Materials and Structures (2008) 41:73–83

DOI 10.1617/s11527-006-9220-y

O R I G I N A L A RT I C L E

A simple way to mitigate alkali-silica reaction

André Carles-Gibergues Æ Martin Cyr Æ
Marc Moisson Æ Erick Ringot

Received: 3 November 2006 / Accepted: 22 December 2006 / Published online: 30 January 2007

RILEM 2007

Abstract A possible practice to prevent disor- Keywords Alkali-silica reaction
Mineral

ders due to alkali-silica reaction (ASR) in con- admixtures
Reactive aggregate powders

cretes containing reactive aggregates is the use of Surface cracking
Expansion
chemical admixtures (lithium salts) or, more
commonly, mineral admixtures such as fly ash,
1 Introduction
silica fume, ground granulated slag or metakaolin.
An analysis of the literature concerning ASR
In the decade following the discovery of the alkali
revealed some papers devoted to the mitigating
silica reaction (ASR) by Stanton, recommenda-
activity of reactive aggregate powders (RAP)
tions, such as the use of non-reactive aggregate or
when associated with their parent aggregates.
the use of low alkali cement, were made to avoid
These RAP result from the grinding of the
or to mitigate the disorders due to this reaction.
reactive aggregate. To verify the efficiency of this
At the same time, it was established that blended
method of mitigation, tests were performed on
cements made of OPC mixed with pozzolanic
concrete prisms cured at 60C and 100% R.H.,
admixtures were efficient in reducing ASR expan-
using four reactive aggregates, associated or not
sion. The use of such admixtures (fly ash, silica
with their RAP. The results showed that the use
fume, ground granulated slag) is now current
of RAP reduced or suppressed ASR-expansion,
practice to prevent disorders due to ASR.
suppressed surface cracking and counteracted the
An analysis of the literature concerning other
loss of compressive strength due to ASR. If these
admixtures obtained by grinding rocks shows that
findings are confirmed by trials on other types of
the papers can be classified in two categories:
aggregates, the addition of RAP would be a
simple way of avoiding ASR disorders. (i) Papers related to ground admixtures added
to mixtures containing a reactive aggregate
coming from a different source. Many refer-
ences concern this category, particularly
Hanna [1], Stanton [2], Bennett and Vivian
A. Carles-Gibergues
M. Cyr (&)
M. Moisson
[3], Pepper and Mather [4], Abdun-Nur [5],
E. Ringot Ming-Shu et al. [6, 7], Asgeirsson [8] and
Laboratoire Matériaux et Durabilité des
Kodama and Nishino [9].
Constructions, INSA – UPS, 135 Avenue de Rangueil,
Toulouse Cedex 04 31077, France (ii) Papers devoted to fine admixtures associated
e-mail: cyr@insa-toulouse.fr with the reactive aggregate from which they
74 Materials and Structures (2008) 41:73–83

are obtained by grinding. Only a few refer-
ences are found in this category. For exam-
ple, Qinghan et al. [10] worked on mortars
made of andesite, and replaced between 30
and 70% of cement by this ground aggregate
(fineness ranging between 300 and 1100 m2/
kg). They showed that ground andesite
reduced the expansion of mortars. However,
high replacement rates and high finenesses
were needed for the powders to be efficient,
thus limiting the interest of the solution in
practice. Guédon-Dubied et al. [11] found
that, when filler from reactive siliceous
limestone was added to mixtures containing
the same reactive aggregate, expansions
were reduced to about the half of those
occurring in concrete with no filler.
The aim of this paper is to verify whether
reactive aggregate powder (RAP) is an efficient
and general solution to counteract ASR expan-
sion due to the coarse aggregate. Tests were
performed on concrete using four reactive aggre-
gates with their ground fillers.
2 Materials
The binder was an ordinary Portland cement,
CEM I 52.5R as specified by European Standard
EN 197-1, with a specific weight of 3.15 and a
fineness of 350 m2/kg (Blaine). Its chemical com-
position (Table 1) shows a moderately high
Na2Oeq content (0.8%).
Four coarse aggregates were chosen:
– A siliceous aggregate B, a metaquartzite con-
taining very small quantities of muscovite and
amorphous silica in the intergranular cement.
This rock, chosen in France as a standard
reactive aggregate, has low kinetics of silica
dissolution.
– A siliceous limestone T, mainly composed of
calcite with a few percent of dolomite. When
observed at microscopic level, diagenetic silica
is seen to consist mainly of quartz and traces of
feldspars, muscovite and kaolinite. This aggre-
gate is also a French standard reactive rock,
with a slow rate of silica dissolution.
– An opaline aggregate O, which is very delete-
rious and commonly reported in laboratory
experiments as a reference mineral. The aggre-
gate used in this study was an ‘‘opal CT’’,
according to the classification of Jones [12]. The
glassy phase was predominant, the crystallised
phase being composed essentially of tridymite
and a few percent of quartz.
– A crushed waste glass G, resulting from the
crushing of window glasses. It was chosen
because the reuse of waste glasses has been
envisaged by several workers in the last decade
[13–17]. It had an amorphous structure as
determined by X-ray diffraction.
Reactive aggregate powders b, t, o and g
were obtained by crushing and sieving aggre-
gates B, T, O and G respectively at 80 lm. The
chemical compositions of these powders appear
in Table 1. In order to study the influence of
fineness in counteracting expansion, the pow-
ders were ground to a specific surface area
(Blaine) ranging from 100 to 650 m2/kg,
depending on the aggregate (100, 200 and 400
for B and G, 200, 400 and 600 for T, 200, 400
and 650 for O).

Table 1 Chemical compositions of cement, aggregates and aggregate powders (%)

Cement B b T t O o G g

SiO2 20.1 94.1 87.7 20.0 15.4 93 92.7 72.1 66.7

Al2O3 5.6 1.6 4.0 1.3 1.7 0.0 0.0 1.0 0.7
Fe2O3 2.0 0.3 1.0 0.6 1.1 0.2 0.3 0.1 0.1
CaO 62.5 0.2 0.4 40.6 43.6 0.2 0.2 6.3 7.2
MgO 3.1 0.1 0.2 1.2 1.5 0.1 0.1 3.7 4.1
K2O 0.9 0.7 0.9 0.4 0.5 0.1 0.1 0.1 0.2
Na2O 0.2 0.2 0.1 0.4 0.5 0.2 0.2 12.7 13.4
SO3 3.2 0.1 0.1 0.3 0.2 0.9 1.1 0.1 0.2
L.O.I 1.7 0.7 1.1 34.7 34.9 5.5 6.0 0.4 0.7
Materials and Structures (2008) 41:73–83 75

3 Experimental methods the average of three measurements. The contain-

3.1 Mortar tests
The mortar test was an accelerated method using
an autoclave apparatus (French Standard P18–
594 [18]). This test consisted of measuring the
length variations of mortar prisms
(4 · 4 · 16 cm) doped with alkalis (4% of cement
mass) after a treatment lasting 5 h at 127C and
0.15 MPa. Mortars had a water/cement ratio of
0.5 and an aggregate/cement ratio of 2.0. The first
series of tests were performed on mortars con-
taining only cement, water and reactive aggre-
gates (0.08–4 mm sand). For the second series, a
fraction of the 0.08–4 mm sand (by mass) was
replaced by the aggregate powders: 10 and 20%
of different finenesses ranging from 100 to
650 m2/kg (Blaine).
3.2 Concrete tests
The concrete samples were made, stored and
tested according to French Standard P18-454 [19].
This test consisted of evaluating the performance
of a concrete with respect to ASR by measuring
its expansion on 7 · 7 · 28 cm prisms stored at
60C and 100% relative humidity. After mixing,
the samples were stored in a moist room at 20C
for 24 h. Then, the concrete prisms were placed
vertically on grids in watertight containers con-
taining 35 mm of water (the concrete was not in
contact with the water). Each concrete was tested
on three prisms and the expansion data represent
ers were placed in a reactor generating an
atmosphere at 60C and 100% R.H. Expansion
measurements were taken after the containers
and their prisms had been cooled for 24 h at 20C.
Immediately after each measurement, the prisms
were put back into their containers, which were
kept in the reactor at 60C until the next
measurement. The last measurement was taken
at 88 weeks. At 34 weeks, map cracking was
measured in order to quantify the degree of
damage due to ASR. At the same age, compres-
sive strength (fc) was measured on three cubes
(7 · 7 · 7 cm) cut from a 7 · 7 · 28 cm prism.
Details of the concrete mixtures, including the
slump (modified Abram’s cone of 200 mm) and
density, are given in Table 2. Each mixture is
designated by a capital letter representing the
reactive aggregate used, followed by a small letter
for mixtures containing fine powders. The water
content of the concrete was set between 190 and
210 l/m3. A significant amount of cement was
used (between 429 and 467 kg/m3) in order to
obtain high alkali content. However, considering
the value of Na2Oeq in the cement, it was decided
to enhance the alkali content to 5 kg/m3 by
adding NaOH to the mixing water.
For concretes B and T, the coarse aggregate
(4–16 mm) was composed of reactive aggregate
and the sand (0.08–4 mm) was non-reactive
quartz. For concrete O, the amount of opal was
limited to 36% of the coarse aggregate (8–
16 mm). The rest of the aggregate was non-
reactive. In concrete G, the reactive aggregate

Table 2 Composition, density and slump of concrete mixtures

B T O G B-b T-t O-o G-g

Cement (kg/m3) 477 477 469 452 462 467 438 442
Water (l) 212 212 208 201 205 208 195 197
W/C 0.44 0.44 0.44 0.44 0.44 0.44 0.44 0.44
Superplasticizer (l) 0 0 0 0 2.46 2.49 4.87 4.91
Agg. powder (kg/m3) 0 0 0 0 122.4 113.5 119.4 136.8
Sand 0.08–4 (kg/m3) 631 (NR) 617 (NR) 639 (NR) 699 (R) 490 (NR) 454 (NR) 478 (NR) 546 (R)
Aggregate (kg/m3)
4–8 235 (R) 248 (R) 138 (NR) 200 (NR) 228 (R) 242 (R) 129 (NR) 195 (NR)
8–16 NR 0 0 604 766 0 0 565 749
8–16 R 888 929 345 0 861 910 323 0
Density (kg/m3) 2443 2445 2403 2319 2371 2397 2252 2272
Slump (cm) 4 4 4 4 3 7 11 5
(R) = reactive aggregate; (NR) = non-reactive aggregate
76 Materials and Structures (2008) 41:73–83

resulted from the crushing of waste window glass,
so its maximum size was forced by the thickness
of the window glass, i.e. 4 mm. Hence, the coarse
aggregate > 4 mm was non-reactive quartz.
Aggregate powders used to replace 20% of the
sand (by weight) had finenesses of 400, 600, 650
and 400 m2/kg for powders b, t, o and g, respec-
tively.
– Powder o (fines of opal) was efficient in
reducing the expansion but only for a large
amount of fines having a high fineness.
– Powder g (fines of glass) was very efficient,
whatever the amount and the fineness of the
fines.
4.2 Concretes

4 Experimental results The length variations of the concrete prisms

4.1 Mortars
The results of the autoclave expansion tests
performed on the mortars are given in Table 3.
They confirm that the four aggregates used can be
classified as reactive, since the measured expan-
sions varied between 0.30 and 0.42%. These
expansions were much higher than 0.15%, the
value under which the aggregate is considered as
non-reactive according to French standard P18–
594 [18].
The use of all aggregate powders decreased the
expansion of mortars. However, their activity
depended on the origin, fineness and quantity of
the fines used. A comparison of the performances
of the powders regarding the non-reactive crite-
rion (0.15%) and the decrease of expansion due
to ASR led to the following observations:
cured for 88 weeks at 60C and 100% R.H. are
shown in Fig. 1. In a first analysis, it is noteworthy
that all the aggregate powders reduced the
expansion of concrete.
After 34 weeks of curing, when the expansion
of the prisms had practically stabilized, all faces of
the 7 · 7 · 28 cm samples were photographed.
The apparatus, comprising the camera, its CCD
sensor and the lens, enabled a resolution of
570 lm per pixel. Each face was numbered as
shown on Fig. 2, number one corresponding to
the upper face of the sample. Crack patterns are
shown on Fig 3. Cracks were segmented and
analyzed for the specific length of cracks per pair
of parallel faces, expressed in cm/cm2 . The
results, reported in Table 4, give information on
both the intensity of the damage and the possible
isotropic distribution of cracking. Two obvious
conclusions can be drawn from this Table:

– Powder b (fines of quartzite) was efficient
without a large amount or a significant fineness
of fines having to be used.
– Powder t (fines of siliceous limestone) gave a
small decrease in the expansion, but its effi-
ciency was not sufficient to satisfy the criterion
of non-reactivity.
(a) The presence of fines dramatically decreased
the ASR crack development, as was con-
firmed by the swelling measurements.
(b) 2D specific length LA was systematically
higher in the vertical planes (with reference
to the casting position) so that it can be
concluded that cracks developed preferen-

Table 3 Expansion (in %) of mortar versus amount and fineness of powder aggregate
Aggregate B T O G

Aggregate powder
Amount (%) 0 10 20 0 10 20 0 10 20 0 10 20
Low 0.30 0.30 0.26 0.38 0.38 0.29 0.42 0.27 0.17 0.38 0.15 0.06
Fineness medium 0.29 0.16 0.34 0.30 0.35 0.28 0.11 0.15
High 0.13 0.08 0.35 0.27 0.24 0.06 0.12 0.03
Mean value of expansion (%) 0.30 0.24 0.17 0.38 0.36 0.32 0.42 0.29 0.17 0.38 0.13 0.08
Reduction of expansion (%) 20 43 5 16 31 60 66 79
Mean reduction (%) 31 8 45 72
Materials and Structures (2008) 41:73–83 77

Fig. 1 Expansion of
concretes versus time,
cured at 60C and 100%
RH

At the same age of 34 weeks, the compressive

Fig. 2 Indexation of the prism faces in map cracking
measurements
tially in the horizontal plane. These results
are in accordance with the conclusions
presented by Larive [20].
strength (fc) of each concrete formulation was
measured on three 7 · 7 · 7 cm cubes cut from a
prism. These values are reported in Table 5. The
supplementary data fc = 61 MPa relates to a
concrete with same proportioning and curing as
the B, T, O, G series but with innocuous aggre-
gates. It can be seen that addition of aggregate
powder enhances the compressive strength of all
concretes.
Freshly broken surfaces (obtained when testing
compressive strength) were observed by scanning

Table 4 Influence of aggregate powders on specific cracking length (LA) in concretes (in cm/cm2)
Face B B-b T T-t O O-o G G-g

1–3 0.28 0 0.12 0.01 0.06 0 0.22 0

2–4 0.55 0 0.30 0.03 0.10 0 0.50 0
5–6 0.25 0 0.38 0.03 0.31 0 0.44 0
78 Materials and Structures (2008) 41:73–83

Table 5 Influence of aggregate powder on compressive strength of concretes (age 34 weeks)

Aggregate B T O G Ra

Without powder aggregate (MPa) 60 58 47 45 61

With powder aggregate (MPa) 71 67 81 71 –
Increase in compressive strength (%) 18 15 72 58 –
a
Concrete with innocuous aggregate having the same mixture design and the same curing conditions as other concretes
with reactive aggregates

electron microscopy (SEM) coupled with energy

dispersive spectrometry (EDS). All concretes
without aggregate powder reacted and produced
large amounts of ASR gel. These gels had various
morphologies, from smooth cracked surfaces to
lamellar and acicular crystals. Concretes contain-
ing aggregate powder did not present any ASR
gel on the broken surfaces and only very small
amounts of gel were detected in some voids. The
scarcity of these sites did not allow us to deter-
mine if reactive aggregate powders significantly
altered the chemistry and the expansion of the
gel.

5 Discussion

An assessment of the beneficial activity of aggre-

gate powders against ASR deterioration needs to
correlate the different components: expansion as
defined by its maximum value and maximum
speed, surface cracking, compressive strength and
gel detected at the microscopic level. Table 6 will
be used as a basis for this discussion.
Since the addition of aggregate powders Fig. 3 Influence of aggregate powders on surface cracking
increased the compressive strength of concretes of concrete
affected by ASR, it is necessary to distinguish
between the variation of strength1 due to ASR experimental compressive strength fc measured
(DASR) and that due to a pozzolanic-like effect on concrete prisms after curing for 34 weeks at
(DPozz) assumed for the aggregate powders [25]. 60C and 100% RH. The experimental compres-
The decoupling of these effects was carried out by sive strength fc of each concrete mixture is related
calculating a theoretical strength fth for each (equation 1) to the theoretical strength fth and to
concrete mixture and comparing this value to the a variation in strength Dfc due to an external
effect other than cement hydration. In our case,
Dfc was related either to the effect of ASR alone
1
Loss of strength in our case. In fact, many authors such as (mixtures without fines) or to the combined
Nixon et al. [21] and Boyd et al. [22] have observed that
losses of compressive strength are, in most cases, much less
effects of ASR and pozzolanic-like reaction
marked than decreases in tensile strength. In a few cases, (mixtures with fines).
no losses of strength were observed [20,23]. Some rare
authors report a slight strengthening despite continuous fc = fth + Dfc = fth + DASR + DPozz (1)
expansion [24].
Materials and Structures (2008) 41:73–83 79

Table 6 Influence of aggregate powders on the properties of concrete (relative to ASR)

B T O G

Expansion (88 weeks) Without powder Final expansion (%) 0.234 0.200 0.245 1.07
Maximum expansion rate (lm/m/week) 212 395 2070 1025
With powder Reduction of expansion (%) 89 36 96 90
Reduction of expansion rate (%) 75 63 98 97
Acceptance criteria P18-454a yes no yes yes
Cracking (34 weeks) Without powder Cracking length (cm/cm2) 0.40 0.23 0.10 0.36
With powder Reduction (%) ~100 91 ~100 ~100
Compressive strength Without powder Compressive strength (MPa) 60 58 47 45
With powder Increase in compressive strength (%) 18 15 72 58
Microstructure Without powder ASR products visible by SEM yes yes yes yes
With powder ASR products visible by SEM no no no no
a
expansion at 3 months lower than 0.02%

where DASR is the loss of strength due to ASR
and DPozz is the rise of strength due to a
pozzolanic-like effect of aggregate powder.
Many formulas exist for predicting the com-
pressive strength (fth) of concrete, such as those
given by Féret [26], Abrams [27] and Bolomey
[28]. Most of them depend partly on the quanti-
ties of cement and water in the mixture. In our
case, since the water-cement ratio was the same
for all concretes, fth was taken as the compressive
strength of the reference concrete R already
defined in Table 5. We recall that this concrete
was made with innocuous aggregates, having no
ASR or pozzolanic activities.
For concretes without aggregate powder,
DPozz = 0. So DASR can be obtained from the
difference between the experimental and
theoretical strengths. For concretes containing
aggregate powders, DPozz can be calculated by
subtracting fth and DASR from the experimental
strength fc. It is assumed that DASR is similar for
concrete containing the same type of aggregate
(e.g. DASRB = DASRB-b).
The values of fth, DASR and DPozz, reported in
Table 7 for all mixtures, are used in the following
analysis.
5.1 Analysis of concrete results
5.1.1 Concretes B and B-b
The reduction in the final expansion due to
aggregate powder b reached a significant value
of 89%. A delayed start of the reaction and a
lowering of the expansion speed were also
observed (Table 6). Surface cracking, which was
initially very significant for concrete B, was
suppressed for concrete B-b containing aggregate
powder. The concrete B-b meets the recommen-
dations of the standard NF P18–454 [19], its
expansion at the age of 3 months being lower
than the limit of 0.02%. Moreover, the addition of
aggregate powder acted like a pozzolan and
increased the compressive strength by 18%.
5.1.2 Concretes T and T-t
The activity of the aggregate powder t was
moderate, resulting in an expansion reduction of
36% (Table 6). Expansion at the age of 3 months
remained far above the limit of 0.02%. Thus the
use of concretes T and T-t should be prohibited.
However, it is worth noting that surface cracking

Table 7 Values, in MPa, of experimental compressive strength (fc), theoretical strength fth, loss of strength due to ASR
(DASR) and rise of strength due to a pozzolanic-like effect of aggregate powder (DPozz)
R B B-b T T-t O O-o G G-g

fc 61 60 71 58 67 47 81 45 71
fth 61 61 61 61 61 61 61 61 61
Dfc 0 –1 10 –3 6 –14 20 –16 10
DASR 0 –1 –1 –3 –3 –14 –14 –16 –16
DPozz 0 0 11 0 9 0 34 0 26
80 Materials and Structures (2008) 41:73–83

was divided by 10 for concrete T-t. In a work started after a very long delay (1 year) due
carried out to determine the alkali-reactivity of to the aggregate powder, it still reached a
another limestone, Guédon-Dubied et al. [11] high level: the expansion at 88 weeks was
showed that incorporation of reactive limestone fivefold the limit suggested by the standard
filler (15 at 30% by weight) reduced expansion at at 5 months.
1 year. The observed reduction of 50% was
greater than in the present work, but the expan-
5.2 Hypothesis concerning the mechanism
sion of their concrete without filler was 0.065% at
by which aggregate powders counteract
1 year, a much lower value than in our case
the deleterious effects of reactive
(0.20%).
aggregates
5.1.3 Concretes O and O-o
There is an evident lack of correlation between
the three parameters chosen to characterize the
Aggregate powder o was very efficient in reduc-
severity of damage in ASR-affected concretes.
ing both expansion (96%) and surface cracking
On the one hand, Fig. 4 clearly shows that the
(totally suppressed). It was noted that ASR
cracking length and the loss of compressive
caused a large decrease in the compressive
strength due to ASR (DASR) are not correlated
strength of concrete O (evaluated at 14 MPa,
with expansion. The existence of a threshold may
Table 7). The beneficial contribution of aggre-
be presumed, but not enough tests were carried
gate powder, typically due to its pozzolanic
out to affirm that cracks would not appear for
behaviour, raised the compressive strength by
expansions lower than 0.2%. Compressive
some 70%. Practically, the concrete O-o meets
strength was chosen as mechanical criterion of
the requirements of standard NF P18–454 [19],
concrete. We are conscious that this choice is
its expansion being only the half the authorized
open to criticism: particularly measurement of
level.
elastic modulus would have better chance to
correlate with density of cracking.
5.1.4 Concretes G and G-g On the other hand, as can be seen in Fig. 5, the
correlation between the three indicators of the
The analysis of the experimental results was not aggregate powder activity is far from clear.
easy at all, because three points of view had to be In fact, Fig. 6 is more instructive. With respect
taken into account. to the mechanical properties of both series of
concretes (with and without powders), it shows
(i) On a phenomenological level, the expansion
of concretes G was very high (1.07%), the
surface cracking was significant and the loss
of compressive strength was moderate
(13%). The addition of aggregate powder
was very interesting since it led to a great
reduction in expansion (90%), a suppression
of surface cracking and an enhancement of
compressive strength (58%).
(ii) A strict examination of the performance of
concrete G-g in relation to standard NFP18-
454 [19] led to an approval of this mixture
since its expansion was lower than 0 at
5 months.
(iii) From a practical point of view, it would be
negligent to implement such a concrete. Fig. 4 Length of surface cracking and loss of strength
The reason is that, even though expansion versus final expansion, in ASR affected concretes
Materials and Structures (2008) 41:73–83
Fig. 5 Combined effects of reduction of expansion, reduc-
tion of cracking and pozzolanic gain of strength due to the
addition of aggregate powders
symmetry in the behaviour of aggregates, the
most deleterious of them in a coarse state being
the most effective in reducing expansion when
powdered. These findings are in agreement with
the results found by Yamamoto and Kamaru [29].
These authors show a good correlation between
the activity index of fly ash (determined by
measurement of the compressive strength of
mortars) and its preventive effect on ASR
expansion.
It can be suggested that the activity of an
aggregate (deleterious or beneficial) depends on
its rate of silica release. This hypothesis is in
accordance with previous data obtained on the
dissolution of powders placed in 1M NaOH
solution at 60C: the maximum dissolution rates,
expressed in mmole SiO2/g/day, were 1.7 for b; 0.9
for t; 24.4 for o and 15.5 for g [30]. Several authors
have expressed such an opinion. For example
Hudec et al. [16], when comparing the effective-
ness of fly ash, silica fume and ground waste
Fig. 6 Effects of ASR and pozzolanic activity of aggregate
powders on the compressive strength of concretes
81
glasses to reduce ASR, concluded that ‘‘waste
bottle glass, or any glass, can be considered to be
a pozzolan or a reactive aggregate, depending on
the particle size’’. Taylor expressed the same idea
when he wrote that ‘‘the chemistry of ASR is
essentially that of pozzolanic reaction, the differ-
ent effect on concrete properties arising mainly
from the difference in particle size of the siliceous
material.’’ [31]
Pursuing this line of thought leads us to put
these aggregate powders and typical pozzolans in
the same category. Thus their activity as ASR-
reducers could be explained using the same
mechanisms as proposed for pozzolans. When
aggregate powders are dispersed in a cement
paste they release silica, resulting in a lowering of
the Ca/Si ratio in C–S–H. Many authors [32–34]
have established that the ability of these low
Ca/Si C–S–H to fix alkalis is enhanced. The
depletion of free alkalis lowers the pH of the pore
solution and, consequently, reduces the attack of
reactive aggregates. Finally the expansion is
reduced or suppressed.
To apply the conclusions of this work to the
practice of civil engineering it will be necessary to
verify whether these results of laboratory tests are
in good agreement with the long-term perfor-
mance of real concrete in the field. This verifica-
tion is of prime importance, as confirmed by the
observation of Pedersen et al. [35] and several
other authors, since the balance between ASR
and pozzolanic-like reaction of aggregate powder
may be modified by elevated temperatures in the
laboratory test compared to the field situation.
The conclusions of Bokern and Siebel [36] rein-
force the uncertainty about the reliability of
accelerated concrete tests when the effectiveness
of a pozzolan used as an expansion reducer is
tested.
6 Conclusion
The experiments reported in this paper were
intended to verify the following hypothesis: the
harmful effects of ASR induced by an alkali-
reactive aggregate can be counteracted by the
beneficial activity resulting from the addition of a
fraction of this aggregate ground into a fine state.
82
The work was carried out on concretes incor-
porating four alkali-reactive aggregates, with or
without aggregate powder added. The concretes
were cured at 60C and 100% RH and their
expansion was measured for up to 88 weeks. In
these conditions, which are different from actual
conditions in the field, the following conclusions
can be drawn:
(i) The addition of a reactive aggregate powder
to concrete containing the same coarse reac-
tive aggregate can reduce or suppress ASR
expansion.
(ii) The efficiency of such an aggregate powder
depends on its petrographic composition,
content and fineness.
(iii) Surface cracking is strongly reduced or
suppressed when aggregate powders are
used in concretes.
(iv) Aggregate powders act as pozzolans do and
counteract, more or less, the decrease of
compressive strength caused by ASR.
(v) Another consequence of the addition of
aggregate powder is the invisibility of ASR
products on the broken surfaces of concretes
examined by scanning electron microscopy.
However, it cannot be asserted that the
production of gel is effectively prevented;
perhaps it forms in particles too small and
diffuse to be detectible at this level of
magnification.
Before any attempt to extrapolate from
laboratory to field concrete, our findings have
to be verified on other aggregates and more
attention must be paid to the influence of the
amount of aggregate powder on its activity. In
the event of successful verifications, the method
of counteracting the ASR expansion due to a
reactive aggregate by addition of a finely
ground fraction of this aggregate would be
attractive from a practical point of view, since
this addition could be made at the quarry or at
the concreting site.
Of course it would be interesting to compare
the efficiency of this method with that of mitiga-
tion due to mineral admixtures such as silica
fume, fly ash, slag and metakaolin.
Materials and Structures (2008) 41:73–83
References
1. Hanna WC (1947) Unfavorable chemical reaction of
aggregates in concrete and a suggestive corrective.
Proc ASTM 47:986
2. Stanton TE (1950) Studies of use of pozzolanas for
counteracting excessive concrete expansion resulting
from reaction between aggregates and the alkalis of
cement. In: Symposium on use of pozzolanic materials
in mortars and concretes, ASTM Special Technical
Publication No. 99, 1950,178–201
3. Bennett IC, Vivian HE (1955) Studies in cement
aggregate reaction XXII. The effect of fine-ground
opaline material on mortar expansion. Aust J Appl Sci
6:88–93
4. Pepper L, Mather B (1959) Effectiveness of mineral
admixtures in preventing excessive expansion due to
alkali-aggregate reaction. Proc ASTM 59:1178–1202
5. Abdun-Nur EA (1961) Fly ash in concrete—an
evaluation. Highway research board bulletin 284.
National Research Council, Washington, DC
6. Ming-Shu T, Yu-Feng Y, Mei-Qi Y, Shi-Hua Z (1983)
The preventive effect of mineral admixtures on alkali-
silica reaction and its mechanism. Cem Concr Res
13(2):171–176
7. Ming-Shu T, Han S (1983) Rapid method for
determining the preventive effect of mineral
admixtures on alkali-silica reaction. In: Proceedings
of the 6th International Conference on AAR in
concrete, Copenhagen, Denmark, 383–386
8. Asgeirsson H (1986) Silica fume in cement and silane
for counteracting of alkali-silica reactions in Iceland.
Cem Concr Res 16(3):423–428
9. Kodama K, Nishino T (1986) Observation around the
cracked region due to alkali-aggregate reaction by
analytical electron microscope. In: Proceedings of the
7th International Conference on AAR in concrete,
Ottawa, Canada, 398–402
10. Qinghan B, Xuequan W, Ming-Shu T (1996) Effect of
reactive aggregate powder on suppressing expansion
due to alkali-silica reaction. In: Proceedings of the 10th
International Conference on AAR in concrete,
Melbourne, Australia, 546–553
11. Guédon-Dubied JS, Cadoret G, Durieux V, Martineau
F, Fasseu F, Van Overkeke V (2000) Etude du calcaire
Tournaisien de la carrière Cimescaut à Antoing
(Belgique) Analyse pétrographique et chimique et
réactivité aux alcalins. Bull Lab Ponts et Chaussées
226:57–66
12. Jones JB, Segnit ER (1971) The nature of opal. I
Nomenclature and constituent phases. J Geol Soc
Austr 18(1):57–68
13. Nishikawa T, Takatsu M, Daimon M (1995)
Mechanical properties of mortars containing glass
powder. 49th annual meeting of Japan Cement
Association 49:114–119
14. Polley C, Cramer SM, de la Cruz RV (1998) Potential
for using waste glass in Portland cement concrete. J
Mat Civ Eng 10(4):210–219
Materials and Structures (2008) 41:73–83
15. Yixin S, Lefort T, Moras S, Rodrigues D (1998) Waste
glass: a possible pozzolanic material for concrete.
CANMET/ACI Int Symp on sustainable
development of cement and concrete, Ottawa, 317–326
16. Hudec PH, Ghamari C (2000) Ground waste glass as
an alkali silica reactivity inhibitor. In: Proceedings of
the 11th International Conference on AAR in
Concrete, Quebec, Canada, 663–672
17. Kojima T, Takagi N, Haruta K (2000) Expanding
characteristics of mortar with glass powder produced
from waste bottles. In: Proceedings of the 11th
International Conference on AAR in concrete,
Quebec, Canada, 673–682
18. AFNOR (2004) NF XP P 18–594, Aggregates -Test
methods on reactivity to alkalies
19. AFNOR (2004) NF P 18–454, Concrete—Reactivity of
a concrete formula with regard to the alkali-aggregate
reaction - Performance test
20. Larive C (1998) Apports combinés de
l’expérimentation et de la modélisation à la
compréhension de l’alcali-réaction et de ses effets
mécaniques, Laboratoire Central des Ponts et
Chaussées, OA28
21. Nixon PJ, Bollinghaus R (1985) The effect of AAR on
the tensile and compressive strength of concrete.
Durability building mater, 2:243–248
22. Boyd S, Bremmer T, Holt T (2000) Addition of
lightweight aggregates reduces expansion in concrete
containing a highly reactive normal weight aggregate.
In: Proceedings of the 11th International Conference
on AAR in concrete, Quebec, Canada, 593–602
23. Multon S (2004) Evaluation expérimentale et
théorique des effets mécaniques de l’alcali-réaction
sur des structures modèles. Laboratoire Central des
Ponts et Chaussées, OA46
24. Dunstan ER (1983) Parker dam concrete
investigations. Bureau of Reclamation Engineering
and Research Center, Denver
25. Moisson M (2005) Contribution à la maı̂trise de la
réaction alcali-silice par ajout de fines de granulats
réactifs dans le béton (in French). Ph.D. thesis, INSA,
Toulouse
83
26. Féret R (1892) Sur la compacité des mortiers
hydrauliques. Annale des Ponts et Chaussées, Série
7(4–5):161
27. Abrams DA (1918) Design of concrete mixtures.
Bulletin 1. Structural Materials Research Laboratory,
Lewis Institute, Chicago, 22 pp
28. Bolomey J (1927) Determination of the compressive
strength of mortars and concretes. Bulletin Technique
de la Suisse Romande, 16:22–24
29. Yamamoto T, Kanazu T (2000) Chemical assessment
method for the preventive effect of fly ash on alkali-
silica reaction. In: Proceedings of the 11th
International Conference on AAR in concrete,
Quebec, Canada, 801–810
30. Moisson M, Cyr M, Ringot E, Carles-Gibergues A
(2004) Efficiency of reactive aggregate powder in
controlling the expansion of concrete affected by
alkali-silica reaction (ASR). In: 12th International
Conference on AAR in concrete, Beijing, China,
617–624
31. Taylor HFW (2003) Cement Chemistry, 2nd edn.
Thomas Telford, p 363
32. Taylor HFW (1987) A method for predicting alkali ion
concentrations in cement pore solutions. Adv Cem Res
1(1):5–16
33. Bhatty MSY, Greening NR (1987) Some long time
studies of blended cements with emphasis on alkali-
aggregate reaction. In: Proceedings of the 7th
International Conference on AAR in concrete,
Ottawa, Canada, 85–92
34. Hong SY, Glasser FP (1999) Alkali binding in cement
pastes – Part I. The C-S-H phase. Cem Concr Res
29:1893–1903
35. Pedersen B, Mørtsell E, Jensen V (2004) Effects of
alkali-reactive crushed fillers on expansions due to
ASR. In: 12th International Conference on AAR in
concrete, Beijing, China, 764–772
36. Bokern J, Siebel E (2004) Alkali-silica reaction in
Germany – Transfer of laboratory results to practice.
In: 12th International Conference on AAR in
Concrete, Beijing, China, 490–498