Construction and Building Materials 58 (2014) 193205

Contents lists available at ScienceDirect

Construction and Building Materials

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

Potential use of binary and composite limestone cements in concrete

production
Mohammed Seddik Meddah a,, Mukesh C. Lmbachiya b, Ravindra K. Dhir c
a

Department of Civil and Architectural Engineering, College of Engineering, Sultan Qaboos University, 123 Al-khod, Muscat, Oman
School of Civil Engineering & Construction, Kingston University, London KT1 2EE, UK
c
School of Engineering, Physics and Mathematics, University of Dundee, Nethergate DD1 4HN, UK
b

h i g h l i g h t s
 The use of LS in binary, ternary and quaternary cementitious systems was investigated.
 Using LS permits the design of concrete with low environmental impact.
 Composite cement with LS seems to perform better than binary LS-cement.
 An optimal replacement level of 15% LS is recommended.
 LS composite cements provide a durability improvement compared to binary LS-cement.

a r t i c l e

i n f o

Article history:
Received 28 September 2012
Received in revised form 12 November 2013
Accepted 1 December 2013
Available online 7 March 2014
Keywords:
Binary and composite cements
Carbonation
Chloride diffusion
Compressive strength
Drying shrinkage
Embodied CO2
Freezethaw
Limestone
Portland cement

a b s t r a c t
Over the last decades, the use of various by-products and pozzolanic materials in concrete production has
become a common practice, not only to reduce the environmental impact of Portland cement (PC) manufacturing and to save natural resources but also to enhance the mechanical and durability performance
of concrete.
The present study highlights the main performance properties of 50 concrete mixes designed with binary, ternary and quaternary cementitious systems, including the use of various proportions of slag (S), y
ash (FA), limestone (LS), silica fume (SF) and metakaolin (MK) as a partial replacement by weight of PC.
The binary cements were designed with various LS proportions ranging from 10% to 45%, while the ternary system consisted of 29% slag and 21% FA as a partial substitute of PC. The three quaternary systems
were designed with 25% FA and slag (50.1% or 47.5%) combined with either 4.9% SF, 4.9% MK or 2.5% LS.
The concrete mixes were designed with a wide range of water-to-cementitious ratios (w/c) ranging from
0.45 to 0.79.
The main objective of this paper is to design a concrete with low environmental impact using various
types and proportions of cementitious materials.
It has been observed that the use of composite cements improves concrete workability and reduces the
amount of superplasticizer required to reach the same slump value compared with LS and PC cements,
while the setting time is decreased for both LS-cement and composite cements. The strength results indicate that LS could lead to signicant strength loss compared with PC and composite cement concretes. In
addition, the quaternary PCSFALS mix appears to perform better than the binary LS-cement in terms of
strength development and durability performance.
The results indicate that PCLS15 is freezethaw durable (durability factor over 80%); however, with
replacement levels higher than 15%, the durability factor decreased. However, the composite cements
generally exhibited a satisfactory durability factor of approximately 80% or a slightly lower DF. Moreover,
the composite cements exhibited improved resistance to chloride ingress, while a negative effect on carbonation depth was observed.
Overall, the results indicate that the mechanical and durability performance of both binary and composite cement concretes are strongly linked to the chemical composition, neness, particle size distribution and potential reactivity of the cementing materials used.
2014 Elsevier Ltd. All rights reserved.

Corresponding author. Tel.: +968 2414 2672.

E-mail address: seddikm@squ.edu.om (M.S. Meddah).
http://dx.doi.org/10.1016/j.conbuildmat.2013.12.012
0950-0618/ 2014 Elsevier Ltd. All rights reserved.

194

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

1. Introduction
Since new environmental regulations have been implemented,
the pressure to reduce CO2 emissions generated by the concrete
industry has increased. Over the last decades, extensive research
has been undertaken to minimize the use of Portland cement by
increasing the amount of various supplementary cementing materials (SCM) or llers embedded in concrete as a partial replacement
of PC. Indeed, it is well recognized that Portland cement manufacturing, with an annual production of 3.7 billion tons in 2012, accounts for approximately 7% of the global CO2 emissions [13].
Various supplementary cementing materials, whether natural
pozzolans or those derived from industrial by-product waste materials such as silica fume, y ash and ground granulated blast furnace slag, have now been used for many years to develop
composite cements not only to reduce the environmental load
but also to improve concrete durability.
The use of limestone as a construction material dates back to
ancient times when calcined limestone or gypsum was used to
make mortar [4]. Limestone consists mainly of calcium carbonate
(CaCO3), which reacts with the tricalcium aluminate (C3A) of
Portland cement to form calcium carboaluminate (CCA). The effects of LS, as well as other mineral admixtures such as slag,
FA, SF and MK used as partial replacements for Portland cement,
have been widely discussed and are now well established and
documented [514]. Natural pozzolana, y ash, slag, silica fume
and limestone are the main cementing materials that are permitted by the EN 197-1 [15]. According to BS EN 197-1:2000 [15],
type II cements (CEMII/A-LL 32,5/42,5) may contain various
materials as their main constituents in percentages ranging from
6% to 35%.
Although limestone has been widely used as a ller material, it
is also used in blended cement as a partial substitution for Portland
cement [15], with a recommended amount ranging between 6%
and 20% [16]. It has been reported that ne limestone powder
could promote the early age hydration of Portland cement
[17,18] and may reduce the total porosity and delay the initial
and nal setting time as well [9]. While the setting time might
be slightly shortened and the tendency to bleeding signicantly reduced, the fresh properties, such as plasticity and water retention,
in Portland-Limestone mortar and concrete are slightly improved
or similar to those of a bulk Portland CEM I concrete [19].
In addition to its ller effect, LS also has a chemical effect: the
calcium carbonate of the limestone powder can interact with the
aluminate hydrates formed by the hydration reactions of Portland
cement [18,20]. This interaction leads to the stabilization of the
ettringite and could increase the total volume of the hydration
products, decrease the porosity of the concrete and consequently
increase its strength. Limestone powder could also interact with
the AFm and AFt hydration phases, leading to the formation of carboaluminates at the expense of monosulfate, thereby stabilizing
the ettringite, as reported by De Weerdt et al. [21].
It has been reported that the addition of LS to Portland cement
increases the rate of hydration at early ages and consequently enhances the early strength; however, LS can reduce the 28-day and
long-term strength compared with concrete prepared with CEM I
due to the dilution effect [11,22]. Meanwhile, at the same concrete
strength, Portland-LS cement concrete exhibits similar performance
as CEM I concrete with respect to the carbonation rate, chloride ingress and resistance to freezing and thawing (both air-entrained
and non-air-entrained concrete). It has been observed that depending on the amount used, limestone in concrete increases chloride
ion diffusion [14] while signicantly reducing the peak rate of heat
evolution [16]. Moreover, Ghrici et al. [11] reported that a ternary
cementitious system containing 20% LS ller and 30% natural

pozzolans exhibited improved early and long-term compressive

and exural strengths and enhanced durability against sulfate, acid
and chloride ion ingress.
However, some studies have focused on the thaumasite
problem linked with the use of limestone-cement concrete and calcareous aggregates. The risk of thaumasite formation (CaSiO3CaCO3CaSO415H2O) is a serious problem associated with the use
of limestone in cement and concrete exposed for a few months to
sulfate solutions at low temperature (approximately 5 C) [19,23
25]. It is believed that this form of sulfate attack completely destroys
the binding ability of the cement by transforming the CSH gel into
a mush, weakening the CSH and leading to lower strength.
Thaumasite formation requires a source of calcium silicate, sulfate
and carbonate ions, excess humidity and preferably low temperature [24,26].
In this paper, the effect of various limestone contents (1545%)
in binary cement and various amounts of y ash, slag, silica fume,
metakaolin and limestone in composite cements (ternary and quaternary systems) on the resulting concrete performance is studied.
The results presented herein are part of an extensive research project aiming to develop concrete made with various Portland-composite cements to reduce the environmental load and design more
sustainable and durable concrete.

2. Experimental work
2.1. Materials
General use Portland cement CEM I 42.5 N meeting the requirements of EN 1971:2000 was used in all the mixes with and without SCMs. Five different SCMs,
namely, slag, y ash, silica fume, metakaolin and limestone, were used in various
proportions as substitutes for Portland cement to produce binary, ternary and quaternary binders. The chemical and mineralogical compositions and physical properties of Portland cement and the ve cementing admixtures (S, FA, SF, MK and LS)
used are listed in Table 1, and their particle size distribution is shown in Fig. 1.
Natural siliceous sand and crushed granite with maximum sizes of 5 mm and
20 mm, respectively, were used as ne and coarse aggregates. A superplasticizer
(SP) was used to obtain a nominal target slump value of 75 5 mm, while an airentraining agent was employed to investigate the freezethaw resistance of the
air-entrained concrete mixes.

2.2. Details of mixtures, concrete mixing and specimens

Both the Portland cement and blended cement concretes were designed with
a wide range of w/c ratios of 0.79, 0.65, 0.60, 0.52 and 0.45. Six types of mixes
designed with ve different SCMs were investigated in this study. A single content of 4.9% (by weight) of SF and MK was introduced in the quaternary systems,
while FA was used in proportions varying from 20% to 25% and slag was incorporated at two different proportions of 29% and 25%. In addition, LS was introduced at 2.5% in the quaternary system, while the LS-binary cements were
formulated by varying the replacement (by weight) of PC by LS from 15% to
45%. In all the mixes, the free water (185 kg/m3) and coarse aggregate
(1200 kg/m3) contents were kept constant, while the ne aggregate content
was adjusted depending on the type and content of SCM incorporated. Tables
2 and 3 provide the mix proportions of the PCLS concretes and the various composite cement concretes investigated, respectively.
All the concrete mixtures were produced using a horizontal forced-action pan
mixer with a 0.045 m3 capacity, and each mix was appropriately labeled. The concrete mixes were referred to as control Portland cement (PC), binary cement-limestone (PCLS), ternary cementslagy ash (PCSFA) and quaternary systems
cementslagy ashsilica fume (PCSFASF), cementslagy ashmetakaolin (PCSFAMK) and cementslagy ashlimestone (PCSFALS).
After mixing, a slump test was performed, and then, the concrete was cast into
steel molds (cubes, cylinders and prisms) in three layers and compacted using a
plate vibrator, as specied by BS 1881: Part 108: 1983. All the concrete specimens
were stored for the rst 24 h under a plastic sheet in a laboratory environment.
The specimens were then demolded and wet-cured at 20 2 C for the rst 28 days,
and afterwards, specic curing conditions were applied depending on the durability
test. For all the concrete types, three samples of Portland and blended cement concretes (binary, ternary and quaternary) were tested.

195

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

Table 1
Chemical composition and physical properties of the cements used.
Cementing materials

Physical properties

Chemical compositions

PC
LS
FA
GGBS
SF
MK

Density (kg/m ) Fineness (m /kg) Residue 45 lm (%)

SiO2

Al2O3

Fe2O3

CaO

MgO

SO3

TiO2

Alkali

LOI

3140
2700
2276
2900
2200
2590

21.4
1.1
51.1
34.2
95.3
55.1

4.7
0.06
24.9
13.9
0.65
40.4

2.7
0.09
9.0
0.6
0.28
0.64

65.2
54.7
1.5
41.6
0.27
0.03

1.0
0.3
1.4

0.41
0.36

2.9

0.7

0.25

0.01
1.0

0.01

0.55
0.053
3.97
0.463
0.767
1.392

0.9
43.5
5.7
0.9

1.2

381
638
450
509
15,750
3474

Bogue composition of PC C3S

C2S
C3A
C4AF

6.2
8.8
12.8
1.8
0.2
0.4

67.3
10.6
7.9
8.2

(Na2Oeq) = Na2O + 0.658 (K2O) = 0.55.

Table 2
Mixture composition for 1 m3 of the PCLS binary concretes.

100

Percentage passing (%)

90

Constituent materials, kg/m3

80
70

w/c
(MPa)

PC
Slag

60

PC
PC
content (%)

15
25
35
45

700

400

800

0.30

185

0.65(35) 285

100
85
75
65

15
25
35

730

400

800

0.24

185

0.60(40) 310

100
85
75
65
55
100
85
75
65
55

15
25
35
45

15
25
35
45

710

400

800

0.13

185

670

400

800

0.11

185

100
85
75
65
55

15
25
35
45

625

400

800

0.21

185

MK
SF

30
20
10
0
1

10

100

1000

Sieve size (m)

Fig. 1. Particle size distribution of Portland cement and the ve cementing
materials used.

0.52(50) 355

2.3. Test procedures

0.45(60) 410
2.3.1. Mechanical properties
Compression tests were performed on cubic specimens of 100 mm size at different ages of 1, 2, 7, 28, 60, 180 and 365 days in accordance with BS EN 12390-3. The
exural strength test was determined under three point bending loading on prismatic specimens with dimensions of 100  100  500 mm in accordance with EN
12390-5, while the modulus of elasticity was determined on 150  300 mm cylinder specimens at 28 days of age in accordance with BS EN 1352:1997. All the specimens were wet cured at 20 C from demolding at 24 h of age until testing. The
reported results are the averages obtained for three specimens.
2.3.2. Shrinkage
The drying shrinkage was measured on 75  75  300 mm prisms with stainless steel studs xed over a 200 mm gauge length at either end of the specimen.
The concrete test samples were stored in dry air conditions (20 2 C and 60 5%
RH), and measurements were taken for as long as 90 days. Length change readings
were periodically recorded for 90 days using a strain frame sensitive to 2 lm.
2.3.3. FreezingThawing (FT)
Prism specimens with dimensions of 75  75  300 mm were used to assess the
concrete resistance to freezingthawing according to the ASTM C666 test method
[27]. The specimens were cured in water at 20 2 C before their exposure to freezing and thawing cycles. The specimens were subjected to temperature cycling from
4 C to 18 C. The duration of the cycles was 25 h. The freezing portion of the cycle was accomplished by air cooling (similar to air conditioning), while the thawing
portion was performed by submersion in water.
2.3.4. Carbonation
Accelerated carbonation testing was performed on cubes of 100 mm size that
were wet cured during the rst 28 days. The specimens were then pre-conditioned
by drying in the laboratory for 14 days. Before curing in the carbonation chamber,
all the sides of the concrete specimens were sealed using a bituminous coating

5 mm 10 mm 20 mm

SP, % by weight of Water

cement

100
85
75
65
55

LS

40

Aggregates

0.79(25) 235

FA

50

LS
(%)

paint except the top side, which was exposed in a carbonation tank to a CO2 enriched atmosphere containing 4% CO2 at 20 2 C and 55 5% RH. The carbonation
depth in the tested concrete specimens was measured by applying a phenolphthalein color indicator spray on a freshly broken piece of the specimens. This spray
turns non-carbonated concrete pink, while carbonated concrete remains colorless.
The depth of the uncolored zone of the concrete (the carbonated layer) from the
edges of the broken piece was measured at 3 points, and the mean value was reported as the carbonation depth.

2.3.5. Chloride ion ingress

The chloride ion ingress was measured across a slice of the concrete specimens
measuring 100 mm and 25 mm thick that was cut from a 150  300 mm cylinder. The concrete cylinders were subjected to two curing systems, wet and dry
air curing, for the rst 28 days, and the results of both curing systems are presented.
The concrete slices were located in standard two compartment diffusion test cells.
The inner part of the cell was lled with calcium hydroxide solution, and the cell
was partially immersed in a chloride tank, thereby exposing the outer face of the
concrete to the chloride solution. The electrochemical chloride ingress test method
used was called potential difference, where a potential difference of 7.5 V was
passed across the specimen. The chloride transport rate was evaluated from frequent chloride analyses of the liquid in the two chambers. Ficks rst law was used
to calculate the chloride diffusion coefcient; it should be noted that the coefcients were derived from steady-state diffusion experiments:

196

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

Table 3
Mixture composition for 1 m3 of the composite cements concretes.
w/c

PC content

PC (%)

Slag (%)

Constituent materials, kg/m3

FA (%)

SF (%)

MK (%)

LS (%)

Aggregates
5 mm

10 mm

20 mm

SP,% by weight of cement

Water

0.79

235

100
50
50.1
50.1
47.5

29
25
25
25

21
20
20
25

4.9

4.9

2.5

700

400

800

0.30
0.23
0.30
0.26
0.24

185

0.65

285

100
50
50.1
50.1
47.5

29
25
25
25

21
20
20
25

4.9

4.9

2.5

730

400

800

0.30
0.17
0.23
0.18
0.16

185

0.60

310

100
50
50.1
50.1
47.5

29
25
25
25

21
20
20
25

4.9

4.9

2.5

710

400

800

0.13
0.10
0.13
0.11
0.08

185

0.52

355

100
50
50.1
50.1
47.5

29
25
25
25

21
20
20
25

4.9

4.9

2.5

670

400

800

0.11
0.07
0.10
0.08
0.06

185

0.45

410

100
50
50.1
50.1
47.5

29
25
25
25

21
20
20
25

4.9

4.9

2.5

625

400

800

0.21
0.15
0.17
0.16
0.14

185

Q Dc

@c
;
dx

where Q is the mass transport rate for one square-meter of concrete (mol/m2 s), Dc is
the diffusion coefcient (m2/s), c is the total chloride concentration in the pore water
(mol/m3), t is the time and x is the distance from the exposed surface (m).
2.3.6. Initial surface absorption (ISA)
The ISA test measures the rate at which water ows into the capillary pore network of concrete through a known surface area. The volume ow is estimated by
measuring the length of ow along a capillary of known dimension. The initial surface absorption of the various mixes was determined on 150 mm cubes pre-cured in
water for the rst 28 days as per BS 1881-208, Part 5. The concrete specimens were
then oven-dried at 105 C to constant weight before the test and left to cool to the
laboratory temperature (1720 C) in a desiccator jar for a period of 24 h. The contact area was dened by a plastic cell sealed onto the concrete surface and should
not be less than 5000 mm2. Water was introduced into the cell via a connecting
point and maintained at a head of 200 mm using a lter funnel. A second connection point to the cap led to a horizontal capillary tube. The connection to the reservoir was closed, and the absorption was measured by observing the movement of
the end of the water line in the capillary tube with an afxed scale after 10 min.
The ISA measurements were recorded 10 min after the initiation of the test. The
measurements are referred to as ISA-10 at 10 min. More details about the testing
and setup used can be found in an earlier publication by the authors [28].
Due to the considerable number of mixes investigated, the exural strength
testing and some of the durability testing were only conducted on selected mixes.

3. Test results and discussion

3.1. Fresh properties
For each concrete mix investigated, the targeted nominal slump
value of approximately 75 5 mm was reached by adjusting the
amount of superplasticizer used. The mix design of concretes
investigated presented in Tables 2 and 3 revealed that the presence
of different SCMs at various contents in the concretes was benecial for workability enhancement. Generally, all the blended mixes
required almost the same or even less amount of superplasticizer
to reach a slump value similar to that of the control mix. As observed in Table 2, regardless of the w/c of concrete, the inclusion

of various amounts of LS in the binary concrete (PCLS) designed

with different w/c did not require any increase of the SP content
to reach the targeted slump value compared with that of the Portland cement concrete. In addition, the presence of LS in the quaternary cement system (PCSFALS) signicantly reduced the SP
demand and required the lowest amount of SP compared with all
the other binary, ternary and quaternary systems investigated.
Overall, it appears that the presence of LS in composite cements
could provide better improvement of concrete workability than
when it is used in a binary system LS-cement. In fact, while no
change in the SP content was recorded for the binary system LS-cement to reach the targeted slump value, the reduction in the
amount of SP for the quaternary system was substantial and ranged between 20% and 46% depending on the w/c of the concrete
mix. In line with these results, several previous investigators have
concluded that LS-cement requires less water/SP demand than
Portland cement and could improve concrete workability [2932].
In addition, the ternary (PCSFA) and quaternary (PCSFASF and
PCSFAMK) cementitious systems demonstrated a signicant reduction in the amount of SP required to reach the targeted slump value
(75 5 mm). This reduction in the required amount of SP generally
ranged from 23% to 43% for the ternary systems and from 9% to 40%
for the quaternary systems. Moreover, the quaternary systems designed with MK exhibited better workability enhancement (reduction of SP content) compared with their corresponding mixes
designed with SF. In fact, the quaternary mixes with MK led to a
reduction between 5% and 18% in the SP amount required compared with their corresponding quaternary mixes with SF. Both
MK and SF are ultrane cementing materials with high specic surface areas that require more SP compared with slag and FA, and, as
observed in Table 1, the neness of SF is 4 times higher than that of
MK, which resulted in the higher demand of SP for SF mixes compared with MK mixes.
The mix compositions given in Table 3 indicate that in addition
to the quaternary systems (PCSFALS), which require the lowest PS

197

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

amount for the required slump value, the ternary system (PCSFA) is
very effective in enhancing concrete workability and reducing SP
demand. Indeed, FA is known to react slowly during the hydration
process, and the spherical particles retain much more free water
than PC and other SCMs and hence reduce the water/SP demand.
However, the consistency results of the various cementing materials used did not exhibit a uniform trend. The increase of water consistency when increasing the LS content is related to the relatively
high water absorption capacity of LS, as also noted by Heikal et al.
[9].
In fact, the major factor affecting the SP demand in such a
cementing system is mainly the physical properties of the materials used, including shape, particle size distribution, specic surface,
the amount of ultrane particles in the mix and the chemical composition, especially the amount of unburned carbon. The proportion of particles retained on a 45 lm sieve also appears to be an
important parameter regarding the workability and PS demand.
A higher amount of particles passing the 45 lm sieve is correlated
with a higher PS demand, as observed in Tables 13.
Moreover, all the blended cements tested exhibited a longer initial setting time compared with Portland cement, as observed in
Fig. 2. The delay of the setting time was more pronounced in all
the composite cements containing FA compared with the binary
LS-cements. The ternary system PCSFA had the longest setting time
at 205 min, while PC had the shortest setting time at only 105 min.
The delay in the setting time of the LS-cements, especially the composite cements containing FA, is attributed to the low early-age
reactivity of both LS and FA compared with clinker as well as other
SCMs (SF, MK and slag). In fact, the setting time and hydration rate
of composite cements are strongly linked to the reactivity of the
pozzolanic materials and hence to their chemical composition.

3.2. Mechanical properties

Tables 4 and 5 present the compressive strength results of the
different tested combinations of binary PCLS and composite cements. For both the control and blended cement concretes,
decreasing the waterbinder ratio (w/b) of concrete and increasing
the curing time resulted in a substantial increase in the compressive strength. Regardless of the w/c and age of concrete, the use
of various amounts of LS in the binary and quaternary systems,
as well as the use of various amounts of slag, FA, MK, SF and LS
in the composite cements, led to a reduction in the compressive
strength. In fact, the reduction in the compressive strength of the
binary LS-cement concretes was more pronounced compared with

250

40

200

30
25

150

20
100

15
10

Consistency

Setting time (min)

Water of consistency (%)

35

50

Setting time

LS
S

PC

FA
S

PC

FA

M
K

F
S

FA

FA
S

S
PC

LS
45

PC

LS
35

LS
25

LS
15

0
PC

Concrete mixes
Fig. 2. Setting time and consistency of the concrete mixes investigated.

Table 4
Compressive strength of the PCLS concretes.
PC/LS

Mix

LS (%)

Age (days)
2

28

60

180

365

w/c = 0.79
235 kg/m3

PC
LS15
LS25
LS35
LS45

0
15
25
35
45

7.5
6.5
4.5
3.0
1.0

12.0
10.0
7.5
4.5
2.0

18.5
16.0
12.0
8.5
4.5

27.5
24.0
18.0
13.0
6.5

31.0
27.0
19.5
14.0
7.0

34.0
28.5
20.5
14.5
7.5

35.0
29.5
21.0
14.5
7.5

w/c = 0.65
285 kg/m3

PC
LS15
LS25
LS35
LS45

0
15
25
35
45

10.5
9.0
7.0
5.0
3.5

17.5
15.0
11.5
8.0
5.0

26.5
23.0
18.5
14.0
10.0

37.0
32.5
26.5
20.0
14.0

41.0
35.5
29.0
21.5
14.5

44.5
39.0
30.5
22.5
15.5

46.0
40.0
31.0
23.0
15.5

w/c = 0.60
310 kg/m3

PC
LS15
LS25
LS35
LS45

0
15
25
35
45

12.0
10.5
8.0
6.0
4.0

20.0
17.0
13.0
10.0
6.5

30.5
26.5
21.5
17.0
12.5

41.0
36.5
30.5
23.5
17.0

45.0
39.5
33.0
25.0
18.0

49.0
42.5
34.0
26.0
18.5

50.5
44.0
35.0
26.5
18.5

w/c = 0.52
355 kg/m3

PC
LS15
LS25
LS35
LS45

0
15
25
35
45

16.0
14.0
10.5
8.0
5.5

26.0
22.5
17.5
13.5
9.0

38.5
34.5
28.0
22.5
16.5

50.0
45.5
37.5
30.0
22.0

54.0
48.5
39.5
31.5
23.0

58.5
52.0
42.0
33.0
23.5

61.0
53.5
43.0
33.5
23.5

w/c = 0.45
410 kg/m3

PC
LS15
LS25
LS35
LS45

0
15
25
35
45

20.5
18.0
14.0
10.5
7.0

32.0
28.0
21.5
15.5
11.0

47.5
42.5
34.5
28.5
21.0

59.0
53.5
45.0
36.0
27.0

63.5
57.0
47.0
38.5
28.5

68.0
61.0
49.5
39.5
29.0

71.0
63.0
51.0
40.0
29.0

the composite cements (ternary and quaternary). Higher LS contents in the binary cementitious system were associated with
greater reductions of the compressive strength. For all the PCLS
mixes, depending on the w/c and age of concrete, the strength loss
ranged from 9% up to 86%. As observed in Table 4, embedding up to
15% LS as a replacement for PC resulted in a moderate strength loss
of approximately 12%, whereas increasing the replacement level of
PC by LS beyond 15% always resulted in a substantial strength loss.
The results indicate that increasing the LS content from 25% to 45%
almost doubled the strength loss, which was more pronounced for
mixes with a high w/c while tending to decrease when decreasing
the w/c of concrete at 0.45.
Regardless of the w/c of concrete and the testing age, the LS45
mixes exhibited an important strength loss over 60% compared
with the control PC mix. This reduction could reach 80% for mixes
with high w/c values of 0.79 and 0.65 compared with mixes with
lower w/c values (under 0.60). It was observed that using LS in a
binary system resulted in the largest reduction in the compressive
strength compared with the other cementing systems (ternary and
quaternary).
The effect of the proportion of LS on the compressive strength
was greater than that of w/c. This result was also reported in previous research by Svermova et al. [30].
It is clear that a high content of LS (over 15%) can signicantly
alter the compressive strength of the PCLS mix. This strengthreducing effect induced through the use of LS was more perceptible
at early ages (up to 7 days) compared with long-term use. Due to
the signicant strength loss generated using LS, several studies
and national standards have set the optimum replacement level
of PC by LS to 20%. The relative strength results presented in Table 6
support these restrictions in terms of the maximum LS content
used. The relative strength of the LS15 mixes is approximately
90% of the PC strength but decreases when the LS content is increased to 25% or more.
It is believed that the strength loss of mixes made with various
proportions of LS may be due to the signicant reduction of the potential cementitious material content in cement, which is known

198

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

Table 5
Compressive strength of the composite cement concretes.
Composites

Mix

PC (%)

Age (days)
1

28

60

180

365

w/c = 0.79
235 kg/m3

PC
PCSFA
PCSFASF
PCSFAMK
PCSFALS

100
50
50.1
50.1
47.5

7.5

12.0
4.5
4.0
3.5
4.0

18.5
9.5
10.0
9.5
9.0

27.5
20.5
23.0
21.5
18.5

31.0
25.5
28.5
26.0
23.0

34.0
30.0
33.0
31.0
29.5

35.0

w/c = 0.65
285 kg/m3

PC
PCSFA
PCSFASF
PCSFAMK
PCSFALS

100
50
50.1
50.1
47.5

10.5

17.5
7.0
6.0
5.5
6.0

26.5
14.0
14.5
14.5
13.5

37.0
29.0
32.5
31.0
27.0

41.0
35.5
39.5
37.0
32.5

44.5
40.5
44.5
42.5
38.5

46.0

w/c = 0.60
310 kg/m3

PC
PCSFA
PCSFASF
PCSFAMK
PCSFALS

100
50
50.1
50.1
47.5

12.0

20.0
8.0
7.5
7.5
7.0

30.5
16.5
17.0
17.0
16.0

41.0
33.0
37.0
35.0
30.0

45.0
39.5
43.0
41.0
35.5

49.0
44.5
49.5
47.0
43.0

50.5
50.5
57.0
51.0
48.0

w/c = 0.52
355 kg/m3

PC
PCSFA
PCSFASF
PCSFAMK
PCSFALS

100
50
50.1
50.1
47.5

16.0

26.0
11.0
11.0
10.0
9.5

38.5
21.0
21.5
21.5
20.5

50.0
40.5
45.0
43.0
37.0

54.0
47.5
52.5
50.0
43.0

58.5
54.5
60.5
57.5
52.0

61.0

w/c = 0.45
410 kg/m3

PC
PCSFA
PCSFASF
PCSFAMK
PCSFALS

100
50
50.1
50.1
47.5

20.5

32.0
14.5
15.5
14.0
13.5

47.5
26.0
27.0
27.0
25.0

59.0
47.5
54.0
51.5
43.5

63.5
55.0
61.5
59.0
51.0

68.0
62.5
70.0
66.5
62.0

71.0

Table 6
Relative strength of the binary and composite cement concretes investigated.
LS-cements

Relative strength
7 day

28 day

w/c = 0.79
LS15
LS25
LS35
LS45

0.86
0.64
0.46
0.24

0.87
0.65
0.47
0.23

w/c = 0.65
LS15
LS25
LS35
LS45

0.86
0.69
0.52
0.37

w/c = 0.60
LS15
LS25
LS35
LS45

Composite cements

Relative strength
7 day

28 day

PCSFA
PCSFASF
PCSFAMK
PCSFALS

0.51
0.54
0.51
0.48

0.74
0.83
0.78
0.67

0.87
0.71
0.75
0.37

PCSFA
PCSFASF
PCSFAMK
PCSFALS

0.52
0.54
0.54
0.50

0.78
0.87
0.83
0.73

0.86
0.70
0.55
0.41

0.89
0.74
0.57
0.41

PCSFA
PCSFASF
PCSFAMK
PCSFALS

0.54
0.55
0.55
0.52

0.80
0.90
0.85
0.73

w/c = 0.52
LS15
LS25
LS35
LS45

0.89
0.72
0.58
0.42

0.91
0.75
0.60
0.44

PCSFA
PCSFASF
PCSFAMK
PCSFALS

0.54
0.55
0.55
0.53

0.81
0.90
0.86
0.74

w/c = 0.45
LS15
LS25
LS35
LS45

0.89
0.72
0.60
0.44

0.90
0.76
0.61
0.45

PCSFA
PCSFASF
PCSFAMK
PCSFALS

0.54
0.56
0.56
0.52

0.80
0.91
0.87
0.73

as dilution. It has been speculated that increasing the neness of LS

could enhance its reactivity and consequently improve the compressive strength of the PCLS mix. In fact, for the binary system
with a moderate replacement level of 15%, the PCLS concretes
exhibited a slight decrease of the compressive strength ranging between 10.5% and 16.7% compared with the control PC mix. The obtained results indicate that to limit the reduction effect on the
compressive strength, using a maximum of 15% or less LS as a

partial substitute for Portland cement is recommended. This recommended amount is also supported by the results of several previous studies [3339].
Livesey [33] observed that the addition of 5% LS had little effect
on blended cement performance; however, beyond this limit, the
properties of limestone used resulted in a signicant effect. In a
previous study, De Weerdt et al. [34] observed that up to 15% of
PC could be replaced by limestone powder without impairing the
compressive strength development. Beyond this limit, a reduction
in strength was recorded, and a high replacement level (e.g., 35%)
resulted in a signicant strength loss. Similar results were also obtained by other researchers [35,29]. Bentz [36] considered that the
effect of limestone powder on the compressive strength is likely to
be more pronounced at lower w/c ratios. Findings by Bentz et al.
[37] indicated that depending on the particle size of LS, the
strength reduction at early age varied from 11% to 28%, while at
28 days, the average strength loss was approximately 7%.
A low content of LS (2.5%) appears to yield much better performance when combined with other SCMs in the quaternary systems
(PCSFALS) compared with a higher content of LS (between 15% and
45%) used in the binary system PCLS. In fact, while a signicant
strength reduction (over 50%) was recorded at early age for these
quaternary PCSFALS mixes, strength development recovery at later
age with only approximately 10% strength loss was recorded.
In addition, it is well recognized that when blended cements are
produced by intergrinding clinker together with the cementitious
materials (genuine cement), the resulting performance surpasses
that of the one mixed separately in the concrete mixer. The results
obtained by other researchers [38,39] indicated that pre-mixed
blended LS or natural pozzolana cements with low PC replacement
levels up to 15% could slightly improve the compressive strength
or at least lead to a strength comparable to that of PC concrete. Indeed, the behavior of binary cements with LS or other pozzolanic
materials is not only linked to the replacement level but also to
the physical properties and chemical composition of the SCMs
used, especially their neness.

199

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

The resulting mechanical performance of any cementing system

used in a concrete mix is strongly related to its reactivity, which is
in turn linked to the chemical composition and intrinsic characteristics of each SCM used. The chemical composition given in Table 1
reveals that LS is the poorest material in terms of silicates and aluminates, which are the two main compounds of pozzolanic materials and are required for the formation of additional CSH via
pozzolanic reactions. In addition, the reactivity of cementing materials is also linked to their neness and particle size distribution.
Finer particles result in higher reactivity of the SCMs. The compressive strength results presented in Fig. 3 and in Tables 4 and 5 indicate that MK and SF were much more reactive than LS and slightly
more reactive than FA and slag. The relative strength results
(Table 6) indicate that the quaternary systems designed with SF
and MK generated the highest relative strength compared with
their corresponding LS mix. A simple comparison between the
ternary mix (PCSFA) and the LS quaternary mix (PCSFALS), with
quite similar proportions of slag and FA, demonstrates that the
addition of LS had a negative effect on the strength development,
as observed in Table 6.
One major distinction between the ve mineral admixtures
used (S, FA, MK, SF and LS) is the kinetics of the reactions. The
use of LS in the binary system resulted in a continuous increasing
trend of the compressive strength loss over time, while a less negative effect on the compressive strength was exhibited when LS
was used in the quaternary system in a small amount (2.5%). In
contrast, a decreasing trend of the strength loss was observed
when a combination of slag and FA was used. Embedding either
MK or SF in the ternary system PCSFA resulted in a slight decrease/increase of the compressive strength or a comparable
strength to that of the PC concrete.
It could be speculated that LS used in a binary system at a
replacement level ranging between 15% and 45% does not provide
any signicant strength improvement either at early or later ages,
while using LS at a low replacement level (2.5%) in a quaternary
cementitious system has a limited negative effect while having a
positive effect on the cement matrix densication and hence
strength development. Moreover, the addition of SCM is expected
to improve the interface properties via the pozzolanic reaction
and the lling effect and thus improve the strength and impermeability of the concrete. This improvement is intimately dependent
on the SCM parameters, including the proportion, reactivity and
oxide composition.
Overall, it appears that LS did not contribute to the pozzolanic
reaction but seemed to act more as additional nucleation points
for cement hydration and ller material and thus contributed little
to the strength development. Therefore, LS is not efcient in a

28 d

Compressive strength (MPa)

70

PCLS

0.79
0.65

60

0.60

50

0.52
0.45

40
30
20
10
0
0

10

15

20

25

30

35

40

45

50

LS content (%)
Fig. 3. Effect of limestone content on the 28-day compressive strength of the
concretes.

Table 7
The 28-day compressive and exural strengths and modulus of elasticity of the
concrete mixtures investigated (reduction proportions).
Mix code (w/
c = 0.60)

28-day cube
strength

Flexural strength
(MPa)

Modulus of
elasticity (GPa)

PC
LS15
LS25
LS35
LS45
PCSFA
PCSFASF
PCSFAMK
PCSFALS

41.0
36.5
30.5
23.5
17.0
33.0
37.0
35.0
30.0

5.0

4.0

4.3
4.8
4.6
4.1

30.0
28.5
25.5
22.5
20.0
27.7
27.0
27.0
25.0

(11)
(25.6)
(42.7)
(58.5)
(19.5)
(9.7)
(14.6)
(26.8)

(20)
(14)
(4)
(8)
(18)

(5)
(15)
(25)
(33)
(7.7)
(10)
(10)
(16.6)

binary system and is likely to be more appropriate as a ller material in a ternary or quaternary cementitious system.
3.3. Flexural strength and modulus of elasticity
Similar to the compressive strength, the use of LS, FA, S, MK and
SF in binary, ternary and quaternary systems resulted in a reduction in both the exural strength and modulus of elasticity, as observed in Table 7. The exural strength reduction was less
pronounced, even insignicant, in the quaternary systems (PCSFASF and PCSFASF) compared with the other blended cement concrete mixes. In fact, the improved compactness of the quaternary
cement pastes contributed to limit the exural strength loss while
improving the compressive strength.
The reduction in the modulus of elasticity of the binary and
quaternary LS-cements was proportional to the replacement level
and ranged between 5% and 33%. For the compressive and exural
strengths, the use of LS in both the binary and quaternary cements
resulted in a negative effect on the modulus of elasticity compared
with the PC concrete. While using 15% LS as a partial replacement
of PC had an insignicant effect on the modulus of elasticity (5%
reduction compared with PC), increasing the replacement level beyond 15% led to further reduction in the modulus of elasticity of
the LS-cement concrete.
In fact, while the modulus of elasticity is intimately linked to
the aggregate bulk, it is clear that the modulus of elasticity is also
affected by the cement paste and the interfacial transition zone
(ITZ). The results appear to indicate that the blended cement paste,
and especially the ITZ were weaker than their corresponding PC cement pastes, which in turn led to the reduction in the modulus of
elasticity. In addition, as discussed previously, the results for the
PCLS mixes suggested that LS was likely acting as a ller material
rather than as a pozzolanic material, which creates a second interfacial transition zone in the cement matrix and hence weakens the
modulus of elasticity of the concrete composite.
Moreover, because the strength of cement is a function of the
hydrated part, the strength and modulus of elasticity development
are affected by mineralogical features of the clinker, pozzolanic
reactions, neness, reactive SiO2 ratio and water demand of the cement mixtures. Table 1 shows that LS is very poor in reactive silica
(SiO2) and alumina (Al2O3), which, in the presence of water, react
with (Ca(OH)2) and form an additional CSH gel similar to that
produced during the hydration of PC. Therefore, the low reactivity
and hence negative effect on the strengths for LS are mainly attributed to its chemical composition and the very low content of reactive silica and alumina compared with the other cementing
materials used. In addition, the loss on ignition could indicate
the pozzolanicity of such a material. LS exhibited a very high LOI
of approximately 43.5%. It has been stated that loss on ignition
should be less than 20% (maximum) for proper pozzolanicity of a
material [40].

200

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

3.4. Durability performance

3.4.2. Initial surface absorption (ISA)

Initial surface absorption testing was performed to evaluate the
porosity of the concrete and cement based-materials. Figs. 57
present the test results of the ISA-10 of the PC, PCLS and composite
cement concretes to water after 10 min. For the three w/c ratios
tested, the results indicate that for the PCLS mixes, increasing the
LS content in the concrete led to signicant increases of the ISA10, especially when using up to 20% LS. The increase of ISA-10 appears to be more pronounced for a high w/c value of 0.79 compared with 0.60 and 0.45, as observed in Fig. 5. Meanwhile, for
both the binary and composite cement concretes, regardless of
the type and content of SCMs used, the ISA-10 decreases when
increasing the curing time, as observed in Figs. 6 and 7. Indeed,
increasing the curing time up to 6 months results in the formation
of more hydrate products, less porosity, a disconnected pore network, higher strength and, hence, lower ISA-10 values, especially
for composite cement concretes. Moreover, the use of ternary
and quaternary cementitious systems reduced the ISA-10 values
compared with the Portland and binary LS-cement concrete.
As expected, the PCSFASF mix followed by the PCSFAMK mix
exhibited the lowest ISA-10 values, while the PCSFALS mix

0.79

100

ISA-10 (m/m2/sec 10-2 )

3.4.1. Drying shrinkage

The ultimate magnitude of drying shrinkage strains measured
after 90 days of exposure to dry air (20 2 C and 55 5% RH)
are summarized in Fig. 4. Note that LS caused a slight reduction
in the total magnitude of shrinkage by an average of 1017%, while
the composite cements generally exhibited increased drying
shrinkage magnitudes by approximately 8% or developed the same
magnitude of shrinkage as the control mix. For the binary cements,
increasing the LS content produced a reducing effect on the drying
shrinkage magnitude. As discussed previously, LS appears to be
signicantly less reactive than PC and the other SCMs (S, FA, SF
and MK) used and may behave more as an inert ller material
(non-reactive material), which tends to reduce shrinkage strains.
Whereas the quaternary system with LS (PCSFALS) slightly reduced
the drying shrinkage magnitude, the presence of the other cementing materials (S, FA, SF and MK) in the ternary and quaternary
cementitious systems generated a slight increase in the drying
shrinkage magnitude. This increase in the drying shrinkage strains
may be due to the high neness of MK and especially SF, as well as
their chemical reactivity, compared with both PC and LS. Previous
results by other investigators have suggested that the use of LS
with suitable neness and content could reduce the drying shrinkage strains of concrete [41].

120

0.60
0.45

80

60

40

20

0
0

10

400
300
200
100
0

LS

70
60
50
40
30
20
10

PC
LS35

0
0

Fig. 4. Drying shrinkage of Portland and blended cement concretes investigated.

50

3.4.3. Carbonation
The carbonation of concrete consists of the reaction of carbon
dioxide (CO2) from air with the calcium hydroxide (CH) in concrete
to form calcium carbonate. This process may result in a reduction
of the alkalinity (pH) of the concrete to below 10 and could lead
to the initiation of the corrosion process of the steel reinforcement
bar and degradation of the concrete structure.
For the LS-cement mixes with the two w/c ratios examined,
decreasing w/c from 0.65 to 0.52 resulted in a signicant reduction
in the carbonation depth varying from 5 mm to 15 mm, depending
on the LS content, as observed in Fig. 8. This result indicates that LS
produced a negative effect on carbonation resistance; increasing
the LS content always led to a signicant increase of the carbonation depth. It is suggested that the LS content in concrete used

ISA-10 (m/m2/sec 10-2)

Drying shrinkage
strains (microstrains)

500

40

exhibited the highest ISA-10 values compared with the other composite cements. Beyond 6 months of curing, the composite cement
PCSFASF mix generated the lowest ISA-10 value of 25 m/m2/s 102,
while the highest value of 78 m/m2/s 102 was recorded for the
binary PCLS45 mix.
As aforementioned, the ISA-10 indicates the porosity of the cement paste, which is strongly linked to the concrete mix design
and the type of cementitious materials used. Embedding a combination of various cementing materials with different intrinsic
properties may induce a synergistic effect and result in a lower
porosity and ner pores compared with the PC concrete.

80

600

30

Fig. 5. Effect of limestone content incorporated on the ISA.

800
700

20

LS content (%)

50

LS15
LS45
100

LS25
150

Fig. 6. ISA versus time for the limestone-cement concretes.

200

201

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

50

PCSFA

Carbonation depth (mm)

40

-2

ISA-10 (m/m2/sec10 )

PC

70

45

35
30
25
PC

20

PCSFA

15

PCSFASF

10

PCSFAMK

PCSFALS

60

PCSFASF

50

PCSFAMK
PCSFALS

40
30
20
10
0
0

0
0

50

100

150

10

200

15

20

25

30

Curing time (days)

Age (days)
Fig. 9. Concrete carbonation depth versus time of exposure for the composite
cements.

20-week exposure

45

w/b = 0.65

Carbonation depth (mm)

40
35
30

w/b = 0.52

25
20

70

PCSFA
PCSFASF

50

PCSFAMK
PCSFALS

40
30
20
10
0
0.40

0.50

0.60

0.70

0.80

w/c
Fig. 10. Concrete carbonation depth versus the w/b of concrete.

60

LS45

PCLS

50

LS35
40

LS25
30

LS15
PC

20
10
0
0.40

15

PC

60

Carbonation depth (mm)

as a replacement for PC should not exceed 15%; beyond this limit,

a substantial increase in the carbonation depth (over 80%) was
recorded. However, Fig. 9 shows that all the four composite cement
concretes investigated increased the carbonation depth compared
with the PC concrete. The increase in the carbonation depth
of composite cement concretes is more pronounced at an early
age than at 28 days. Indeed, the carbonation depth was
signicantly reduced when increasing the curing time up to
28 days. While the gap in the carbonation depth between the composite cements and the control mix is large, no signicant differences in the carbonation depth could be observed among the
four composite cements investigated, as demonstrated in Figs. 9
and 10.
In fact, the increase in the carbonation depth for the composite
cements, especially at early age, might be attributed to the slow
reaction of FA used in these cementitious systems at the approximately 20% replacement level. In addition, the carbonation depth
appears to continuously decrease over time, as observed in the
trend in Figs. 9 and 10. It is expected that in the long term, when
FA and the other incorporated SCMs reach their full reactive potential, the carbonation resistance of the composite cement concretes
might be substantially enhanced and, consequently, the carbonation depth would decrease.
Indeed, as discussed previously, LS appears to act as a ller
material, which has a limited positive effect on carbonation resistance. However, the use of composite cements with a combination

Coefficient of diffusion ( 10-11 m2/s)

Fig. 7. ISA versus time for the composite cement concretes.

0.45

0.50

0.55

0.60

0.65

0.70

w/c

10
Fig. 11. Effect of LS content and w/c ratio on the coefcient of chloride diffusion of
concrete.

5
0
0

10

20

30

40

50

LS content (%)
Fig. 8. Concrete carbonation depth versus limestone content after 20-week
exposure.

of various cementing material types and contents with different

reactivities would result in the formation of additional hydrate
products in the system (secondary CSH), which contribute to lling the capillary pores, rene and disconnect the pore network and,

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

Coefficient of diffusion ( 10-11 m2/s)

202

30

PC

Water cured at 20 C

PCSFA
25

PCSFASF
PCSFAMK

20

PCSFALS

15
10
5
0
0.40

0.45

0.50

0.55

0.60

0.65

0.70

Fig. 12. Coefcient of diffusion of concrete prepared with composite cements and
water cured at 20 C.

hence, reduce the permeability and penetration of CO2 into the

composite cement concrete.

Coefficient of diffusion ( 10-11 m2/s)

3.4.4. Chloride ingress resistance

The chloride diffusion coefcients (D) for both the binary and
composite cement mixes, as well as the PC, are determined based
on Ficks second law, and the results are presented in Figs. 11
13. Regardless of the amount and type of SCMs used, the results
indicate that increasing the w/c ratio of concrete led to a signicant
increase in the chloride diffusion coefcient. Similarly, increasing
the amount of LS in the binary system PCLS led to a signicant increase in the chloride diffusion coefcient, especially when the LS
content reached 35% and 45%.
For the PC mixes cured underwater, the chloride diffusion coefcient ranged from 11  1011 m2/s to 21.5  1011 m2/s, while
that of the PC mixes cured in dry air ranged from
14.8  1011 m2/s to 27  1011 m2/s, depending on the w/c ratio.
It is clear that the PC mixes cured under dry air exhibit high chloride diffusion coefcients compared with their corresponding
water-cured mixes. The water curing may allow the mixes to develop higher strength and rene their pore network characteristics,
which reduces their chloride coefcient diffusion.
Regardless of the curing regime used, the composite cement
concretes exhibit signicantly reduced chloride diffusion coefcients compared with both the PC and binary PCLS mixes. In addition, as for the PC mixes, the composite cement concretes cured

30

PC
Air cured at 20 C

PCSFA
25

PCSFASF
PCSFAMK

20

PCSFALS

15
10
5
0
0.40

0.45

0.50

0.55

0.60

0.65

0.70

Fig. 13. Coefcient of diffusion of concrete prepared with composite cements and
dry air cured at 20 C.

under water performed better than their corresponding mixes

cured in dry air. PC or composite cement mixes cured under water
could clearly enhance their cement and pozzolanic reactions and,
consequently, may achieve better cement matrix compactness
and strength and a ner pore system. Therefore, enhanced permeability to chloride ingress compared with the mixes cured under
dry air could be achieved.
The results presented in Figs. 12 and 13 indicate that the composite cement concretes exhibit signicantly improved chloride ingress resistance with a maximum chloride diffusion coefcient
value of 15  1011 m2/s for the mixes that were dry air cured; this
coefcient was reduced by half (8.6  1011 m2/s) when the same
concrete mixes were water cured. Indeed, the improved chloride
ingress resistance generated by the composite cements compared
with both the PC and binary PCLS cements is mainly attributed
to the presence of a combination of various SCMs with different
hydration kinetics and the additional CSH formed, which lls
and renes the coarse capillary pores and, hence, reduces the
permeability.
In fact, this nding suggests that the differences between the
transport coefcients of the PC concrete and blended cement concrete are mainly due to the different pore structures developed in
the presence of the SCM particles. Mineral admixtures with higher
reactivities lead to larger amounts of secondary CSH that are
formed more rapidly and hence lower total pore volumes and ner
pore structures. In turn, the fraction of interconnected pores, permeability and diffusivity within the composite cement matrix are
reduced.
As aforementioned and observed in Fig. 11, the actual trends
suggest that using up to 15% LS has no signicant effect on the
chloride diffusion coefcients, while all the composite cements
had substantially reduced chloride diffusion coefcients regardless
of the w/c of concrete and the curing regime applied. These results
are consistent with previous ndings reported by Baron and Douvre [42], which stipulate that the use of LS-cement increases the
chloride diffusion coefcient compared with PC. An earlier nding
by Tsivilis et al. [43] indicated that using up to 15% LS did not signicantly affect the porosity of concrete and hence its permeability. Recently, Thomas et al. [44] observed that there was no
signicant difference in the rapid chloride permeability of concretes containing SCMs produced with Portland cement or limestone cement. However, it has been concluded that using up to
15% LS, which was ground together with clinker and gypsum, leads
to an improved particle size distribution and particle packing,
which results in a positive effect on the concrete resistance against
chloride ingress [45].
Although limestone is often considered as ller when it is added
to Portland cement, it is not completely inert and can contribute to
the concretes microstructure. The ne limestone particles act as
nucleation sites, increasing the rate of hydration of the silicates.
The accelerating effect on the hydration of C3A and C3S, the
changes in the CSH and the better development of the microstructure are some of the specic factors leading to improved clinker reactivity with limestone when the appropriate amount and
neness are used. Nevertheless, as LS is likely to behave as a ller
material, this addition would create an additional interfacial transition zone (ITZ) and may affect the chloride diffusion coefcient.
Experimental results by Persson and Boubitsas [46,47] indicated
that the addition of limestone ller to concrete with unchanged
w/c might increase the chloride ingress of concrete.
In fact, the conicting results regarding the effect of LS on permeability, diffusion and transport properties and even strength
and carbonation are strongly linked to the intrinsic properties of
LS used, in particular the neness and particle size distribution.
In general, the use of SCMs such as slag, FA, LS, SF and MK can
physically improve the pore structure of the blended cement

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

Satisfactory DF, ASTM C494

Durability factor DF (%)

w/c = 0.52
Fresh air content = 5%

Concrete mixes
Fig. 14. Durability factors of the binary and composite cement concretes
investigated.

matrix particle packing by lling the voids between the clinker

particles as well as by being reactive materials and consequently
decreasing chloride diffusion and other harmful substances.
Permeability to chloride ions is intimately linked to the mix design parameters and, more precisely, to the capillary pore network
characteristics. Larger volumes of coarser and connected pores result in higher permeabilities and chloride diffusion coefcients.
The selection of an appropriate mix design that guarantees minimum porosity and a ne pore structure would enhance resistance
to chloride ingress.
Note that both carbonation and chloride ingress are properties
to be examined for reinforced concrete elements, while for unreinforced concrete elements, no risk of corrosion is expected, and
therefore, carbonation and chloride ingress might be ignored.
3.4.5. Resistance to freezingthawing
The durability factors (DFs) of PC, binary PCLS and composite
cement concrete mixes designed with various amount and types
of SCMs are provided in Fig. 14. The binary PCLS15 mix developed
good freezingthawing resistance (durability factor P 80%), while
the mixes containing more than 15% LS exhibited poor freezing
thawing resistance (durability factor 6 80%). Nevertheless, except
for the quaternary mix PCSFASF, which has a DF of approximately
80%, all the other composite cement concretes exhibited DF values
lower than 80%. In fact, the decrease in DF was more signicant for
the quaternary mix (PCSFALS) and the binary mixes with 35% and
45% LS content.
Similarly to the chloride ions and compressive strength results,
the DF appears to decrease when the replacement level of PC by LS
was higher than 15% in the binary system PCLS. In fact, the results
presented above reveal that permeability to chloride increases and
compressive strength decreases when increasing the replacement
Table 8
Contribution of the binary and composite cements used to reducing the embodied
CO2 (ECO2) of concrete.
Materials
ECO2/ton
Mix (w/c = 0.45)

MK
Slag
330
52
ECO2/kg

PC
LS15
LS25
LS35
LS45
PCSFA
PCSFASF
PCSFAMK
PCSFALS

381.3
326
289.3
252.6
216
256
250.4
314
242

LS
32

SF
14

FA
4

Portland cement
930
Reduction in ECO2 (%)

14.5
24.1
33.8
43.4
32.9
34.3
17.7
36.6

203

level of PC by LS beyond 15%. As discussed earlier, LS appears to

act as a ller material and is not actively involved in the pozzolanic
reaction and the formation of additional CSH gel, which could
improve the compactness and strength of the cement paste and
hence improve the FT resistance. However, for the composite cements, it appears that the combination of slowly reactive materials
(FA) with slag without SF or MK (highly reactive materials) may
have contributed to reduce the FT resistance and, hence, the DF.
The freezingthawing durability factor clearly increases as both
impermeability and strength increase. It could be speculated that
the combination PCSFASF exhibited appropriate particle packing
and strength development, which led to the satisfactory DF obtained with this mix, while the corresponding mix with MK exhibited a slightly lower DF (76%) that was very close to the acceptable
DF (80%).
Freezingthawing resistance, which is expressed by the DF, is
intimately linked to the air system parameters, including the total
volume of air, spacing factor and specic surface area of the air
bubbles. These parameters are in turn affected by the mix design
parameters, particularly the cementitious materials type and content used. The results indicate that for the mixes investigated herein with a w/c of 0.52 and a fresh air content of 5%, the amount and
type of cementitious materials used appears to be the major factor
affecting the DF of air-entrained blended cement concrete mixes.
4. Contribution of LS/composite cements to the embodied CO2
reduction
Many years of widespread use have demonstrated that these
SCMs mainly derived from industrial by-products can, when properly designed within the cementitious system, not only improve
the quality of the concrete but also provide multiple environmental benets. Table 8 summarizes the total reduced embodied CO2
(ECO2) resulting from the use of various content and types of SCMs
as a partial replacement of Portland cement in 1 m3 of concrete.
The results indicate that using various contents and types of SCMs
in concrete has led to signicant reduction of the (ECO2), ranging
from 14% to 43%. The reduction of the (ECO2) in the binary concrete
was proportional to the LS content used. While a high LS content of
45% resulted in a signicant reduction (approximately 43%) in the
ECO2, the overall performance of this mix LS45 was below the required concrete quality, as discussed earlier. However, the use of
all the composite cements generated a substantial reduction of
approximately 33% in the ECO2, except the PCSFAMK mix, which
only reduced the ECO2 by 17%.
Overall, it can be concluded that the use of these SCMs removes
an industrial by-product from the waste stream and converts it
into a benecial additive for sustainable construction, greatly
improving concrete quality, lowering its carbon footprint, saving
more non-renewable materials and providing economy to concrete
structures.
5. Conclusions
The present study highlights the performance of concrete mixes
designed with various w/c ratios and four composite and binary cements. Five types of supplementary cementitious materials, slag,
y ash, limestone, silica fume and metakaolin, were used at various
replacement levels to prepare binary, ternary and quaternary cements. Based on the test results reported herein, the following conclusions may be drawn:
1. The type and composition of blended cement appear to have
a signicant effect on both the fresh and hardened properties of concrete.

204

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

2. The use of LS appears to have no signicant effect or slightly

improves the concrete slump, while the use of composite
cements reduced the amount of superplasticizer required
compared with the PC mix.
3. The composite cements prolonged the initial setting time of
all the concretes tested compared with the PC and PCLS
mixes.
4. At all the tested w/c values and ages, the compressive
strength of the binary PCLS concrete decreased when
increasing the PC replacement level. This strength loss was
more pronounced when the replacement level of PC by LS
was higher than 15%.
5. The compressive strength of the composite cement concretes was strongly linked to the type and proportions of
the SCMs used. The ternary (PCSFA) and quaternary (PCSFALS) cementitious systems always led to a strength loss,
especially at early age, while the two other quaternary systems (PCSFASF and PCSFAMK) developed similar or slightly
higher compressive strengths compared with the PC mixes.
6. For both the binary and composite cements, a slight
decrease of the 28-day exural strength and modulus of
elasticity was recorded. The reduction was more signicant
in the binary LS-cement with a replacement level higher
than 15%.
7. The use of LS-cement resulted in a reduction of the total
magnitude of drying shrinkage strains of approximately
15%, while a slight increase of the drying shrinkage strains
was recorded when using the composite cements.
8. The use of all the composite cements improved the resistance to chloride ingress and resulted in much better performance than both the binary PCLS and the PC concretes.
However, the carbonation depth results suggested that both
the binary and composite cements exhibited increased carbonation depth compared with the PC mix.
9. The experimental results demonstrated that the binary concrete mixes designed with a w/c ratio of 0.52 and 15% LS
with approximately 5% air content would be expected to
provide good concrete freezethaw resistance with a durability factor higher than 80%; this DF decreased when a
higher amount of LS was incorporated.
10. The quaternary cementitious system PCSFASF reached a satisfactory DF (approximately 80%), while all the other composite cements concretes exhibited a DF lower than 80%.
11. Even though the LS particles are relatively reactive compared with other ller materials, the results indicate that
the LS used in this study likely acted as an inert ller material rather than as a pozzolanic one. The results propose a
maximum replacement level of PC by LS of 15% to limit
the strength loss and decrease in durability performance.
12. Composite cements (ternary and quaternary), when properly optimized, are the most advantage cementitious systems that signicantly reduce the environmental impacts
of the concrete industry while also improving strength and
durability performance.
13. The use of both binary and composite cements signicantly
contributes to reducing the embodied CO2 emissions compared with Portland cement.

References
[1] Damtoft JS, Lukasik J, Herfort D, Sorrentino D, Gartner EM. Sustainable
development and climate change initiatives. Cem Concr Res 2008;38(2):
11527.
[2] United States Geological Survey. USGS mineral program cement report,
January 2013. p. 389.

[3] Turanli L, Uzal B, Bektas F. Effect of large amounts of natural pozzolan addition
on properties of blended cements. Cem Concr Res 2005;35(6):110611.
[4] Mayeld LL. Limestone additions to Portland cement an old controversy
revisited. Cem Concr Aggr 1988;10(1):38.
[5] Livesey P. Performance of limestone-lled cements. In: Swamy RN, editor.
Blended cements in construction. London: Elsevier Science; 1991. p. 115.
[6] Matthews JD. Performance of limestone ller cement concrete. In: Dhir RK,
Jones MR, editors. Impact of ENV 197 on concrete construction. London: E & FN
Spon; 1994. p. 11347.
[7] Gonzlez AM, Irassar EF. Effect of limestone ller on the sulphate resistance of
low C3A Portland cement. Cem Concr Res 1998;28(11):165567.
[8] Tsivilis S, Batis G, Chaniotakis E, Grigoriadis Gr, Theodossis D. Properties and
behavior of limestone cement concrete and mortar. Cem Concr Res
2000;30(10):167983.
[9] Heikal M, El-Didamony H, Morsy MS. Limestone-lled pozzolanic cement. Cem
Concr Res 2000;30(11):182734.
[10] Menndez G, Bonavetti V, Irassar EF. Strength development of ternary blended
cement with limestone ller and blast-furnace slag. Cem Concr Compos
2003;25(1):617.
[11] Ghrici M, Kenai S, Said-Mansour M. Mechanical properties and durability of
mortar and concrete containing natural pozzolana and limestone blended
cements. Cem Concr Compos 2007;29(7):5429.
[12] Chindaprasirt P, Homwuttiwong S, Sirivivatnanon V. Inuence of y ash
neness on strength, drying shrinkage and sulfate resistance of blended
cement mortar. Cem Concr Res 2004;34(7):108792.
[13] Tsivilis S, Chaniotakis E, Kakali G, Batis G. An analysis of the properties of
limestone cements and concrete. Cem Concr Comp 2002;24(34):3718.
[14] Bonavetti V, Donza H, Rahhal V, Irassar E. Inuence of initial curing on the
properties of concrete containing limestone blended cement. Cem Concr Res
2000;30(5):7038.
[15] European Committee for Standardization, Cement: Composition, Specications
and Conformity Criteria, Part 1: Common Cements, EN 197-1, EN/TC51/WG 6
rev; 2000.
[16] Lafarge product data sheet, Portland limestone cement, Lafarge Cement UK,
March 2009.
[17] Soroka I, Setter N. The effect of llers on strength of cement mortars. Cem
Concr Res 1977;7(4):44956.
[18] Bonavetti VL, Rahhal VF, Irassar EF. Studies on the carboaluminate formation
in limestone ller-blended cements. Cem Concr Res 2001;31(6):8539.
[19] Hawthorm F. Ten years experience with composite cements in France. In:
Proceedings of a seminar of the BRE/BCA/Cement Industry Working Party,
Building Research Establishment, Garston, November 1989.
[20] Lothenbach B, Le Saout G, Gallucci E, Scrivener K. Inuence of limestone on the
hydration of Portland cements. Cem Concr Res 2008;38(6):84860.
[21] De Weerdt K, Kjellsen KO, Sellevold E, Justnes H. Synergy between y ash and
limestone powder in ternary cements. Cem Concr Compos 2011;33(1):308.
[22] Lawrence P, Cyr M, Ringot E. Mineral admixtures in mortars: effect of inert
materials on short-term hydration. Cem Concr Res 2003;33(12):193947.
[23] Barker AP, Hobbs DW. Performance of Portland limestone cements in mortar
prisms immersed in sulfate solutions at 5 C. Cem Concr Compos
1999;21(2):12937.
[24] Bensted J. Thaumasite background and nature in deterioration of cements,
mortars and concretes. Cem Concr Compos 1999;21(2):11721.
[25] Skaropoulou A, Tsivilis S, Kakali G, Sharp JH, Swamy RN. Long term behavior of
Portland limestone cement mortars exposed to magnesium sulfate attack. Cem
Concr Compos 2009;31(9):62836.
[26] Aguilera J, Blanco Varela MT, Vzquez T. Procedure of synthesis of thaumasite.
Cem Concr Res 2001;31(8):11638.
[27] ASTM C666/C666M-03. Standard test method for resistance of concrete to
rapid freezing and thawing. West Conshohocken, PA: ASTM International;
2003.
[28] Limbachiya MC, Meddah MS, Ouchagour Y. Use of recycled concrete aggregate
in y-ash concrete. Constr Build Mater 2012;27(1):43949.
[29] Tsivilis S, Chaniotakis E, Badogiannis E, Pahoulas G, Ilias A. A study on the
parameters affecting the properties of Portland limestone cements. Cem Concr
Compos 1999;21(2):10716.
[30] Svermova L, Sonebi M, Bartos PJM. Inuence of mix proportions on rheology of
cement grouts containing limestone powder. Cem Concr Compos
2003;25(7):73749.
[31] Selvamony C, Ravikumar MS, Kannan SU, Basil Gnanappa S. Investigation on
self-compacted self-curing concrete using limestone powder and clinker.
ARPN J Eng Appl Sci 2010;5(3):16.
[32] Bdrina M, Khenfer MM, Dheilly RM, Quneudec M. Reuse of local sand:
effect of limestone ller proportion on the rheological and mechanical
properties of different sand concretes. Cem Concr Res 2005;35(6):11729.
[33] Livesey P. Strength characteristics of Portland-limestone cements. Constr Build
Mater 1991;5(3):14750.
[34] De Weerdt K, Justnes H, Kjellsen KO. Fly ashlimestone ternary composite
cements: synergetic effect at 28 days. Nordic Concr Res 2010;42(2):20.
[35] Soroka I, Stern N. Calcareous llers and the compressive strength of Portland
cement. Cem Concr Res 1976;6(3):36776.
[36] Bentz DP. Modeling the inuence of limestone ller on cement hydration using
CEMHYD3D. Cem Concr Compos 2006;28(2):1249.
[37] Bentz DP, Irassar EF, Bucher BE, Weiss WJ. Limestone llers conserve cement,
Part 2: durability issues and the effects of limestone neness on the mixtures.
ACI Concr Int 2009;31(12):359.

M.S. Meddah et al. / Construction and Building Materials 58 (2014) 193205

[38] Voglis N, Kakali G, Chaniotakis E, Tsivilis S. Portland-limestone cements. Their
properties and hydration compared to those of other composite cements. Cem
Concr Compos 2005;27(2):1916.
[39] Gneyisi E, Gesoglu M, zturan T, Mermerdas K, zbay E. Properties of
mortars with natural pozzolana and limestone-based blended cements. ACI
Mater J 2011;108(5):493500.
[40] Practical action, testing methods for pozzolanas, The Schumacher centre for
technology & development, Warwickshire, UK.
[41] Khanh Bui V, Montgomery D. Drying shrinkage of self-compacting concrete
containing milled limestone. In: Skarendahl , Petersson , editors.
Proceedings PRO 7: 1st International RILEM symposium on self-compacting
concrete; 1999. p. 22738.
[42] Baron J, Douvre C. Technical and economical aspects of the use of limestone
ller additions in cement. World Cem 1987;18(4):1004.

205

[43] Tsivilis S, Tsantilas J, Kakali G, Chaniotakis E, Sakellariou A. The permeability of

Portland limestone cement concrete. Cem Concr Res 2003;33(9):146571.
[44] Thomas MDA, Hooton D, Cail K, Smith BA, De Wal J, Kazanis KG. Field trials of
concretes produced with Portland limestone cement. ACI Concr Int
2010;32(1):3541.
[45] Tsivilis S, Asprogerakas A. A study on the chloride diffusion into Portland
cement limestone cement concrete. Mater Sci Forums 2010;636637:135561.
[46] Persson B. Assessment of the chloride migration coefcient, internal frost
resistance, salt frost scaling and sulphate resistance of self-compacting
concrete with some interrelated properties. Div. Building Research. Report
TVBM-3100, Lund; 2001.
[47] Boubitsas D. Studies on the efciency of granulated blast furnace slag and
limestone ller in mortars. Long-term strength and chloride penetration. Div.
Building Research. Report TVBM-3125, Lund; 2005.