You are on page 1of 8

Construction and Building Materials 93 (2015) 49–56

Contents lists available at ScienceDirect

Construction and Building Materials


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

Alkali-activated concrete: Engineering properties and stress–strain


behavior
Robert J. Thomas, Sulapha Peethamparan ⇑
Department of Civil and Environmental Engineering, Clarkson University, Potsdam, NY 13699, USA

h i g h l i g h t s

 The engineering properties of alkali-activated concretes (AAC) are determined.


 Results are compared to existing models for portland cement concrete (PCC).
 Models are proposed to predict tensile strength and modulus of AAC.
 AAC exhibits higher tensile strength and lower Poisson’s ratio than PCC.
 Alkali-activated slag concrete is much more brittle than PCC.

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

Article history: This paper presents an investigation into the tensile strength, modulus of elasticity, Poisson’s ratio, and
Received 18 November 2014 stress–strain relationships of alkali-activated portland-cement-free concrete made with fly ash or ground
Received in revised form 6 March 2015 granulated blast furnace slag (GGBFS) as the sole binder. Alkali-activated concrete is shown to be stronger
Accepted 22 April 2015
in tension and have lower Poisson’s ratio than portland cement concrete. Relationships are proposed to
estimate the tensile strength and modulus of elasticity based on the compressive strength of
alkali-activated concrete, which are of the same form as those currently employed for portland cement
Keywords:
concrete.
Alkali-activated concrete
Modulus of elasticity
Ó 2015 Elsevier Ltd. All rights reserved.
Poisson’s ratio
Stress–strain curve
Tensile strength

1. Introduction improve the sustainability of concrete, but have also been shown
to improve rheology, mechanical strength, and durability [7–10].
Portland cement concrete (PCC) is the most widely used con- As an added benefit, repurposing industrial wastes reduces land-
struction material, with production reaching one hundred million filling costs and associated environmental impact. Despite numer-
tonnes in the United States in 2005 [1,2]. Even as domestic ous and well-established benefits of partial cement replacement,
production declines, global consumption of portland cement grows fly ash and other pozzolans are typically limited to about 25% of
significantly. Modern advancements to the portland cement man- total cementitious material, while GGBFS is often used in quanti-
ufacturing process have reduced associated energy consumption ties as high as 50% of total cementitious material. A few studies
and emissions, but even the most modern operations use as much have investigated the feasibility of high-volume cement replace-
as 4 GJ of energy and emit nearly one tonne of carbon dioxide per ment with some success, but these mixtures still require the use
tonne of cement produced [3–6]. This is by no means a new reve- of a significant amount of portland cement [11,12].
lation; the concrete industry has long sought means to supplement More recently, a number of studies have demonstrated the syn-
or replace portland cement in concrete. Partial cement replace- thesis of portland cement-free binders by alkali activation of
ment with cementitious or pozzolanic industrial wastes like fly industrial byproducts and natural pozzolans. Turner and Collins
ashes and ground granulated blast furnace slag (GGBFS) has been [13] recently suggested that alkali-activated binders can result in
employed with measurable success. These modifications not only as much as 45% reduction in total carbon emissions associated with
concrete production. The activation of GGBFS, fly ashes, and natu-
⇑ Corresponding author. Tel.: +1 315 268 4435. ral pozzolans with sodium and potassium alkalis has been shown
E-mail addresses: thomasrj@clarkson.edu (R.J. Thomas), speetham@clarkson.edu by several studies to result in strong and durable binders [14–
(S. Peethamparan). 19]. Arguably the most common and effective activator used in

http://dx.doi.org/10.1016/j.conbuildmat.2015.04.039
0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.
50 R.J. Thomas, S. Peethamparan / Construction and Building Materials 93 (2015) 49–56

relevant studies is sodium silicate, i.e., Na2O +mSiO2 where m is  n


r ¼ r0 ð6Þ
the silica modulus. Sodium silicate-activated fly ash and slag con- 0 n  1 þ 0 n
cretes have been shown to develop compressive strengths in
excess of 30 MPa. Some formulations, particularly those based on 1
0 ¼ 2:7  104 r0 4
ð7Þ
fly ashes, require elevated temperature curing to develop apprecia-
ble early-age strength [20–25]. Although the literature is some-
what conflicted with regard to durability, most available studies n ¼ 0:4  103 r0 þ 1 ð8Þ
have shown activated concrete to exhibit increased resistance to A few studies have discussed the tensile strength of
chloride penetration [20,26,27], acid attack [28–30], and sulfate alkali-activated and geopolymer concrete [22,44,45,52,53]. Several
attack [31], and reduced potential for deleterious alkali–silica reac- have suggested, based on somewhat limited data and without any
tion [32–34]. Despite the wealth of knowledge regarding the per- proposed mathematical relationship, that alkali-activated concrete
formance of alkali-activated concrete, very few studies have exhibits improved tensile strength over that typical of portland
addressed the engineering properties of activated concrete, viz. cement concrete [18,22,52,53]. Specifically, Duran Atisß et al. [52]
the modulus of elasticity, Poisson’s ratio, tensile strength, and investigated the mechanical properties of alkali-activated GGBFS
stress–strain relationships. Thorough understanding of these prop- mortar, concluding that sodium silicate-activated GGBFS mortar
erties is essential to fully characterize the performance of exhibited the best tensile strength, and that optimum solution
alkali-activated concrete for the purpose of design and field parameters (i.e. silica modulus m and sodium oxide content) exist
implementation. that provide the best tensile strength performance. A few other
The engineering properties of PCC are generally well under- studies have shown contradicting results. Lee and Lee [43] found
stood. The tensile strength f ct , measured by the splitting tension that the tensile strength of concrete made with alkali-activated fly
(Brazilian) test, is typically 10–15% of the compressive strength ash/GGBFS mixtures was about 20% lower than that of portland
0
f c . A number of more specific estimates have been proposed which cement concrete. Similarly, Sofi et al. [44] found that the tensile
have generally been of the form of Eq. (1), where the splitting ten- strength of high-compressive-strength geopolymer concrete made
0
sile strength f ct and the compressive strength f c are in MPa. Typical with fly ash and GGBFS was significantly lower than that of both
fitting parameters are: 5 6 A 6 2 and 9 6 B 6 79 [35–40], although
1 1 5 normal and high strength portland cement concrete. The latter
the power B is frequently constrained to a value of 12, giving a func- two studies proposed two very similar mathematical models to
0
tion of the square root of the compressive strength [41–44]. Most relate splitting tensile strength f ct to compressive strength f c , which
notable is the equation for estimated splitting tensile strength are given as Eqs. (9) [43] and (10) [44].
given by ACI building code 318 [41] (Eq. (2)). High strength con- qffiffiffiffi
0
crete often exhibits marginally higher tensile strength, and there- f ct ¼ 0:45 f c ð9Þ
fore the relationship proposed by Carrasquilio and Nilson [42] qffiffiffiffi
(Eq. (3)) applies. 0
f ct ¼ 0:48 f c ð10Þ
 0 B
f ct ¼ A fc ð1Þ Several authors have reported the modulus of elasticity of var-
qffiffiffiffi ious types of alkali-activated concrete in conjunction with other
f ct
0
¼ 0:56 f c ð2Þ studies, but few have offered any substantial analysis or discussion
of significant factors or their effects. These studies have all
qffiffiffiffi reported moduli of elasticity in the range of 30–40 GPa
0
f ct ¼ 0:59 f c ð3Þ [14,18,22] for activated fly ash and GGBFS concrete, which is in
the same range as expected for portland cement concrete. Based
Young’s modulus of elasticity E (MPa) of portland cement con-
on the results from experiments on alkali-activated natural
crete is generally estimated as a function of the compressive
0
geopolymer concrete, Duxson et al. [53] suggested that the modu-
strength f c (MPa) and the unit weight wc (kg/m3) as in Eq. (4), after lus of elasticity of alkali-activated natural geopolymer concrete is
[45] and ACI building code 318 [41]. For normalweight concrete strength–independent, at least within the normal range of com-
this reduces to Eq. (5) which, it should be noted, is of the same pressive strengths for structural concrete. Instead, it is suggested
form as Eq. (1) with B ¼ 12. that the microstructural development has a more significant effect
0 1=2 on the modulus of elasticity, which may be more or less the same
E ¼ 0:043wc3=2 ðf c Þ ð4Þ for some alkali-activated binders [53]. Lee and Lee [43] proposed
the relationship shown in Eq. (11) for the modulus of elasticity of
0 1=2
E ¼ 4700ðf c Þ ð5Þ alkali-activated fly ash/GGBFS mixtures, finding that the ACI equa-
tion (Eq. (4) overestimated the modulus of elasticity. Poisson’s
The stress–strain behavior of concrete under uniaxial compres-
ratio has gone almost entirely unmentioned in conjunction with
sion is relatively well understood, and is marked by nearly perfect
alkali-activated concrete in available literature, with the exception
linear elasticity up to the ultimate strength, followed by distinct
of Sofi et al. [44], who found Poisson’s ratios in the range of 0.24–
strain softening [46,47]. As the compressive strength increases,
0.26 for inorganic polymer concretes.
so does the brittleness; this manifests as an increasingly steep soft-
ening curve. Very high strength concrete exhibits highly brittle  0 1
E ¼ 5300 f c 3 ð11Þ
failure, with little or no softening behavior prior to complete rup-
ture. Several authors have proposed analytical relationships to This study seeks to address the aforementioned gaps in knowl-
describe the stress–strain curve of portland cement concrete, edge and to cultivate a more thorough understanding of the engi-
although these relationships tend to be highly dependent on test neering properties and stress–strain behavior of alkali-activated
conditions [48–50]. A reliable numerical stress–strain relationship concrete for the purpose of design and field implementation.
was determined by Popovics [47] in 1973, wherein the stress r Specifically, this study evaluates the compressive strength, split-
(psi) at a given strain  is estimated by a function of the ultimate ting tensile strength, modulus of elasticity, Poisson’s ratio, and
stress r0 (psi), the corresponding strain 0 , and a constant n (Eqs. stress–strain response for alkali-activated class-C fly ash (FC) and
(6)–(8)). GGBFS concrete; the results from this study are compared to the
R.J. Thomas, S. Peethamparan / Construction and Building Materials 93 (2015) 49–56 51

Table 1
and specific gravity of 2.71 was used as coarse aggregate, and was
Chemical oxide composition and selected physical properties of binders.
used in SSD condition. Mixture proportions were selected based
GGBFS FC on preliminary investigations which indicated acceptable worka-
% by mass bility and appreciable compressive strength [54], and are detailed
SiO2 36.0 37.7 in Table 2. Alkali-activated FC and GGBFS mixtures included 0.1%
Al2O3 10.5 20.0 dosage of rosin-based air-entraining agent for the purpose of work-
CaO 39.8 23.4 ability improvement, as demonstrated elsewhere [54]. ASTM C150
MgO 7.9 4.3 Type I/II ordinary portland cement concrete (PCC) control mixtures
Na2O 0.3 1.7
were selected to produce similar compressive strengths as expected
SO3 2.1 2.4
K2O 0.2 0.6 for the selected alkali-activated FC and GGBFS mixtures.
Fe2O3 0.7 5.6
3.2. Specimen preparation
typical behavior of portland cement concrete (PCC). Additionally,
24 cylindrical concrete specimens measuring 150 mm in length
this study evaluates the efficacy of using existing models for PCC
and 75 mm in diameter were cast in accordance with the specifica-
to predict the behavior of alkali-activated FC and GGBFS concrete.
tions of ASTM C192 for each mixture proportion as listed in
Finally, improved models are proposed to predict the splitting ten-
Table 2. The specimen size was limited to 75 mm due to the
sile strength and modulus of elasticity of alkali-activated con-
expense of the reagent-grade activators. Half of the specimens
cretes. It should be noted that the models presented herein relate
were cured at elevated temperature (48 ± 0.25 h @ 50 ± 0.1 °C)
the splitting tensile strength and modulus of elasticity of
and the other half were cured at ambient temperature
alkali-activated concrete to the compressive strength, as is typical
(28 ± 0.5 days @ 22 ± 1 °C and >95% RH). All specimens remained
for cementitious composites. Since the strong dependence of com-
sealed for the duration of curing to prevent moisture loss. After
pressive strength of alkali-activated concrete on the binder, activa-
curing, specimens were de-molded and test faces were ground flat
tor, and curing condition have previously been established, the
and plane using a wet silicon carbide grinding belt.
authors do not purport to address the effect of these mixture param-
eters on the engineering properties of alkali-activated concrete. That
3.3. Test methods
is to say, the aim of this paper is to propose relationships with which
engineers can easily estimate the engineering properties of
The compressive strength, splitting tensile strength, modulus of
alkali-activated concrete without any detailed knowledge of the
elasticity, Poisson’s ratio, and stress–strain relationship were
complex chemical makeup of alkali-activated formulations.
experimentally evaluated for each mixture at each curing condi-
tion, providing four replicates for each measurement. The average
2. Research significance unit weight of each batch was recorded to an accuracy of 1 kg/m3.
Compressive strength was determined in accordance with the
Although a wide body of literature is available describing the specifications of ASTM C39 using a stress application rate of
hydration products, reaction kinetics, mechanical strength, and 15 MPa/min. Splitting tensile strength, which is known to overesti-
durability of alkali-activated concrete, the engineering properties, mate the tensile strength of concrete when compared to direct ten-
viz., tensile strength, modulus of elasticity, and Poisson’s ratio, sion testing [37], was determined in accordance with the
have received only cursory attention, if any at all. Several papers specifications of ASTM C496 using a stress application rate of about
have specifically called for additional investigations into these 1 MPa/min. The width of the bearing strips for the splitting test
areas [44,53]. Additionally, the authors are unaware of any studies was less than 8% of the cylinder diameter in order to achieve
that have investigated the stress–strain relationships of activated repeatable results, as [55] showed significant variation in tests
concrete. A complete understanding of these properties and how results when bearing strips exceeded this limit. The static chord
they relate to one another is vital for design and field implementa- modulus of elasticity and Poisson’s ratio were determined in gen-
tion of alkali-activated concrete structures. eral agreement with the specifications of ASTM C469, except for
that the testing was performed under closed-loop displacement
3. Experimental investigation control. This modification in test procedure was made in order to
allow evaluation of the post-failure strain response. Specimens
3.1. Materials were loaded in pure uniaxial compression at a constant displace-
ment rate of 0.1 mm/min and were instrumented with an axial
The binder materials used in this study were high-calcium ASTM and radial extensometer, as shown in Fig. 1. Post-failure strain
C618 Class C coal fly ash (FC) and ASTM C989 Grade 100 granulated measurements were also desired, and because failure typically
blast furnace slag (GGBFS), both from reputable commercial suppli- causes the mounted gages to slip, two LVDT were also deployed
ers. The oxide composition, as determined by X-ray fluorescence in diametrically opposed positions in the axial direction.
spectrometry, and selected physiochemical properties of the
as-received binders are listed in Table 1. The hydration products, 4. Results and discussion
reaction kinetics, and mechanical strengths of these binders have
bene described by Deir et al. [19], and Gebregziabiher et al. [25]. The average compressive strength, tensile strength, elastic
The activator was a mixed solution of sodium silicate (Na2O modulus, and Poisson’s ratio for each mixture (4 replicates) are
+mSiO2) and sodium hydroxide (NaOH). The Na2O equivalent of reported in Table 3. The unit weights of all concretes were within
the activator is reported relative to binder mass (%Na2O, by mass 2400 ± 27 kg/m3. Individual unit weights are not reported due to
of binder), and the relative concentration of silica to sodium oxide their similarity between mixtures.
equivalent is reported as the silica modulus m. Saturated
surface-dry (SSD) natural fine aggregate with fineness modulus of 4.1. Compressive strength
2.42 and specific gravity of 2.65 was used for all mixtures.
Crushed stone composed predominately of rose quartz with a nom- Ambient-temperature-cured (28 d @ 22 °C) activated fly ash
inal maximum particle size of 9.5 mm, bulk density of 1560 kg/m3, concrete exhibits very low compressive strength compared to
52 R.J. Thomas, S. Peethamparan / Construction and Building Materials 93 (2015) 49–56

Table 2
Mixture proportions

ID %Na2O (binder mass) m %SiO2 (binder mass) s/b (by mass) Solution (kg/m3) Binder (kg/m3) Fine Aggr. (kg/m3) Coarse Aggr. (kg/m3)
PCC1 – – – 0.40 228 570 658 780
PCC2 – – – 0.36 228 620 610 780
FC1 4.0 1.5 6.0 0.40 228 570 658 780
FC2 5.0 1.5 7.5 0.40 228 570 658 780
FC3 6.0 1.5 9.0 0.40 228 570 658 780
FC4 6.0 1.5 9.0 0.40 248 620 610 780
GGBFS1 4.0 0.75 3.0 0.40 228 570 658 780
GGBFS2 5.0 0.75 3.75 0.40 228 570 658 780
GGBFS3 6.0 0.75 4.0 0.40 228 570 658 780
GGBFS4 5.0 0.75 3.75 0.40 248 620 610 780
GGBFS5 2.0 1.5 5.0 0.40 228 570 658 780
GGBFS6 2.5 1.5 6.25 0.40 228 570 658 780

Table 3
0
Summary of compressive strength f c , tensile strength f ct , elastic modulus E, and
Load cell Poisson’s ratio l results for all mixtures.
0
ID Curing condition f c (MPa) f ct (MPa) E (GPa) l

Spherical seat PCC1 28 d @ 22 °C 39.5 5.9 33.1 0.184


PCC2 28 d @ 22 °C 40.9 6.2 32.7 0.175
FC1 28 d @ 22 °C 16.2 2.9 17.7 0.126
Top platen 48 h @ 50 °C 31.5 6.1 28.8 0.128
FC2 28 d @ 22 °C 28.9 6.7 21.5 0.128
48 h @ 50 °C 47.7 7.6 26.0 0.124
FC3 28 d @ 22 °C 22.9 3.8 22.6 0.127
LVDT LVDT 48 h @ 50 °C 50.3 7.5 35.5 0.124
FC4 28 d @ 22 °C 21.3 4.1 10.5 0.127
48 h @ 50 °C 40.9 6.5 30.9 0.126
Radial gage GGBFS1 28 d @ 22 °C 33.7 6.8 34.2 0.129
Axial gage 48 h @ 50 °C 29.5 6.1 31.7 0.125
GGBFS2 28 d @ 22 °C 44.7 7.2 26.2 0.130
48 h @ 50 °C 44.0 7.4 28.2 0.126
GGBFS3 28 d @ 22 °C 46.7 6.3 27.0 0.127
48 h @ 50 °C 45.6 7.7 25.4 0.124
Specimen
GGBFS4 28 d @ 22 °C 35.0 6.3 27.9 0.134
48 h @ 50 °C 37.7 6.2 28.9 0.129
GGBFS5 28 d @ 22 °C 45.7 8.3 22.4 0.125
48 h @ 50 °C 48.7 7.4 22.9 0.128
Bottom platen GGBFS6 28 d @ 22 °C 52.6 8.4 33.5 0.127
48 h @ 50 °C 50.8 7.7 33.5 0.126

Fig. 1. Test configuration for determination of modulus of elasticity, Poisson’s ratio, marginally higher than those of PCC control mixtures. The average
and stress–strain relationship. splitting tensile strength of alkali-activated GGBFS concrete is
17.0 ± 2.1% of the corresponding compressive strength, which is
significantly higher than observed for PCC mixtures in a similar
identical specimens cured at elevated temperature (48 h @ 50 °C).
strength range. The sensitivity of the splitting tensile strength of
No such effect is observed in compressive strength results for
activated fly ash concrete to the compressive strength appears to
alkali-activated GGBFS concrete, as has been reported by a few
be affected by the curing condition; when cured at ambient tem-
studies [19,54]. For both binders, increased binder content results
perature, the average splitting tensile strength is 19.4 ± 2.8% of
in reduced compressive strength as indicated by the strength
the corresponding compressive strength, while the ratio is
reduction between FC formulations FC3 and FC4 and GGBFS formu-
16.5 ± 1.9% when cured at elevated temperature.
lations GGBFS2 and GGBFS4. Improved strength with increased
The splitting tensile strength and corresponding compressive
binder content has been observed in PCC [56]. The propensity of
strength of activated fly ash and GGBFS concrete are plotted in
activated concrete for microcracking, particularly under
Fig. 2. Two prediction models are proposed which relate the split-
elevated-temperature-curing, has been discussed by Collins and
ting tensile strength to the compressive strength. The first, Eq. (12),
Sanjayan [57]. It is likely that this microcracking, combined with
is of the general form of Eq. (1) and fits the data quite well
increased volume fraction of the paste phase, leads to the reduc-
tion in strength of alkali-activated concrete with increased binder (R2 ¼ 0:86). A second proposed equation is also given, following
content. the common convention of limiting the power B in Eq. (1) to 12, as
in [42–45]. This square-root model (Eq. (13)) fits the data reason-
ably well (R2 ¼ 0:80), but not quite as well as with the unrestricted
4.2. Tensile strength power (Eq. (12)). The goodness of fit of both proposed equations is
considered quite acceptable due to the previously established vari-
The average tensile strength of PCC control specimens is ability of splitting tensile test results [55].
14.1 ± 0.6% of the corresponding compressive strength, which is
within the expected range of 10–15%. The tensile strengths of 2  0 79
f ct ¼ f ð12Þ
alkali-activated FC and GGBFS concrete observed in this study is 5 c
R.J. Thomas, S. Peethamparan / Construction and Building Materials 93 (2015) 49–56 53

18

16

14 Lateral Axial

12

Stress (MPa)
10

6
FC2 22◦ C
4 FC2 50 ◦ C
FC4 22◦ C
Fig. 2. Splitting tensile strength f ct and compressive strength
0
fcof alkali activated 2 FC4 50◦ C
FC and GGBFS concrete along with proposed models (Eq. (12) and (13)) and and PC1
existing models proposed by Lee and Lee [43], Sofi et al. [44], ACI 318 [41], 0
Carrasquilio and Nilson [42]. −100 0 100 200 300 400 500 600 700 800
Strain (μ)
qffiffiffiffi
0 Fig. 3. Axial and lateral stress–strain relationships for alkali-activated fly ash
f ct ¼ 1:08 f c ð13Þ concrete in pure uniaxial compression within the proportional limit.

In addition to the proposed models (Eqs. (12) and (13)), several


previously proposed models for splitting tensile strength of con-
18
crete are also included for comparison on Fig. 2. These include
the model proposed by Lee and Lee [43] for alkali-activated mixed 16
fly ash/slag concrete, that proposed by Sofi et al. [44] for
high-strength inorganic polymer concrete, the ACI 318 [41] model 14 Lateral Axial
for normal strength PCC, and a model proposed by Carrasquilio and
Nilson [42] for high strength PCC. The splitting tensile strengths 12
Stress (MPa)

observed here for alkali-activated FC and GGBFS concretes far


10
exceed the predictions for both portland cement concrete [41,42]
and for alkali-activated and geopolymer concrete [43,44] by a fac- 8
tor of nearly two. Lee and Lee [43] and Sofi et al. [44], despite using
similar materials as used in the present study, used comparatively 6
high solution-to-binder (s=b) ratios; the former did not specifically GGBFS3 22◦ C
4 GGBFS3 50 ◦ C
specify s=b, but in most cases specified mass ratios of water to fly
GGBFS5 22◦ C
ash greater than 1, and the latter used s=b ¼ 0:56. The compara- GGBFS5 50◦ C
2
tively low ratio used in the present study (s=b ¼ 0:40) is a likely PC1
reason for the discrepancy. A few other studies have suggested 0
improved tensile strength in alkali-activated concretes compared −100 0 100 200 300 400 500 600 700 800
to PCC [22,51,52], as is seen in the present study. Strain (μ)

Fig. 4. Axial and lateral stress–strain relationships for alkali-activated GGBFS


4.3. Modulus of elasticity and Poisson’s ratio concrete in pure uniaxial compression within the proportional limit.

The axial and lateral strain relationships in the elastic region


of the form of Eq. (1); the second, given by Eq. (15), is of the same
under uniaxial compression for alkali-activated concrete are highly
linear, as expected. Representative curves are presented in Figs. 3 form with the constraint that the power B ¼ 12, as in Eq. (5). The
0
and 4 for activated fly ash and GGBFS concrete, respectively. modulus of elasticity E and compressive strength f c are of course
Poisson’s ratio for alkali-activated FC and GGBFS concrete mixtures in MPa. The fitted equations vary only slightly from the ACI 318
is 0.127 ± 0.003, while PCC control specimens exhibited Poisson’s equation (Eq. (5) in terms of both predicted values and goodness
ratio very near the expected value of 0.18. This is a very surprising of fit. Coefficients of determination (R2 ) for Eqs. (5), (14), and
result which has gone unreported in the existing literature, and can (15) are 0.54, 0.58, and 0.60, respectively. Despite these low coef-
be seen in Figs. 3 and 4 where the lateral strain curve for PCC is ficients of correlation which occur as a result of highly variable
very similar to those for alkali-activated FC and GGBFS concrete data and several potential outliers, all three equations do indeed
formulations even though the axial strain curve is much steeper. provide a good representation of the trend. Additionally, because
The modulus of elasticity of all alkali-activated FC and GGBFS con- of the significant variation in the data, it is really impossible to con-
crete mixtures is within the expected range for concrete; observa- clude that one of these equations is a better predictor than another
tions are within 10 6 E 6 35 GPa. The individual values observed without additional studies. The model proposed by Lee and Lee
for the modulus of elasticity of activated fly ash and GGBFS con- [43] for alkali-activated fly ash/GGBFS mixtures is also plotted; this
crete are plotted against the corresponding compressive strengths model drastically underestimates the modulus of elasticity of the
in Figs. 5 and 6. alkali-activated FC mixtures tested here, which could again be
The modulus of elasticity of activated fly ash concrete appears due to the higher water content of those mixtures, or purely due
to vary directly with the compressive strength, and this variation to the use of mixed fly ash/GGBFS formulations. Additional models
shows no discernible difference from that of PCC, as modeled by fit using curing condition as an additional predictor for the modu-
Eq. (5). Two models are proposed: the first, given by Eq. (14), is lus of elasticity of activated fly ash concrete only marginally
54 R.J. Thomas, S. Peethamparan / Construction and Building Materials 93 (2015) 49–56

Fig. 5. Modulus of elasticity and compressive strength of alkali-activated fly ash Fig. 7. Representative stress–strain curves for alkali-activated fly ash concrete.
concrete. Models for PCC after Popovics [47] are based on ultimate strength of given alkali-
activated FC specimens.

improve the goodness of fit. However, since the curing condition


has been established as a significant effect in the compressive any significant variation with curing condition observed.
strength, it is likely that its use as a parameter in determining Attempting to fit a linear model to these data results in a negative
the modulus of elasticity represents the addition of a confounding coefficient of determination (R2 < 0). The practical meaning of this
variable. Furthermore, it should be noted that because mixture is that a constant relationship (horizontal line) fits these data bet-
proportions in Table 2 are by weight and because the specific grav- ter than the least-squares linear regression equation. Additionally,
ities of the binders are not the same, the volume fraction of aggre- the ACI 318 equation (Eq. (5)) fits the data poorly. This is reminis-
gate is not constant between PCC and alkali-activated FC or GGBFS cent of the conclusions drawn by Duxson et al. [53] regarding the
binders (ranging from 0.279 to 0.305). The aggregate volume frac- modulus of elasticity of geopolymer concretes made with natural
tion, as discussed by Hansen [58] is one of the main factors affect- pozzolan binders, where it was suggested that the microstructure,
ing the modulus of elasticity of concrete, although theoretically, composition, and degree of hydration had a more significant effect
the effect of such a slight change in aggregate volume is fairly on the modulus of elasticity than the compressive strength or solu-
small. tion parameters. This is due to the fact that the activator and curing
 0 3 condition affect the composition, density, volume fraction, and
E ¼ 2900 f c 5 ð14Þ physical size of the product phases within the binder [19,25],
which can have significant effects on the mechanical properties.
 0 1
E ¼ 4400 f c 2 ð15Þ This result could also be a result of the significant microcracking
that has been demonstrated in sodium silicate-activated GGBFS
The modulus of elasticity of activated GGBFS concrete, averages concrete [57], which may be increasingly prevalent as compressive
about 30 GPa, does not appear to vary with compressive strength strength increases. No significant correlation was found between
0
over the range 20 MPa 6 f c 6 60 MPa, as shown in Fig. 6, nor is the modulus of elasticity of activated GGBFS concrete and the acti-
vator concentrations (sodium oxide content %Na2O, silica modulus
m, or silica content mNa2O).

4.4. Stress–strain relationships

A few representative stress–strain curves are shown for


alkali-activated FC and GGBFS concrete in Figs. 7 and 8. These
experimental curves are compared with the numerically-derived
curves for PCC after Popovics [47] (Eq. (6)). The stress–strain curves
for alkali-activated FC concrete resemble the numerical estimates
for PCC fairly closely. Minor differences in the strain corresponding
to the ultimate stress are observed under elevated-temperature
curing, and the behavior is slightly more brittle. Additionally,
alkali-activated FC concrete shows a more rapid decline in stress
during post-peak strain softening. This suggests decreased tough-
ness as compared to PCC. Overall, however, the stress–strain
behavior of alkali-activated FC concrete is not much different than
that typical of PCC. On the other hand, the stress–strain behavior of
alkali-activated GGBFS concrete is starkly different than that of
PCC. The strain corresponding to the ultimate stress is quite simi-
Fig. 6. Modulus of elasticity and compressive strength of alkali-activated GGBFS lar, but the post-peak behavior is drastically dissimilar. Where PCC
concrete. displays gradual strain softening beyond the ultimate stress,
R.J. Thomas, S. Peethamparan / Construction and Building Materials 93 (2015) 49–56 55

the entire range of compressive strengths represented in this


study. Existing equations, viz., Eqs. (4) and (5) (ACI 318), fit rea-
sonably well despite wide variation in the data.
 Alkali-activated class-C fly ash concrete exhibits similar stress–
strain behavior to ordinary portland cement concrete, marked
by imperfect linear elasticity to failure followed by post-peak
strain softening. Conversely, alkali-activated GGBFS concrete
exhibits highly brittle behavior marked by near perfect linear
elasticity followed by sudden and total failure.

Acknowledgment

The authors gratefully acknowledge the financial support of the


University Transportation Research Center, Region 2 (UTRC2) and
of the National Science Foundation through CMMI Award No.
1055641.

Fig. 8. Representative stress–strain curves for alkali-activated GGBFS concrete. References


Models for PCC after Popovics [47] are based on ultimate strength of given alkali-
activated GGBFS specimens. [1] United States Geological Survey. Mineral commodity summaries, cement;
2009.
[2] Madlool NA, Saidur R, Hossain MS, Rahim NA. A critical review on energy use
and savings in the cement industries. Renew Sust Energ Rev
2011;15(4):2042–60.
alkali-activated GGBFS displays brittle fracture immediately fol- [3] Worrell E, Martin N, Price L. Potentials for energy efficiency improvement in
the us cement industry. Energy 2000;25(12):1189–214.
lowing the peak. This increased brittleness has been bserved by [4] Worrell E, Price L, Martin N, Hendriks C, Meida LO. Carbon dioxide emissions
previous studies [52], and has generally been attributed to the high from the global cement industry 1. Annu Rev Energy Environ
prevalence of microcracking in alkali-activated GGBFS [57]. Due to 2001;26(1):303–29.
[5] Khurana S, Banerjee R, Gaitonde U. Energy balance and cogeneration for a
the limited number of test specimens evaluated here, no attempt is
cement plant. Appl Therm Eng 2002;22(5):485–94.
made to develop a numerical model of the stress–strain curve for [6] Choate WT. Energy and emission reduction opportunities for the cement
alkali-activated FC and GGBFS concrete; some researchers have industry. Energy and emission reduction opportunities for the cement
industry. BCS, Incorporated; 2003.
done so for PCC, but have generally used a very high sample size.
[7] ACI Committee 232. Use of fly ash in concrete (ACI 232.2R-03). American
For the purpose of this investigation, it is sufficient to conclude Concrete Institute; 2003.
that the stress–strain behavior of alkali-activated fly ash concrete [8] ACI Committee 232. Slag cement in concrete and mortar (ACI 233R–03).
is similar to that of PCC, but that of alkali-activated GGBFS concrete American Concrete Institute; 2003 (Reapproved 2011).
[9] Geiseler J, Kollo H, Lang E. Influence of blast furnace cements on durability of
exhibits high brittleness that is uncharacteristic of PCC. concrete structures. ACI Mater J 1995;92(3):252–7.
[10] Bouzoubaa N, Zhang MH, Malhotra VM, Golden DM. Blended fly ash cements a
review. ACI Mater J 1999;96(6):641–50.
5. Summary of conclusions [11] Bilodeau A, Malhotra VM. High-volume fly ash system: concrete solution for
sustainable development. ACI Mater J 2000;97(1):41–7.
The splitting tensile strength, modulus of elasticity, Poisson’s [12] Malhotra VM. High-performance high-volume fly ash concrete. Concr Int
2002;24(7):30–4.
ratio, and stress–strain relationships for alkali-activated FC and
[13] Turner LK, Collins FG. Carbon dioxide equivalent (CO2-e) emissions: A
GGBFS concrete were studied experimentally. The following con- comparison between geopolymer and OPC cement concrete. Constr Build
clusions are drawn from the results of this study: Mater 2013;43:125–30.
[14] Douglas E, Bilodeau A, Malhotra VM. Properties and durability of alkali-
activated slag concrete. ACI Mater J 1992;89(5):509–16.
 The mechanical strength of alkali-activated fly ash concrete [15] Fernández-Jiménez AM, Palomo JG, Puertas F. Alkali-activated slag mortars:
shows a significant effect from curing condition; ambient- Mechanical strength behaviour. Cem Concr Res 1999;29(8):1313–21.
temperature-curing (28 d @ 22 °C) resulted in significantly [16] Palomo A, Grutzeck MW, Blanco MT. Alkali-activated fly ashes: A cement for
the future. Cem Concr Res 1999;29(8):1323–9.
lower mechanical strength than elevated-temperature-curing [17] Hardjito D, Wallah SE, Sumajouw DMJ, Rangan BV. On the development of fly
(48 h @ 50 °C). In alkali-activated GGBFS concrete, the two cur- ash-based geopolymer concrete. ACI Mater J 2004;101(6):467–72.
ing conditions resulted in no difference in mechanical strength. [18] Fernández-Jiménez AM, Palomo A, Lopez-Hombrados C. Engineering
properties of alkali-activated fly ash concrete. ACI Mater J
 The tensile strength of alkali-activated fly ash and GGBFS 2006;103(2):106–12.
concrete is significantly higher than that of portland cement [19] Deir E, Gebregziabiher BS, Peethamparan S. Influence of starting material on
concrete with similar compressive strength. the early age hydration kinetics, microstructure and composition of binding
gel in alkali activated binder systems. Cem Concr Compos 2014;48:108–17.
 The splitting tensile strength of alkali-activated concrete is [20] Douglas E, Bilodeau A, Brandstetr J, Malhotra VM. Alkali activated ground
related to the corresponding compressive strength by a power granulated blast-furnace slag concrete: preliminary investigation. Cem Concr
function of similar form to those previously proposed for port- Res 1991;21(1):101–8.
[21] Wang SD, Scrivener KL, Pratt PL. Factors affecting the strength of alkali-
land cement concrete. A relationship with 86% correlation is
activated slag. Cem Concr Res 1994;24(6):1033–43.
proposed relating the tensile and compressive strengths of [22] Collins FG, Sanjayan JG. Workability and mechanical properties of alkali
alkali-activated FC and GGBFS concrete. activated slag concrete. Cem Concr Res 1999;29(3):455–8.
 Poisson’s ratio for alkali-activated fly ash and slag concrete is [23] Bakharev T, Sanjayan JG, Cheng YB. Effect of elevated temperature curing on
properties of alkali-activated slag concrete. Cem Concr Res
about two-thirds that typical of portland cement concrete with 1999;29(10):1619–25.
excellent repeatability. [24] Palomo A, Grutzeck MW, Blanco MT. Alkali-activated fly ashes: A cement for
 Young’s modulus of elasticity varies linearly with compressive the future. Cem Concr Res 1999;29(8):1323–9.
[25] Gebregziabiher BS, Thomas RJ, Peethamparan S. Very early-age reaction
strength for alkali-activated fly ash concrete, while the modulus kinetics and microstructural development in alkali-activated slag. Cem Concr
remains relatively constant for activated GGBFS concrete over Compos 2015;55(1):91–102.
56 R.J. Thomas, S. Peethamparan / Construction and Building Materials 93 (2015) 49–56

[26] Byfors K, Klingstedt G, Lehtonen V, Pyy H, Romben L. Durability of concrete [42] Carrasquilio RL, Nilson AH. Properties of high strength concrete subject to
made with alkali-activated slag. ACI Spec Publ 1989;114:1429–66. short-term loads. ACI J Proc 1981;78(3):179–86.
[27] Ravikumar D, Neithalath N. Electrically induced chloride ion transport in alkali [43] Lee NK, Lee HK. Setting and mechanical properties of alkali-activated fly ash/
activated slag concretes and the influence of microstructure. Cem Concr Res slag concrete manufactured at room temperature. Constr Build Mater
2013;47:31–42. 2013;47:1201–9.
[28] Bakharev T, Sanjayan JG, Cheng YB. Resistance of alkali-activated slag concrete [44] Sofi M, Van Deventer JSJ, Mendis PA, Lukey GC. Engineering properties of
to acid attack. Cem Concr Res 2003;33(10):1607–11. inorganic polymer concretes (IPCS). Cem Concr Res 2007;37(2):251–7.
[29] Fernández-Jiménez AM, Garcia-Lodeiro I, Palomo A. Durability of alkali- [45] Pauw A. Static modulus of elasticity of concrete as affected by density. ACI J
activated fly ash cementitious materials. J Mater Sci 2007;42(9):3055–65. Proc 1960;57(12):679–88.
[30] Ariffin MAM, Bhutta MAR, Hussin MW, Mohd Tahir M, Aziah N. Sulfuric acid [46] Hsu TC, Slate FO, Sturman GM, Winter G. Microcracking of plain concrete and
resistance of blended ash geopolymer concrete. Constr Build Mater the shape of the stress-strain curve. ACI J Proc 1963;60(2):209–24.
2013;43:80–6. [47] Popovics S. A numerical approach to the complete stress-strain curve of
[31] Donatello S, Palomo A, Fernández-Jiménez AM. Durability of very high volume concrete. Cem Concr Res 1973;3(5):583–99.
fly ash cement pastes and mortars in aggressive solutions. Cem Concr Compos [48] Desayi P, Krishnan S. Equation for the stress-strain curve of concrete. ACI J Proc
2013;38:12–20. 1964;61(3):345–50.
[32] Gifford PM, Gillott JE. Alkali-silica reaction (ASR) and alkali-carbonate reaction [49] Wang PT, Shah SP, Naaman AE. Stress-strain curves of normal and lightweight
(ACR) in activated blast furnace slag cement (ABFSC) concrete. Cem Concr Res concrete in compression. ACI J Proc 1978;75(11):603–11.
1996;26(1):21–6. [50] Carreira DJ, Chu KH. Stress-strain relationship for plain concrete in
[33] Fernández-Jiménez AM, Puertas F. The alkali–silica reaction in alkali-activated compression. ACI J Proc 1985;82(6):797–804.
granulated slag mortars with reactive aggregate. Cem Concr Res [51] Collins F, Sanjayan JG. Cracking tendency of alkali-activated slag concrete
2002;32(7):1019–24. subjected to restrained shrinkage. Cem Concr Res 2000;30(5):791–8.
[34] García-Lodeiro I, Palomo A, Fernández-Jiménez AM. Alkali–aggregate reaction [52] Duran Atisß C, Bilim C, Çelik Ö, Karahan O. Influence of activator on the strength
in activated fly ash systems. Cem Concr Res 2007;37(2):175–83. and drying shrinkage of alkali-activated slag mortar. Constr Build Mat
[35] Akazawa T. Tension test method for concrete. RILEM Bull 1953;16. 2009;23(1):548–55.
[36] N.J. Carino, Lew HS. Re-examination of the relation between splitting tensile [53] Duxson P, Provis JL, Lukey GC, Mallicoat SW, Kriven WM, Van Deventer JSJ.
and compressive strength of normal weight concrete. ACI J Proc Understanding the relationship between geopolymer composition,
1953;79(3):214–8. microstructure and mechanical properties. Colloids Surf A 2005;269(1):47–58.
[37] Raphael JM. Tensile strength of concrete. ACI J Proc 1984;81(2):158–65. [54] Thomas RJ, Howe A, Peethamparan S. Alkali-activated cement free concrete:
[38] Ahmad SH, Shah SP. Structural properties of high-strength concrete and its development of practical mixtures for construction. In: Proceedings of the
implications for precast concrete. PCI J 1985;30(6):92–119. 93rd annual meeting of the transportation research board. TRB; 2014.
[39] Oluokun FA, Burdette EG, Deatherage JH. Splitting tensile strength and [55] Rocco C, Guinea GV, Planas J, Elices M. Size effect and boundary conditions in
compressive strength relationships at early ages. ACI Mater J the brazilian test: experimental verification. Mater Struct 1999;32(3):210–7.
1991;88(2):115–21. [56] Mehta PK, Monteiro PJM. Concrete: microstructure, properties, and materials,
[40] Oluokun F. Prediction of concrete tensile strength from its compressive 3. New York: McGraw-Hill; 2006.
strength: an evaluation of existing relations for normal weight concrete. ACI [57] Collins F, Sanjayan JG. Microcracking and strength development of alkali
Mater J 1991;88(3):302–9. activated slag concrete. Cem Concr Compos 2001;23(4):345–52.
[41] A.C.I. Committee 318. Building code requirements for structural concrete (ACI [58] Hansen TC. Influence of aggregate and voids on modulus of elasticity of
318–08) and commentary (ACI 318R–08). American Concrete Institute; 2005. concrete, cement mortar, and cement paste. ACI J Proc 1965;62(2):193–216.

You might also like