Cement and Concrete Research 34 (2004) 2237 – 2243

Why do HPC and SCC require a longer mixing time?

David Chopin, François de Larrard*, Bogdan Cazacliu
Laboratoire Central des Ponts et Chaussées-Centre de Nantes Route de Bouaye, BP 4129, Bouguenais Cedex, 44341, France
Received 17 March 2003; accepted 17 February 2004

Abstract

The industrial production of superplasticized concrete is slowed down because of the long mixing time often necessary for these materials.
The aim of the study presented here was to find out the mix-design peculiarities of high-performance concrete (HPC) and self-consolidating
concrete (SCC) that could be responsible for the long mixing time.
To compare the behaviour of various mixes, a mathematical model is proposed to fit the power consumption curve of the mixer and to
choose a criterion to determine the stabilisation time of the curve.
The parameters studied were the paste content, its components (silica fume, limestone filler, etc.), the high-range water reducing admixture
(HRWRA) type and dosage and the water/cement (w/c) ratio. The comparison of stabilisation time for 36 different HPC and SCC mixes allowed
to highlight the major influence of w/c ratio, on the stabilisation time. The higher the w/c ratio, the lower the stabilisation time. It also appears that
fine content, HRWRA dosage and the use of silica fume have significant effects on the time necessary to homogenise the material. This result can
also be expressed by using the actual and maximum solid content concepts, which are dependent on all the mix-design parameters.
These results were quantified, for our experimental conditions, through two models which are proposed in this paper. An empirical one
using the components’ amount and the other, the actual and maximum solid content of the concrete.
The validation of such a model with other sets of data obtained with different materials and mixers could lead to its use in mix-design
software programs; thus, it would become possible to take the mixing time into account at the laboratory step of mix design.
D 2004 Elsevier Ltd. All rights reserved.

Keywords: High-performance concrete; Mixing; Mixture-proportioning; Rheology; High-range water reducer

1. Introduction (HRWRA), silica fume or filler materials are used for

controlling the packing density of HPC and SCC. This
During the last 20 years, high-performance concrete paper presents the parameters that can explain the long
(HPC) and self-consolidating concrete (SCC) have become mixing time of new generation concretes by observing the
more and more commonly used. But the expansion of these stabilisation time of mixer power consumption for 40
new materials is slowed down by difficulties of industrial different mix designs.
production. The mixing time is one of the problems high-
lighted by ready-mix concrete firms. Indeed, mixing time
often reaches 2 min for SCC produced in concrete plants 2. Experimental program
and 5 min for some HPC, while, in France, ordinary
concrete is mixed for 35 s. The solution to this problem is To understand why HPC and SCC require a longer
researched by the authors within the mix-design peculiari- mixing time, it is necessary to study the influence on
ties of those concretes. Indeed, the components are not mixing time of the mix design. Most of these concretes
anymore the four basic materials (cement, water, sand and have a higher paste volume (30 –40%), a lower water/
coarse aggregate), but, to reach the mechanical and rheo- cement (w/c) ratio, use silica fume, and a high HRWRA
logical properties, high-range water reducing admixture dosage.
An experimental program was set up to study the
* Corresponding author. Tel.: +33-2-40-84-5638; fax: +33-2-40-84-
influence of these parameters on the mixing time. The
5992. selection of parameters came from industrial experience
E-mail address: larrard@lcpc.fr (F. de Larrard). and published literature [1,4].

0008-8846/$ – see front matter D 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cemconres.2004.02.012
2238 D. Chopin et al. / Cement and Concrete Research 34 (2004) 2237–2243

2.3. HRWRA type and content

Finally, the use of various HRWRAs at various dosages is

one of the peculiarities of SCC and HPC. Two different
molecules have been used in this program: polycarboxylic
type and melamine type. The first one is a more commonly
used molecule which acts by steric effect, while melamine
molecules, which have an electric repulsion effect, have been
used for a longer time. The content of HRWRA in this
experimental program is 40% or 100% of the saturation
dosage [3].
Fig. 1. Definition of the stabilisation time.
2.4. Granular skeleton
2.1. Fine particles content
The size distribution of aggregate (diameter over 80 Am)
The time necessary to reach a uniform dispersion of a is the same for all mixes. It has been determined by
granular material into a mix is assumed to be related to the optimisation of the packing density using the Compressible
mean diameter of the material considered [1]. This implicates Packing Model [2]. The gravel/sand ratio obtained for the
that the repartition of coarse aggregates is reached faster than maximum packing density is 1.18.
the repartition of sand or fine particles. Thus, as SCC and The coarse aggregates are crushed materials of two
HPC contain more fine particles than ordinary concrete, high granular fractions: 5– 12.5 and 10 –16 mm. The sand is a
mixing time should be necessary. The fine particles content, river natural sand with particles from 80 Am to 4 mm.
for this study, was fixed to either 425 or 550 kg/m3.
2.5. Mixtures and procedures
2.2. Paste composition
The 36 mixes designed were then batched on a 330-
The second parameter studied is the paste composition. l vertical orbital mixer set up in an experimental, full-scale
In SCC and HPC, supplementary cementitious materials concrete plant. During mixing, the power consumption of the
(SCMs) are often used. Therefore, the effect of silica fume mixer was recorded. All mixtures were designed to display a
and limestone filler have been studied. slump value of 20 F 4 cm for HPC mixes and a slump flow
The dosage of silica fume is set between 0% and 8% of of 60 F 5 cm for SCC mixes. Detailed mix design and
the mass of cement. Three different forms of silica fume properties of mixtures are summarised in Appendix A [5].
have been chosen:

(i) undensified, 3. Definition of stabilisation time

(ii) densified (by a pneumatic process) and
(iii) premixed and crushed with cement. For industrial production of concrete, the power consump-
tion curve versus time is often recorded to determine the time
The influence of filler material has been studied by for emptying the mixer. As a general rule, concrete is dis-
comparing SCC with 0% or 30% of limestone filler by charged when the stabilisation time [5,6] is reached. It is
mass of cement. defined as the time at which the power –time curve reaches

Fig. 2. Power consumption curve fitting and calculation of stabilisation time ts.
 D. Chopin et al. / Cement and Concrete Research 34 (2004) 2237–2243 2239

Fig. 3. Examples of power curves fitted with Eq. (1) (with two different concrete recipes).

the asymptote. For the present study, a visual determination when w/c ratio decreases within each group. We can notice
of stabilisation time lacks objectivity to analyse the concrete the plots fall into two groups depending on the fine element
behaviour. Therefore, the following method was adopted: the content of the mix. In these experiments, the highest fine
power– time curves is fitted with Eq. (1), and stabilisation content leads to the lowest stabilisation time. Moreover, it
time is taken as the time where the derivative of the power seems that HRWRA dosage slightly changes the slope of the
function (normalized by the constant Pl), is equal to a curve.
constant called e (see Fig. 1). The fitting equation is the By making the same analysis for mixes with melamine
following: HRWRA, the same tendencies are found, although less
      systematically (see Section 4.2).
t
td 2Pm ta  t p
PðtÞ ¼ Pl þ Pd e þ Arctan þ Another observation deals with the comparison of stabi-
p tm 2 lisation time for concretes with and without silica fume. It
ð1Þ appears that the mixing time is significantly reduced (by
where P(t) is the power consumed by the mixer, as a function 40%) for the mixes containing silica fume (Fig. 5). More-
of time t (see Fig. 2). Pl is the power consumed after over, if we compare the stabilisation time for the different
stabilisation of the curve, Pd and Pm are adjusted coefficients, types of silica fume, no significant difference appears
td, ta and tm are time coefficients between undensified and densified, while blended silica
   fume leads to slightly higher stabilisation times.
 B Pðts Þ  Also in Fig. 5, the influence of the HRWRA type can be
ts so that ¼e ð2Þ
Bt Pl  studied by comparing mean values of stabilisation time for
mixes designed with melamine and polycarboxylic types.
where ts is the stabilisation time of power curve. e is fixed to
Concretes designed with polycarboxylic HRWRA display
get a minimum value of 35 s for ts among the mixtures
considered. As a matter of fact, this time is the minimum
mixing time required by French standard NF P 18-305.
Two examples are given in Fig. 3 to show how the
power– time curve can be fitted by Eq. (1).

4. Results and discussions

Using the method highlighted here, we can examine the

influence of various mix-design parameters on ts. An
attempt is then made to develop a relationship between ts
and the mix-design parameters using two models.

4.1. Stabilisation time of power consumption curve versus

mix design
Fig. 4. Relationship between stabilisation time of power consumption and
Fig. 4 shows the stabilisation time ts as a function of w/c
w/c ratio, for mixes with polycarboxylic HRWRA. HPC, high-performance
ratio, for several groups of mixes containing polycarboxylic concrete; SCC, self-consolidating concrete; 425 or 550, fine element
HRWRA and having the same binder content and workabil- content; PC, polycarboxylic; 0.4 and 1, superplastcizer amount in terms of
ity. The first observation is that stabilisation time increases saturation degree.
2240 D. Chopin et al. / Cement and Concrete Research 34 (2004) 2237–2243

Table 1
Values of coefficients used in Eq. (3) to fit experimental stabilisation time
results
HRWRA Polycarboxylic Melamin
T0 392 110
aw  2.09  0.91
ac 0.351 0.325
asp  9.82  3.48
Mean error (s) 14.54 17.57
Correlation 0.92 0.68
Mean of aE (s)  454  169
Mean of acC (s) 165 165
Mean of aspSP (s)  10  14

content (kg/m3); SP is the HRWRA content (kg/m3); T0, aw,

Fig. 5. Influence of silica fume on mixing time of concrete. ac, asp are the coefficients of the model.
After an optimization consisting in minimizing the
slightly longer stabilisation times than those which are mean error of the model, the results obtained are dis-
designed with melamine HRWRA. But it is unclear whether played in Fig. 6 and Table 1. The model is quite
this result is due to differences in the chemical nature of satisfactory for polycarboxylic HRWRA but not really
HRWRA, or to the concrete rheological behaviour. Hence, satisfying for melamine. Therefore, it seems necessary to
because the comparisons between both natures of HRWRA find an explanation of the effects of the different param-
are made at fixed water/binder ratio, consistency differs from eters to have a more accurate model, valid for both
a product to the other one. This difference in consistency acts HRWRA types.
on the level of power consumed, which is accounted for in the Most of the differences between experimental and pre-
calculation of stabilisation time (see Eq. (2)). dicted stabilisation time values can be explained by the lack
of precision when weighing constituents on a full-scale
4.2. Analysing influences by a linear model plant, but this cannot explain everything. This means that
the parameters considered in this model are not the only
The relationship we have observed between stabilisation ones that can affect the mixing time. Thus, it is necessary to
time, w/c ratio and fine content leads to try to separate the look for the physical parameters of the concrete which are
effects of water and fine content. Moreover, the small behind the changes of mix-design parameters.
difference due to HRWRA dosage is to be taken into
account. The following linear model is proposed to under- 4.3. Stabilisation time of power versus packing parameters
stand the influence of each parameter:

ts ¼ T0 þ aw W þ ac C þ asp SP ð3Þ de Larrard [2] has shown that packing concepts can be
used successfully to model various fresh concrete properties.
where ts is the stabilisation time of power consumption Among those concepts, the actual solid content / is the sum
curve (s); W is the water content (l/m3); C is the cement

Fig. 6. Comparison between predicted (Eq. (3)) and experimental values of Fig. 7. Comparison between predicted (Eq. (4)) and experimental values of
stabilisation time. stabilisation time.
 D. Chopin et al. / Cement and Concrete Research 34 (2004) 2237–2243 2241

of all solid per unit volume, whereas the maximum solid 5. Conclusions
content / * is the maximum solid concentration which could
be obtained with a compaction index equal to 9 for given To understand why HPC and SCC are so long to mix,
proportion of dry constituents. The compaction index an experimental plan has been set up to study the
describes the compaction energy used to place the granular influence of mix-design parameters, through 36 different
mixture. mixes. The parameters that have been studied are the fine
Using these definitions, a relation exists between mix- element content, the use of limestone filler or silica fume
ture composition and the packing parameters of concrete. (in various forms) and the HRWRA type and dosage. To
For instance, an increase in water content, for given analyse the influence of those parameters on mixing time,
proportions of other constituents, will lead to a decrease it has been necessary to define a stabilisation time of
of actual solid content while the maximum solid content power consumption which is recorded during mixing. An
remains constant. The relative solid concentration of con- accurate empirical model is then proposed to fit the power
crete /// * will then be reduced. In the same way, curves.
replacing part of the cement by silica fume increases the A first linear, empirical model has been proposed,
maximum solid content, but does not change the actual directly accounting for the mix-design ratios. Further-
solid content (if the replacement is performed on a volume more, it has been established that the mixing time of a
basis). This leads to a decrease of the relative solid given mix is linked to its actual and its maximum solid
concentration. content. Those two concepts, defined by de Larrard [2],
Thus, the influence of the various parameters we have are related to mix design of concrete. An experimental
studied on stabilisation time can be related to the concrete model, based on those conclusions, is proposed to eval-
packing parameters following Eq. (4): uate the stabilisation time of a concrete, for a given
mixer. Such a model could be used in a mix-design
software such as BetonLabPro 2 [7], which would
/
Ts ¼ Tm þ A   ð4Þ provide a prediction of mixing time at the mix-design
Sp
/*  Bi step.
Sp* In practice, the effect of packing density can be con-
trolled by knowing the effect of some mix design data on
where Ts is the stabilisation time; / is the actual solid this parameter. The mixing time can then be reduced by the
content; / * is the maximum solid content; Sp/Sp* is the following means:
relative HRWRA dosage (Sp is the actual dosage and Sp* is
the saturation dosage; Ref. [3]); Bi is a coefficient function – increasing the fine particle content, with a constant w/c
of the HRWRA type, expressing the difference in fine ratio,
elements dispersion obtained with different molecules; Tm – increasing the total water amount (the compressive
and A are constants. strength will be modified),
Fig. 7 and Table 2 show the comparison between actual – optimising the granular skeleton will lead to a higher
and theoretical data, after a numerical optimization. maximum solid content, for a constant total aggregate
Comparing the two models reveals that both have content and
comparable precisions, but the latter seems more phys- – replacing part of the cement by silica fume will have
ical. Furthermore, as it has been assessed that the /// * the same effect (reduced packing density). The use of
ratio is linked to plastic viscosity of fresh concrete [2], different forms of silica fume (undensified, densified
it could therefore be deduced that mixing time is likely or premixed) can also have a small effect on mixing
to increase with this rheological parameter. This time.
explains why HPC, which is known to display higher
values of plastic viscosity, often requires long mixing It would be interesting to check the validity of this
times. semiempirical model with the help of another independent
set of data, dealing with different materials and mixer. In case
of success, a further step towards scientific concrete mix
Table 2
design would have been made through the research related in
Values of coefficients used in Eq. (4) to fit experimental stabilisation time this article.
results
Coefficient Value
Tm (s)  920.8 Acknowledgements
A 1062.3
BPC 0.04 The collaboration in this project of ‘‘Béton de France/
BML 0 RMC’’ and ‘‘Liebherr France’’ companies is gratefully
Mean error (s) 16.90
acknowledged.
 2242
Appendix A. Mix design and stabilisation time for the 40 mixes

Crushed Natural OPC Limestone Silica HRWRA Total w/c Yield Plastic Slump Actual Maximum Stabilisation
aggregate, sand, CEM, filler fume (kg/m3) water stress viscosity (cm) solid solid time (s)
5/16 mm 0/4 mm 52.5 R (kg/m3) (kg/m3) (kg/m3) (Pa) (Pa s) content content
(kg/m3) (kg/m3) (kg/m3)
HPC425-PC-0,4 901 766 430 0 0 2.6 206 0.47 385 23 23.5 0.7916 0.849 89.7
HPC425-PC-1 935 795 429 0 0 6.4 177 0.41 397 76 24.5 0.8212 0.856 177.3
HPC425-ND-PC-0,4 868 738 396 0 34 2.4 225 0.6 533 9 20 0.7725 0.862 63.62
HPC425-DP-PC-0,4 868 738 396 0 34 2.4 225 0.6 480 1 20 0.7725 0.862 50.06
HPC425-ND-PC-1 887 755 396 0 34 5.9 207 0.5 407 5 20 0.7905 0.869 103.3

D. Chopin et al. / Cement and Concrete Research 34 (2004) 2237–2243

HPC425-DP-PC-1 887 755 396 0 34 5.9 207 0.5 603 10 20 0.7905 0.869 81.35
HPC425-CJ-PC-0,4 868 738 396 0 34 2.4 225 0.6 284 4 25 0.7730 0.8542 54.24
HPC425-CJ-PC-1 912 776 394 0 34 5.9 187 0.5 424 11 21 0.8107 0.87 104.3
HPC550-CJ-PC0,4 782 666 512 0 45 3.1 244 0.5 291 11 26 0.7541 0.84 60.49
HPC550-CJ-PC-1 829 705 511 0 44 7.7 205 0.4 492 43 23 0.7928 0.852 102.2
HPC550-PC-0,4 814 692 558 0 0 3.3 230 0.4 162 18 25 0.7680 0.824 64.67
HPC550-PC-1 860 732 556 0 0 8.3 190 0.4 830 99 17 0.8074 0.834 196.1
HPC550-ND-PC-0,4 776 661 513 0 45 3.1 250 0.5 588 5 20 0.7475 0.84 51.11
HPC550-DP-PC-0,4 776 661 513 0 45 3.1 250 0.5 801 1 21 0.7476 0.84 47.98
HPC550-ND-PC-1 810 689 512 0 45 7.7 221 0.43 618 6 20 0.7764 0.851 88.66
HPC550-DP-PC-1 810 689 512 0 45 7.7 221 0.43 676 11 19 0.7763 0.851 76.14
SCC550-PC 813 691 558 0 0 8.4 227 0.4 467 16 – 0.7712 0.8295 94.91
SCC550-F-PC 809 689 391 168 0 5.9 223 0.6 127 30 – 0.7754 0.8435 135.6
SCC425-AV-PC 888 756 431 0 0 6.5 213 0.5 366 19 – 0.7853 0.853 109.5
SCC425-F-AV-PC 901 166 302 129 0 4.5 199 0.7 304 25 – 0.8000 0.864 97
HPC425-PC-1-S 927 788 429 0 0 6.4 182 0.4 401 108 21 0.8157 0.856 136.6
HPC425-PC-1-S2 1122 628 435 0 0 6.5 169 0.4 1679 236 9 0.8292 0.856 140.8
HPC425-ML-0,4 913 776 429 0 0 6.7 192 0.5 326 52 23 0.8056 0.861 62.58
HPC425-ML-1 933 793 427 0 0 17 168 0.4 592 60 20 0.8289 0.871 112.6
HPC425-ND-ML-0,4 904 769 395 0 34 6.2 195 0.5 554 32 20 0.8024 0.868 58.41
HPC425-DP-ML-0,4 904 769 395 0 34 6.2 195 0.5 550 24 20 0.8024 0.868 47.98
HPC425-ND-ML-1 934 795 393 0 34 15 163 0.4 534 60 20 0.8341 0.881 56.32
HPC425-DP-ML-1 934 795 393 0 34 15 163 0.4 530 58 19 0.8340 0.881 56.32
HPC550-ML-0,4 838 714 556 0 0 8.7 206 0.37 406 52 23 0.7915 0.84 95.96
HPC550-ML-1 860 732 552 0 0 21.6 176 0.34 671 95 20 0.8207 0.854 109.5
HPC550-ND-ML-0,4 823 700 511 0 44 8 211 0.4 487 17 23 0.7865 0.851 70.92
HPC550-DP-ML-0,4 823 700 511 0 44 8 211 0.4 508 14 21 0.7864 0.851 75.1
HPC550-ND-ML-1 865 736 508 0 44 20 167 0.4 644 46 20 0.8300 0.869 62.58
HPC550-DP-ML-1 865 736 508 0 44 20 167 0.4 1057 80 15 0.8296 0.8695 133.5
SCC550-ML 821 699 554 0 0 22 206 0.4 197 18 – 0.7911 0.851 69.88
SCC550-F-ML 838 713 388 166 0 15 191 0.5 269 56 – 0.8068 0.864 81.35
SCC425-AV-ML 911 775 428 0 0 17 185 0.5 85 24 – 0.8125 0.87 75.1
SCC425-F-AV-ML 924 786 300 129 0 12 174 0.6 227 43 – 0.8240 0.879 79.27
HPC425-ML-1-S 1138 637 429 0 0 17 149 0.4 1913 215 9 0.8489 0.87 113.7
HPC425-ML-1-S2 1123 629 430 0 0 17 160 0.4 410 170 21 0.8383 0.87 90.74
 D. Chopin et al. / Cement and Concrete Research 34 (2004) 2237–2243 2243

References [5] D. Chopin, Malaxage des bétons à hautes performances et des bétons
autoplacßants. Optimisation du temps de fabrication (Mixing of High
[1] A. Johansson, The relationship between mixing time and type of con- Performance Concrete and SelfConsolidating Concrete—Optimisation
crete mixer, Swedish Cement and Concrete Research Institute, Proceed- of mixing time ), PhD thesis of Ecole Centrale de Nantes, January, 2002
ings HANDLINGAR, Stockholm. (in French). Etudes et Recherches des Laboratoires des Ponts et Chaus-
sées, OA 41, 2003, edited by LCPC, 58 bd Lefebvre, 75732 PARIS
[2] F. de Larrard, Concrete mixture proportionning—A scientific ap-
proach, Modern Concrete Technology, vol. 9, E&FN Spon, London, CEDEX 15.
1999 (441 pp.). [6] D. Chopin, B. Cazacliu, F. de Larrard, Monitoring of concrete homog-
[3] F. de Larrard, F. Bosc, C. Catherine, F. Deflorenne, The AFREM meth- enisation with the power consumption curve, submitted to Nordic Con-
crete Research.
od for the mix design of High Performance Concrete, Materials and
structures, vol. 30, No. 201, RILEM, Paris, 1997 (August – September), [7] T. Sedran, F. de Larrard, BetonlabPro 2, ComputerAided MixDe-
pp. 439 – 446. sign Software, Presses de l’Ecole Nationale des Ponts et Chaussées,
28 rue des Saints-Pères, 75007 PARIS, France, 2000. Internet ad-
[4] P.O. Vandanjon, F. de Larrard, B. Dehousse, G. Villain, R. Maillot,
dress: http://www.lcpc.fr/fr/presentation/organigramme/div_tgce/result/
P. Laplante, Homogenisation of concrete in a batch plant: Influence of
mixing time and method on the introduction of mineral admixtures, betonlabpro.php.
Mag. Concr. Res. 55 (2) (2003 April) 105 – 116.