See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/272790983

RPC mix optimization by determination of the minimum water

requirement of binary and polydisperse mixtures

Conference Paper · November 2005

CITATIONS READS

10 564

2 authors:

Andrzej Cwirzen Vesa Penttala

Luleå University of Technology Aalto University
121 PUBLICATIONS 2,614 CITATIONS 73 PUBLICATIONS 2,290 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Effects of Fly Ash on the properties of Alkali Activated Slag Concrete View project

Mechanochemical activation of slag View project

All content following this page was uploaded by Vesa Penttala on 28 February 2015.

The user has requested enhancement of the downloaded file.

 RPC MIX OPTIMIZATION BY DETERMINATION OF
THE MINIMUM WATER REQUIREMENT OF BINARY
AND POLYDISPERSE MIXTURES

A. Cwirzen, V. Penttala and C. Vornanen

Helsinki University of Technology

Laboratory of Building Materials Technology
P.O.BOX 2100, FIN-02915 HUT, Finland

ABSTRACT

Reactive powder concrete (RPC) is an ultra high performance material which development
was strongly associated with the maximization of the packing density. This paper presents an
alternative approach for the optimization of the mix design of RPC by using the
determination of the water demand of binary, ternary and polydisperse mixes.
Measurements of the water demand of quartz/sand and quartz/sand/cement/silica fume mixes
were performed. The control mixes - having water to binder ratio of 0.187, 25% of silica
fume and 5% of superplasticizer measured from cement content - were produced for the
assessment of fresh concrete properties and compressive strength. The concretes were heat-
treated at 90 °C. The results showed that maximum compressive strength was obtained in
concretes having binary quartz/sand mixes characterized by the highest packing density. The
best rheorogical properties were obtained in mixes, which had the highest packing density
when measured on polydisperse mixes containing all RPC components.

KEYWORDS
Reactive powder concrete, water demand, packing density, binary mixes, polydisperse mixes
INTRODUCTION

Reactive powder concrete (RPC) is an ultra high strength material which was developed in
80’s and 90’s, [1,2,3,5,12,15] The compressive strength can reach more than 200 MPa when
an appropriate curing procedure is applied. The innovation of RPC was a result of several
changes that were done in comparison to high performance concrete. The most important
were improvements of the packing density, more extensive usage of pozzolanic materials,
higher cement content and limitation of the maximum aggregate size to 600 μm to eliminate
the interfacial transition zone (ITZ) around larger aggregates.

The most commonly used pozzolanic materials are various types of silica fume e.g. gray,
zirconium or amorphous silica. Additional quartz fillers are used to improve the packing
density and to enhance pozzolanic reactions during heat-treatment. Recently, the
applicability of other fillers like for instance lime stone, metakaolin, pulverized fly ash or
micronized phonolith, was investigated [8]. In general the basic RPC mix consist of cement,
silica fume, fine sand, quartz and often short steel fibers [12].
The optimization of the RPC mix is a very complex task due to large amounts of cement,
silica fume and superplasticizers which result in a number of complicated chemical
interactions. These interactions can influence the properties in fresh state as well as hydration
products, being both important factors in defining the final mechanical properties of
concrete. The common approach in optimization of the concrete mix is to choose a right
grading curve. The studies on this subject were done in the beginning of the twentieth
century by Fuller and Thompson [9], who developed a concept of ideal grading curves. The
curves were based on the packing density of the particular materials. Later Weymouth
developed a model based on the theory of particle interference [14]. More recently
DeLarrard [5-7] introduced linear and compressible packing density models. Currently, there
are a number of research projects which aim to develop an universal model for mix
optimization by approximating the dry mix by 3-dimensional systems of spherical particles
[10]. Most of the models use an unrealistic spherical simplification of the particle shape and
also do not account for the interstitial liquid phase that fills the spaces between the particles.
In our opinion the lack of liquid phase in the model can have a considerable effect on the
prediction of the packing density. Therefore, the main objective in this research project was
to use an approach which takes water and the unregular shape of the particles into account.

As it is known, water contained in the fresh mix consists of the filling water and the water
that is adsorbed on the surfaces of the particles [4]. The first type fills the spaces between the
particles and does not contribute to the fluidity of the mix. The filling water is related to the
packing density of the system. The second type creates the water film on the particles and its
thickness determines the fluidity. An interesting and simple experimental method for
determination of the packing density was developed by Puntke [11]. The method allows the
approximation of the packing density by determination of the water demand of the “dry”
particle system. Unfortunately, the water demand that is determined consists of both filling
and adsorbed water. To obtain exclusively the filling water, an additional estimation of the
adsorbed water dependency on the particle size distribution curves would have to be done.
For the purpose of this research a simplification has been made that the water demand
determined by the Puntke method corresponds directly to the packing density.
The primary objective of this research was to determine the optimum dry mix composition
that would result in production of RPC characterised by the largest flow, shortest breaking
time and the highest compressive strength. The breaking time denotes the time measured
from addition of water to the dry mix until the moment when moistening occurs. In order to
diminish the amount of influencing factors the cement type and its amount, silica fume (SF)
type and its dosage, superplasticizer (SP) type and its dosage were constant in all of the test
mixes. The investigated variables were the total amount of fillers (including sand, quartz) to
the cement content, the type of fillers and their proportions.

MATERIALS

The reactive powder concretes were produced using sulfate resistant cement CEM I 42.5N
(SR), four types of quartz fillers, sand (160-600 µm) and densified gray silica fume type
Elkem 920. The content of polycarboxylate superplasticizer was constant for all of the mixes.
The particle size distribution curves were obtained by using the Culver type of the laser
particle size analyzer. The results are shown in Figure 1.

SAD

Figure 1; Particle size distribution curves

EXPERIMENTAL SETUP

The minimum water demand of the dry mix was determined by using a method based on the
approach presented by Puntke [11]. In this method the dry materials are mixed manually in a
steel vessel until uniform mix is obtained. Thereafter, water or water with SP is added and
mixed together with the dry material. In the next step the vessel is dropped repeatedly from a
constant height until visible moistening of the surface of the material is obtained. If such
wetting does not occur a small amount of water is being added and the whole procedure is
repeated. If the dry mix contains also cement and silica fume then the superplasticizer has to
be added. This method was used to determine the water demand for the basic binary mixtures
(sand and quartz) and polydisperse mixes containing additionally cement and silica fume.
For some of the polydisperse mixtures trial concretes were done using a 5l Hobart mixer.
Mix proportions are shown in Table 1.

Table 1: Mix proportions (ratios given according to the cement weight)

Cement Wate Super- Silica Quartz Sand W/B W/C

r plasticizer Fume ratio ratio
1 0,187 0,05 0,25 See Table 2 0,176 0,22

The concrete specimens with size 10x10x10 cm were cured as follows; de-molding after 24
hours, heat-treatment for 48 hours in 90oC, storage in 95% RH until the compression test.
The compressive strength was determined after 7 days for heat-treated samples and after 28
days for non heat-treated. The mix compositions are shown in Table 1 and Table 2. The flow
was measured by the “mini cone” method.

Table 2: Filler proportions, fresh concrete properties and mechanical properties

7day 28 day
air breaking
flow compr. compr.
mix description* content time
[cm] strength strength
[%] [min]
[MPa] [MPa]
1.0sand 4,2 25 2,8 177,1 138,6
0.8sand/0.2sad 4,7 25 4,3 181,3 132,6
0.6sand/0.4sad not det. 17,5 11,5 189,9 142,7
0.5sand/0.5sad not det. 10 32 172
0.8sand/0.2ehk not det. 25 10 198,4 150,1
0.6sand/0.4ehk not measured (mix did not get wet)
0.8sand/0.2kemiö 4,8 21 9 197,8 151,4
0.6sand/0.4kemiö 5,4 12,5 15,75 183,8 156,4
0.2sand/0.6kemiö not measured (mix did not get wet)
0.8sand/0.2nilsiä 3,4 23 4,5 172,1 141,7
0.6sand/0.4nilsiä 4,8 18,5 8 193,8 140,8
0.5sand/0.5nilsiä not det. 10 36 197
0.5sand/0.2nilsiä/0.3kemiö not det. 10 45 163
0.6sand/0.16nilsiä/0.24kemiö not det. 17 10,8 204,25 133,4
0.8sand/0.08nilsiä/0.12kemiö 4 23 5 185,1 140,2
0.6sand/0.32nilsiä/0.08kemiö not measured (mix did not get wet)
0.8sand/0.16nilsiä/0.04kemiö 6 24 5 184,4 131,8

*denotes sand to quartz filler ratio

TEST RESULTS

Figure 2 shows the water demand and calculated packing densities determined by the Puntke
method. The highest water demand was obtained for the finest materials; EHK quartz and
silica fume. This can be attributed to the larger surface area resulting in flocculation of the
particles and increased amount of adsorbed water. Lower water demand was obtained for
coarser fillers and sand for which - despite the increased amount of filling water and less
adsorbed water - a decrease in the total amount of water was observed, see also Chengzhi et
al. [4]. The calculated packing densities are obtained assuming that the measured water
demand corresponds only to the filling water, see Equation 1.

WT = WF + WA (1)

In which WT is water demand according to Puntke

WF is filling water, which determine packing density
WA is adsorbed water, which contributes to the flowability of concrete
25 0,80
w ater demand
packing density (Puntke)
20
water demand [g]

packing density
0,60

15
0,40
10
0,20
5

0 0,00
EHK
SAD

Nilsiä

SF
Kemiö

Sand
Cement

Figure 2: Water demand and packing density calculated according to Puntke method for the
basic materials

Figure 3 shows the water demand determined for different combinations of binary mixtures
consisting of sand and quartz and for water demand determined for polydisperse mix
containing all of RPC ingredients. The figure shows also flow test results and 7-day
compressive strength results of heat-treated concretes. The results revealed that in two out of
four binary mixes the optimum dosage of quartz, by means of the lowest water demand was
obtained by combination of 20% of quartz and 80% of sand. In the case of SAD and Nilsiä
quartz the optimum amount appeared to be between 30 and 40%. These two quartz fillers
were the coarsest and the particle size distribution curves were partly overlapping with sand,
Figure 1. Insufficient discrepancy of the diameters could cause nearly total interaction
between these materials. As a result, quartz could not fully fill the free spaces between the
sand particles and improvement of the packing density was less pronounced, also described
by DeLarrad [6]. On the contrary, much finer EHK and Kemiö quartz appeared to
significantly improve the packing density.
The flow in all cases was decreasing with higher quartz content, which can be explained by
the increased amount of water that was adsorbed on the quartz particles. The tests included
also the derivation of the lowest water uptake of ternary mixes consisting of two types of
quartz and sand. The results are shown also in Figure 3. It can be seen that ratios of
Nilsiä/Kemiö quartz 0.2/0.8 and 0.4/0.6 gave nearly the same water demand. This tendency
was similar when a “full mix” consisting of all of the ingredients was done. The test results
revealed also that compressive strength measured after 7 days for heat-treated concrete
specimens was increasing with decreasing water demand measured on binary and ternary
sand/quartz mixtures. However, flow depended on the water demand measured for the “full
mix”, (Figure 3). In case of EHK/sand and Nilsiä/sand combinations no moistening of the
dry mix was observed in 30 minutes after the addition of water with superplasticizer.

13 195 Kemiö quartz / sand

SAD quartz / sand 13 210
12 12
190 205
water demand [g]

11 11

water demand[g]

strength [MPa]
/ flow *2 [cm]

200

Compressive
flow *2 [cm]
10 10
185

strength [MP]
Compressive
9 195
9
8 190
8 180
7
185
7 6
175
6 5 180

5 170 4 175
0 0,2 0,4 0,6 0,8 1 0,0 0,2 0,4 0,6 0,8 1,0
filler content [%] filler content [%]

13 200 18 200
Nilsiä quartz EHK quartz / sand
12

Compressive strength
/ sand 195 16
11 195
strength [MPa]

water demand [g]

Compressive
[g]/ flow *2[cm]
water demand

14
/ flow *2 [cm]

10 190

[MPa]
9 12 190
185
8
10 185
7 180
6 8
175 180
5 6
4 170
4 175
0 0,2 0,4 0,6 0,8 1
0 0,2 0,4 0,6 0,8 1
filler content [%]
filler content [%] Compressive strength
quartz/sand mix
all components mix
flow

14 210
Nilsiä/Kemiö quartz
compressive strength
200 0.4kemiö/0.6nilsiä
12
compressive strength
water demand [g]

190 nilsiä/kemiö mix

10
[MPa]

180 nilsiä(0.8)+kemiö(0.2)
8 /sand
170
all componenets mix
6 0.8/0.2
160
nilsiä(0.6)+kemiö(0.4)
/sand
4 150
0 0,2 0,4 0,6 0,8 1 all components mix
filler ratio 0.6/0.4

Figure 3: Water demand of basic binary quartz/sand systems, “all components” mixes and 7-
day compressive strength of heat-treated concretes
ANALYSIS OF THE RESULTS

Water demand and properties of the fresh concrete

Analysis of the results revealed a fairly good correlation between the water demand
determined for the full mix (including all components) and the flow of the fresh mix. The
flow tended to be higher with lower water demand, (Figure 4). This tendency can be
explained by the amounts of absorbed and filling water. Mixes with higher packing density
by means of lower water demand had less filling water entrapped in voids between the
particles. As a result more available water could increase the layer of adsorbed water and
thus improve the workability.

The results revealed also a tendency of increase in the breaking time with higher water
demand. This can be also explained by the proportions of filling and adsorbed water. In the
case of lower water demand there is less entrapped filling water thus more water is available
during the wetting process. As a result wetting is faster and breaking time shorter.

A big advantage of the determination of the minimum water demand of the full mix is the
possibility to obtain a low water-to-binder ratio with the lowest amount of superplasticizer
needed. High dosage of SP, which is often required to achieve good workability of RPC, can
result in slower hydration and decrease of the final strength. This is attributed mainly to the
changes in the hydration products of C-S-H.

30 20

2
R = 0,7 18
25
16
breaking time [min]

14
20
12
flow [cm]

15 10

8
10
6

4
5
R2 = 0,3 2

0 0
5,5 6 6,5 7 7,5 8
flow
water demand [g]
breaking time
Figure 4; Correlation between water demand , flow and breaking time
Water demand and mechanical properties

The direct comparison between the water demand of polydisperse “dry” mixtures and 7 or
28-day compressive strength values revealed no statistical correlation. However, when
comparing the minimum water demand determined for the binary or ternary quartz/sand
mixtures estimated from Figure 3, a good correlation was found, (Figure 5). The 7-day
compressive strength of heat-treated concretes appeared to increase proportionally to the
decreasing water demand of the quartz/sand mix. In other words, the highest compressive
strength value after 7 days was obtained for those concretes, which had the optimum
proportions of quartz to sand by means of the lowest water demand. Keeping in mind that
proportions of the other materials were constant this correlation can be related to the
pozzolanic activity of quartz in high temperatures. In order to obtain the highest pozzolanic
efficiency of quartz, the material has to be added in right proportions. As described by Taylor
[13], too low amount of quartz causes the formation of a weak and porous α-C2S. However,
too high addition causes that not all quartz is consumed and as a results it may act as a
diluent thus causing a decrease in the compressive strength probably due to formation of
weak α-C2S. Addition of the right amount of fine quartz is proposed to cause portlandite to
be consumed and the formation of C-S-H having a low Ca/Si ratio.

Furthermore, higher compressive strengths were obtained for concretes produced with quartz
fillers having a mean diameter of the same or smaller in comparison with cement particle
size. This complies with earlier results described by Taylor [13].

205
Compressive strength [MPa]

200

195

190 Quartz fillers

185
3 4 5 6 7 8
Minimum water demand [g]
Figure 5: Relationship between the minimum water demand determined for the quartz/sand
dry mix and 7-day compressive strength for heat-treated concretes having different quartz
fillers

Optimization of the RPC mix

The optimization of the RPC mix according to the measurements of the water demand of
binary and polydisperse mixes can be divided into two parts. The first part includes
measurements of the minimum water demand (approximating the maximum packing density)
of binary or ternary sand/filler mixes. As a result an optimum proportion can be obtained that
would give the highest compressive strength. The second part includes measurements of the
minimum water demand of polydisperse mixes containing all of the ingredients. The
minimum value should correspond to the mix proportions giving the best rheorogical
properties. The final step of the optimization is to combine these two contradicting
tendencies; the highest compressive strength and low workability or the best workability but
lower compressive strength. The choice of the final mix depends on many factors like type of
superplasticizer, its compatibility with the cement, type of cement, type of the silica fume
etc. In the best case it should be possible to obtain a mix of acceptable workability for
proportions near to the optimum corresponding to the highest compressive strength. In the
worst case scenario it might be necessary to compromise the strength in order to sustain a
good workability.

CONCLUSIONS

The presented test results revealed that determination of the water demand of basic
binary/ternary and polydisperse “full” mixes can be used for the optimization of the RPC
mix. Specifically, the determination of the minimum water demand of basic binary or ternary
dry mixtures using the Puntke method can determine the optimum ratio of quartz to sand in
regard to its highest packing density and pozzolanic activity which results in the highest
compressive strength after the heat-treatment. The determination of the water demand for the
polydisperse mix containing all concrete ingredients allows optimization of the mix
proportions in regard to its workability.

ACKNOWLEDGMENT

This work being a part of a bigger project dealing with reactive powder concrete has received
financial support from TEKES (Technology Development Center of Finland), which is
gratefully acknowledged.

LITERATURE

1. Aïtcin, P.C. and Richard, P., (1996), The footbridge of Sherbrooke, In Proc. 4th Int.
Symposium on Utilization of HPC, Paris.
2. Bonneau, O, Vernet, C. and Moranville, (1998), Optimization of the rheorogical
behaviour of reactive powder concrete, Proc. 1st Int. Symp. On HPC and RPC,
Sherbrooke, Canada, Vol.3, 99-118
3. Bonneau, O., Lachemi, M., Dallaire, J., Dugat, P. and Aïtcin, P.C., (1997), Mechanical
properties and durability of two industrial reactive powder concretes, ACI Mater J94 (4),
286-290
4. Chengzhi, Z., Aiqin, W., Mingshu, T., (1996), The filling role of pozzolanic material,
Cement and Concrete Research, Vol.26, No.6, 943-947
5. De Larrard, F., (1988), Ultrafine particles for making of very high strength concrete,
Cement and Concrete Research, Vol.19, 161-172
6. De Larrard, F., (1999), Concrete mix proportioning- a scientific approach, E&FN SPON
 7. De Larrard, F., Sedran, T., (2002), Mixture proportioning of high-performance concrete,
Cement and Concrete Research, Vol. 32, 1699-1704
8. Droll, K., (2004), Influence of additions on ultrahigh performance concretes – grain size
optimization, Proc. Int. Symposium on Ultrahigh Performance Concrete, Kassel,
Germany, 285-301
9. Fuller, W.B. and Thompson, S.E., (1907), The laws of proportioning of concrete, Trans,
ASCE, 59, 67-143
10. Geisenhansluke, C., Schmidt, M., (2004), Methods for modeling and calculation of high
density packing density for cement and fillers in UHPC, Proc. Int. Symposium on
Ultrahigh Performance Concrete, Kassel, Germany, 303-312
11. Puntke, W., (2002), Wasseranspruch von feinen Kornhaufwerken, Beton (5/02), 242-
248
12. Richard, P. and Cheryrezy, M., (1995), Composition of reactive powder concrete,
Cement and Concrete Resarch, Vol.25, No.7, 1501-1511.
13. Taylor, H.F.W., (1997), Cement Chemistry, Reedwood Books, Trowbridge, 2nd Edition
14. Weymouth, C.A.G., (1933), Effect of particle interface in mortars and concrete, Rock
Prod. 36(2), 26-30
15. Zanni, H., Cheryrezy, M., Maret, V., Philippot, S. and Nieto, P., (1996), Investigation of
hydration and pozzolanic reaction in reactive powder concrete (RPC) using Si NMR,
Cement and Concrete Research, Vol.26, No.1, 93-100

View publication stats