Cement and Concrete Research 33 (2003) 1481 – 1486

Communication
Investigation of hydraulic activity of ground granulated
blast furnace slag in concrete
S.C. Pala, A. Mukherjeeb, S.R. Pathakc,*
a
Faculty of Civil Engineering, College of Military Engineering, Pune 4110031, India
b
Faculty of Civil Engineering, Indian Institute of Technology, Powai, Mumbai 400076, India
c
Department of Civil Engineering, Government College of Engineering, Pune 411005, India
Received 11 September 2002; accepted 3 February 2003

Abstract

Ground granulated blast furnace slag (GGBFS), a by-product of the steel manufacturing industry, being used as an effective partial cement
replacement material, has already been proven to improve several performance characteristics of concrete. The reactivity of GGBFS has been
found to depend on the properties of slag, which vary with the source of slag, type of raw material used, method and the rate of cooling. The
present work aims at bringing out a novel relationship between the Hydraulic Index (HI) of slag at 7 and 28 days (HI7 and HI28) and the
influencing properties of slag, namely, glass content, fineness and chemical composition by employing multiple regression analysis on 37
slag samples from various sources. HI7 and HI28, thus obtained, have been mapped onto a Slag Activity Index (SAI) plot, giving an
indication of the ranges of strength of slag.
D 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Cement; Compressive strength; CaO; Granulated blast furnace slag; MgO

1. Introduction performance expected from a slag, since the hydration

reactions taking place are far more complex than indicated
Blast furnace slag is a by-product obtained in the by these formulas [3]. From earlier research work, it has
manufacture of pig iron in the blast furnace and is formed been accepted that the reactivity of slag is influenced by the
by the combination of earthy constituents of iron ore with slag properties such as glass content, chemical composition,
limestone flux. When the molten slag is swiftly quenched mineralogical composition, fineness and the type of activa-
with water in a pond, or cooled with powerful water jets, it tion provided. Hence, the present work aims at obtaining the
forms into a fine, granular, almost fully noncrystalline, Hydraulic Index (HI) at 7 and 28 days (HI7 and HI28) by
glassy form known as granulated slag, having latent considering these crucial parameters influencing slag char-
hydraulic properties. Such granulated slag, when finely acteristics.
ground and combined with Portland cement (PC), has been
found to exhibit excellent cementitious properties [1]. The
reactivity of ground granulated blast furnace slag (GGBFS) 2. Physical and chemical characteristics of GGBFS
is considered to be an important parameter to assess the
effectiveness of GGBFS in concrete composites. 2.1. Physical characteristics
In order to predict the hydraulic activity of a blast furnace
slag, various hydraulicity formulas have been proposed [2] The specific gravity of the slag is approximately 2.90
as summarized in Table 1. However, it has been observed with its bulk density varying in the range of 1200 – 1300 kg/
that these formulas do not adequately predict the strength m3. The color of GGBFS is normally whitish (off-white).

2.1.1. Fineness
* Corresponding author. Tel.: +91-20-553-7062. As with all cementing materials, the reactivity of slag is
E-mail address: [email protected] (S.R. Pathak). determined by its surface area. In general, increased fineness

0008-8846/03/$ – see front matter D 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0008-8846(03)00062-0
1482 S.C. Pal et al. / Cement and Concrete Research 33 (2003) 1481–1486

Table 1 2.2. Chemical characteristics

Formulas proposed for assessment of hydraulicity of GGBFS
Serial Formula Requirement for Preference The chemical composition of the slag plays a key role
number good performance
upon which the HI has a bearing. From a chemical stand-
1 CaO/SiO2 1.3 – 1.4 1 point, slags can be classified into two types according to
2 (CaO + MgO)/SiO2 >1.4 1
their basicity index. Several basicity indices have been
3 (CaO + MgO)/ 1.0 – 1.3 1
(SiO2 + Al2O3) defined by different authors, the simplest one being the
4 (CaO + 0.56 Al2O3 + 1.4 MgO)/
1.65 2 CaO/SiO2 ratio given by Nkinamubanzi [7] as given in
SiO2 Table 1. Metallurgists classify slag as either basic or acidic:
5 (CaO + MgO + Al2O3)/SiO2
1.0 3 the more basic the slag, the greater its hydraulic activity in
the presence of alkaline activators [8]. At constant basicity
results in better strength development, but in practice, the strength increases with the Al2O3 content, and a defi-
fineness is limited by economic and performance consid- ciency in CaO can be compensated by a larger amount of
erations and factors such as setting times and shrinkage. In alumina (MgO). The influence of MgO as a replacement for
the United Kingdom, GGBFS is marketed at a surface area CaO seems to depend both on the basicity and the MgO
of 375 –425 m2/kg Blaine’s, whereas some slags in the content of the slag. Variations in the MgO content up to
United States have a surface area in the range of 450 –550 about 8 – 10% may have little effect on strength devel-
m2/kg; Canadian slags are about 450 m2/kg, while in India it opment, but high contents have an adverse effect [8].
is found to vary from 350 to 450 m2/kg Blaine’s. Further, Frearson [9] has mentioned that the presence of
The fineness of GGBFS is a very important parameter, merwinite crystallites within the glass structure would
which is dependent on energy-saving and economic con- improve the reactivity of slag. Moreover, it was observed
siderations, influences the reactivity of GGBFS in concrete, that hydraulic activity increases with increasing CaO, Al2O3
early strength development of concrete and water require- and MgO and decreases with increasing SiO2 content.
ment. Swamy [4] reported that an increase in fineness of According to European Standard ENV 197-1:1992 and
two to three times that of normal PC can preserve the British Standards, the ratio of the mass of CaO plus MgO
benefits of material fineness on a variety of engineering to the mass of SiO2 must exceed 1.0. This ratio assures high
properties such as bleeding, time of setting, heat evolution, alkalinity, without which the slag would be hydraulically
high strength and excellent durability. Thus, for better inactive [10]. Lea [8] obtained similar trends on the CaO
performance, the fineness of GGBFS must be greater than content and reported that the hydraulic value increases with
that of cement. the CaO/SiO2 ratio up to a limiting point (not defined
precisely), but CaO content beyond the limiting point makes
2.1.2. Glass content granulation difficult and results in lower glass content as
The glass content of slag is considered to be the most reported by Frearson [9]. In a later study, Frearson and
significant variable and certainly the most critical to hydrau- Higgins [11] reported that the Al2O3 content of the slag
licity. Several factors influence the degree of vitrification influences the sulfate resistance of slag concrete and noted
achieved during quenching, but the most important variable that an MgO level of about 13% is required for a satisfactory
influencing the nature of slag is the temperature at which the performance against the sulfate attack. The percentage of
furnace is tapped. The rate of quenching, which influences soluble sulfate (expressed as SO3) is stipulated to be no
the glass content, is thus the predominant factor affecting greater than 4% and the percentage of total sulfur to be no
the strengths of slag cements. Increasing crystalline contents
reduce hydraulicity, but there is no well-defined or single
relationship between strength and glass content, although
some research has shown linear glass content – strength
relationship [5]. Although a glassy structure is essential to
reactivity, research has shown that there is no exact correla-
tion of glass content to hydraulicity, and therefore, there is
no guarantee that a high glass content will produce a highly
reactive slag. Research data show that slag samples with as
little as 30– 65% glass contents are still suitable, but no
specific minimum required glass content appears to emerge
from these tests. Because of these uncertainties, most
international standards judge slag activity by direct strength
performance tests rather than include minimum glass con-
tent criteria [6]; although it has been heuristically reported
that generally, the glass content of the slag should be in Fig. 1. Ternary diagram indicating composition of PC and GGBFS in the
excess of 90% to show satisfactory properties. CaO – SiO2 – Al2O3.
 S.C. Pal et al. / Cement and Concrete Research 33 (2003) 1481–1486 1483

greater than 2.5% for reasonable durability requirements. Table 2

SAI standards for various grades as prescribed in ASTM C989
Moreover, according to Ganesh Babu and Sree Rama
Kumar [12], the reactive glass content and fineness of Age and grade SAI, minimum percent
GGBFS alone influence the cementitious/pozzolanic effi- Average of last Any individual
ciency, or its reactivity in concrete composites, significantly five consecutive samples sample
within the limit of the above parameters. 7-day index
Grade 80 – –
Grade 100 75 70
Grade 120 95 90
3. Hydraulic reactivity of slag
28-day index
Research carried out so far reveals that the hydration Grade 80 75 70
product that is formed when GGBFS is mixed with PC and Grade 100 95 90
Grade 120 115 110
water is essentially the same as the principal product formed
when PC hydrates, i.e., calcium silicate hydrate (CSH) [2].
As seen in the ternary diagram in Fig. 1, PC and GGBFS lie relative compressive strength. Classification is in accord-
in the same general field, although PC is essentially in the ance with Table 2 (ASTM C989).
C3S field, whereas GGBFS is found essentially in the C2S
field of the diagram. This is why GGBFS hydrates are
generally found to be more gel-like than the products of 5. Hydraulic Index
hydration of PC, and so add denseness to the cement paste.
The hydration mechanism of GGBFS is different from To assess more critically towards high volume replace-
that of cement. When GGBFS is mixed with water, initial ment of cement by slag and compare the same with other
hydration is much slower than PC mixed with water. pozzolanic material, a new parameter, namely, Hydraulic
Hydration of GGBFS in the presence of PC depends upon Index (HI) could be introduced.
the breakdown and dissolution of the glassy slag structure Keil [16] has defined the HI of the slag based on 70/30 as
by hydroxyl ions released during the hydration of PC and HI 70/30=((a  c)/(b  c))  100 where: a = the strength of
also the alkali content in cement. The hydration of GGBFS 70% slag/30% PC at time t; b = the strength of 100% PC at
consumes calcium hydroxide and uses it for additional CSH time t; c = the strength of 70% ground quartz/30% PC at
formation. Research by Regourd [13], Vanden Bosch [14] time t.
and Roy and Idorn [15] have suggested that, in general, This index gives a range of values from 0 to 100 or even
hydration of GGBFS, in combination with PC, at normal greater, leading to a better range than that obtained in the
stage is a two-stage reaction. Initially and during the early ASTM SAI test, since the HI relates to the reactivity of the
hydration, the predominant reaction is with alkali hydrox- supplementary cementing material alone. The use of 70%
ide, but subsequent reaction is predominantly with calcium slag content, higher than normally used, helps to distinguish
hydroxide. the poor quality of slags from the good ones.
The complexity of the influencing factors suggest that
direct performance evaluations of workability, strength
characteristics and durability are the most satisfactory meas- 6. Novel relationships between HI7, HI28 and slag
ures of the effectiveness of GGBFS use. The ASTM C989 properties
Slag Activity Index (SAI) is therefore recommended as a
basic criterion for evaluating the relative cementitious Authors of this paper have studied a few more slag
potential of GGBFS. samples obtained from various sources in India. After
determining the physical and chemical properties of these
slag samples, it was observed that the oxide contents of each
4. Slag Activity Index slag vary widely. Hence, the effect of each of these
parameters on hydraulic index at 7 and 28 days (HI7 and
ASTM C989 defines SAI as the percentage ratio of the HI28) need to be studied. Therefore, an effort is made to
average compressive strength of slag cement (50 – 50%) obtain such correlation among HI7, HI28 and the significant
mortar cubes to the average compressive strength of ref- slag characteristics, namely, SiO2, CaO, MgO, Al2O3, glass
erence cement mortar cubes at a designated age, expressed content and Blaine’s fineness.
as: SAI = Slag Activity Index, percent=(SP/P  100) where, Hooton and Emery [17] have reported mortar cube
SP = average compressive strength of slag-reference cement strength data for 37 slag samples collected from various
mortar cubes [psi]; P = average compressive strength of sources like America, Canada, Australia, France, Britain,
reference cement mortar cubes [psi]. etc. For this purpose, 50-mm mortar cubes were cast using
Based on this, slag was classified into three grades— ASTM Type I PC, commercially ground quartz and sepa-
Grade 80, Grade 100 and Grade 120 depending upon the rately ground slag with relative densities in the range of 2.89
1484 S.C. Pal et al. / Cement and Concrete Research 33 (2003) 1481–1486

where Blaine represents Blaine’s fineness in units of 500

m2/kg.
The standard deviations for Eqs. (1), (2), (3) and (4) are
9.77, 8.64, 8.07 and 7.68, respectively.
From the values of the coefficient of correlation for each
of these equations, the best correlation for HI7 has been
found to be Eq. (4) with R=.941. A scatter (normalized
between 0 and 1) of predicted HI7 values against their
experimental counterparts, as obtained by authors and
reported by Hooton and Emery [17], is depicted in Fig. 2.
Similarly, the relations for HI28 have been obtained as:
HI28 ¼ 143:496 þ 1:025ðglassÞ þ 132:21ðC=SÞ

R ¼ :801 ð5Þ
Fig. 2. Scatter of HI7 values (normalized). HI28 ¼ 155:817 þ 0:976ðglassÞ

and 2.97. A w/c ratio of 0.485 used for PC was reduced to þ 99:603ððC þ M þ AÞ=SÞ
0.439 for mortars incorporating slag. For finely ground slag
(  498 m2/kg Blaine), a w/c ratio was further reduced to R ¼ :813 ð6Þ
0.419. Mortar cubes were moist cured at 23 °C and 100%
HI28 ¼ 36:908 þ 3:112ðCaOÞ  3:909ðSiO2 Þ
relative humidity (RH) and then tested in compression at 7
and 28 days. The HIs were calculated from these strength
þ2:989ðAl2 O3 Þ þ 2:425ðMgOÞ þ 0:966ðglassÞ
results of slag cubes in comparison to those of PC and
quartz cubes as per Keil’s [16] approach. þ12:5ðBlaineÞ
Multiple regression analysis was then employed on this
data to compute HI7 and HI28. The combination of different R ¼ :815 ð7Þ
slag parameters affecting the HI has been tested. The first
where again, Blaine represents Blaine’s fineness in units of
equation was developed using SiO2 and glass content of
500 m2/kg.
slag. Further, two of the compositional moduli, namely, (C/
The standard deviations for Eqs. (5), (6) and (7) are
S) and ((C + M + A)/S), often cited in the literature, have
15.36, 14.95 and 15.84, respectively. The relationship of
been incorporated along with the glass content to form Eqs.
HI28 with glass content and SiO2 gives an R value even less
(2) and (3), respectively. Ultimately, the influence of all the
than .8 and hence, is not considered here.
significant physical and chemical characteristics of slag
Eq. (7) with R=.815, best predicts HI28. Fig. 3 shows a
(SiO2, CaO, MgO, Al2O3, glass content and Blaine’s fine-
graph of a scatter of normalized values of HI28 from Eq. (7)
ness) have been simultaneously taken into account to obtain
vis-à-vis their corresponding experimental values.
HI7, giving rise to Eq. (4).
It is then obvious from the exhaustive HI7 and HI28
HI7 ¼ 325:627  9:144ðSiO2 Þ þ 0:735ðglassÞ relationships (Eqs. (4) and (7) above), that physical and

R ¼ :888 ð1Þ
HI7 ¼ 231:026 þ 0:72ðglassÞ þ 205:738ðC=SÞ
R ¼ :914 ð2Þ

HI7 ¼ 241:472 þ 0:638ðglassÞ þ 149:729

ððC þ M þ AÞ=SÞ

R ¼ :925 ð3Þ

HI7 ¼ 46:991 þ 4:589ðCaOÞ  5:733ðSiO2 Þ

þ 4:582ðAl2 O3 Þ þ 2:93ðMgOÞ þ 0:633ðglassÞ

þ 3:5ðBlaineÞ

R ¼ :941 ð4Þ Fig. 3. Scatter of HI28 values (normalized).

 S.C. Pal et al. / Cement and Concrete Research 33 (2003) 1481–1486 1485

Fig. 4. Graph of SAI grades and HI7, HI28 values.

chemical properties, i.e., glass content, fineness and all 3. Using the equations developed in this work, the reactivity
major oxides such as CaO, SiO2, Al2O3 and MgO of slag, of slag can be determined knowing the physical and
in fact, do significantly influence the HI of slag both at 7 chemical properties of slag, even in the absence of mortar
and 28 days. Further, the glass content and fineness have a compressive strengths.
remarkable effect on HI7 and HI28, as is evident from the 4. A wide range of variations in the slag characteristics can
significant values of their corresponding coefficients in be accounted for in these equations, which not only
these relationships. predict the HI but also the strength performance of slag at
It is now instructive to study the connection between HIs 7 and 28 days.
as obtained from Keil’s [16] approach and SAI values as 5. HI7 has been found to numerically correspond to the SAI
defined by ASTM C989. Fig. 4 portrays the curves for Grades Grade 100 and Grade 120 slag.
80,100 and 120 as per SAI. The values of HI7 and HI28
computed from the equations developed herein have been
mapped on this plot. It has been observed that HI7 values in Acknowledgements
the range from 95 to 120 lie between the Grade 100 and Grade
120 curves, whereas HI7 of 125.22 (>120) lie above the One of the authors, S.R. Pathak, is indebted to Dr. Sudhir
Grade 120 curve. This means that at 7 days, SAI Grade 100 A. Kulkarni, Consultant, Mahindra British Telecom, Pune,
and Grade 120 numerically correspond to the HI (percent of India, for helpful discussions regarding the data analysis
controls) values of 100 and 120, respectively, as obtained carried out in the present work.
from the HI7 equation developed in this work, leading to a
direct correlation between SAI at 7 days and HI7. However, at
28 days, all the HI28 values lie within the Grade 100 and References
Grade 120 curves. Thus, Eqs. (4) and (7) could be gainfully
employed for computing HI7 and HI28, given the significant [1] C.L. Hwang, C.Y. Lin, Strength development of blended blast furnace
physical and chemical characteristics of slag. slag cement mortars, SP 91-65, V.M. Malhotra (Ed.), Proceedings of
the 2nd International Conference on Fly Ash, Silica Fume, Slag and
Natural Pozzolans in Concrete, Madrid, Spain, vol. 2. American Con-
7. Conclusions crete Institute, Detroit, MI, USA, 1986, pp. 1323 – 1340.
[2] H.G. Smolczyk, The effect of chemistry of slag on the strength of
blast furnace cements, Zem.-Kalk-Gips 31 (6) (1978) 294 – 296.
1. The HIs are strongly correlated with the most significant [3] D.G. Mantel, Investigation into the hydraulic activity of five granu-
physical and chemical properties of slag, namely, SiO2, lated blast furnace slags with eight different portland cements, ACI
CaO, MgO, Al2O3, glass content and Blaine’s fineness at Mater. J. 91 (5) (1994) 471 – 477.
[4] R.N. Swamy, Design for Durability and Strength Through the Use of
both 7 and 28 days.
Fly Ash and Slag in Concrete, CANMET/ACI International Workshop
2. HI defined on 70/30 basis can be related with SAI having on Supplementary Cementing Materials, Superplasticizers and Other
50/50 basis, which in turn indicates the strength Chemical Admixtures in Concrete, Toronto, Canada, American Con-
performance of slag. crete Institute, 1998, pp. 1 – 72.
1486 S.C. Pal et al. / Cement and Concrete Research 33 (2003) 1481–1486

[5] H.E. Schwiete, The effect of cooling conditions and the chemical [12] K. Ganesh Babu, V. Sree Rama Kumar, Efficiency of GGBS in con-
composition on the hydraulic properties of the haematitic slags, For- crete, Cem. Concr. Res. 30 (2000) 1031 – 1036.
schungsber. Landes Nordrh.-Westfal., 1963, pp. 23 – 29. [13] M. Regourd, Characterization of thermal activation of slag cements,
[6] R.N. Swamy, A. Bouikni, Some engineering properties of slag con- Proceedings of the 7th International Congress on the Chemistry of
crete as influenced by mix proportioning and curing, ACI Mater. J. 87 Cements (Paris) 2 (III-3) (1980) 105 – 111.
(1990) 210 – 220. [14] V.D. Vanden Bosch, Performance of mortar specimens in chemical
[7] P.C. Nkinamubanzi, M. Baalbaki, J. Bickley, P.C. Aitcin, The use of and accelerated marine exposure, Performance of Concrete in Marine
slag for making high performance concrete, Sixth NCB International Environment, ACI SP-65, American Concrete Institute, Detroit,
Seminar on Cement and Building Materials, XIII, NCB, New Delhi, 1980, pp. 487 – 507.
1998, pp. 13 – 39. [15] D.M. Roy, G.M. Idorn, Hydration, structure and properties of blast
[8] F.M. Lea, The Chemistry of Cement and Concrete, 3rd ed., Edward furnace slag cements, mortars and concrete, Proc.-Am. Concr. Inst. 79
Arnold, London, 1970, p. 727. (6) (1982) 445 – 457.
[9] J.P.H. Frearson, Sulfate resistance of combination of portland cement [16] F. Keil, Slag cements, Proceedings of the Third International Sympo-
and blast furnace slag, ACI Publ. SP-91 2 (1986) 1495 – 1524. sium on the Chemistry of Cements, London, Cement and Concrete
[10] A.M. Neville, Properties of Concrete, 4th ed., ELBS Publication, Association, UK, 1952, pp. 530 – 571.
Addison-Wesley Longman Limited, Essex, UK, 1996, p. 80. [17] R.D. Hooton, J.J. Emery, Glass content determination and strength
[11] J.P.H. Frearson, D.D. Higgins, Sulfate resistance of mortars containing development predictions for vitrified blast furnace slag, ACI SP 79,
ground granulated blast furnace slag with variable alumina content, in: American Concrete Institute, Detroit, 1983, pp. 943 – 962.
V.M. Malhotra (Ed.), ACI Publ. SP-132, American Concrete Institute,
Istanbul, 1992, pp. 1525 – 1542.