Effect of Red Mud Content on Strength and Efflorescence in Pavement using Alkali-Activated Slag Cement
International Journal of Concrete Structures and Materials
Effect of Red Mud Content on Strength and Efflorescence in Pavement using Alkali-Activated Slag Cement
Efflorescence which severely occurs in alkali-activated slag cement can cause reduction of strength and durability due to calcium leaching. In the work, efflorescence characteristics in pavement containing red mud which can be affected by strong alkaline were investigated through various tests such as compressive strength, porosity, absorption, efflorescence area, alkali leaching content, and properties of the efflorescence compound. The compressive strength of pavement was evaluated to be higher over 15.0 MPa in all cases regardless of replacement ratio of red-mud and binder type, which can provide a reasonable strength for walking and bike lanes. The pavement with red mud was applicable to parking lots only when the replacement ratio of red-mud is within 10%. The efflorescence area increased with a higher replacement ratio of red mud and its propagation appeared though the efflorescence was removed through evaporation of moisture. However, the area of efflorescence gradually decreased with the repetition of the test.
pavement; alkali-activated slag cement; red mud; efflorescence; compaction
With increasing significance of human-friendly
environment and sustainability, eco-friendly pavement materials
have been used for bike trails, local roads, and parking lots,
which can give emotional stability. Natural loess and soils
are conventionally used for soil-pavement as a
binder-cement or a stabilizing agent
(Adebisi et al. 2013; Dipti et al.
2015; Kang 2016)
. However, the soil texture may cause
imperfect integration due to cement hydration, which
negatively affects both durability and aesthetic condition,
producing surface cracks and dust scattering, propagates to
economic losses. As an alternative to the expensive soil
pavement, various studies have been performed on the red
mud which is produced as a byproduct in the alumina
production process from the bauxite ore. Fe2O3 accounts for
about 22% in the chemical formation of red mud, playing an
important role in forming the reddish-yellow color that can
give the feeling of natural loess. For the reason, red mud is
reported to be capable of replacing conventional soil
pavement with sufficient quality
(Liu et al. 2011; David et al.
2017; Kang 2016)
. Another effort for utilizing red mud as an
activator has been made for alkali slag red mud cement in
the construction industry, which is so called clinker-free
(Gonga and Yang 2000; Pan et al. 2002, 2003)
the other hand, several studies have been carried out for the
use of a powdery industrial byproduct as a replacement for
expensive liquid activator
Efflorescence has been considered be relatively
insignificant since it does not cause a rapid structural problem in
normal Reinforced Concrete (RC) structure or member,
however, alkali-activated cement is very vulnerable to
efflorescence so that extreme damages from efflorescence
has been reported
(Kresse 1987; Bensted 2000; Zhang et al.
. In particular, alkali-activated binders that contain
sodium-based activators have been reported to be more
vulnerable to efflorescence. This is due to a increase in
Na2O/Al2O3 ratio. When Na2O exists in an unreacted state
increases, it can accelerate relatively easy movement of
sodium ions in the alumino-silicate structure which is a
product of the alkali-activated binder. Consequently, the
alkali-activated binder is known to be more vulnerable to
efflorescence in comparison to Ordinary Portland Cement
(Dow and Glasser 2003; Kresse 1987; Kani et al.
In the work, alkali-activated slag red-mud pavement is
developed for feasible application to construction industry
using red mud-based binder. The efflorescence
characteristics which can be caused by strong alkaline in red-mud are
investigated considering evaluation of pavement
compaction, compressive strength, porosity, absorption,
efflorescence area, alkali leaching content, and properties of the
efflorescence compound. In addition, the results are
compared with those from OPC-based pavement.
2. Materials and Experimental Program
2.1.1 Binder Properties
The physical and chemical properties of OPC and AAS
(Alkali-Activated Slag Cement) are shown in Table 1.
AAS based on Ground Granulated Blast Furnace Slag
(GGBFS) is clinker free since it is made from industrial
byproducts, including desulfurized gypsum and
high-calcium fly-ash. The chemical formation of the AAS cement is
SiO2 21.2%, Al2O3 8.8%, CaO 46.6%, and SO3 11.5%.
Compared with OPC, it has less CaO, and more SO3 and
Al2O3. In addition, it has a specific surface area of
4058 cm2/g with 2.83 g/cm3 of density. It implies that it is
lighter and finer than OPC. Figure 1 shows the results of
XRD analysis of hydration reaction of pastes by ages. As an
alkaline activator for GGBFS, only industrial byproducts
such as desulfurized gypsum and high-calcium fly ash are
used, so the hydration products consist of ettringite,
portlandite, and unreacted gypsum
(Moon et al. 2009)
2.1.2 Red Mud Properties
Approximately two tons of red mud are produced in
sludge with 40.0–60.0% moisture content for one ton of
Al2O3 production through Bayer process. An annual
production amount in Korea is approximately thirty tons. The
sludge with 40.0–60.0% moisture content is dried to about
10.0%, then ground into granulated type. The physical and
chemical properties of the dried red mud are listed in
Table 2. In the chemical composition, the summation of
SiO2, Al2O3, and Fe2O3 reaches about 80% of the total
weight. Fe2O3 accounting for red mud color (red brown) is
evaluated to be 22.8% of the weight. The weight ratio of
Na2O which produces strong alkali and mainly causes
efflorescence is observed to be 10.0%. The density and
moisture content show 3.5 g/cm3 and 10.2%, respectively.
Since sodium hydroxide is used in the extracting process
of aluminum from Bauxite ores, sodium hydroxide solution
still remains, which causes strong alkali over 11.0 of pH.
The strong alkaline property in red-mud means that it can
utilized as an activator for GGBFS. The specific surface area
of the dried red mud is 23.53 m2/g and the average grain
diameter is 2.75 lm. The average grain diameter of red mud
is even smaller than that of micro-cement (4–6 lm). The
small particles can provide a packing effect on binder matrix
and strength development so that strength reduction due to
efflorescence can be compensated
(Al-Akhras and Smadi
. Figure 2 shows the particle size distribution of the red
mud, indicating that the particle size is only 10.0% level
compared with the average grain diameter for OPC and
2.1.3 Aggregate Properties
The physical properties of the crushed fine aggregate are
shown in Table 3 and Fig. 3 shows the comparison of the
size distribution with the standard size distribution specified
in Unified Standard Specifications for Concrete
Construction. The size distribution of fine aggregate used in this study
satisfies the standard.
2.2 Experimental Program
2.2.1 Mix Design
The mix proportions for red mud pavement are listed in
Table 4. In order to evaluate the efflorescence in AAS
redmud pavement with replacement ratio of red-mud, red-mud
was replaced with total mass of cement by 0, 10, 20, and
30%. Dry pavement was manufactured by mixing 10% of
binder and 90% of crushed sand of the total mass of
pavement except for mass of water. The mass of water was
determined by using the optimum water content.
2.2.2 OMC Evalaution Test
The pavement properties differ with the effect of
compaction, dry density, and optimum water content. With
higher water content in the binder, the moisture in the
particles serves as a lubricant, which increases compaction
effect and dry density. In the test, the moisture content at the
highest dry density is set as the optimum water content. For
calculating Optimum Moisture Content (OMC), compaction
was repeated referred to KS F 2312 and the period to
Chemical composition (%)
hardening after the mixing was measured. The draft angle,
height, and volume of the mold are 10.0 cm, 12.7 cm,
1000 cm3, respectively. The diameter, drop height and
weight of the rammer are 5 cm, 30 cm, and 2.5 kgf,
respectively. Compaction was conducted 25 times per each
layer for three layers. OMC (x001) is evaluated referred to
KS F 2312 A taking into account the period from mixing to
hardening of the pavement. The water content of the
pavement in compaction was calculated following which the
OPC and the highest dry density were determined in
accordance with the compaction curve after wet and dry
density were calculated using Eq. (1)
where Yt is wet density (g/cm3), M2 is the total mass (g), M1
is the mass of the plate, and V is the volume of the mold
(cm3) Dry density is obtained from Eq. (2).
Yd ¼ 1 þ 100
where Yd is dry density (g/cm3), Yt is wet density (g/cm3),
and x is water content.
2.2.3 Compressive Strength Test
For compressive strength evaluation, the dry soil
pavement was mixed with OMC based on the Standards in Korea
(SPSKCICO-001:2003) using a pan-type mixer for three
minutes and compacted in U10 9 20 cm molds. The
rammer whose diameter, falling height and weight are 5 cm,
30 cm, and 2.5 kgf was used, respectively. Compaction was
performed 25 times per layer for three layers. Dry soil
pavement samples were prepared and cured for 28 days in
the room condition with 25 C and 60% of R.H. After curing
for 14 days and 28 days, compressive strength was
evaluated referred to KS F 2405. Figure 4 shows the compaction
of the pavement for compressive strength test and the
2.2.4 Efflorescence Acceleration Test
The specimens for efflorescence test were produced with a
size of U100 9 50 mm which is identical to the one used for
compressive strength test. After 28 days of curing, the side
surface of the sample is sealed with epoxy and about 6.0 mm
of the bottom is immersed in distilled water for 14 days in
the condition of 7 C of temperature and 50% of R.H. The
photos for efflorescence area are taken after 14 days of
accelerating period and the area is quantitatively evaluated
through Paint.NET software. The efflorescence grading is
evaluated based on the recommendation of As/NZS 6656.6
2.2.5 Alkali Leaching Test
The pavement samples are crushed after the accelerated
efflorescence test and the powder is mixed with distilled
water in a weight ratio of 1:50. They are exposed to the same
conditions of the accelerated efflorescence test for 48 h.
After 48 h, 20 mL of the mixed water is obtained through
filtering. Since soluble Na? and Ca2? are the major ions for
Type of binder
Unit weight (kg/ton)
Ordinary Portland Cement (OPC)
Alkali-activated slag cement (AAS)
Water content (%)
Optimum water content Changed by optimum
efflorescence, the concentrations of them are measured
through ICP (Inductively Coupled Plasma) analysis. The
results are compared with the results before accelerated
3. Result and Discussions
3.1 Compactability Evaluation
Figure 5 shows the compaction curve of the pavement in
terms of replacement ratio of red mud by binder type. In the
compaction curve, the dry density changes gradually
depending on the water content, which implies that there is
no distinct difference in compaction according to the
replacement ratio of red mud. In all binders, OMC increases
and dry density decreases in the compaction curve. The
OMC is calculated and plotted in Fig. 6 considering the
replacement ratio of red mud and binder type.
The OMC appears to increase as the replacement ratio of
red mud is getting higher. The OMC is relatively higher in
ASS pavement compared with that in OPC. The results with
binder type are shown to be 9.1–10.2% in AAS and
Fig. 5 Compaction characteristics with red mud replacement
8.4–9.9% in OPC, respectively. In other words, the OMC in
ASS is 0.3–0.8% higher than that in OPC. As shown in
Fig. 2, the size distribution of AAS is relatively gradual,
which implies that there are more microfines whose surfaces
are wide enough to hold a large quantity of hygroscopic
water. In addition, the OMC by mix proportion of red mud is
observed to be 12% higher in AAS 30 than in AAS 0, while
it is 17% higher in OPC 30 than in OPC 0. Similarly, the
reason for increasing OMC with higher red mud replacement
ratio is that the red-mud is silt and clay with an average
particle size of 2.75 lm with high micro-fine content. Red
mud has relatively higher density and a larger specific
surface area compared with the binders such as OPC and ASS,
so that its physical properties are similar to those of silt and
(Bahmani et al. 2014)
. Therefore with increasing
replacement ratio of red mud, it moves to the right in the
compaction curve, from which it can be interpreted that the
OMC increases while the highest dry density decreases.
3.2 Compressive Strength Evaluation
Figure 7 shows the results of compressive strength with
varying red mud replacement ratios. The compressive
strength at 28 days is evaluated to be 18.9–27.0 MPa in
AAS, and 18.4–28.8 MPa in OPC, showing a similar level
between the cases, which shows a slight different trend from
the previous results
(Kang and Kwon 2017)
compressive strength of ASS mortar was observed to be
lower than that of OPC mortar. It is inferred that industrial
byproducts like high-calcium fly ash and desulfurized
gypsum are easily expandable after compacting by the hydration
reaction, which can induce swelling pressure to soil
Regardless of binder type, the higher replacement ratio of
red mud is in OPC or in AAS, the lower the compressive
strength is measured. The degradation of the compressive
strength caused by the mix proportion of red-mud appeared
to be comparatively less in ASS than in OPC. In the related
Standards in Korea (SPSKCICO-001:2003), the
performance of compressive strength is classified into three
stages. It is specified that a strength over 18 MPa for parking
lots, a strength higher over 15 MPa for bike lanes, and a
strength over 12 MPa for walking lanes are required
considering user’s needs. In the results, the compressive strength
was evaluated over 15 MPa in all specimens regardless of
the replacement ratios of red-mud and binder types.
However, it is applicable to parking lots only when the
replacement ratio of red mud is within the range of 10%.
3.3 Porosity Evaluation
Figure 8 shows the cumulative porosity with replacement
ratio of red mud by binder type. Total porosity appears to be
higher in ASS than in OPC. The total porosity volume
appeares to be great in OPC with replacement ratio of
redmud, while there was no significant effect of red mud
replacement on porosity in AAS. In the case if AAS 0, the
cumulative porosity appears to be 0.94 ml/g, which is 34.2%
higher compared to 0.70 ml/g of control case (OPC 0).
However, as the replacement ratio of red mud becomes
higher, there is no big change in the result in AAS, from 0.94
to 0.93 ml/g, while it is increasing in OPC, from 0.70 to
0.93 ml/g. This, as shown in Fig. 9, seems to be due to the
fact that AAS has more 10–1000 nm capillary pores
compared with OPC. For this reason, even though red-mud is
more mixed with AAS, the volume of capillary pores is
almost unaffected in AAS, while the volume of capillary
pores increases in OPC. The volume along
1000–100,000 nm pores decreases in AAS, but there is no
significant difference found in OPC. Therefore, if red mud is
mixed instead of a binder, the amount of cement needed for
hydration reaction decreases in OPC, which leads to a
reduction in porosity. Meanwhile, AAS is affected less by
red mud content with almost constant porosity due to the
filling effect of red-mud. The results of pore size distribution
are shown in Fig. 9.
3.4 Efflorescence Characterization
As shown in Table 5, the degree of efflorescence is
quantitatively categorized into 5 levels considering the
efflorescence area. XRD analyses are performed in order to
characterize the efflorescence compounds. The efflorescence
Fig. 9 Pore size distribution with red mud replacement ratio.
area and class are listed in Table 6 from the test. The
efflorescence area increases with a higher replacement ratio
of red mud. The efflorescence area with repeating tests after
removing the efflorescence is shown in Fig. 10.
As shown in in Fig. 10, the efflorescence continues to
appear even after removal of efflorescence, however the area
of efflorescence tends to gradually decrease with the
repetition of the test. In the first test, the efflorescence area varies
from 15 to 75% in the specimens with a replacement ratio of
20% of red mud. However, after the fourth test, the area
appeared to be less than 10% overall. In the first test for 30%
replacement ratio, the area is 75% but after the fourth test, it
decreases to around 10–20%. Through the results, it is found
that the efflorescence cannot be completely removed, but the
area can be dramatically reduced with few times of
efflorescence removal. In order to examine the main mechanism
of efflorescence, the concentrations of Na? and Ca2?
leached are measured and the results are shown in Fig. 11
where the concentrations of soluble Na? and Ca2? leached
after the fifth efflorescence test and the amount of Na? and
Ca2? before the first test are compared.
The concentrations of Na? leached from OPC 0 is
0.46 mg/l, and from AAS 0 is 0.81 mg/l. However, as more
red-mud is replaced, the Na? content increases both in OPC
30 (to 24.26 mg/l) and in AAS 30 to (22.04 mg/l). It was
52.7 times higher in OPC 30 compared with OPC 0 and 27.2
times higher in AAS 30 compared with AAS 0. This is
because OPC and AAS had a small quantity of Na2O, 0.4
and 0.2%, respectively but red mud contains about 10.0%
Na2O. When red mud with sufficient Na2O is replaced with
OPC, the supply of Na2O increases. Additionally in all
specimens, the content of Na? leached decreases after the
fifth efflorescence test compared with initial condition. In
particular, the decrease appears to be greater with more red
mud content. The leached Na? becomes the main ingredient
of the efflorescence. As more efflorescence appears,
considerable eluted Na? is also removed during the removal
The efflorescence in OPC-based concrete is generated by
free CaO or Ca(OH)2, a product of hydration reaction
(Kresse 1987; Bensted 2000)
, so that more efflorescence is
generated when sufficient amount of Ca(OH)2 is produced
through the hydration reaction. Therefore, it is expected that
more Ca2? ions are leached out from the pavement.
Regarding the leached amount of Ca2?, OPC 0 is 1.9 times
higher before the first efflorescence test while it is 1.7 times
higher after the fifth test. In particular, the higher mix
proportion of red-mud is, the less Ca2? is eluted. This is because
CaO content in OPC and AAS is 63.1 and 54.9%,
respectively, while CaO content in red mud is only 3.4%. If red
mud is replaced with OPC, the source of CaO decreases. In
addition, after the fifth efflorescence test, it is found that the
amount of Ca2? leached is slightly decreases compared with
that before the first test. But the Ca2? appeares to decrease
less compared with Na?. It imples that the content of Ca2?
removed along with the efflorescence compound is lower.
Compared with the Na?, it has a relatively small effect on
To identify how the leaching content of Na? and Ca2?
affected efflorescence, XRD analysis of efflorescent
compound on the pavement is performed. The results of XRD
analysis are shown in Fig. 12.
Red mud content (%)
Red mud content (%)
11.043.5a0-I efflorescence area—efflorescence degree.
In the efflorescence from OPC binder with red mud (30%
of replacement ratio), Na2CO3H2O is mainly observed while
Na2SO4 is dominant in AAS binder (30% of replacement
ratio). This reveals that the efflorescence is mainly
comprised of sodium compounds regardless of the binder type.
The efflorescence from binder with red mud shows that
CaCO3 is partially observable
(Perez-Villarejo et al. 2012)
but sodium compounds are dominantly observable. The
efflorescence from binder OPC with red mud is generic
alkaline carbonate efflorescence (Na2CO3H2O) produced
from the reaction between Na? from the red mud and CO2
gas in the atmosphere. This is similar to the mechanism for
geo-polymer efflorescence generation, which is caused by
abundant Na? from a sodium activator such as NaOH and
water–glass. The formulations of the efflorescence
generation in geo-polymer are expressed as Eqs. (3) and (4),
CO2ðgÞ þ 2OH ðaqÞ !
CO23 ðaqÞ þ H2O
The efflorescence from binder without OPC is mainly
made up with Na2SO4 from the reaction between Na?
supplied from red mud and SO3 from desulfurization gypsum
used as the activator for GGBFS.
Fig. 11 Results of alkali leaching test with red mud
Fig. 12 XRD analysis results of the efflorescence compound
The conclusions on effects of red mud and alkali-activated
slag cement on efflorescence in pavement are as follows.
(1) The OMP (Optimum Moisture Content) increases with
higher replacement ratio of red-mud. The OMC is
relatively high in AAS pavement compared with that in
OPC. The OMC with binder type is shown to be
9.1–10.2% in AAS and 8.4–9.9% in OPC, respectively.
(2) The compressive strength at 28 days is evaluated to be
18.9–27.0 MPa in AAS, and 18.4–28.8 MPa in OPC,
showing a similar level between the cases. The
compressive strength is evaluated to be higher than
15.0 MPa in all specimens regardless of the
replacement ratio of red-mud and binder type, which enabled
application of red mud for making it appropriate for
walking and bike lanes. The optimum replacement
ratio for parking lots is evaluated to be within the range
(3) Total porosity in ASS is measured to higher than that in
OPC. In addition, regarding in terms of the replacement
ratio of red-mud, the total porosity volume is evaluated
to be higher in OPC but significant difference is not
found in AAS. If red-mud is replaced with binder, the
cement amount for hydration reaction decreases in
OPC, which leads to a reduction in porosity and coarse
structure. Meanwhile, AAS is affected less by red mud
content, and there is no significance difference in
porosity because of a filling effect of red-mud on
(4) The efflorescence area increases with increasing
replacement ratio of red mud and still appeared after
1st cleaning of surface, however it gradually decreases
with the repetition of the test.
This research was supported by a Grant
(16CTAP-C11520601#) from Infrastructure and transportation technology
promotion research Program funded by Ministry of Land,
Infrastructure and Transport of Korean Government. This
work also was supported by the National Research
Foundation of Korea (NRF) grant funded by the Korea
This article is distributed under the terms of the Creative
Commons Attribution 4.0 International License (http://
creativecommons.org/licenses/by/4.0/), which permits unre
stricted use, distribution, and reproduction in any medium,
provided you give appropriate credit to the original
author(s) and the source, provide a link to the Creative
Commons license, and indicate if changes were made.
Adebisi , N. O. , Adeyemi , G. O. , Oluwafemi , O. S. , & Songeca , S. P. ( 2013 ). Important properties of clay content of lateritic soils for engineering project . Journal of Geography and Geology , 5 ( 2 ), 99 - 115 .
Al-Akhras , N. M. , & Smadi , M. N. ( 2004 ). Properties of tire rubber ash mortar . Cement & Concrete Composites , 26 , 821 - 826 .
Bahmani , S. H. , Huat , B. K. , Asadi , A. , & Farzadnia , N. ( 2014 ). Stabilization of residual soil using SiO2 nanoparticles and cement . Construction and Building Materials , 64 , 350 - 359 .
Bensted , J. ( 2000 ). Efflorescence-prevention is better than cure . Concrete , 34 , 40 - 41 .
David , D. A. , Rania , A. N. , Benjamin , A. T. , David , O. O. , & Abu , Y. A. ( 2017 ). Bauxite red mud-cement based composites: Characterisation for pavement applications . Case Studies in Construction Materials , 7 , 45 - 55 .
Dipti , R. B. , Umesh , C. S. , & Suresh , R. D. ( 2015 ). Characterization of granular lateritic soils as pavement material . Transportation Geotechnics , 6 , 108 - 122 .
Dow , C. , & Glasser , F. P. ( 2003 ). Calcium carbonate efflorescence on Portland cement and building materials . Cement and Concrete Research , 33 ( 1 ), 147 - 154 .
Gonga , C. , & Yang , N. ( 2000 ). Effect of phosphate on the hydration of alkali-activated red mud slag cementitious material . Cement and Concrete Research , 30 , 1013 - 1016 .
Kang , S. P. ( 2012 ). A study on the usability of red mud as activator of alkali-activated cementless binder . Journal of the Architectural Institute of Korea Structure & Construction , 28 , 133 - 140 .
Kang , S. P. ( 2016 ). Properties of alkali-activated slag-red mud soil pavement using recycled aggregate . Journal of the Korean Recycled Construction Resources Institute , 4 ( 3 ), 276 - 283 .
Kang , S. P. , & Kwon , S. J. ( 2017 ). Effects of red mud and alkali-activated slag cement on efflorescence in cement mortar . Construction and Building Materials , 133 , 459 - 467 .
Kani , E. N. , Allahverdi , A. , & Provis , J. L. ( 2012 ). Efflorescence control in geopolymer binders based on natural pozzolan . Cement & Concrete Composites , 34 ( 1 ), 25 - 33 .
Kresse , P. ( 1987 ). Efflorescence-mechanism of occurrence and possibilities of prevention . Betonw Fert Tl .-Tech., 53 , 160 - 168 .
Liu , X. , Zhang , N., Sun , H. , Zhang, J. , & Li , L. ( 2011 ). Structural investigation relating to the cementitious activity of bauxite residue-Red mud . Cement and Concrete Research , 41 , 847 - 853 .
Moon , K. J. , Hyoung , W. K. , Ju , H. S. , Park , W. C. , So , S. Y. , & Soh , Y. S. ( 2009 ). Pore structure of non-sintering cement using inorganic industrial waste . Journal of Korea Society Waste Management , 26 , 128 - 134 .
Pan , Z. , Cheng, L., Lu , Y. , & Yang , N. ( 2002 ). Hydration products of alkali-activated slag-red mud cementitious material . Cement and Concrete Research , 32 , 357 - 362 .
Pan , Z. , Li , D. , Yu , J. , & Yang , N. ( 2003 ). Properties and microstructure of the hardened alkali-activated red mudslag cementitious material . Cement and Concrete Research , 33 , 1437 - 1441 .
Perez-Villarejo , L. , Corpas-Iglesias , F. A. , Martinez-Martinez , S. , Artiaga , R. , & Pascual-Cosp , J. ( 2012 ). Manufacturing new ceramic materials from clay and red mud derived from the aluminium industry . Construction and Building Materials , 35 , 656 - 665 .
Zhang , Z. , Provis , J. L. , Reid , A. , & Wang , H. ( 2014 ). Fly ashbased geopolymers: the relationship between composition, pore structure and efflorescence . Cement and Concrete Research , 64 , 30 - 41 .