitting of EM and CS prediction of concrete for Case 1: (a) PC30, (b) PC50, (c) FA30, and (d) FA50.

Table 2
Determination of fc and coefficient “s” for two cases.
Case No. EM Concrete

Label Direct fitting of CS Label EMM Direct fitting of CS

fc [MPa] s [-] R2 fc [MPa] R2 fc [MPa] s [-] R2

1 PC30-M 38.6 0.283 0.955 PC30-C 30.5 0.991 31.4 0.311 0.989
PC50-M 54.4 0.215 0.971 PC50-C 49.5 0.982 51.2 0.192 0.986
FA30-M 25.9 0.365 0.991 FA30-C 34.6 0.996 34.1 0.360 0.996
FA50-M 46.7 0.268 0.975 FA50-C 50.5 0.995 50.3 0.297 0.993
2 M40 38.6 0.192 0.999 C40 38.2 0.994 39.0 0.205 0.993
M40-FA 45.0 0.198 0.985 C40-FA 43.0 0.988 42.8 0.182 0.989
M40-C 39.9 0.181 0.994 C40-C 42.0 0.958 42.47 0.181 0.958

Fig. 4. CS fitting of EM (a) and CS prediction of concrete (b) for Case 2.

for Case 2 are presented in Fig. 4 (a), and obtained fcm and coefficient “s” the CS estimation performance using EMM for two cases. In addition, we
for EM are summarized in Table 2. By utilizing these “s” values derived also directly fit the CS results of the concrete under investigation using
from EM, accurate predictions of the CS development for the investi­ Eq. (1) to get the 28d CS values and their actual coefficient “s”, as shown
gated concrete in Case 2 are well achieved, as depicted in Fig. 4(b). in Table 2. It is evident that when EMM is applied, the R2 values are
In Table 2, the coefficient of determination (R2) is provided to assess remarkably high. Apart from C40-C (0.958) from Case 2, R2 vales for the

6
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

Table 3
“s” values from different methods for two cases.
Case No. Method Concrete EMM Eq. (3) Eq. (4) Eq. (5) Eq. (6) Eq. (7)

1 PC30 0.311 0.283 0.20 0.227 0.268 0.317 0.429

PC50 0.192 0.215 0.20 0.227 0.268 0.221 0.353
FA30 0.360 0.365 0.25 0.335 0.397 0.382 0.631
FA50 0.297 0.268 0.25 0.335 0.398 0.301 0.510
2 C40 0.205 0.192 0.2 0.249 0.268 0.192 0.328
C40-FA 0.182 0.198 0.2 0.249 0.349 0.281 0.527
C40-C 0.181 0.181 0.2 0.249 0.268 0.192 0.328

other six investigated concrete specimens exceed 0.98, indicating the the investigated concrete, addition of Ca(NO3)2 admixture has
high accuracy of the proposed EMM for CS estimation. Besides, the “s” decreased the coefficient “s”, and the reduction of “s” values might be
values determined by EMM for the investigated concrete closely align because Ca(NO3)2 can promote early-age strength development of
with their actual values. The absolute errors between the “s” values cement-based materials. While some formulas still achieve good CS
calculated using EMM and the actual “s” values are less than 0.03. This estimation for C40-C, we cannot guarantee they can still work well for
demonstrates the applicability of EMM in determining the coefficient “s” all similar situations. Another noteworthy observation is that for C40-
for the investigated concrete with or without FA. FA, despite the partial replacement of cement with FA, the calculated
“s” value using Eq. (4) is the same as that for C40. This is because Eq. (4)
4.1.3. Accuracy of EMM assumes the coefficient “s” is related to CaO/(SiO2 + Al2O3) [14], which
EMM is also compared with other “s”-related formulas listed in the does not significantly change with the incorporation of 20 % FA content
Introduction. The “s” values for two cases, determined by different in Case 2.
formulas, have been presented in Table 3, and the absolute errors (AEs) In addition to the AEs of the “s” value, the relative errors (REs) of the
between the best-fit “s” value obtained from direct fitting of concrete CS CS data for two cases are also calculated, and the range of REs for
data and “s”-related methods are shown in Fig A in Appendix. different methods is displayed in the form of boxplot, as shown in Fig B
In Case 1, it can be observed that Eq. (6) from Vollpracht et al. [16] in Appendix. The REs for all six methods nearly follow a normal distri­
has achieved the lowest AEs, all of which are positive values. The EMM bution. In Case 1, the average RE from our proposed EMM is approxi­
follows with the second lowest AEs. Unlike Eq. (6), the AEs from EMM mately zero. Moreover, the majority of REs for EMM in Case 1 are well
include both positive and negative ones. On the other hand, the AEs controlled within ±12.5 %. While RE for EMM is slightly higher than
from the other formulas are significantly higher compared to EMM and those from Eq. (6), EMM shows evident advantage over the other for­
Eq. (6). The largest AEs come from Sun et al. [17] (Eq. (7)), all of which mulas, indicating the suitability of the proposed EMM for “s” determi­
are positive values. In Case 2, EMM has achieved the best estimation nation. For Case 2, the superiority of EMM becomes more apparent. As
performance on “s” value. For 40C-C, EMM yields a zero AE value as the shown in Fig. B(b), EMM maintains an average RE value close to zero,
determined “s” value matches the best-fit one from the direct fitting of similar to Case 1. In contrast, the REs for the “s”-related formulas have
CS data for concrete. Eq. (3) also demonstrates relatively low AEs for the the average RE values below zero, indicating a consistent tendency to
coefficient “s”. While AEs from Eq. (6) remain low for concrete with or favor evident underestimation on CS value. In addition, compared to
without Ca(NO3)2, it produces a high AE for C40-FA. Upon careful ex­ these formulas, RE values from EMM have the smallest level of vari­
amination of the “s” values in Table 3, we find that there is no significant ability around the average RE.
increase in the coefficient “s” for FA concrete in Case 2 compared to
concrete without FA. Typically, it is assumed that the “s” value increases
4.2. Analysis of Ea
with higher FA content [11,13,16,17]. The decrease in the “s” value
could be attributed to the uncertainties of CS and high reactivity of FA.
4.2.1. Fitting the modified strength function
Nevertheless, the coefficient “s” derived from EMM for C40-FA remains
Case 3 and 4 are utilized to analyze Ea value for the fib model code’s
very close to the best-fit value obtained from the investigated concrete
maturity method. A reference curing temperature of 20℃ is set for Case
data. Among all the methods for “s” determination, Eq. (7) exhibits the
3, and 22.8℃ for Case 4. The modified strength function, as proposed in
largest absolute errors, similar to the results in Case 1.
Eq. (10), is employed to fit the CS development and determine Ea values
The most representative mixture in Case 2 is C40-C containing Ca
for the investigated concrete under different curing temperatures. The
(NO3)2 as the antifreeze admixture. Notably, the existing “s”-related
fitted curves for different concrete mixtures from two cases are pre­
formulas have never quantitatively described the Ca(NO3)2 effect on
sented in Figs. C–E in Appendix. The CS evolution for cement concrete is
variation of “s” value. Hence, it can only be assumed that concrete with
found to be well fitted by Eq. (10), regardless of the curing temperature.
or without Ca(NO3)2 addition shares the same “s” value, when these
However, for FA concrete/mortar, there are slight differences in the
formulas are utilized. However, as shown in Table 2, whether for EM or
fitting performance of both cases. At the curing temperature of 30℃, Eq.

Table 4
Determined kinetic parameters for two cases.
No. Case 3 Case 4

Mix ID M1 M2 M3 M4 M5 FA25-0.5 FA25-0.6 FA35-0.6 FA35-0.7

OPC FA OPC FA OPC FA OPC FA OPC FA

fcm[MPa] 14.9 15.2 23.7 23.8 37.0 37.7 50.9 52.3 63.2 64.6 33.2 24.1 18.8 14.3
s [-] 0.414 0.465 0.467 0.481 0.379 0.439 0.372 0.399 0.349 0.337 0.491 0.477 0.565 0.583
kl[d-1] 0.48 0.47 0.47 0.51 0.47 0.53 0.54 0.54 0.55 0.54 0.43 0.34 0.51 0.61
kr[d-1] 1 1 1 1 1 1 1 1 1 1 1 1 1 1
kh[d-1] 1.46 1.67 1.20 1.53 1.25 1.47 1.53 1.49 1.47 1.56 1.27 1.18 1.39 1.30

Note: kl, kr and kh represents the reaction rate constants for the CS development at the curing temperature lower, equal to, or higher than the reference curing
temperature, respectively.

7
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

Fig. 5. LAP for Case 3 (a-e) and Case 4 (f-g).

(10) struggles to accurately fit the late-age CS values, especially at 28d reference curing temperature has already exceeded 28d, rendering the
for Case 3, and 21 and 28d for Case 4. The fitted values are lower than fixed coefficient “s” unsuitable for FA concrete/mortar. However, when
the tested CS, possibly because the pozzolanic reaction of FA is enhanced curing temperature is below the reference temperature, Eq. (10) can still
under higher curing temperature, and the modified strength function get a good fitting performance for FA concrete in both cases, for its
fails to fit the late-age CS development for FA concrete. Another reason equivalent age at the reference curing temperature is still smaller than
could be the time effect on the coefficient “s” [11,14]. With a temper­ 28d.
ature of 30℃ and a curing time of 28d, the equivalent age at the The corresponding kinetic parameters for both cases are summarized

8
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

Table 5
Comparison of different calculation approach-derived Ea values.
Case No. Label ISM LAP Default value by fib 2010
2
Ea [kJ/mol] SE [MPa] Ea [kJ/mol] R SE [MPa] Ea [kJ/mol] SE [MPa]

3 M1 OPC 31.82 0.210 31.22 0.998 0.213 33.33 0.227

FA 33.43 0.692 35.74 0.999 0.705 0.692
M2 OPC 29.39 0.889 27.15 0.979 0.914 0.964
FA 30.51 0.966 30.63 0.999 0.966 1.001
M3 OPC 30.35 1.106 28.12 0.973 1.149 1.181
FA 28.72 1.105 28.68 0.999 1.105 1.293
M4 OPC 28.39 0.526 28.89 0.999 0.535 1.073
FA 28.52 0.929 28.65 0.999 0.930 1.327
M5 OPC 27.52 0.557 27.72 0.999 0.559 1.437
FA 29.03 0.936 29.67 0.999 0.948 1.364
4 FA25-0.5 31.13 1.142 30.17 0.999 1.150 1.183
FA25-0.6 37.41 1.259 35.67 0.987 1.270 1.322
FA35-0.6 29.18 1.683 27.27 0.992 1.690 1.716
FA35-0.7 22.27 1.396 20.46 0.988 1.401 1.563

Fig. 6. ISM to find out the best Ea for cement concrete from Case 3 (a)-(e), FA concrete from Case 3 (f)-(j), and FA mortar from Case 4 (k)-(n).

9
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

Fig. 6. (continued).

in Table 4. As mentioned in Section 2.2, the values of the reaction rate 30℃, the obtained k values are higher than 1, indicating a much faster
constant k, which reflect the rates of CS gain, are kept 1 for all concrete strength development of concrete at higher curing temperature.
specimens cured at the reference curing temperature. When the curing
temperature is below the reference curing temperature, k values are 4.2.2. Comparison of different Ea values
decreased for all specimens in both cases compared with those cured at The Ea values for the investigated concrete from Case 3 and 4 are
reference temperature, indicating a slower rate of strength development. determined using LAP, as shown in Fig. 5. Good linear relationships are
Conversely, for the investigated concrete at the curing temperature of observed for all specimens between ln(k) and 1/T, when the curing

10
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

temperature varies from 5 to 30℃ in Case 3, and from 4.4 and 29.4℃ in shows more stable and more accurate predictive performance on CS of
Case 4, and all R2 values (in Table 5) for the fitted lines are higher than concrete with or without FA. There are two advantages for EMM. One is
0.97, indicating that the rates of strength gain obtained by Eq. (10) that EMM considers the strength development characteristics of
closely follow the Arrhenius equation. On this basis, ISM is applied to cementitious materials instead of relying solely on the variation rules of
find out the best-fit Ea values for the two cases. The search range of the “s” values under different conditions. Some of the existing “s”-related
Ea value is set from 20 to 40 kJ/mol in this study. Fig. 6 presents the formulas might be too simple to reflect the true relationships between
results of ISM for concrete from Case 3 and 4. The Ea values obtained by the potential influencing factors and coefficient “s”. The other advantage
LAP (in Table 5) and the default Ea value (33.33 kJ/mol) given by fib is that the prepared EM shares the same constituent materials as its
model code are also marked for better comparison. corresponding concrete, ensuring that the physicochemical properties of
It can be observed that in Fig. 6, the SE value initially decreases with cement and FA are fully considered in this method. Actually, EMM can
an increase in the Ea value. Once the SE value reaches its lowest point, also be extended to concrete incorporating special cements, such as high
further increase in the Ea value leads to an increase of the SE value. The belite cement, or novel admixtures like temperature rising inhibitors.
corresponding SE values for different Ea are also summarized in Table 5. However, EMM is still laborious and time-consuming compared to the
Even with the best-fit Ea determined by ISM, the SE is not zero, sug­ “s”-related formulas. Shortening the experimental time of EMM or
gesting that the best-fit Ea value cannot completely eliminate the pre­ proposal of more robust machine learning models to determine coeffi­
diction error. The smallest SE for the best-fit Ea mainly comes from two cient “s” is needed for the fib model code’s strength function. The related
sources: the variability of CS and the limitation of the model itself (fib work is ongoing in our lab.
model code’s strength function may not perfectly reflect the strength In analogy to ASTM C1074 [19], the original fib model code’s
development of concrete). In addition, it is evident, in Fig. 6, that the strength function is modified to incorporate the reaction rate constant k,
searched-out best-fit Ea value varies from one mixture to another. so that the mixture-specific Ea value for the fib model code’s maturity
However, the Ea value determined by LAP closely aligns with the best-fit method can be calculated by LAP. ISM is also utilized to determine the
Ea value. The slight difference between these two Ea values arises from best-fit Ea value for different concrete/mortar specimens. In contrast, the
the selection of the prediction benchmark. ISM is based on the fitted Ea values obtained from these two methods are close to each other, and
curve for the investigated concrete at the reference curing temperature, are not always align with the default Ea value (33.33 kJ/mol) provided
but LAP considers the comprehensive influence of three different curing by the fib model code. Based on the calculated prediction errors for the
temperatures. This discrepancy is directly reflected in LAP. As shown in two cases presented in our study, the default Ea value is still found to
Fig. 5, the fitted lines do not necessarily pass through the point at 20℃, induce relatively low SE on strength estimation for concrete with or
leading to the Ea value determined by LAP differing from the best-fit without FA addition, and the Ea values determined from LAP and ISM
values. only slightly improves prediction performance. We will not assert that
When the default Ea value is compared with the Ea values determined the default Ea value is still applicable for FA concrete here, because the
from ISM and LAP, we find that the default Ea value generally deviates purpose of our study is mainly to provide alternative methods to
further from the best-fit value, compared to the LAP-derived one. Only determine the Ea value for the fib model code’s maturity method.
for M1-FA in Case 3, the default Ea value closely matches its best-fit Further mathematical analysis is still required to explain this phenom­
value. For M5-OPC, default Ea value is even 5.81 kJ/mol higher than enon and assess the suitability of the default Ea value on CS estimation
the best-fit Ea. In comparison, LAP-derived Ea value for M5-OPC is for FA concrete. Furthermore, it is worth mentioning that the prediction
almost identical to the best-fit one. Similar results are also observed in performance of the fib model code’s maturity-based strength function is
Case 4. For FA35-0.7, the default Ea value is even 11 kJ/mol higher, but relied not only on the Ea value, but also the coefficient “s”. The vari­
the Ea value obtained from LAP is only 1.8 kJ/mol lower compared to ability of coefficient “s” could also contribute to the prediction error.
the best-fit Ea value. These results have indicated that compared to the However, our study only presents the calculation method of the Ea value,
default Ea value, LAP-derived Ea value appears to be more appropriate with the “s” value already determined by direct fitting of the CS data
for the fib model code’s maturity method. When turning to the SE values from the investigated concrete. If we use the formula-derived “s” and the
presented in Table 5, it is surprisingly to find that the default Ea do not default Ea to predict the CS of FA concrete, the errors may become more
cause significant prediction deviation. For M5-OPC in Case 3, its SE significant. Therefore, considering this, we still recommend using our
value corresponding to the best-fit Ea is only 0.557 MPa. However, when proposed LAP or ISM to calculate Ea.
the default Ea value is applied, SE value is increased to 1.437 MPa. This
seems to be acceptable for strength estimation in engineering terms. For 5. Conclusion
Case 4, this phenomenon is more evident. Specifically, the SE values
induced by the default Ea value are only 0.04, 0.06, 0.03, and 0.17 MPa In this study, in order to extend the maturity-based strength function
higher than those obtained using ISM, for FA25-0.5, FA25-0.6, FA35- recommended by the fib model code to FA concrete, practical ap­
0.6, and FA35-0.7, respectively. In view of the SE values for the two proaches to determine coefficient “s” and apparent activation energy Ea
cases, the default Ea value appears to be reasonably applicable for are proposed. Four cases are adopted to validate the feasibility of the
strength estimation using fib model code’s maturity method, while it approaches. The conclusions drawn from the study are as follows.
sometimes deviates too much from the best-fit Ea value. However, There are several “s”-related formulas proposed by different re­
considering the limited case studies, it cannot be asserted that the searchers. However, different formulas might derive similar or signifi­
default Ea can be applied to all concrete mixtures. To achieve more ac­ cantly different “s” values for the same FA concrete mixture. Meanwhile,
curate prediction of CS development, especially for FA concrete, it is still there is no formula that considers all the influencing factors. The
recommended to utilize LAP or ISM to determine the Ea value for the equivalent mortar method, proposed in this study, can help to determine
investigated concrete. the coefficient “s” for the investigated concrete, when the fib model
code’s strength function is applied to FA concrete. The equivalent
4.3. Further discussion mortar contains the same constituents as its corresponding concrete, and
the physical, chemical, and mineralogical properties of FA and cement
In this study, EMM given in ASTM C1074 [19] has been utilized to on coefficient “s” and characteristics of strength gain have also been
help determine the coefficient “s” for the fib model code’s strength taken into consideration in the equivalent mortar method. This causes
function. Seven mixtures, from two cases with different w/b, FA/b, and the equivalent mortar method superior over the other “s”-related for­
chemical composition of FA and cement, have well validated the mulas. While EMM is accurate, this method still requires labor and time.
applicability of EMM. Compared with the “s”-related formulas, EMM More robust “s”-related formula with all possible variables is still

11
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

Fig. A. AEs of “s” values from different methods for (a) Case 1 and (b) Case 2. (1: EMM; 2: Eq. (3), Bamforth’s formula; 3: Eq. (4), Chen’s formula; 4: Eq. (5),
Bhakara’s formula; 5: Eq. (6), Vollpracht’s formula; 6: Eq. (7), Sun’s formula).

needed. CRediT authorship contribution statement

Two calculation procedures have been supplemented to fib model
code’s maturity method to determine the Ea value. One is to modify the Y. Sun: Writing – review & editing, Writing – original draft, Visu­
original strength function to let it incorporate a reaction rate constant, alization, Validation, Supervision, Software, Resources, Project admin­
so that linear Arrhenius plot can be used to determine the Ea value. The istration, Methodology, Investigation, Funding acquisition, Formal
other is the iterative searching method to find out the best-fit Ea value. analysis, Data curation, Conceptualization. H.S. Lee: Supervision,
The results show that the default Ea value deviates much from the best- Project administration.
fit Ea value from the iterative searching method or the one from the
linear Arrhenius plot, while only a slight increase of the standard error is
Declaration of Competing Interest
induced by the default Ea value. Further analysis still needs to be con­
ducted to understand the relationship between the default Ea and pre­
The authors declare that they have no known competing financial
dictive error of the fib model code’s maturity method, and understand
interests or personal relationships that could have appeared to influence
the applicability of the default Ea value. At present, in order to guarantee
the work reported in this paper.
accurate estimation on strength development of FA concrete, it is rec­
ommended that the suitable Ea values should be determined from our
Data availability
proposed calculation procedures before application of the fib model
code’s maturity method.
Data will be made available on request.

Fig. B. Boxplots of REs for (a) Case 1 and (b) 2 (1: EMM; 2: Bamforth formula; 3: Chen’s formula; 4: Bhakara’s formula; 5: Vollpracht’s formula; 6: Sun’s formula).

12
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

Fig. C. Fitting the CS development of cement concrete under different curing temperatures for Case 3: (a) M1-OPC, (b) M2-OPC, (c) M3-OPC, (d) M4-OPC, (e)
M5-OPC.

Fig. D. Fitting the CS development of FA concrete under different curing temperatures for Case 3: (a) M1-FA, (b) M2-FA, (c) M3-FA, (d) M4-FA, (e) M5-FA.

13
Y. Sun and H.S. Lee Construction and Building Materials 409 (2023) 133643

Fig. E. Fitting the CS development of FA mortar cured under different temperatures for Case 4: (a) FA25-0.5, (b) FA25-0.6, (c) FA35-0.6, and (d) FA35-0.7.

Acknowledgements [14] J. Chen, K.G. Kuder, D.E. Lehman, C.W. Roeder, L.N. Lowes, Creep modeling of
concretes with high volumes of supplementary cementitious materials and its
application to concrete-filled tubes, Mater. Struct. 50 (2017) 1–20.
This research was supported by research fund of Hanyang University [15] X.-Y. Wang, Effect of fly ash on properties evolution of cement based materials,
(HY-2023-202300000000534). Constr. Build. Mater. 69 (2014) 32–40.
[16] A. Vollpracht, M. Soutsos, F. Kanavaris, Strength development of GGBS and fly ash
concretes and applicability of fib model code’s maturity function–A critical review,
Appendix Constr. Build. Mater. 162 (2018) 830–846.
[17] B. Sun, W. Zhao, G. Cai, T. Noguchi, W. Wang, A novel strength prediction model of
mortars with different types of cement and SCMs, Struct. Concr. 23 (2022)
1214–1225.
[18] C.V. Nielsen, M. Kaasgaard, Activation Energy for the Concrete Maturity Model-
References Part 1: Compressive Strength Tests at Different Curing Temperatures, Nord. Concr.
Res. 62 (2020) 87–106.
[1] A. Adesina, Recent advances in the concrete industry to reduce its carbon dioxide [19] C. ASTM, 1074-19e1, Standard practice for estimating concrete strength by the
emissions, Environ. Challenges. 1 (2020), 100004. maturity method, 2021. https://doi.org/10.1520/C1074-19E01.
[2] Z.T. Yao, X.S. Ji, P.K. Sarker, J.H. Tang, L.Q. Ge, M.S. Xia, Y.Q. Xi, [20] R.W. Nurse, Steam curing of concrete, Mag. Concr. Res. 1 (1949) 79–88.
A comprehensive review on the applications of coal fly ash, Earth-Science Rev. 141 [21] A.G.A. Saul, Principles underlying the steam curing of concrete at atmospheric
(2015) 105–121. pressure, Mag. Concr. Res. 2 (1951) 127–140.
[3] C.S. Poon, L. Lam, Y.L. Wong, A study on high strength concrete prepared with [22] H. Freiesleben, P. Ej, Måleinstrument til control af betons hærdning, Nord. Betong.
large volumes of low calcium fly ash, Cem. Concr. Res. 30 (2000) 447–455. (1977) 21–25.
[4] C.D. Atiş, Heat evolution of high-volume fly ash concrete, Cem. Concr. Res. 32 [23] M. Soutsos, A. Vollpracht, F. Kanavaris, Applicability of fib model code’s maturity
(2002) 751–756. function for estimating the strength development of GGBS concretes, Constr. Build.
[5] K. Hwang, T. Noguchi, F. Tomosawa, Prediction model of compressive strength Mater. 264 (2020), 120157.
development of fly-ash concrete, Cem. Concr. Res. 34 (2004) 2269–2276. [24] N.J. Carino, R.C. Tank, 17 Maturity functions for concrete made with various
[6] D. Ravina, P.K. Mehta, Compressive strength of low cement/high fly ash concrete, cements and admixtures, in: Test. Dur. Concr. Constr. Proc. RILEM Colloquium,
Cem. Concr. Res. 18 (1988) 571–583. Darmstadt, March 1990, CRC Press, 1991: p. 192.
[7] P. Nath, P. Sarker, Effect of fly ash on the durability properties of high strength [25] M. Soutsos, A. Hatzitheodorou, F. Kanavaris, J. Kwasny, Effect of temperature on
concrete, Procedia Eng. 14 (2011) 1149–1156. the strength development of mortar mixes with GGBS and fly ash, Mag. Concr. Res.
[8] N. Bouzoubaa, M.-H. Zhang, V.M. Malhotra, Mechanical properties and durability 69 (2017) 787–801.
of concrete made with high-volume fly ash blended cements using a coarse fly ash, [26] S.J. Barnett, M.N. Soutsos, S.G. Millard, J.H. Bungey, Strength development of
Cem. Concr. Res. 31 (2001) 1393–1402. mortars containing ground granulated blast-furnace slag: Effect of curing
[9] T.R. Naik, S.S. Singh, M.M. Hossain, Permeability of concrete containing large temperature and determination of apparent activation energies, Cem. Concr. Res.
amounts of fly ash, Cem. Concr. Res. 24 (1994) 913–922. 36 (2006) 434–440.
[10] B. Sun, T. Noguchi, G. Cai, Q. Chen, Prediction of early compressive strength of [27] C.H. Lee, K.C. Hover, Influence of datum temperature and activation energy on
mortars at different curing temperature and relative humidity by a modified maturity strength predictions, ACI Mater. J. 112 (2015) 781.
maturity method, Struct. Concr. 22 (2021) E732–E744. [28] Q. Ya, Maturity method to predict the strength of concrete in winter construction,
[11] G.S. Vijaya Bhaskara, K. Balaji Rao, M.B. Anoop, Model for compressive strength J. Harbin Inst. Tech. (2016).
development of OPC concrete and fly ash concrete with time, Mag. Concr. Res. 70 [29] R.K. Dhir, J.G.L. Munday, L.T. Ong, Investigations of the engineering properties of
(2018) 541–557. OPC/pulverised fueL AS-H concrete: strength development and maturity, Proc.
[12] fib Model Code for Concrete Structures, Ernst & Sohn publishing company, 2010. Inst. Civ. Eng. 77 (1984) 239–254.
[13] P. Bamforth, D. Chisholm, J. Gibbs, T. Harrison, Properties of concrete for use in [30] S.A. Ragan, Compressive strength-maturity relationships of mortar containing fly
Eurocode 2, MPA the Concrete Centre (2008). ash, Structures Laboratory (US) (1984). https://apps.dtic.mil/sti/citations/
ADA148705.

14