Combined effect of ne y ash and packing density on the properties

of high performance concrete: An experimental approach

Mohamed Abd Elrahman
a,b,
, Bernd Hillemeier
a
a
Institute of Civil Engineering, TU Berlin, Germany
b
Structural Engineering Department, Mansoura University, Egypt
h i g h l i g h t s
HPC with superior properties has been optimized using 312 kg/m
3
binder.
Fine FA is more effective than normal FA on improving concrete properties.
Total porosity is reduced to 3% by combination of ne FA and SF.
The optimized HPC has a chloride diffusion coefcient of 1.4 10
13
m
2
/s.
a r t i c l e i n f o
Article history:
Received 16 October 2013
Received in revised form 5 February 2014
Accepted 8 February 2014
Available online 13 March 2014
Keywords:
Packing density
Ideal Fuller curve
Fine y ash
Porosity
Durability
a b s t r a c t
High performance concrete often contains large amount of cement which makes ecological, economical
and technical problems. This study provides a new approach to optimize high performance concrete with
low cementitious materials content. The ideal grading curve according to Fuller has been used in concrete
mix design to ensure high packing density of concrete mixtures and to reduce the required binder con-
tent. Several systems comprising various pozzolanic materials (silica fume, y ash and ne y ash) have
been prepared and tested. The role of ne y ash on concrete performance has been estimated by mea-
suring the concrete mechanical properties, porosity and durability. The mechanical properties were
assessed from compressive strength and modulus of elasticity, whilst the durability characteristics were
investigated in terms of water permeability, water absorption and chloride diffusion. The results showed
that ne y ash performed better than normal y ash for the strength development and durability
aspects. The ternary system containing slag cement, ne y ash and silica fume with low w/b ratio per-
formed the best amongst all the systems regarding concrete mechanical properties and durability. Com-
bination of ne y ash and silica fume with OPC or with slag cement resulted in a signicant reduction in
concrete porosity. All mixes containing ne y ash exhibited high performance concrete with excellent
durability properties.
2014 Elsevier Ltd. All rights reserved.
1. Introduction
Nowadays, progress in science and technology in the eld of
construction industries and usage of new materials have resulted
in the use of reinforced concrete in special structures such as
sewage systems, nuclear power containments, cooling towers of
power plants, high way bridges and tunnels. In such aggressive
environments, high durability, stability and resistance to chemical
attack are of more concern. High performance concrete (HPC)
provides an attractive option for such conditions. According to
Mehta and Monteiro [1], three main characteristics make the
concrete with high performance: high workability, high strength
and high durability. Because of the various requirements including
workability, durability, strength and dimensional stability, the mix
design of HPC is a challenging task. Unlike normal concrete, it is no
longer sufcient to base the mix composition on the principle of
compressive strength and w/c ratio relationship. HPC is produced
using carefully selected ingredients and low water/binder (w/b) ra-
tio. In addition, it requires high amount of cementitious materials
(400550 kg/m
3
) and also high dosage of superplasticizer [2].
However, the use of high amount of cement can lead to environ-
mental, economic and technical problems. For example, using high
cement content increases the hydration heat and shrinkage which
are critical issues for concrete [3]. To cope with these problems,
http://dx.doi.org/10.1016/j.conbuildmat.2014.02.024
0950-0618/ 2014 Elsevier Ltd. All rights reserved.

Corresponding author at: Institute of Civil Engineering, TU Berlin, Germany.

Tel.: +49 30 314 72 109.
E-mail address: mohattia76@gmail.com (M. Abd Elrahman).
Construction and Building Materials 58 (2014) 225233
Contents lists available at ScienceDirect
Construction and Building Materials
j our nal homepage: www. el sevi er . com/ l ocat e/ conbui l dmat
more optimization is needed to minimize the cement content in
order to reduce the ecological and economic impact of HPC. One
possibility of optimizing the HPC mixture is the selection of con-
crete constituents in such a way that the packing density of the
whole granulometric assemblage is maximized. Basically, increas-
ing the packing density of aggregate would decrease the volume of
paste needed to ll up the voids and increase the amount of addi-
tional paste that could be utilized to improve the workability.
Additionally, the concrete will have less durability problems such
as permeability, shrinkage, and thermal degradation.
2. Particle packing optimization
The packing density can be dened as the ratio of volume frac-
tion occupied by the solids to the volume of the surrounding con-
tainer. It is a matter of interest in many elds of material science
such as packed beds, ceramics, asphalts and concrete [4]. The opti-
mum packing density of the system could be attained only if
spheres with smaller sizes are added to the assemblage. The small
size spheres can ll the voids between the large spheres, and there-
by the packing density is signicantly increased. In 1961, McGeary
reported that it is possible to achieve a packing density of 95.1
using four sizes of spheres with diameter ratios of 1, 7, 38 and
316 with fraction volumes of 6.1%, 10.2%, 23% and 60.7% respec-
tively. However, the maximum density of innite differences in
sizes can be attained is 97.5% [5]. For concrete, the situation is
more complex since the system is composed of various particle
sizes with different shapes and sizes. Effective packing can be at-
tained by selecting proper proportions and sizes of small particles
to ll in the voids between the bigger particles. The important ef-
fects of aggregate grading on the properties of concrete have been
emphasized in very early reports [6]. In 1892, Feret concluded that
the maximum strength can be attained when the voids in the mix-
ture is minimum [7]. Fuller reported that the best grading curve of
aggregate to get the maximum density is a parabolic shape [8].
Both Feret and Fuller conrmed that concrete properties can be
signicantly improved by using continuous grading [7]. In 1923,
Talbot developed the well-known equation [9]:
P d=D
q
1
where P is the total percent passing through a sieve, d is the diam-
eter of the current sieve, D is the maximum aggregate size and q is
the gradation ratio. The maximum packing density can be achieved
when q = 0.5, which is close to the Fuller curve [10,11], but the
resulting concrete is harsh and unsuitable. In 1930, Andreasen tried
to improve the Fuller curve. He suggested using the exponent q in
the range of 0.330.5, because ne particles are not able to pack
similar to bigger particles [12].
3. The work of Fuller and Thompson
Fuller and Thompson studied the grading analysis for a wide
variety of aggregate types and mixtures to achieve the maximum
packing density. They found that the best density of aggregate
can be attained when the particle size distribution of aggregate is
continuously graded and the grading curve takes a parabolic shape
(Fig. 1). This curve is known as Fuller parabola and can be applied
for calculating the optimum grading of aggregate only (not for a
mixture of aggregate and binder) as Fuller mentioned later [8]. This
is because the mixture of aggregate which gives the maximum
density in the dry state does not necessarily achieve the greatest
density when combined with cement and water. The very low void
content between the aggregate prevent the cement and water to t
in perfectly [8]. In addition, Fuller parabola leads to low powder
content, whereas, more nes are needed to maintain good cohe-
sion and to prevent segregation.
In 1903, Fuller and Thompson began an intensive work to
achieve the greatest packing density for mixtures of aggregate
and ne materials with maximum size of 2.25 inch. The ideal grad-
ing curve has been obtained by trial mixes without referring to
mathematical basis. To get this curve, at least 7% of the solid mate-
rials should be ner than the No. 200 sieve. It composed basically
of an ellipse at the lower part and merging into a straight line tan-
gent to the elliptical part [11]. The ellipse begins from 0.0029 inch
(sieve No. 200) and runs to a value of x equals to one-tenth the
maximum grain size (Fig. 1). At this point, the straight line begins
and continues to y = 100% and x = D (where D is the maximum
grain size). After nding the ideal curve, equation was tted to this
curve. The equation covering this ideal grading curve is divided
into two parts:
For the elliptical part:
y 7
2
b
2

x a
2
a
2
1 2
For the straight line part:
y
100 y
1
D x
1
x
0
x
1
y
1
3
where a and b are the axis of the ellipse and their values depend
mainly on the shape of the particles and the maximum aggregate
size [11], x
0
= D/10 to D, y
1
= y of the ellipse at D/10 and x
1
= D/10
[13].
In 1992, Puntke reviewed Fuller work and redraw the ideal
curve in a semi-logarithmic scale for the sake of simplicity
(Fig. 2) [13]. This curve has been used for designing concrete mixes
for several applications, particularly those, which need high den-
sity and high resistance to acid attack. For example, in 2000, the
highest cooling tower in the world (200 m, Niederauem, Ger-
many) has been constructed of acid resistant concrete. The used
concrete has been designed on the basis of the ideal Fuller curve.
By applying this concrete in the cooling tower, the tank did not
need any internal protective layer (as normal) because the used
concrete has high density as well as high resistance to acid attack
[14,15].
4. Mix design
In this investigation, the concrete mixture proportioning is
based on the granular optimization of all concrete constituents
according to the ideal Fuller curve. The maximum grain size of
coarse aggregate was 16 mm. For this size and according to the
ideal Fuller curve (Fig. 2), the required aggregate volume
(d > 125 lm) is 85.13%, while the binder volume (d < 125 lm) is
14.87%. According to this calculation, the required amount of
cementitious materials and aggregates are 312 and 1984 kg/m
3
respectively. On the other hand, to obtain a good size distribution,
the skeleton of aggregate size fractions should be viewed as a
whole rather than two separate entities; coarse and ne aggregate.
Aggregate as they come from the quarry do not normally have size
distributions that t the dense packing curve as can be seen in
Fig. 2. There are some deviations between the non-optimized mix-
ture and the ideal Fuller curve. The mixture has more materials in
the range of 0.52 mm. Therefore, sieve analysis of aggregate is
necessary to be done and the required amount of each size is taken.
On the contrary, the mixture has low ne materials in the size
range 0.0630.25 mm. Thus, quartz sand and quartz powder are
added to ll this gap and to densify the matrix.
226 M. Abd Elrahman, B. Hillemeier / Construction and Building Materials 58 (2014) 225233
5. Experimental details
5.1. Materials
Natural quartzite aggregates with maximum size of 16 mm has been used in
this investigation. CEM I 32.5 R and CEM III B 32.5 N-LH/HS/NA with slag content
of about 68% according to DIN EN 197 have been implemented. Normal y ash
(FA), ne y ash (M20) according to DIN EN 450-1, and silica fume (SF) comply-
ing with DIN EN 13263-1 have been also used as cement supplementary materi-
als. Quartz sand (QS) and quartz powder (QP) were used as correcting aggregate
to adjust the mixture with ideal grading curve. Fig. 3 shows the particle size dis-
tribution (PSD) of all nes measured with laser diffraction spectrometry. Physical
properties and chemical composition of the used materials are shown in Table 1.
On the other hand, it is theoretically known that, capillary porosity begin to form
at water/binder ratio higher than 0.42. Therefore, two values are chosen to re-
duce the probability of capillary pores formation: 0.42 and 0.27. The effectiveness
factor (k factor) is assumed to be 0.4 and 1 for y ash and silica fume respectively
(K factor represents the efciency of supplementary materials compared to OPC.
The quantity of the SCM in the mixture can be multiplied by the k value to esti-
mate the equivalent cement content, which can be added to the existing cement
0
10
20
30
40
50
60
70
80
90
100
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
P
a
s
s
i
n
g

%

(
V
o
l
u
m
e
)
Size (d/D)
Fuller Parabola
Ideal Fuller curve
D/10
b
a
7
Fig. 1. Comparison between Fuller parabola for best grading of aggregate and the ideal Fuller curve for grading of aggregate and ne materials.
Fig. 2. The grading of non-optimized mixture, Fuller parabola and the ideal Fuller curve in semi logarithmic scale.
Fig. 3. Particle size distribution of ne materials measured with laser granulometry.
M. Abd Elrahman, B. Hillemeier / Construction and Building Materials 58 (2014) 225233 227
content for the determination of the water-to-cement ratio [16,17]). In addition,
a polycarboxylate-based superplasticizer complies with DIN EN 934-2 has been
used to ensure a desirable consistency (class F3F4 according to DIN EN 206-1).
5.2. Mix composition and mixing
In order to optimize a high performance concrete mixture, several supplemen-
tary materials have been implemented. In this investigation, in order to achieve
dense and homogeneous cement matrix, it is suggested to use CEM III/B. One con-
trol mix incorporating OPC only and eight other mixes containing supplementary
materials were prepared. To estimate the inuence of ne y ash on concrete prop-
erties, normal y ash with the same replacement level (30%) has been investigated.
In addition, silica fume with 10% replacement level has been used. Likewise, to en-
hance the packing density of ne materials which can be achieved by using a wide
range of particle sizes, combination of 25% ne y ash with 10% silica fume with
OPC as well as with slag cement have been implemented. To evaluate the perfor-
mance of ne y ash at low w/b ratio (0.27), two mixes were prepared; the rst
with 30% of ne y ash and the second with combination of 25% ne y ash with
8% silica fume. The details of concrete mixtures proportioning are given in Table 2.
The concrete was mixed in 60 liters mixer. After mixing, fresh concrete properties
were measured, and the moulds were cast and compacted according to EN
12390-2. After 24 h of adding the water, the test specimens were demoulded. Then,
the curing process took place in water basin at room temperature (20 1 C).
5.3. Determination of concrete properties
5.3.1. Mechanical properties
The compressive strength has been determined using 100 mm cubes at ages of
28 and 91 days according to EN 12390-3. In addition, the modulus of elasticity of
concrete is determined on cylinders (150 mm 300 mm) according to European
standard EN 1048-5 at age of 91 days.
5.3.2. Concrete porosity
The concrete porosity and pore size distribution have been measured using the
mercury intrusion porosimetry (MIP) at age of 91 days. The principle of this tech-
nique is based on penetrating a non-reacting liquid such as mercury, into the pores
of a dried and evacuated porous medium. The Washburn equation is used to calcu-
late the porosity. The relationship between the pore size and the exerted pressure is
expressed as:
r
2c cos h
p
m
4
where r is the radius of the intruded pore (nm), c is surface tension of mercury (N/
m), p
m
is the applied pressure (bar), and h is contact angle between mercury and the
pore walls. Details of measurements procedures can be found in [18].
5.3.3. Durability
In this investigation in order to assess the durability of concrete, permeability,
absorption and diffusion tests have been made. The water penetration depth under
pressure according to EN 12390-8 was used to measure the permeability. The water
absorption coefcient has been measured according to EN ISO 15148. In addition,
the chloride diffusion coefcient was measured according to BAW procedure [19].
The following equation was used to determine the diffusion coefcient:
D
cl

RTh
zFU

x
d
a
d

x
d
p
t
5
With
a
d
2

RTh
zFU
erf
1
1
2c
d
c
o

s
6
where D
cl
is the migration coefcient (m
2
/s), Z is valency, for chloride ions Z = 1, F is
Faraday constant, F = 9.649 10
4
J (V* mol)
1
, U is absolute potential difference (V),
R is gas constant, R = 8.315 J (K* mol)
1
, T is absolute mean temperature of the solu-
tions during the test (K), h is height of the test specimen (m), x
d
is mean penetration
depth of the chloride ions in each half of the test specimen (m), t is duration of the
test (s), erf
1
is inverse error function, c
d
is chloride concentration at which the col-
our changes, c
d
= 0.07 (mol/L ).
6. Results
6.1. Mechanical properties
6.1.1. Compressive strength
Table 3 shows the results of compressive strength of concrete
mixes at age of 28 and 91 days. It is clear that, the main factor,
which controls the compressive strength, is the w/b ratio. Mixes
8 and 9 with w/b ratio of 0.27 had the highest compressive
strength at both 28 days and 91 days. It was 82 MPa for mix 8 at
28 days and increased to 99 MPa after 91 days. In addition, the ce-
ment type has also an important role on the development of
strength especially at early ages. At 28 days, the compressive
strength of mix 1 with OPC was 62 MPa, however for mix 3 with
slag cement it was 52 MPa. Nevertheless, at 91 days both mixes
exhibited similar strengths (70 2 MPa). On the other hand, the
addition of pozzolanic materials has an important inuence on
concrete compressive strength; depending on its type, content
and properties. Combination of pozzolanic materials with Portland
cement enhances the strength of concrete more than with slag
cement as can observed from the results of mixes 2 and 7 respec-
tively. In addition, the strength improvement is more pronounced
Table 1
Physical and chemical properties of the used ne materials.
Material CaO SiO
2
Al
2
O
3
Fe
2
O
3
MgO Na
2
O K
2
O SO
3
Cl Specic density Surface area (cm
2
/g)
CEM I 63.25 20.8 4.61 2.59 4.17 0.16 0.50 2.70 3.17 3450
CEM III B 48 29.7 6.5 1.2 8 0.31 0.65 2.49 0.02 2.96 4156
Fly ash 3.1 49.2 27.6 7.6 2.1 0.9 5 5.0 2.29 2877
M20 5 52 25 6 1 1.5 1.9 1.1 2.45 6000
Silica fume 0.2 98.4 0.2 0.01 0.10 0.15 0.20 0.10 0.01 2.2 20,000
Quartz powder 98.5 0.7 0.06 2.69 2683
Quartz sand 0.013 99.6 0.11 0.012 0.004 0.006 0.024 2.67 760
Table 2
Composition of concrete mixes.
Mix Cement FA M20 SF SP Aggregate QS QP w/b Flow diameter Air content Fresh density
Type (%) (%) (%) (%) (kg/m
3
) (kg/m
3
) (kg/m
3
) (kg/m
3
) ratio (cm) (%) (t/m
3
)
1 CEM I 100 2.8 1854 46 84 0.42 52 1.5 2.45
2 CEM I 65 25 10 3 1854 46 84 0.42 46 1 2.46
3 CEM III 100 2.2 1854 46 84 0.42 53 1.5 2.43
4 CEM III 70 30 3.3 1854 46 84 0.42 52 1.3 2.45
5 CEM III 70 30 3 1854 46 84 0.42 51 1.3 2.44
6 CEM III 90 10 2.5 1854 46 84 0.42 48 2.3 2.42
7 CEM III 65 25 10 3 1854 46 84 0.42 50 2 2.44
8 CEM III 67 25 8 8.8 1947 48 88 0.27 44 1.5 2.50
9 CEM III 70 30 6.6 1947 48 88 0.27 48 1.4 2.49
228 M. Abd Elrahman, B. Hillemeier / Construction and Building Materials 58 (2014) 225233
in mixes with silica fume (mix 2 and 6). However, the strength
development of mix 4 with normal y ash is slow compared to
other mixes, particularly at early ages.
6.1.2. Modulus of elasticity
The experimental results of elastic modulus of concrete are
found in Table 3. All mixes have modulus of elasticity more than
40,000 MPa. Mix 8 with low w/b ratio has the highest elasticity
of about 51,000 MPa. Mix 1 and mix 3 showed roughly the same
elasticity, that mean slag cement and OPC have the same effect
on elasticity [20]. On the other hand, the elastic modulus of mixes
with silica fume is higher than mixes without silica fume. Mix 6
with silica fume showed the highest elasticity at w/b of 0.42, it
was 45,200 MPa. However, mix 4 with normal y ash showed the
lowest elastic modulus, it was 40,200 MPa.
6.2. Porosity measured with MIP
Table 4 presents the experimental results of concrete total
porosity at age of 91 days. Compared to literature, all mixes have
lower porosity than traditional concrete which lies between 15%
and 20% [21]. Mix 1 with OPC has the highest porosity (10%), while
the use of slag cement reduces the porosity to about 8.9% (mix 3).
However, replacing OPC by 25% ne y ash and 10% silica fume as
in mix 2 dramatically reduces the porosity to 6.2%. Moreover,
mixes 8 and 9 with low w/b ratio exhibited porosity in the range
of UHPC (about 46%) [21]. The total porosities of mix 8 and mix
9 were 3% and 4.9% respectively.
On the other hand, regarding durability and deterioration of
concrete, capillary porosity is of most concern. Most of the trans-
port mechanisms take place via the capillary pore framework.
Therefore, capillary porosity of concrete mixes has been calculated
from the data sheet of MIP measurements. It is assumed to be in
the range of 30 nm to 10 lm [21]. Table 4 shows also the capillary
porosity of different concrete mixes. Similar to total porosity re-
sults, mix 1 with w/b of 0.42 has the highest capillary porosity
(5.6%). At w/b ratio of 0.42, the capillary porosity of mix 2 was
4.5%. Mixes 8 and 9 with w/b ratio of 0.27 have very low capillary
porosity of 1.9% and 2.96% respectively. Fig. 4 presents the pore
size distribution of concrete mixes. Mix 1 has the largest volume
of pores in the size range of 0.010.1 lm. It is clear from the gure
also that combined addition of ne y ash and silica fume along
with the use of low w/b ratio (mix 8) resulted in large reduction
in pore sizes, particularly in the capillarity range.
6.3. Durability
6.3.1. Water penetration depth (permeability)
Table 5 illustrates the results of water penetration depth tests.
All mixes exhibited low permeability (depth < 20 mm). The results
revealed also that the main factor governs the permeability is the
w/b ratio. The penetration depth of mix 8 with w/b ratio of 0.27
was about 3 mm which is the lowest penetration depth, while
for mix 9 it was 5 mm. In addition, cementitious materials have
crucial inuences on pore volume and connectivity which control
the water permeation. Mix 5 with ne y ash has lower penetra-
tion than mix 4 with normal y ash; it was 8 and 18 mm respec-
tively. At the same time, mix 5 has lower depth than mixes 6
and 7 with silica fume. Mix 3 with slag cement showed higher
resistance to penetration than mix 1 with OPC (15 and 18 mm
respectively).
6.3.2. Absorption (capillary suction)
Table 5 shows the results of the absorption coefcient of differ-
ent concrete mixes. Mix 8 made with slag cement, ne y ash and
with w/b ratio of 0.27 has the lowest absorption coefcient of
about 0.08 kg/m
2
h
0.5
. However, for mix 1 with w/b ratio of 0.42
and with OPC, the water absorption coefcient was 0.32 kg/
m
2
h
0.5
, which is about 4 times higher than that of mix 8. All con-
crete mixes with slag cement have lower absorption than mix 1.
For all these mixes, the absorption coefcient ranges between 0.1
to 0.2 kg/m
2
h
0.5
. Compared to mix 1 with OPC, replacing a part
of OPC with 25% of ne y ash and 10% silica fume (mix 2) reduced
the absorption to about 0.1 kg/m
2
h
0.5
.
6.3.3. Chloride diffusion
Table 5 shows the experimental results of chloride diffusion
coefcient of concrete. Mixes with slag cement exhibited higher
resistance to chloride diffusion than mix 1 with OPC alone. How-
ever, mix 2 with partial replacement of OPC with silica fume and
ne y ash showed signicant reduction in the diffusion coef-
cient. It reduced from 28 10
13
m
2
/s for mix 1 to
4.5 10
13
m
2
/s for mix 2. At w/b ratio of 0.42, mix 5 with ne
y ash exhibited higher resistance to chloride diffusion than mix
6 with silica fume alone and higher than mix 7 with both silica
fume and ne y ash. Furthermore, due to incorporation of ne
y ash (mix 5), the chloride diffusion is reduced to around the half
of the value of mix 3 with slag cement only. As expected, reducing
the w/b ratio resulted in high resistance to chloride diffusion. With
w/b ratio of 0.27, combination of silica fume and ne y ash (mix
8) resulted in a very high reduction in chloride diffusion to
1.39 10
13
m
2
/s compared to mix 1 with w/b of 0.42 and OPC,
which is about 20 times less. On the other hand, contrary to the re-
sults of mixes with w/b ratio of 0.42, combination of silica fume
with ne y ash is more efcient in reducing the chloride diffusion
than ne y ash alone at w/b ratio of 0.27. Proof for that are the re-
sults of mixes 5, 7, 8 and 9. Similar conclusion has been empha-
sized by Bentz, who reported that silica fume is more efcient
for reducing diffusivity in low w/c ratio concretes [22].
Table 3
Compressive strength and modulus of elasticity of concrete mixes.
Mix Compressive strength (MPa) Modulus of elasticity (MPa)
28 days 91 days
Mix 1 62 72 43,100
Mix 2 78 82 44,500
Mix 3 52 68 43,500
Mix 4 56 64 40,200
Mix 5 68 72 42,900
Mix 6 66 78 45,200
Mix 7 66 71 40,800
Mix 8 82 99 50,200
Mix 9 80 91 46,900
Table 4
Total and capillary porosity of concrete mixes measured at the age of 91 days using MIP.
Mix Mix 1 Mix 2 Mix 3 Mix 4 Mix 5 Mix 6 Mix 7 Mix 8 Mix 9
Total porosity (%) 10.30 6.20 8.90 7.80 6.98 7.20 6.50 3 4.90
Capillary porosity
a
(%) 5.60 4.50 4.50 3.89 3.50 3.90 3.80 1.90 2.96
a
The capillary porosity is assumed to be in the range of 30 nm to 10 lm.
M. Abd Elrahman, B. Hillemeier / Construction and Building Materials 58 (2014) 225233 229
Fig. 4. Pore size distribution of concrete mixes measured at age of 91 days using MIP.
Table 5
Durability properties of concrete measured at age of 91 days.
Mix Water penetration depth (mm) Water absorption coefcient (kg/m
2
h
0.5
) Chloride diffusion coefcient 10
13
(m
2
/s)
Mix 1 18 0.32 28
Mix 2 14 0.095 4.5
Mix 3 15 0.178 9
Mix 4 18 0.132 8
Mix 5 8 0.105 3.81
Mix 6 11 0.148 6
Mix 7 13 0.15 5
Mix 8 3 0.074 1.39
Mix 9 5 0.088 4.33
230 M. Abd Elrahman, B. Hillemeier / Construction and Building Materials 58 (2014) 225233
7. Discussion
7.1. Mechanical properties
By having a close look to the experimental results, it can be ob-
served that the compressive strength and modulus of elasticity of
all mixes are higher than 60 MPa and 40 GPa respectively at
91 days with total cementitious materials of about 312 kg/m
3
.
However, for traditional concrete without optimization, the com-
pressive strength and modulus of elasticity are in the range of
about 40 MPa [21] and 25 GPA respectively. This signicant
enhancement in mechanical properties of concrete can be directly
attributed to the high packing density of solid particle system due
to applying ideal Fuller curve. On the other hand, the addition of
supplementary materials has important inuences on concrete
mechanical properties. The pozzolanic reaction turns CH into a
new pozzolanic CSH gel which tightens and strengthens the
aggregate-paste interface, therefore modulus of elasticity and late
strength are markedly improved. In addition, supplementary mate-
rials enhance the packing density which results in better compact-
ness, thus the strength increases [23]. These effects are more
noticeable with using ne y ash and silica fume.
In fact, it should be pointed out here that, at different w/b ratios
the behaviour of ne materials is somewhat different. At w/b ratio
of 0.42, the comparison of mix 5 and 7 showed that the compres-
sive strength and modulus of elasticity of mix 5 (72 and
42,900 MPa) with ne y ash was a little bit higher than mix 7
(71 and 40,800 MPa) with ne y ash and silica fume. While, at
low w/b ratio (0.27), the situation is totally reversed. The addition
of both silica fume and ne y ash (mix 8) enhanced the mechan-
ical properties (99 and 50,200 MPa) more than mix 9 with ne y
ash (91 and 46,900 MPa). On the other hand, although the cement
content is low (can be said lean mixes), the mechanical properties
of mixes 8 and 9 are in the range of high strength concrete which
needs more cement, more silica fume and more superplasticizer
per cubic meter. This can be reasoned by sequential lling of the
voids in the system. The voids between each particle class are lled
with particles from smaller class. Especially at the micro scale, the
gap between the cement particles and silica fume with average
diameter of about 30 lm and 0.5 lm respectively are lled with
ne y ash with diameter of 110 lm. As a consequence, the bond
between aggregate and cement paste is increased which improves
the mechanical properties as the experimental results revealed.
7.2. Porosity
In order to evaluate the experimental results, a small compari-
son with total and capillary porosity of concrete as reported in lit-
erature will be presented. It is stated that the total porosity of
normal concrete, HPC, and UHPC are 15%, 8.3% and 6% respectively,
while the capillary porosity (0.0310 lm) are 8.3%, 5.2% and 1.5%
by volume respectively [21]. The experimental results indicated
that the maximum total porosity was 10% for mix 1. In addition,
the capillary porosity was 5.6% which is lower than porosities of
conventional concrete as reported in the literature. These results
signicantly manifest the inuence of enhancing the packing den-
sity on concrete porosity. The porosity of concrete is resulted from
the porosity of cement paste only, however the voids inside the
aggregate do not participate in the porosity of concrete. By maxi-
mizing the packing density of aggregate, the volume of cement
paste (cementitious materials and water) required to ll the gaps
between the aggregate is reduced to about 23%. In this investiga-
tion, the cement content is about 312 kg/m
3
and the water content
is about 131 liters (w/c = 0.42). So, the total volume of cement and
water (cement paste) equals about 230 liters for one cubic meter of
concrete (23%). However, normal high performance concrete needs
at least 400 kg cement and about 168 liters of water (w/c = 0.42).
The total volume of cement paste (cement and water) is about
300 liters for one cubic meter of concrete (about 30%). So, if the ce-
ment pastes of both concrete have the same porosity, then the total
porosity of the densely packed concrete will be lower by 23.33%
because it has less cement paste.
The use of supplementary materials largely reduces the porosity
of concrete depending on their type and content at w/b ratio of
0.42. The use of slag cement, with higher neness than OPC, re-
duces the total and capillary porosity to about 9% and 4.5% respec-
tively (mix 3). More reduction of porosities was attained by the
addition of normal y ash (mix 4). The total and capillary porosities
were reduced to about 7.8% and 3.9% respectively. Mix 5 with ne
y ash showed lower porosity; about 7% and 3.5% for total and cap-
illary porosity respectively, which are smaller than the aforemen-
tioned porosities of HPC according to Teichmann [21]. This can
be attributed to the efcient packing of ne y ash which has small
size (<10 lm), spherical shape and smooth texture. On the other
hand, the addition of silica fume to the system of slag cement
and ne y ash showed a small effect at w/b ratio of 0.42. At the
same direction, mix 2 made with OPC, ne y ash and silica fume
exhibited about 6.3% and 4.5% for total and capillary porosity
respectively. At low w/b ratio, the porosity is signicantly reduced.
For mix 9 with w/b ratio of 0.27 and with ne y ash, the total and
capillary porosities were 4.9% and 2.9% respectively. More reduc-
tion was gained by the addition of the silica fume to the system.
The total and capillary porosity of mix 8 was about 1.9% 3% respec-
tively which are much lower than normal HPC [24]. These values
are comparable to those of UHPC which needs more nes, cement,
llers and superplasticizer and special mixing procedures.
7.3. Durability
The incorporation of ne y ash or both ne y ash and silica
fume with low w/b ratio result in an effective reduction in the
transport of contaminants into concrete. The results of sorption
and chloride diffusion conrm this inuence. The water absorption
coefcient of mix 9 was about 0.09 kg/m
2
h
0.5
. However, combina-
tion of ne y ash and silica fume leaded to more reduction in
water absorption coefcient to 0.07 kg/m
2
h
0.5
. Similar results
were obtained for the chloride diffusion which is considered the
most important factor that can be used to evaluate the concrete
durability. The chloride diffusion coefcient of mix 1 without
pozzolanic materials was about 28 10
13
m
2
/s. However, the
use of slag cement (mix 3) reduced the diffusion to around
9.3 10
13
m
2
/s. The use of ne y ash at the same w/b ratio of
0.42 (mix 5) reduced the diffusion to about 3.8 10
13
m
2
/s. How-
ever, mix 8 with w/b ratio of 0.27 showed the lowest chloride dif-
fusion (about 1.4 10
13
m
2
/s) which is in the range of UHPC [21].
The role of high packing density in reducing the chloride diffusion
coefcient can be elucidated by comparing the result of mix 5 with
literature [25]. In spite of both concretes made with ne y ash, the
chloride diffusion of mix 5 is 3.8 10
13
m
2
/s which is 5 times
lower than that of Brandburger which was 1.5 10
12
m
2
/s. Fur-
ther, by the use of silica fume it is reported that the diffusion coef-
cient reduced to 1.5 10
12
which is half the resistance of mix 7.
The transport of liquids and gases takes place through the con-
tinuous pore system inside concrete. So, enhancing the packing
density reduces the penetration of contaminates into concrete.
The increased volume of aggregate closes the transport passes
and makes it longer. The experimental results of water penetration
test clarify the important role of the increased packing density in
transport mechanisms. All concrete mixes exhibited water pene-
tration depth lower than 20 mm which can be classied as imper-
meable concrete under aggressive conditions (<30 mm) according
M. Abd Elrahman, B. Hillemeier / Construction and Building Materials 58 (2014) 225233 231
to Neville [26]. Because of its higher neness, the use of slag ce-
ment results in lower permeability compared to OPC concrete
[20]. This is attributed to the dense microstructure and the low
CH concentration in the transition zone in slag cement concrete.
The addition of both ne y ash and silica fume notably reduces
the permeability and diffusivity [27]. The main effect of ne mate-
rials is the enhancement of transition zone properties and making
it denser. This zone is known as the locus of micro-cracks which
inuences not only the mechanical properties but also the perme-
ability and durability. Due to its high pozzolanity, ne y ash and
silica fume increase the homogeneity of the microstructure by
replacing CH crystals with additional CSH gel. Therefore, the
probability of micro-cracking is reduced and the transition zone
becomes thinner. Moreover, the size and content of capillary pores,
as well as the CH crystals concentration are reduced with progress
of the pozzolanic reaction. Furthermore, the produced CSH gel
blocks the pores, reduces its size and interrupts its connectivity.
In addition, it reduces the wall effect around the aggregate sur-
faces, thus allowing better packing of cement particles at the inter-
face between cement paste and aggregate.
7.4. The role of dense packing and ne y ash
To explain the role of the packing density and ne y ash on
concrete properties, it is worth to repeat the fact that all of the
studied mixtures prepared roughly with the same volume fractions
of aggregate and cementitious materials. Applying ideal Fuller
curve in the concrete mix design minimized voids content. These
voids should be also lled with dense cement paste. However, ce-
ment particles with sizes of about 30 lm cannot achieve this tar-
get alone. Therefore, another material with smaller size which
can t in the voids between cement particles should be imple-
mented. Fine y ash is the ideal in these conditions because it
has smaller size, spherical shape, smooth texture and pozzolanic
reactivity. Going deeper, the voids between ne y ash should be
lled with smaller size particles. Silica fume with particle size of
about 0.10.5 lm could be the best material lling the voids be-
tween ne y ash. By this mechanism, cementitious materials
are needed only to ll the voids between the aggregates, which
are minimized, and also to cover the aggregate surface to keep
the workability at a satised level. As a result, dense microstruc-
ture are generated which enhances both mechanical properties
and durability of concrete using low binder content.
The role of ne y ash in reducing the porosity and enhancing
the concrete durability can be interpreted basically from several
aspects. Firstly, it enhances the packing density of the system by
its ball bearing effect which reduces the particle interlocking
[28]. Secondly, it enhances the packing of particles at the aggregate
surfaces. Thirdly, it reacts with CH, therefore, the transition zone
become denser and the pores become ner [29]. Fourthly, the poz-
zolanic reaction takes place within the capillary pores, and the
hydration products block and reduce the size of capillary pores
and prevent their connectivity. Fifthly, the pozzolanic reaction
takes some water of the free water in the system, which indirectly
reduces the porosity. Sixthly, it enhances the cohesion of fresh con-
crete and reduces the amount of bleeding water beneath the aggre-
gate, and as a result the transition zone is densied. Seventhly,
because of its spherical shape and smooth texture, it reduces the
water demand and makes the microstructure more homogenous.
Finally, because of its low hydration rate compared to cement,
the hydration heat is low which result in smaller thermal stresses
at early ages. Both normal y ash and ne y ash has similar inu-
ences on concrete from chemical point of view because they have
similar chemical composition as can be seen from Table 1. The only
difference is the mean particle size (neness): for normal y ash it
is in the range of cement (about 30 lm), while ne y ash it is
lower than 10 lm as can be seen in Fig. 3. The benets of ne y
ash instead of normal y ash can be drawn as follow:
1. Fine y ash is more reactive than normal y ash, therefore,
mechanical properties are better enhanced particularly at early
ages [2931].
2. Normal y ash has relatively large particle size; therefore, it
cannot ll the voids between cement particles which resulted
in poor packing density on the cement matrix.
3. Fine y ash enhances the packing density on the aggregate sur-
face which cannot be achieved by normal y ash.
On the other hand, the use of silica fume with cement (binary
system) result in low packing density because of the high inter-
particle forces and the possibility of agglomeration. Silica fume
should be uniformly and homogeneously dispersed in order to
achieve its pozzolanic and ller effects [32]. Therefore, the use of
adequate cementitious materials with optimum composition, with
the use of appropriate superplasticizer that is sufcient to ef-
ciently disperse the ne particles will lead to concrete with dense
microstructure and superior properties.
8. Conclusion
Based on the above results and discussions, the following con-
clusions can be drawn:
1. By applying ideal grading curve, it is possible to produce HPC
with superior properties using only 312 kg/m
3
of cementitious
materials.
2. In the densely packed system, replacing OPC with 25% ne y
ash and 10% silica fume (mix 2) exhibited compressive strength
higher than 80 MPa with w/b ratio of 0.42, while the elastic
modulus reached around 45 GPa. However, the compressive
strength reached 99 MPa and the elastic modulus was more
than 50 GPa for mix 8 with combination of slag cement, ne
y ash and silica fume along with w/b ratio of 0.27.
3. The maximum porosity of concrete mixes was 10% for mix with
OPC only, (signicantly lower than traditional concrete), which
revealed the importance role of packing density. However, a
total porosity of 3% has been achieved by using supplementary
materials with w/b ratio of 0.27.
4. A very low capillary porosity was obtained for all mixes com-
pared to traditional concrete and HPC. In particular, mix 8 with
w/b ratio of 0.27 has a capillary porosity of lower than 2%.
5. The optimized concrete showed very low water penetration
depth. At w/b ratio of 0.42, it ranges between 8 and 14 mm
and it decreased to 38 mm with reducing the w/b ratio to 0.27.
6. The chloride diffusion coefcient was reduced from
28 10
13
m
2
/s for OPC concrete to 4.5 10
13
m
2
/s by using
25% ne y ash and 10 silica fume. More reduction of chloride
diffusion is reached by using both silica fume and ne y ash
with slag cement at w/b ratio of 0.27, it is reduced to 1.39
10
13
m
2
/s (about 20 times lower than OPC concrete).
7. Silica fume is more effective in improving the durability and
strength of concrete in systems with loww/b ratio than systems
with high w/b ratio.
References
[1] Mehta PK, Monteiro PJM. Concrete: microstructure, properties and materials.
McGraw-Hill; 2006.
[2] Neville A, Atcin PC. High performance concretean overview. Mater Struct
1998;31(2):1117.
[3] Larrard Fde. Concrete mixture proportioning: a scientic approach. Taylor &
Francis; 1999.
232 M. Abd Elrahman, B. Hillemeier / Construction and Building Materials 58 (2014) 225233
[4] Andersen PJ, Johansen V. A guide to determining the optimal gradation of
concrete aggregates. Report No. SHRP-C-334., ed. N.A.o. Sciences; 1993.
[5] McGeary RK. Mechanical packing of spherical particles. J Am Ceram Soc
1961;44(10):51322.
[6] Sobolev K, Amirjanov A. Application of genetic algorithm for modeling of dense
packing of concrete aggregates. Constr Build Mater 2010;24(8):144955.
[7] Hsken G, Brouwers HJH. A new mix design concept for earth-moist concrete:
a theoretical and experimental study. Cem Concr Res 2008;38(10):124659.
[8] Fuller WB, Thompson SE. The laws of proportioning concrete. ASCE J Transp
1907;59:67143.
[9] Rached M, Michael DM, Fowler DW. Utilizing aggregates characteristics to
minimize cement content in Portland cement concrete. Report no. ICAR 401.
International Center for Aggregates Research (ICAR), The University of Texas at
Austin, Construction Materials Research Group, 2009.
[10] Brouwers HJH, Radix HJ. Self-compacting concrete: theoretical and
experimental study. Cem Concr Res 2005;35(11):211636.
[11] Powers TC. The properties of fresh concrete. New York: Wiley; 1968.
[12] Andreasen AHM. Ueber die Beziehung zwischen Kornabstufung und
Zwischenraum in Produkten aus losen Krnern (mit einigen Experimenten).
Kolloid-Zeitschrift 1930;50(3):21728.
[13] Puntke W. Grundlagen fr die Matrixoptimierung und Ausfhrung in der
Praxis, Matrixoptimerung mikroskopisches Analysieren von Beton. Technische
Akademie Esslingen, TAE, 1990.
[14] Httl R, Hillemeier B. Hochleistungsbeton-Beispiel Sureresistenz. Betonwerk
und Fertigteiltechnik 2000;66(1):5260.
[15] Busch D, Haselwander B, Hillemeier B, Strau J. Innovative Betontechnologie
fr den Khlturmbau. Beton 1999;49(4):1089.
[16] Aponte DF, Barra M, Vzquez E. Durability and cementing efciency of y ash
in concretes. Constr Build Mater 2012;30:53746.
[17] Papadakis VG, Antiohos S, Tsimas S. Supplementary cementing materials in
concrete. Part II: A fundamental estimation of the efciency factor. Cem Concr
Res 2002;32(10):15338.
[18] Gluth GJ. Die Porenstuktur von Zementstein und seine Eignung zur
Gastrennung. Fakultt VI Planen Bauen Umwelt. Technische Universitt
Berlin; 2011.
[19] BAW. Resistance of concrete to chloride penetration. BAW Codes of Practice
and Guidelines Publisher. Bundesanstatalt fr Wasserbau; 2012.
[20] Virgalitte SJ, Luther MD, et al. Ground granulated blast furnace slag as a
cementitious constituents in concrete and mortar: ACI Report 233R-95. ACI
Mater J; 1995. 18 pp.
[21] Teichmann T, Schmidt M. Inuence of the packing density of ne particles on
structure, strength and durability of UHPC. In: First international symposium
on ultra high performance concrete, Kassel, Germany, 2004. p. 31323.
[22] Bentz DP. Inuence of silica fume on diffusivity in cement-based materials: II.
Multi-scale modeling of concrete diffusivity. Cem Concr Res 2000;30(7):
11219.
[23] Peng Y, Hu S, Ding Q. Dense packing properties of mineral admixtures in
cementitious material. Particuology 2009;7(5):399402.
[24] Poon CS, Kou SC, Lam L. Compressive strength, chloride diffusivity and pore
structure of high performance metakaolin and silica fume concrete. Constr
Build Mater 2006;20(10):85865.
[25] Brandenburger D, Herten, Httl R. Hochleistungsbeton mit Feinstugasche als
Betonzusatzstoff. Beton 2006;56(5):198200.
[26] Neville AM. Properties of concrete. 4th and nal ed. John Wiley & Sons; 2004.
[27] Thomas MDA, Hooton RD, Scott A, Zibara H. The effect of supplementary
cementitious materials on chloride binding in hardened cement paste. Cem
Concr Res 2012;42(1):17.
[28] Hsken G, Brouwers HJH. On the early-age behavior of zero-slump concrete.
Cem Concr Res 2012;42(3):50110.
[29] Chindaprasirt P, Jaturapitakkul C, Sinsiri T. Effect of y ash neness on
compressive strength and pore size of blended cement paste. Cem Concr
Compos 2005;27(4):4258.
[30] Bentz DP, Hansen AS, Guynn JM. Optimization of cement and y ash particle
sizes to produce sustainable concretes. Cem Concr Compos 2011;33(8):
82431.
[31] Maibaum C, Herten, Httl R. Neuer Zusatzstoff fr Hochleistungsbetone. Beton
2004;54(3):1323.
[32] Marchuk V. Dispersibility of the silica fume slurry in cement paste and mortar.
Betontechnische Berichte, Concrete Technology Reports 2001
2003. Dsseldorf, Germany: Verlag Bau + Technik; 2004. p. 12532.
M. Abd Elrahman, B. Hillemeier / Construction and Building Materials 58 (2014) 225233 233