Materials and Structures (2006) 39:189199

DOI 10.1617/s11527-005-9047-y

Multi scale experimental study of concrete rheology:

From water scale to gravel scale
Z. Toutou N. Roussel

Received: 14 February 2005 / Accepted: 29 July 2005

C RILEM 2006


Abstract Rheometric tests are carried out on cement

based suspensions. Four scales are defined for this
study: the saturating fluid scale, the cement paste
scale, the mortar scale and finally the concrete scale.
Each scale is considered as a suspension defined by
its maximum granular class size. Each scale is obtained by adding a higher granular size to the previous scale. For each obtained suspension, an adequate rheometric apparatus is used to study the rheological behaviour for adequate shear rates range.
The obtained experimental results are presented in
this paper and the influence of the inclusions volume
fraction on the rheological properties of the considered scale is investigated. In addition, a multi scale
analysis is carried out on the measured yield stress
values.
Resume Une serie dessais rheometriques a e te
realisee sur des suspensions cimentaires. Quatre
e chelles detude ont e te definies : lechelle du fluide saturant, lechelle de la pate de ciment, lechelle
du mortier et lechelle du beton. Cheque e chelle
est consideree comme une suspension definie par la
taille de son plus gros grain et est obtenue en rajoutant des inclusions de taille superieure a` lechelle
precedente. Le comportement rheologique de chacune
de ces suspensions a e te mesure a` laide dessais

Z. Toutou N. Roussel ()

LCPC Paris, France

rheologiques adequats. Les resultats obtenus sont

presentes ici et linfluence de la fraction volumique
solide en inclusions sur les proprietes rheologiques
est e tudiee. Une e tude multi-echelle portant sur les
valeurs de seuils mesurees a` toutes les e chelles conclut
larticle.

1. Introduction
From a physical point of view, fresh concrete can be
considered as a suspension of grains of various sizes
in a continuous fluid phase. When flowing, particles
interactions occur and modify the apparent rheological
behaviour. These interactions are particularly complex
in modern concretes as the physical or chemical phenomena generated by the increasing use of additives
are both badly defined and misunderstood. It can however be noted that recent progresses have been made in
the field of superplasticizers [1, 2]. These additions are
often used to improve the concrete casting conditions
(workability, self compaction, stability. . .). Empirical
tests (slump test, slump flow test, L-Box. . .) or proper
rheological apparatus (BTRheom [3], BML [4], Two
points test [5] and others) make it possible to qualify
a given mix fitting (consistency, flowability, pumpability, compactability. . .), or even measure some rheological parameters (yield stress, viscosity). However,
only few theoretical models able to describe, quantify
and predict rheological properties in terms of mix fitting can be found. The model developed by [6] is able

190

to predict the yield stress and plastic viscosity of a

given concrete in terms of numerous parameters describing each of the constituents. As it is based on a
global granular approach, each particle, even the smallest, plays a role in the final concrete packing density
and thus in the final concrete rheological behaviour.
This model is particularly suitable to traditional or high
performance concretes as their rheological behaviour
is strongly affected by direct grain contacts. However,
modern fluid concretes or even self-compacting concretes behave more like fluid suspensions. It is then
possible to imagine that they can, on a theoretical point
of view, be divided into a suspending fluid mixed with
more or less coarse inclusions. Such an approach could
simplify the work of the civil engineer. Indeed, most
of the complex phenomena already stated take place
at the cement paste scale. Let us imagine that we are
able to predict the rheological behaviour of the concrete knowing the rheological behaviour of the cement
paste and the distribution of the granular skeleton. By
testing the cement paste only, the civil engineer would
then be able to predict the fresh concrete rheology or
test new additives without the need for real scale tests.
This approach is of course not new [711], but it is only
becoming valid in the case of modern fluid concretes, in
which the physical phenomena linked to a suspension
behaviour dominate the complex grain to grain contacts. The international research community is doing
every day progress in the fresh concrete rheology field
but, at the same time, we have to feel lucky that the
problem is getting simpler as the concretes are industrially mutating from granular solids to non-Newtonian
fluids.
In order to contribute to a better understanding of
these rheological behaviours, we present here a detailed experimental study. It focuses on several observation scales constituting the fresh concrete. Thus, the
saturating fluid, the cement paste, the mortar and the
concrete constitute as many scales of analysis. An adequate experimental procedure is chosen for each scale
and rheometric tests are carried out. The experimental results allow the identification of the yield stress in
terms of the volume fraction of the studied scale inclusions (cement grains, sand or gravel). A multi scale
study ends the analysis of the results. It focuses on the
yield stress of a yield stress fluid mixed with granular
inclusions and gathers all the results obtained on the
various materials.

Materials and Structures (2006) 39:189199

2. Materials description
In this study, white cement, aggregates, ordinary water
and admixtures were used to prepare the concrete and
its subcomponents (cement paste and mortar).
2.1. Cement
White Cement CEM I/52.5 N CE CP2 NF SB from
Gargenville Calcia was used for all the tests. Its specific
gravity is 3.01. Its compressive strength is 62 MPa at
28 days according to NF EN 196-1 test. The size distribution was measured in water using a laser granulometer (according to NF ISO 13320-1 test) for different amount of superplasticizer (see 2.3.) and is
given in Fig. 1.The specific area determined using a
BLAINE permeameter, according to NF EN 196-6 test,
is 4117 cm2 /g. The cement chemical constituents are
summarized in Table 1.
2.2. Aggregates
The aggregates used in this study are from silica-quartz
alluvial deposits from Saint Christophe du Ligneron
Palvadeau sand pit. They are rather rounded in shape
and of white/yellow colour. Three sizes are used:
two fine sands 0/0.315 mm and 0.315/0.600 mm and
4/8 mm coarse gravel. Their specific density is 2.58 and
the water absorption is 0.50%. The dry packing density values are 0.616 for the 0/0.315 mm sand, 0.623
for the 0.315/0.600 mm sand and 0.605 for the 4/8 mm
coarse gravel. The grading curves, obtained by sieving,
are given in Fig. 2for the three used aggregates.

Fig. 1 Cement size distribution curve.

Materials and Structures (2006) 39:189199

Table 1 Cement chemical constituents
Constituents

% by mass

SiO2
Al2 O3
TIO2
Fe2 O3
CaO
MgO
Na2 O
K2 O
SO3
RI
PAF

20.95
4.08
0.14
0.22
65.55
0.49
0.12
0.20
2.60
1.47
3.36

Fig. 2 Gravel and sands granular distribution curves.

191

3. Mix fitting and mixing procedure

The combination of the materials described above allows the preparation of the following suspensions:
nano-silica suspension, cement paste, mortar and finally concrete. Each suspension is considered as a rheological study scale. The smallest one is the nano-silica
suspension scale and the largest one is the concrete
scale. Each scale is obtained by adding larger particles to a smaller scale. For instance, the mortar scale is
obtained by adding sand to the cement paste.
3.1. Concrete scale
The concrete batches were prepared using water,
white cement, 0/0.315 mm sand, 4/8 mm gravel, Nanometric silica slurry and superplasticizer optima 100.
Twenty litres concrete batches were prepared using
a Water/Cement weight ratio (W/C) equal to 0.35, a
solid Nano-Silica/Cement weight ratio (NS/C) equal
to 2%, a solid Superplasticizer/Cement weight ratio
(SP/C) equal to 0.7%, 0/0.315 mm sand volume fraction equal to 5% and gravel volume fractions G that
varied from 27% to 39%. The mix proportions of the
prepared batches are summarized in Table 2.The concrete was mixed using a SCHWELM ZK30E 30 l mixer.
The dry ingredients were first mixed for 2 minutes, then
the fluids (water, NS and SP) were added and the obtained suspension was mixed during 2 min at 70 rpm.
The mixer was then stopped to scrape its edges. A 3 min
mixing phase at the same 70-rpm rotation speed ended
the mixing procedure.
3.2. Mortar scale

2.3. Admixtures
Two admixtures are used in this study. The first one
is a superplasticizer, ChrysoFluid Optima 100. It is a
modified phosphonate based polymer. This superplasticizer was chosen because of its ability to maintain the
concrete rheology for 90 min. The second one is an anti
segregation and anti bleeding agent, Nanometric Silica
slurry RhoximatTM CS 60 SL. This product is amorphous silica in aqueous suspension. Its solid content
is 22.5% and its density is 1.14. This admixture was
used to prevent segregation form occurring during the
rheological measurements.

The mortar mix fitting is derived from the concrete

mix fitting. Two granular sizes sand were studied:
0/0.315 mm and 0.315/0.600 mm. The prepared mortar
suspensions sand volume fractions s vary from 5% to
30% for the fine sand (0/0.315 mm) and from 10% to
35% for the higher size (0.315/0.600 mm). The W/C,
NS/C and SP/C ratios are the same as the ones given in
the section 3.1. One-litre batches were prepared using
a planetary Hobart Kitchen mixer. As for concretes, the
dry ingredients were first mixed for 2 min at the lowest mixer rpm setting (260 rpm), then the fluids were
added and all were mixed for 2 min. The mixer ws then
stopped to scrape its edges. A higher rotation speed
(700 rpm) is applied for 15 min followed by a 15 min

192

Materials and Structures (2006) 39:189199

Table 2 Mix proportions of

concrete batches (kg/m3 )
ID

Gravel volume
fraction G

White
cement

Sand
0/0.315

Gravel
4/8

Water

Optima
100

Nano
Silica

#01
#02
#03
#04
#05
#06

27%
29%
31%
34%
35%
39%

1004
982
960
909
902
838

94
92
90
85
84
78

706
746
787
882
896
1016

266
260
254
241
239
222

23
23
22
21
21
20

89
87
85
81
80
74

final mixing phase at the lowest mixer rpm setting. This

chosen mixing procedure has two advantages: it ensures good particle dispersions in the fluid phase and
gives the superplasticizer enough time to act. On the
other hand, the authors are aware that the flocculation
state in the mortar scale is not the same when the mortar is part of the concrete or when it is a suspension
on its own as the mixing procedures differ. It will be
necessary to keep this in mind while comparing the two
scales rheological behaviour.
3.3. Cement suspension scale
The cement paste mix fitting is derived from the mortar
mix fitting. The NS/C and SP/C ratios are the same as
the ones given in section 3.1. The W/C weight ratio
varies from 25% to 55%. Thus, 11 cement suspensions
were prepared with cement volume fractions c from
38.5% to 55%. The mixing procedure is identical to the
one used for the mortar scale and described in section
3.2. Comparison between the cement paste scale and
the mortar scale rheological behaviours will be easier
than the one between mortar scale and concrete scale as
the mixing procedures are identical. However, it can be
expected that the presence of sand particles enhances
the mixing and that the state of flocculation of the cement paste is lower when it is mixed with sand than
when it is mixed on its own.
3.4. Nano silica suspension scale
This scales components are water, nano-silica slurry
(NS) and superplasticizer (SP) optima 100. This colloidal suspension scale is considered as the reference
one and as the cement paste saturating fluid. This assumption is licit because the NS maximum particle size
(14.109 m) is largely smaller than the cement minimum particle size (1.106 m). The prepared suspension

solid volume fractions N S vary from 1.7% to 3.4%.

The mixing procedure was carried out using the same
kitchen mixer as above. Fifty millilitres batches were
prepared at high rotating speed (500 rpm) during 3 min.

4. Apparatus description
Rheological measurements were carried out using two
rheometric apparatus according to the studied scales: a
Couette type viscometer RHEOLOGICA RM 500
for nano silica suspensions, cement pastes and mortars, and a plan-plan torsional concrete rheometer
BTRHEOM for concrete testing.
4.1. Couette type viscometer
It consists in two concentric cylinders covered with
sand paper. The inner cylinder is rotating at a controlled speed while the outer cylinder is fixed. The
resulting torque is measured. The cylinder gap was
chosen to be at least 5 times greater than the maximum particle size (Table 3).The nano-silica particle
suspensions were tested under controlled shear rates
from 0.1 s1 to 200 s1 and stress-controlled tests were
carried out on cement pastes and mortars. A large
stress range from 0.1 Pa to 350 Pa was used. This corresponded to shear rates range varying from 0.01 s1
to 400 s1 for the tested materials. For each tested formula, three tests were carried out using increasing and
decreasing shear stresses (or shear rates) ramps. Only
the average measurements obtained at decreasing shear
stresses (or shear rates) ramps are reported in this study.
This protocol was retained in order to get free of the
time-dependant behaviour (such as thixotropy) or all
handless risks (suspensions preparation and test sample selection). This means that all measurements began
at the highest structural breakdown state. Throughout

Materials and Structures (2006) 39:189199

193

Table 3 Used rheometer geometries

Suspension

Particle max
diameter. Dmax (mm)

Inner cylinder
diameter (mm)

Outer cylinder
diameter (mm)

Nano-silica suspension
Cement paste
Mortar 0/0.315
Mortar 0.315/0.6mm
Concrete (4/8)

14.106
0.130
0.315
0.600
8

23
23
23
23
Blades separation
(h = 100 mm)

25
27.1
27.1
29.5

these measurements, temperature in the viscometer was

maintained at 20 C.
When submitted to controlled increasing or decreasing shear rates, processes of structural breakdown and
structural rebuild occur in non Newtonian materials
such as cement pastes [12]. These processes of structural rebuild (under decreasing shear rates) may induce
a static zone inside the cylinder gap (i.e. plug flow).
Thus, the applied shear rate is not sufficient to maintain homogeneous flow in the entire gap. MRI measurements [13] confirm this phenomenon inside Couette
viscometer gap. In other words, the shear rate becomes
equal to zero on the outer fixed cylinder vicinity. This
existing dead zone inside the rheometer gap affect
the measured apparent flow curves so that a change of
the curve slope appears below what is called a critical
shear rate. This point is largely described in previous
works reported in [14] where the influence of the solid
volume fraction and the granular size on the critical
shear rate and stress is investigated. In order to get rid of
the measurement points, for which the entire gap is not
sheared, critical shear rate values are first determined
and analysis is only applied here on the significant
measurements.
4.1.1. Concrete rheometer BTRHEOM
The BTRHEOM rheometer is a parallel plate rheometer [3], for soft-to-fluid concrete with a maximum aggregate size up to 25 mm. A 7 l specimen of concrete
having the shape of a hollow cylinder is sheared between a fixed base and a rotating top section. The resulting torque is measured through the upper blades.
The rheometer rotation speed, vibration and the measurements (torque and rotation speed) were controlled
by a special program ADRHEO 13.1a, specially developed for this experimental campaign. The tests were
carried out at controlled rotation speeds varying be-

Gap (mm)

Gap/Dmax ratio

1.0
2.05
2.05
3.25
100

7.104
15.8
6.5
5.4
12.5

tween 0.314 rd/s and 7.85 rd/s. The maximum torque

that can be measured is about 15 N.m.
The test procedure using the BTRHEOM was
largely described in previous works [1517]. Two tests
were carried out on each tested formula using increasing and then decreasing controlled rotation speed
ramps. The final torque value is the average of the
increasing and decreasing torque measurements. The
tests were carried out at the ambient temperature of the
room (23 C).

5. Analysis
Steady state simple shear data for various cement
pastes, mortars or even concretes have often been well
represented using models following the general form
where is the shear stress and is the strain rate:
= 0 0

(1)

= 0 = 0 + f ( )

(2)

0 is the shear yield stress and f is a positive continuous

increasing function of the shear rate with f (0) = 0. It
should be noted here that, despite their effectiveness
in numerous cases and for different types of materials
(from cement paste to concrete) [1821], these models
do not take in account the thixotropic aspect of many
cementitious materials [1213]. It should also be noted
that the above models do not account for flow situations,
which are more complex than simple shear flow [22].
In literature some theoretical models can predict the
yield stress value in terms of the mix fitting parameters.
In the case of cement pastes, Legrand [23] proposes:
0 = aeb(0.5)

(3)

194

Materials and Structures (2006) 39:189199

where a and b are constants depending respectively on

the cement specific surface and the nature and granular
size of the cement particles.
In the case of concrete, De Larrard [6] developed
a powerful semi-empirical method for optimising the
mix design and predicting the yield stress of a given
concrete. It is based on a division of the granular composition in smaller subclasses i. The interactions between theses subclasses are then taken into account to
calculate the yield stress of the mixture using the following relation:
0 = 2.537 +

[0.736 0.216. log(di )].Ki



P 3
+ [0.224 + 0.910. 1 ].K
C
P

(4)

where K i
= 1
, i , i and di are respectively the
i
volume fraction, the maximum packing volume fraction and the size of particles of the class i, subscript c
refers to the cement, P and P refer to the super plasticizer dosage and saturation dosage respectively. However, this model is not adapted to the case of a concrete
containing more than one binder.
More recently, an analysis taking into account the interparticle forces that occur in super plasticized cement
paste allowed Flatt [9] to propose a theoretical model
predicting the cement paste yield stress as a function of
the particle size distribution, the solid volume fraction,
the maximum packing fraction and a volume percolation threshold:

0 = m1

( 0 )2
M (M )

(5)

Where m1 is a function of the particle size distribution and the interparticle forces, 0 the percolation
solid fraction and M the maximum packing solid fraction. This model includes various parameters linked to
the mix design and to physico-chemical interactions between particles that makes it a rather complete analysis.
It appears to correctly describe the volume fraction; particle size and size distribution scaling of cement pastes
yield stress. This approach is very promising but its
generalisation to the case of concrete needs more investigations and research.

Coussot [24] studied coarse particles mixtures and

related the mixture yield stress to the suspending fluid
yield stress.


m
M = F 1
M

(6)

where M is the mixture yield stress, F the suspending fluid yield stress, m is a coefficient. The author,
obtained a value m = 1 in the case of spread particles
size distribution and for particle volume fractions lower
than 0.6. This approach considers that there are no colloidal interactions between the inclusions and that their
size is far above the small particles constitutive of the
suspending fluid. This relation is similar to the Krieger
Dougherty relation [25], which deals with the apparent viscosity of the solution instead of the yield stress.
However, Eq. (6) only deals with yield stress fluids.

6. Experimental results
The rheological measurements gave us the shear stress
as a function of the shear rate for a given solid volume
concentration. In this section, using adequate fitting
methods, the yield stress is calculated for each studied
suspension scale.
6.1. Nano silica suspension scale
The obtained experimental results on the shear rate
range (0.1 s1 200 s1 ) are plotted on Fig. 3.Nanosilica suspensions display a yield stress despite their
low solid volume fraction. This property is the result
of the high colloidal silica particles attractive ability.
The value of this yield stress is extrapolated using a
Bingham model and is between 0.5 Pa and 3.5 Pa. The
Flatt model (see 5) describes correctly the nano-silica
suspensions yield stresses in terms of the solid volume
fraction as shown on Fig. 4.The fitted percolation value
is 0.6% and the maximum packing value is 29%, which
is surprisingly low (the maximum packing fraction for
non colloidal spheres is 0.74). However, these nanometric particles are far from being non-colloidal.
6.2. Cement paste scale
The studied cement pastes display a shear thickening behaviour for all the tested cement volume frac-

Materials and Structures (2006) 39:189199

195

Fig. 3 Recorded nano-silica suspension responses expressed as

shear stress vs. shear rate for various nano-silica volume fractions
(%).
Fig. 5 Imposed shear stress vs. measured shear rate for the cement pastes for various cement volume fractions (%).

Fig. 4 Nano silica suspensions yield stress fitted with the Flatt
model.

tions (Fig. 5).Although the tested mixtures are prepared

with two different particle species (cement particles
and nano-silica particles) with different inter-particles
forces involved, we chose however to try to use again
the Flatt mode (Fig. 6).It proved to be suitable to describe the measured evolution of the yield stress in
terms of the total solid volume fraction. This good correlation may be explained by the fact that the interparticle forces for the two types of particle are in fact
probably similar: the value of the fitted m parameter
is nearly the same (300 and 279) in the case of the

Fig. 6 Cement pastes yield stress vs solid volume fraction. The

Flatt model is also plotted.

nano-silica particles alone and in the case of the cement and nano-silica mixtures. However, it can be noted
that the fitted percolation threshold is largely higher in
the case of cement pastes. As it could have been expected, the value of the maximum packing fraction is
higher than the standard 0.74 value for perfect spheres.

196

Fig. 7 Imposed shear stress vs. measured shear rate for the mortars prepared with various 0/0.315 mm sand volume fractions
(%).

Materials and Structures (2006) 39:189199

Fig. 8 Imposed shear stress vs. measured shear rate for the mortars prepared with various 0.315/0.600 mm sand volume fractions
(%).

This is due to the fact that the small nano-silica particles fill the voids between the non-spherical cement
grains.
6.3. Mortar scale
The obtained experimental results are plotted on
Figs. 7and 8for the two tested sands. First, it can be
noted that the measured mortars behaviour as the cement pastes behaviour is shear thickening. The measured yield stress is an increasing function of the total
solid volume fraction. However, it has to be noted that
the yield stress of the obtained mortars for low sand
volume fraction is lower than the yield stress of the
associated cement paste. Although the ratio between
the cement particles size and sand grains size is high
enough for a multiphasic approach to be licit, this decrease in the yield stress is not predicted by any model
in the literature. For example, Coussots model [24]
always predicts a mixture yield stress higher than the
suspending fluid yield stress. This decrease in yield
stress could be explained by the fact that the flocculation state of the cement particles is lower when it is
mixed with sand grains that can break any forming
clusters. This is in agreement with the fact that, when
the sand particles size increases, the yield stress of the
mixture for the same sand volume fraction is lower as

Fig. 9 Mortars yield stress values vs. the sand volume fraction
for two sand granular sizes.

shown on Fig. 9.Thus, the thixotropic behaviour (i.e.

flocculation state of the cement paste) has an influence
on how the presence of sand particles increases (or not)
the yield stress of the obtained mortar. When the sand
volume fraction increases above a given value that de-

Materials and Structures (2006) 39:189199

Fig. 10 Measured torque vs. imposed rotation speed for the various concretes for varying gravel volume fraction (%).

pends on the sand particles size and probably on the

mixing procedure, the yield stress of the mortar becomes higher than the yield stress of the cement paste
as the thickening effect of the sand particles becomes
higher than their de-flocculating effect.
6.4. Concrete scale
A Bingham model, as shown on Fig. 10,may correctly
describe the behaviour of the obtained concretes. The
behaviour is not shear thickening as the other scales of
analysis. However, it has to be noted that the strain rate
range tested by the BTRHEOM is very small compared
to the range of the rheometer used in the previous sections. In such a narrow range, the Bingham model may
of course properly describe any behaviour.
The yield stress of the obtained concretes is far above
the yield stress of the associated mortar (Fig. 11).The
(possible) de-flocculation phenomenon spotted in the
case of mortars does not appear here probably because
the amount of gravel used to prepare the concretes is
high. Its effect consists mainly in increasing the yield
stress. In this case and as the sizes of the sand particles
and gravel particles are very different (the ratio is of
the order of 10) a multiphasic approach is licit. Thus,
we tried to apply the Coussots model by considering
that the mortar was the suspending fluid and the gravel
particles the considered inclusions. The obtained corre-

197

Fig. 11 Concrete yield stress in terms of the gravel volume fraction. The mortar yield stress and the prediction of the Coussots
model are also plotted.

lation is good (Fig. 11). The value of the fitted parameters are m = 4.2 and M = 0.6, which is the measured
dry packing density value of the gravels. However,
because of the discrepancy between rheometers obtained in [16, 17], the values of the parameters obtained
here may change if an other concrete rheometer was
used.

7. Multi-scale behaviour analysis

In order to carry out a multi-scale study, let us first
check that the ratio between the minimum size of the
inclusions and the maximum size of the grains constitutive of the so called fluid phase is high enough for
each of our studied scales. For the cement paste, this
ratio is equal to 70; for the mortar, this ratio is between
1 (0/0.315 mm sand) and 4 (0.315/0.600 mm sand); for
the concrete, it is equal to 10. Apart from the fine sand
(ratio = 1), there is in our studied scales a gap between
the particles sizes that allows us to use a multi-scale
analysis.
The shear stress vs. strain rate plots are gathered on
Fig. 12for all the studied scales of analysis. The various
strain rates range depend on the used apparatus while
the stress range really depends on the material and its

198

Materials and Structures (2006) 39:189199

Fig. 13 Reduced yield stress evolution in terms of inclusion

volume fraction.

Fig. 12 Shear stress versus shear rate for all the studied scales.

8. Conclusion

non-Newtonian behaviour. It has to be noted that, as the

size of the sand used in this study is rather small, the
behaviours of the mortar and the cement pastes are very
similar. Moreover, It may be worth precising that the
flow curve of the concrete is continuous because the
BTRHEOM only gave us the yield stress and plastic
viscosity of the tested material. We used these values
to plot the concrete flow curve in the appropriate strain
rate range.
Apart from the reduced yield stresses of the cement
pastes, all the reduced yield stresses seems to follow a
similar evolution in terms of the volume fraction of the
studied scale inclusions (Fig. 13).This type of evolution is analogous to an Einstein or Krieger-Dougherty
law for apparent viscosity. It only involves mechanical interactions of the inclusions within the suspending
fluid. In the case of cement pastes, the strong colloidal
properties of the particles prevent them from following
the Coussots model.
It has to be noted that the Krieger-Dougherty relation is suitable to describe cement pastes apparent viscosity in terms of the suspending fluid Newtonian viscosity and the inclusions volume fraction [25] whereas
it seems here that the same type of relations does not
apply to yield stress.

Rheometric tests were carried out on the saturating fluid

scale, the cement paste scale, the mortar scale and finally the concrete scale. For each scale, an adequate
rheometric apparatus was used to study the rheological
behaviour and measure the yield stress.
The obtained experimental results were presented
and described in this paper and the influence of the inclusions volume fraction on the rheological properties
of the considered scale was investigated.
The scales can be divided in two categories:colloidal
or non colloidal inclusions.
In the case of the colloidal inclusions, the interactions between the particle can not be neglected and a
multi scale analysis considering that the cement grains
are the inclusions in a nano-silica + water suspending fluid is not licit. However, in our experiments, the
Flatt model was able to describe well the measured cement pastes and nano-silica suspensions yield stresses
in terms of their solid volume fraction.
In the case of the non colloidal inclusions, two
regimes appeared in our results. In the first regime (low
inclusions volume fraction), the yield stress of the fluid
+ inclusions is lower than the yield stress of the fluid
alone. This can seem awkward at first but can be explained by the fact that, as cementitious material flocculation state has an influence on their rheological be-

Materials and Structures (2006) 39:189199

haviour, the presence of the inclusions may break this

flocculation during mixing or flowing and thus may
reduce the final yield stress of the mixture. In the second regime (high inclusions volume fraction), the yield
stress of the fluid + inclusions is higher than the yield
stress of the fluid alone and it increases with the inclusions volume fraction. The Coussots model is able to
predict this evolution.
Acknowledgements
We want to thank Pr. Christophe Lanos, Dr. Yannick Melinge,
Pr. Philippe Monnet and Dr. Robert Le Roy. Without their help
and cooperation, this work could not have been done. We wish
to thank Calcia for the white cement, ChrysoFluid for the superplasticizer and Rhodia for the nano-silica suspension. The
experimental results presented here are available to anyone interested.

References
1. Flatt RJ (2004) Dispersion forces in cement suspensions,
Cem Concr Res 34(3):399408.
2. Flatt RJ, Houst YF, Bowen P, Hofmann H, Widmer J, Sulser
U, Maeder U, Burge TA (1997) Interaction of superplasticizers with model powders in a highly alkaline medium, in Proc.
5th Canmet/ACI Int. Conf. On Superplasticizers and Others Chemical Admixtures in Concrete, Ed: Malhotra V.M.,
(Americain concrete Institute, Farmington Hills, Mi, USA,
SP-173) 743762.
3. Hu C, De Larrard F (1996) The rheology of fresh high performance concrete. Cem Concr Res 26(2):283294.
4. Mork JH (1996) A presentation of a B.M.L. viscometer.
Glasgow, RILEM 369376.
5. Tattersall GH (1990) Progress in measurement of workability by the two point test. Proc. of International Concrete
Conference on properties of fresh concrete. RILEM, 203
212.
6. De Larrard F (1999) Concrete mixture proportioning, E &
FN Spon, London.
7. Schwartzentruber A, Catherine C (2000) La methode du
mortier de beton e quivalent (MBE). Un nouvel outil daide
a` la formulation des betons adjuvantes. Method of the concrete equivalent mortar (CEM). A new tool to design concrete
containing admixture, Material and Structures/Materiaux et
Constructions, 33:475482.
8. Ferraris CF, Gaidis JM (1992) Connection between the rheology of concrete and rheology of cement paste. ACI Materials
Journal 88(4):388393.

199
9. Flatt RJ (2004) Towards a prediction of superplasticized concrete rheology. Materials and Structures 37:289300.
10. Farris RJ (1968) Prediction of the viscosity of multi modal
suspensions from viscosity dada. Transactions of the society
of the rheology 12(2):281301.
11. Bard Pedersen (2003) Sverre Smeplass, the relationship between the rheological properties if SCC and the corresponding matrix phase, Third International Symposium on SelfCompacting Concrete, Reykjavik, Iceland.
12. Roussel N (In press) Steady and transient flow behaviour of
fresh cement pastes. Cem Concr Res.
13. Jarny S, Roussel N, Le Roy R, Coussot P (2004) MRI measurements of steady state and transient behaviour of fresh
cement pastes, Annual Transactions of the Nordic Rheology
Conference, The Nordic Rheology Conference, Reykjavik,
Iceland 12:6369.
14. Toutou Z, Lanos C, Monnet P, Pantet A (2003) Rheometrie
des geosuspensions:de la pate de ciment au micro-beton.
Revue Rheologie 4:1831.
15. Sedran T (1999) rheologie et rheometrie des betons. Application aux betons auto-nivelants, th`ese ENPC, mars.
16. Ferraris CF, Brower LE, editors (2001) Comparison of
concrete rheometers: International tests at LCPC (Nantes,
France) in October, 2000. National Institute of Standards
and Technology Interagency Report (NISTIR) 6819.
17. Ferraris CF, Brower LE, editors (2004) Comparison of concrete rheometers: International tests at MB (Cleveland OH,
USA) in May, 2003. National Institute of Standards and
Technology Interagency Report (NISTIR) 7154.
18. Mansoutre S, Colombet P, Van Damme H (1999) Water retention and granular rheological behaviour of fresh C3 S paste as
function of concentration Cem Concr Res 29(9):14411453.
19. Cyr M, Legrand C, Mouret M (2000) Study of the shear
thickening effect of superplasticizers on the rheological behaviour of cement pastes containing or not mineral additives.
Cem Concr Res 30:14771483.
20. Ferraris CF, Brower LE (editors) (2001) Comparison of
concrete rheometers: International tests at LCPC (Nantes,
France) in october 2000, NISTIR 6819:1147.
21. Beaupre, D (1994) Rheology of high performance shotcrete,
phD thesis, University of Bristish Columbia.
22. Roussel N (2004) Three-dimensional numerical simulations
of slump tests, Annual Transactions of the Nordic Rheology
Conference, The Nordic Rheology Conference, Reykjavik,
Iceland 12:5162.
23. Legrand C (1982) La structure des suspensions de ciment,
in Le Beton hydraulique, Ed. de lENPC, Paris.
24. Ildefonse B, Allain C, Coussot C (1997) Des grands
e coulements naturels a` la dynamique du tas de sable.
Cemagref e dition.
25. Krieger IM, Dougherty TJ (1959) A mechanism for NonNewtonian flow in suspensions of rigid spheres. Transactions
of the Society of Rheology, III, 137152.