Construction and Building Materials 105 (2016) 503–510

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

An experimental and numerical study on water permeability of concrete

Xinxin Li, Qing Xu, Shenghong Chen ⇑
State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University, Wuhan, Hubei 430072, PR China

h i g h l i g h t s

Water transport behavior in concrete was simulated by a 3D FE algorithm.

ITZ was modeled practically by the zero-thickness interface element.

Laboratory experiment was conducted to provide parameters and validation data for numerical simulation.

The effects of ITZ and coarse aggregates on water permeability were quantitatively studied.

a r t i c l e i n f o a b s t r a c t

Article history: Durability of concrete depends largely on its water permeability, which is dominated by the composition
Received 6 October 2015 and difficultly evaluated. This paper presents a three-dimensional (3D) FE algorithm to investigate the
Received in revised form 22 December 2015 water transport performance in concrete. This FE model possesses three-phase mesostructure consisting
Accepted 24 December 2015
of coarse aggregates, mortar matrix, and the interfacial transition zone (ITZ) which is practically modeled
by the zero-thickness interface element. A series of numerical tests with regard to concrete samples con-
taining different volume fractions of coarse aggregates are conducted. The calculated permeability coef-
Keywords:
ficients are in reasonable accordance with the physical experiment data. Effects of aggregates and ITZ on
Concrete
Water permeability
the concrete permeability are also quantitatively studied by the numerical test. It is proved that the pro-
Mesostructure posed algorithm is capable of modeling the permeable characteristics of concrete which is affected
FE algorithm greatly by its heterogeneous mesostructure.
ITZ Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction variations in the mixture proportions may exert a considerable

Water permeability of concrete is believed as the key property
related to the serviceability and durability of concrete structures
(e.g., bridges, hydraulic structures and marine structures) sub-
jected to aggressive environments, since water acts as either the
major agent responsible for the deterioration of concrete or the
transport medium for aggressive species like chloride or sulfate
ions [1–4]. In general, higher permeable concrete, with more pores
and voids, is less durable [5]. A good understanding of water trans-
port characteristics of concrete will assist the progress in service-
life prediction and optimum material design [6].
Physical experiment has been exercised for many years to mea-
sure water permeability [7–9] and to investigate different variables
affecting water transport in cementitious materials [10–15].
Among these variables are the water–cement ratio [10,11],
aggregate content [12,13], aggregate size distribution [14] and
pore characteristics [15], etc. Particularly, it is revealed that the
⇑ Corresponding author.
E-mail address: chensh@whu.edu.cn (S. Chen).
impact on the permeability of concrete. Although the physical
experiment method provides a direct means to look into these
variables and their effects, it is subject to several inherent limita-
tions. For example, due to the fact that concrete is a complex
heterogeneous multi-phase composite, it is often difficult to
exactly separate the effects of these variables. In addition, a large
number of specimens should be prepared and tested to eliminate
the stochastic variations aforementioned [16], which will be
time-consuming and costly.
With the significant advance in computer technology and
numerical methods, mesoscopic modeling becomes more and
more widely accepted to study the composite behavior of concrete
[17]. Many models have been proposed to simulate the mesostruc-
ture of concrete, which commonly takes into account of three
phases: coarse aggregates, mortar matrix (or cement paste) and
interfacial transition zone (ITZ) between them. Although these
models are extensively employed to estimate the mechanical prop-
erties, such as elastic modulus, tensile strength, and fracture
energy [18,19], thermal properties (e.g., thermal conductivity
[16,20,21]), and diffusion properties (e.g., chloride diffusivity

http://dx.doi.org/10.1016/j.conbuildmat.2015.12.184
0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.
504 X. Li et al. / Construction and Building Materials 105 (2016) 503–510
[22,23]), yet water transport properties of concrete has received
relatively less attention. Recently, Wang and Ueda [24] develop a
two-dimensional (2D) lattice network model to predict water pen-
etration into concrete, whereby the three phases in concrete are
discretized into lattice elements considered as conductive ‘‘pipes”
with different transport properties. Their results show that the
thickness of ITZ exhibits only a small influence on water absorp-
tion. Abyaneh et al. [25] utilize a mesoscale model to investigate
water sorptivity of concrete with 3D lattice network and to study
the effects of the amount, spatial distribution and shape of aggre-
gate particles on capillary absorption, but the impact of ITZ is not
taken into account. Zhou and Li [26] propose a 2D mesoscale model
for simulating the steady permeation in concrete sample based on
finite element method (FEM). Assuming concrete as a random
three-phase composite, they have performed a numerical and sta-
tistical analysis to investigate its water permeability and to quan-
tify the size of representative volume element (RVE). However,
most of these models are 2D, which might not be used to predict
the spatial 3D flow in reality. In addition, in the lack of a precise
modeling of ITZ – the weakness zone in concrete [27,28], the sim-
ulated results may be questionable.
In this paper, a FE algorithm is established to evaluate the water
permeability of concrete and to explore how the individual compo-
nents of the concrete mixture influence its transport behavior. The
model is built on the mesoscale level, whereby the concrete is ide-
alized as a three-phase composite consisting of coarse aggregate,
mortar and ITZ. The geometry entity of the 3D mesostructure is
explicitly discretized by finite elements, in which the zero-
thickness element is used for modeling ITZ. In order to validate
the proposed algorithm, physical experiments are conducted in
parallel. In the experiment, permeability coefficients of specimens
with different aggregate volume fractions are measured, providing
the material parameters and validation data for the numerical sim-
ulation. Finally, the effects of mesostructural heterogeneity such as
the water transport property of ITZ and the volume fraction of
aggregates on water permeability of concrete are investigated with
the help of the numerical simulation algorithm proposed and val-
idated in this paper.
2. Permeability theory of concrete
2.1. Governing equation
Suppose that the flow of liquid in porous medium obeys Darcy’s
law, which may be written as [3,5]:
u ¼ kr/
1 1
in which u (L T ) is the flow velocity, k (L T ) is the permeability
coefficient, / (L) is the hydraulic potential and r (L 1) is the spatial
gradient operator.
Introducing Darcy’s law into continuity equation for incom-
pressible and steady flow, water transport in perfectly saturated
materials is governed by the following partial differential
equation:
rðk  r/Þ ¼ 0
Eq. (2) is subject to appropriate boundary conditions:
/jC1 ¼ /0 ðfirst typeÞ
k  @/=@fngjC2 ¼ q ðsecond typeÞ
T
in which fng ¼ flx ; ly ; lz g are direction cosines of the external nor-
mal to the boundary, /0 and q are specified hydraulic potential at
the first type boundary C1, and the flow rate through the second
type boundary C2, respectively.
2.2. Permeability test
Water permeability characterizes the capability of saturated
concrete to transport water when subjected to an applied hydrau-
lic gradient [3,5]. It is usually measured in the laboratory by per-
meability tests based on the Darcy’s law (Vide Eq. (1)), which can
be written as follows:
QL
k¼ ð5Þ
D/S
in which D/ (L) is the hydraulic potential difference, S (L2) is the
cross-sectional area perpendicular to flow direction, L (L) is the
length of flow path and Q (L3 T1) is the flow rate.
However, laboratory based permeability tests for modern dense
concrete or high performance concrete may encounter practical
difficulties: a high pressure and a long time are required to drive
the steady state flow through the concrete specimen, and there is
high risk of leaking around the specimen if the surface sealing is
not guaranteed. There seems still no widely or fully accepted tech-
niques for measuring the permeability of concrete [9,29], though
some advanced experimental methods [8,30,31] have been pro-
posed. Given the limitations of experimental test, numerical simu-
lation might be an applicable and effective approach.
3. Mesostructure modeling of permeable concrete using FE
algorithm
The FE algorithm is formulated to investigate the water trans-
port behavior of concrete with a heterogeneous mesostructure, in
which coarse aggregates, mortar matrix and ITZ are modeled as
distinct phases with different material properties assigned.
3.1. Numerical sampling of concrete with mesostructure
The 3D random aggregate structure (RAS) is established for
numerical samples of concrete. Spherical particles are used to
model coarse aggregates for simplification. The size distribution
of aggregates is determined following a certain given grading curve
(such as Fuller curve or Bolomey curve [18,32]). In most concretes,
the volume fraction of coarse aggregates is between 0.4 and 0.5
[18,19]. The prescribed volume fractions of aggregate particles
are randomly arranged and placed in a cube, cylinder or frustum
shaped sample, which depends on the type of concrete specimen
to be tested. The approach of ‘‘occupation and removal method”
[33,34] is employed to conduct this packing process. In order to
ð1Þ produce the geometrical configuration which resembles the real
concrete, the random sampling principle of Monte Carlo (MC) sim-
ulation method is used for both the spatial distribution and size
distribution of aggregates. The detailed procedure for generating
the geometrical models can be referred to literature [17]. Fig. 1
shows three concrete samples in different shapes with constant
aggregate volume fraction of 0.4.
An important phase in the mesostructure is ITZ, which has long
been regarded as a zone of weakness in concrete composite, both
ð2Þ in terms of strength and the permeation [15,28]. So the ITZ is
included in the numerical sample as a bonding layer structure sur-
rounding the coarse aggregates.
ð3Þ
3.2. FE algorithm
ð4Þ
It is still a challenge for the mesh generation of such a complex
3D mesostructure in presence of the unique layer structure of ITZ,
which is extremely thin and rather small compared to the normal
size of coarse aggregates. Therefore, many efforts have been made
to model ITZ structure [17,22,24,35]. An elegant FE model is devel-
 X. Li et al. / Construction and Building Materials 105 (2016) 503–510 505

(a) Cube shaped (b) Cylinder shaped (c) Frustum shaped

Fig. 1. Concrete samples in different shapes with spherical aggregate particles.

oped herein to address this problem, in which aggregates and mor- ator restricted to the ITZ’s tangential plane. fm (L T1) and fa (L T1)
tar are discretized into 3D solid elements, while ITZ is approached are the flow rate penetrating into ITZ in the normal direction from
as additional shell element, namely, zero-thickness interface mortar and aggregates respectively, which can be written as
element. 8
< f m ¼ km @/up
@nup
ð7Þ
3.2.1. Aggregates and mortar elements : f ¼ ka @/down
a @n down
Without ITZ, the RAS containing coarse aggregates and mortar
is not difficult to be meshed into tetrahedral elements. Fig. 2 pre- where n is the unit vector normal to the fracture plane. km (L T1) is
sents the typical FE mesh of a concrete sample composed only of the permeability coefficient for mortar and ka (L T1) for aggregates.
aggregates and mortar with aggregate volume fraction of 0.4. It As the thickness of ITZ is extremely small, it can be assumed
contains 698,483 elements and 146,820 nodes. that the hydraulic potentials at two sides of the ITZ plane are iden-
In FE analysis, both the aggregates and mortar are postulated as tical, i.e.,
homogeneous and isotropic porous medium, therefore the water
/up ¼ /down ¼ /i ð8Þ
flow in these two phases is governed by Eq. (2).
where /i (L) is the hydraulic potential in ITZ.
3.2.2. Zero-thickness interface element With these formulations, the zero-thickness interface element
In this paper, the zero-thickness interface element is used to is idealized as a conductive ‘‘layer” within concrete. It can not only
model ITZ structure, as shown in Fig. 3. It is a shell type element accurately simulate the transport behavior of ITZ, but also reduce
with six nodes, of which three are attached to the aggregate and the number of freedom degrees without generating new solid ele-
the others are attached to the mortar. The element is assumed to ments and consequently bring down the computational cost.
be the face between solid elements with zero thickness. Thus,
the three nodes at the up face have the same coordinates with 3.2.3. Discretization of the governing equation by FE algorithm
the down ones. The subscripts ‘‘up” and ‘‘down” respectively mean The overall discretized equations for the FE solution can be
the up and down surface of the ITZ plane (see Fig. 3). expressed by
Similar to the subsurface flow in the fracture of rock mass, the
interface element might be used to describe the transport behavior ½Kf/g ¼ ff g ð9Þ
along the interior boundaries between aggregates and mortar, In Eq. (9):
which is governed by the following equation [36], 0 1
rs ðki  di rs /Þ ¼ f m þ f a ð6Þ XBZ Z Z ZZ
½K ¼ @ rNT krNdX þ rs NTi ki di rs Ni dcC
A ð10Þ
where ki (L T1) is the permeability coefficient of ITZ and di (L) rep- e
X c
resents its hydraulic thickness. rs (L1) denotes the gradient oper-
ZZ
ff g ¼ N T qdC2 ð11Þ
C2

where X is the domain of aggregates and mortar matrix, c is the ITZ

domain, N is the shape function in X while Ni is the shape function
in c.
Eq. (9) is implemented in the commercial FE software, COMSOL
Multiphysics, which is a wellknown tool providing full access to
solve partial differential equations [37].

3.3. Numerical test for the permeability of concrete

Numerical simulation can be performed for modeling the per-

meability test and investigating the permeability of concrete. The
FE model considers a steady water flow process as shown in
Fig. 4. The hydraulic potential /1 and /2 are applied to the top
Fig. 2. FE meshes of aggregates and mortar in a concrete sample. and bottom boundaries of the concrete sample, respectively. Since
506 X. Li et al. / Construction and Building Materials 105 (2016) 503–510

nup
Tetrahedral element of
mortar
di
Zero-thickness
interface element

Tetrahedral element of
Zero-thickness interface
τ2
aggregate
element
τ1
ndown

Fig. 3. Schematic diagram of the zero-thickness interface elements.

the side surface of the cylinder is generally sealed, impervious con-
dition is employed for the numerical model.
The parameters needed in the FE simulation are the transport
properties of three phases in the mesostructure, i.e., the permeabil-
ity coefficient ka for aggregate, the permeability coefficient km for
mortar matrix, the permeability coefficient ki and the thickness di
for ITZ. The permeability of mortar matrix is relatively easier to
be experimentally measured since a much smaller specimen size
is demanded to obtain representative gaugement. Normally, most
aggregates used in concrete is much denser than mortar matrix
[38], so it may be assumed that they are basically impermeable
with permeability coefficient 107 lower than that of mortar in this
study, namely, ka = km/107. So far, it is difficult to evaluate the exact
properties of ITZ by laboratory physical test. Limited researches in
this field are insufficient for reliable parametric study. Sensitivity
or feedback analysis may be the feasible way to assess the proper-
ties of ITZ, when the permeable characteristics of other two phases
and concrete are well measured from physical experiments at
macroscopic level.
Based on the FE simulation, Eq. (5) is used to calculate the effec-
tive permeability coefficient of concrete, keff. The parametric and
sensitive analysis may be further carried out for a better under-
standing with regard to how water movement taking place in con-
crete and how the individual component (e.g., coarse aggregates or
ITZ) influencing its permeability.
crete specimens sized by U61.8 mm  40 mm are prepared and
grouped into six series with different aggregate volume fractions
ranging from 0.0 to 0.5. The raw materials, P.C 32.5 cement, China
ISO Standard sand and tap water are used for mortar. Two types of
equal-sized spherical glass particles (i.e., 6 mm and 12 mm, respec-
tively) are used for coarse aggregates (see Fig. 5(a)). The glass par-
ticles have an apparent density of 2500 kg/m3. More details of
mixture design are shown in Table 1.
In order to minimize the variations in properties of mortar, mix-
ture proportions (water to cement ratio and sand to cement ratio
as 0.55/1 and 5.88/1, respectively) and density (1920 kg/m3) of
mortar are kept the same in different specimens. Aggregates used
in these specimens, glass particles, are almost non-sorptive, i.e.,
aggregates can hardly absorb any water from mortar. Note that
the formation of ITZ structure around aggregates may somewhat
reduce the effective water to cement ratio of mortar. According
to the literature [39], this phenomenon is overlooked since it is
not likely to have a huge impact. Therefore, it can be assumed that
the permeability of mortar matrix basically remains identical,
independent of the coarse aggregate volume fraction.
An impermeable material (mixture of grease and cement) is
employed to fill the gaps between test mold (cutting ring) and
(b)
(a)

4. Experimental and numerical study

4.1. Laboratory experiment

A laboratory experiment is designed to test the water perme-

ability of concrete. This test provides the material parameters
and validation data for the numerical study. Thirty cylinder con-

Hydraulic potential φ1
S
top cap
up porous
outlet tube stone

Flow rate Q seal ring specimen

L Impervious cutting ring steel casing

air release down porous

Z valve stone
Y
steel bed inlet tube
o X

(c)
Hydraulic potential φ2
Fig. 5. Some details of the experiment: (a) 6 mm and 12 mm particles; (b)
Fig. 4. Configuration of numerical test. manufactured specimen; (c) permeation apparatus.
 X. Li et al. / Construction and Building Materials 105 (2016) 503–510 507

specimen for side surface sealing (see Fig. 5(b)). Before testing, the For a deeper understanding of the effects of ITZ in this paper, a
specimens have been cured at 20 ± 1 °C with the relative humidity sensitivity analysis on its permeability and thickness is subse-
(RH) higher than 95% for 7 days. Then the test is carried out using a quently performed using a concrete sample with fagg = 0.4. For
permeameter (see Fig. 5(c)) with the hydraulic potentials of the sake of simplicity, it is assumed that the thickness of ITZ is uni-
1300 mm and 0 mm applied to the bottom and top surface, respec- form for all aggregate particles in concrete. According to the afore-
tively. Considering the relatively small size of these concrete spec- mentioned review of the published literatures, the specified
imens, such a hydraulic potential difference of 1300 mm is high samples will be tested where the thickness di of ITZ ranges from
enough to drive a steady state flow in several or tens of minutes. 0 to 100 lm. The permeability coefficient of ITZ is defined as
After the steady state condition being established, the flow rate ki ¼ kkm with five levels of the permeability ratio k, i.e., k = 2, 5,
is determined by weighing the outflow to calculate the water 10, 20 and 50. Fig. 7 shows the influence of ITZ on the effective per-
permeability. meability coefficient of concrete sample.
The data obtained from the test are listed in Table 2. In order to It is found that the effective permeability coefficient increases
ensure the effectiveness of the results, the maximum and mini- following the increase of the thickness di of ITZ. According to Eq.
mum values of each series are removed. Subsequently, the remain- (6), the water transport behavior of ITZ is directly related to its
ing three ones are used to carry out a statistic analysis to thickness and permeability coefficient. However, the permeability
determine the average, standard deviation (SD) and coefficient of of ITZ seems to play a more important role. When the permeability
variation (CV) (see Table 2). It can be seen that certain variations ratio is lower (k = 2, 5 or 10), the ratio keff/km calculated is basically
do exist in the experimental data with CV ranging from 0.1 to 0.15. proportional to di. For larger ratio (k = 20 and 50), a non-linear rela-
tionship may be observed. The ratio of keff/km may exceed 1.0 when
k = 50, which means that the permeability of concrete is even lar-
4.2. Numerical test ger than the pure mortar matrix because of the much higher water
conductivity of ITZ.
Numerical test is conducted to simulate the above laboratory In consequence, both the permeability and thickness of ITZ may
experiment by the FE algorithm proposed in this paper. Five series greatly affect the permeability of concrete. Referring to the pub-
of samples with different aggregate volume fractions, fagg ranging lished literature [24], the permeability of ITZ is assumed 10 times
from 0.1 to 0.5 are modeled and numerically tested. The configura- higher than that in mortar, i.e., the ratio k = 10 will be fixed in the
tion parameters of aggregates (size, number and gradation) will be hereinafter numerical tests. Meanwhile, considering the much
kept the same with experimental specimens as depicted in Table 1. smaller size of glass particles comparing with actual coarse aggre-
Fig. 6 shows some of the typical samples used in the test. The per- gates, the thickness of 20 lm is selected for ITZ.
meability coefficient of mortar is readily obtained from the labora-
tory experiment as 1.95  102 mm/s, and aggregates are assumed
basically impermeable (e.g., ka = 1.95  109 mm/s). The parame- 4.2.2. Results and discussions
ters of ITZ need more attention and will be discussed in the Sec- The precision of numerical tests relies largely on the meshing
tion 4.2.1 here after. density [18,19]. Therefore, the computed results by mesoscopic
modeling may be mesh-dependent. In order to obtain a highly
accurate solution, mesh sensitivity analysis should be conducted
4.2.1. Water transport property of ITZ to determine the required size of elements. It is confirmed that
As the weakest phase in concrete, it has been recognized that the FE meshes with element size h = 0.247–3.4 mm (see Fig. 2)
ITZ plays an important role in the transport properties of concrete are acceptable, by restricting the variation in the calculated results
due to its lower density and higher porosity [15,28,40]. Water flow within 5%.
in ITZ is thought to be much faster than that in the mortar matrix Variation may also give rise to numerical results due to the ran-
[24]. However, little is known concerning the quantitative differ- dom distribution of aggregates. In order to investigate such varia-
ences between the permeability of this zone and that of the mortar tion, a statistical analysis is carried out for 20 concrete samples
matrix. with fagg = 0.4. The effects of the sample number on the mean
In terms of the thickness of ITZ, a variety of observations have and standard deviation of predicted permeability coefficient are
been reported. The experimental data of Winslow et al. [15] and plotted in Fig. 8. It is shown that 5 samples are sufficient to obtain
Ollivier et al. [40] show that the thickness of ITZ is in a range of a convergent mean effective permeability coefficient with a stable
10–50 lm. Scrivener [41] uses the technique of backscattered elec- fluctuation smaller than 0.15%. Therefore, 5 concrete samples are
tron imaging to explore the cementitious microstructures and he tested for each series in the following numerical tests.
demonstrates that ITZ varies from 1 to 100 lm in thickness. To validate the FE modeling, the results are compared in Fig. 9,
Nemati and Gardoni [42] conduct experiments using Wood’s metal where the simulated results are obtained by averaging the numer-
intrusion, which also indicate a thickness of 30–100 lm for ITZ. ical test data. It is shown that the simulated results are in reason-
able accordance with most of the laboratory experiment data.
Some of the predicted relative permeability coefficients do not
Table 1
agree well with that from experiment, probably partially due to
Mixture design.
the variations existed in the laboratory data, and partially due to
No. Volume Cement Water Sand (g) Number of the fact that the property of ITZ is determined by assumptions
fraction (g) (g) aggregate
and the sensitive analyses. Actually, both the permeability and
particles
thickness of ITZ could be highly variable, associated with the
6 mm 12 mm
amounts of aggregates [15,28]. To improve the model’s predictive
1 0.0 30.95 17.02 182.03 0 0 capability by providing reliable input parameters, the exact prop-
2 0.1 27.85 15.32 163.83 58 6
erties of ITZ need to be directly measured by experimental means
3 0.2 24.76 13.62 145.63 117 12
4 0.3 21.66 11.91 127.42 175 18 in future.
5 0.4 18.57 10.21 109.22 234 24 It is observed that the effective permeability coefficient of con-
6 0.5 15.47 8.51 91.02 292 30 crete decreases with the increase of aggregate volume fraction.
Similar observations also may be found in some laboratory tests
508 X. Li et al. / Construction and Building Materials 105 (2016) 503–510

Table 2
Permeability coefficient from experimental data (Unit: mm/s).

Specimen Volume fraction

0.0 0.1 0.2 0.3 0.4 0.5
1 1.60  102 1.40  102 9.65  103 7.50E  103 3.84  103 2.92  104
2 2.40  102 2.37  102 1.90  102 2.09  102 8.15  103 5.23  103
3 2.30  102 2.00  102 1.20  102 1.12  102 7.00  103 4.22  103
4 1.74  102 1.52  102 1.50  102 1.60  102 6.18  103 3.09  103
5 1.80  102 1.90  102 1.70  102 1.40  102 5.46  103 3.30  103
Mean 1.95  102 1.81  102 1.47  102 1.37  102 6.21  103 3.54  103
SD (%) 2.52  101 2.08  101 2.05  101 0.197  101 6.27  102 4.91  102
CV 0.129 0.115 0.140 0.143 0.101 0.139

(a) f agg =0.1 (b) f agg =0.2 (c) f agg =0.3 (d) f agg =0.4 (e) f agg =0.5

Fig. 6. Numerical samples with variable aggregate volume fraction from 0.1 to 0.5.

1.5 reduce the overall permeability of concrete, known as the dilution

λ=2 effect [26]. Moreover, the randomly distributed aggregates may
1.3 λ=5 cause tortuosity effect that force the water to flow around them
λ=10 [26,38]. As a result, larger content of aggregates may further
1.1
λ=20 lengthen the flow path and thus decrease the permeability of
keff / km

λ=50 concrete.
0.9
The numerical results show that the effective permeability coef-
ficient of concrete varies almost linearly with aggregate volume
0.7
fraction. While the relation becomes slightly nonlinear when con-
0.5 siderably more aggregates are incorporated (fagg = 0.4–0.5), due to
the fact that more ITZ with relatively higher permeability is cre-
0.3 ated in concrete. It seems that the effects of aggregates play a dom-
0 20 40 60 80 100 inant role in this case.
di (μm) From the numerical test, it is easy to make a good description of
the water transport behavior in concrete. Take a sample with
Fig. 7. Influence of ITZ on the effective permeability coefficient of concrete.
fagg = 0.4 for instance, the hydraulic potential and flow velocity
along the central axis and middle section are presented in Fig. 10
[13,23,43] and numerical tests [22,38,44] with respect to the trans- and Fig. 11, respectively.
port properties of concrete. Impermeable or less pervious aggre- In Fig. 10, hydraulic potential along the central axis exhibits
gates embedded in the homogeneous mortar matrix obviously non-linear but continuous distribution, which distinctly shows

0.525 0.15
Standard deviation of keff / km (%)

0.1
Mean of keff / km

0.515

0.05

0.505 0
0 5 10 15 20 0 5 10 15 20
Sample number Sample number
(a) Mean (b) Standard deviation
Fig. 8. Effects of the sample number on the mean and standard deviation of keff/km.
 X. Li et al. / Construction and Building Materials 105 (2016) 503–510 509

1.2 to bypass the aggregates (Fig. 11(b)). Additionally, the compara-

Numerical tively larger flow velocity may emerge in ITZ. It is indicated that
1.0 the zero-thickness interface element has been well used to model
Experimental
the ITZ structure and describe its water transport behavior.
0.8
These results show that numerical simulation with the mesos-
keff / km

0.6 cale model can provide a direct vision of the transport behavior
of water in concrete, which might be a useful way to study the
0.4 complex transport properties of concrete.

0.2
5. Conclusions
0.0
0.0 0.1 0.2 0.3 0.4 0.5 In this paper, a 3D FE algorithm is proposed to investigate the
f agg water transport in saturated concrete with heterogeneous
Fig. 9. Comparison of numerical and experimental results.
mesostructure consisting of coarse aggregates, mortar matrix and
ITZ. In the numerical test sample, ITZ is practically modeled by
the zero-thickness interface element. The experimental and
numerical study on water permeability of concrete are carried
1400 8.0 out. From the study, several important conclusions can be drawn
Hydraulic potential Flow velocity
1200 7.0 as follows:
Flow velocity (mm/s)
Hydraulic head (mm)

1000
800
600
400
200
0 0.0
0 5 10 15 20
Distance (mm)
Fig. 10. Hydraulic potential and flow velocity along the central axis.
the heterogeneity of concrete. Different from hydraulic potential,
the flow velocity is changing dramatically along the central axis
when water permeates through different components. The peak
values on the axis all emerge where the ITZ is located. A relatively
low but stable flow velocity is observed when water flowing
though the mortar. The flow velocities are almost zero in the aggre-
gates. This could attribute to that the permeability of three compo-
nents has the relationship of ki > km  ka.
Fig. 11 shows the similar phenomenon of hydraulic potential
and velocity as depicted in Fig. 10. The distribution of hydraulic
potential is non-uniform due to the heterogeneous structure of
concrete (Fig. 11(a)). It is clear that the water flow has a tendency
6.0
(1) The mesoscopic FE simulation of saturated flow in concrete
5.0
can be significantly simplified by using the zero-thickness
4.0
interface element, which is able to well describe the trans-
3.0 port behavior in the extremely thin layers (i.e., ITZ). It may
2.0 be a promising approach for the simulation of ITZ with
1.0 respect to moisture movement.
(2) Based on the assumed properties of ITZ, the proposed
25 30 35 40 method might be used to predict the water permeability of
concrete relative to that of mortar. The numerically pre-
dicted permeability coefficients agree well with most of
the laboratory experiment data. Therefore, the numerical
test is proved to be capable of simulating the water perme-
ation in concrete and investigating its influential factors,
subject to taking into account of the heterogeneity at mesos-
cale level.
(3) Due to the random distribution of coarse aggregates in the
concrete samples, a certain variation gives rise to numerical
results. It is shown that 5 concrete samples are sufficient to
obtain stationary results from the numerical simulation in
this study.
(4) It is found that heterogeneous mesostructure, particularly
the water transport property of ITZ and the volume fraction
of coarse aggregates, plays an important role in the perme-
ability of concrete. The presence of nearly impermeable

Fig. 11. Hydraulic potential (a) and flow velocity (b) of the middle section.
510 X. Li et al. / Construction and Building Materials 105 (2016) 503–510
aggregates will reduce the permeability of concrete. On the
contrary, ITZ with higher porosity will raise the permeability
of concrete. In the present study, it shows that the effects of
aggregates are in predominance for the overall permeability
of concrete decreases with the combined influences of
aggregates and ITZ.
However, some of the parameters (e.g., properties of ITZ) used
in this study are identified from some published literatures or sen-
sitivity analysis and probably not accurate enough, which limits
the convincing of the numerical test. Therefore, more laboratory
experiments concerning the material properties of the individual
concrete components are demanded before the numerical simula-
tions can be recognized as a competent tool for the physical and
mechanical study of concrete materials.
Acknowledgements
The support of Hubei Province Natural Science Foundation of
China (No. 2015CFB544) is gratefully appreciated.
References
[1] H.W. Reinhardt, Penetration and permeability of concrete: barriers to organic
and contaminating liquids, RILEM Report 16, E&FN Spon, London, 1997.
[2] C. Hall, W.D. Hoff, Water Transport in Brick, Stone and Concrete, Spon Press,
London, 2012.
[3] L. Basheer, J. Kropp, D.J. Cleland, Assessment of the durability of concrete from
its permeation properties: a review, Constr. Build. Mater. 15 (2001) 93–103.
[4] S.H. Chen, Z. He, Status quo and prospects for simulation analysis of service life
for concrete dam, Eng. J. Wuhan Univ. 3 (2011) 273–280 (in Chinese).
[5] S.H. Chen, Hydraulic Structures, Springer Berlin Heidelberg, Berlin, 2015.
[6] A. Boddy, E. Bentz, M.D.A. Thomas, R.D. Hooton, An overview and sensitivity
study of a multimechanistic chloride transport model, Cem. Concr. Res. 29 (6)
(1999) 827–837.
[7] J.W. Figg, Methods of measuring the air and water permeability of concrete,
Mag. Concr. Res. 25 (85) (1973) 213–219.
[8] A.S. El-Dieb, R.D. Hooton, Water permeability measurement of high
performance concrete using high-pressure triaxial cell, Cem. Concr. Res. 25
(6) (1995) 1199–1208.
[9] Q.T. Phung, N. Maes, G. De Schutter, D. Jacques, G. Ye, Determination of water
permeability of cementitious materials using a controlled constant flow
method, Constr. Build. Mater. 47 (2013) 1488–1496.
[10] S. Ahmad, A.K. Azad, K.F. Loughlin, Effect of the key mixture parameters on
tortuosity and permeability of concrete, J. Adv. Concr. Technol. 10 (3) (2012)
86–94.
[11] P. Soongswang, M. Tia, D. Bloomquist, Factors affecting the strength and
permeability of concrete made with porous limestone, ACI Mater. J. 88 (4)
(1991) 400–406.
[12] P. Halamickova, R.J. Detwiler, D.P. Bentz, E.J. Garboczi, Water permeability and
chloride ion diffusion in Portland cement mortars: relationship to sand
content and critical pore diameter, Cem. Concr. Res. 25 (4) (1995) 790–802.
[13] A. Perrot, D. Rangeard, V. Picandet, S. Serhal, Effect of coarse particle volume
fraction on the hydraulic conductivity of fresh cement based material, Mater.
Struct. 7 (2014) 2291–2297.
[14] B.A. Warda, A.N. Munaz, Effects of aggregate gradation on water permeability
of concrete, Adv. Mater. Res. 488 (2012) 248–252.
[15] D.N. Winslow, M.D. Cohen, D.P. Bentz, K.A. Snyder, E.J. Garboczi, Percolation
and pore structure in mortars and concrete, Cem. Concr. Res. 24 (1994) 25–37.
[16] J.Q. Chen, M. Zhang, H. Wang, L. Li, Evaluation of thermal conductivity of
asphalt concrete with heterogeneous microstructure, Appl. Therm. Eng. 84
(2015) 368–374.
[17] Y. Xu, S.H. Chen, A method for modeling the damage behavior of concrete with
a three-phase mesostructure, Constr. Build. Mater. 102 (2016) 26–38.
[18] X.F. Wang, Z.J. Yang, J.R. Yates, A.P. Jivkov, C. Zhang, Monte Carlo simulations of
mesoscale fracture modelling of concrete with random aggregates and pores,
Constr. Build. Mater. 75 (2015) 35–45.
[19] X.F. Wang, Z.J. Yang, A.P. Jivkov, Monte Carlo simulations of mesoscale fracture
of concrete with random aggregates and pores: a size effect study, Constr.
Build. Mater. 80 (2015) 262–272.
[20] W.P. Zhang, H.G. Min, X.L. Gu, Y.P. Xi, Y.S. Xing, Mesoscale model for thermal
conductivity of concrete, Constr. Build. Mater. 98 (2015) 8–16.
[21] L. Shen, Q.W. Ren, N. Xia, L.G. Sun, X.Z. Xia, Mesoscopic numerical simulation
of effective thermal conductivity of tensile cracked concrete, Constr. Build.
Mater. 95 (2015) 467–475.
[22] X.L. Du, L. Jin, G.W. Ma, A meso-scale numerical method for the simulation of
chloride diffusivity in concrete, Finite Elem. Anal. Des. 85 (2014) 87–100.
[23] J.J. Zheng, X.Z. Zhou, Y.F. Wu, X.Y. Jin, A numerical method for the chloride
diffusivity in concrete with aggregate shape effect, Constr. Build. Mater. 31
(2012) 151–156.
[24] L.C. Wang, T. Ueda, Mesoscale modeling of water penetration into concrete by
capillary absorption, Ocean Eng. 38 (2011) 519–528.
[25] S.D. Abyaneh, H.S. Wong, N.R. Buenfeld, Computational investigation of
capillary absorption in concrete using a three-dimensional mesoscale
approach, Comput. Mater. Sci. 87 (2014) 54–64.
[26] C.S. Zhou, K.F. Li, Numerical and statistical analysis of permeability of concrete
as a random heterogeneous composite, Comput. Concr. 7 (5) (2010) 469–482.
[27] P.K. Mehta, P.J. Monteiro, Concrete: Microstructure, Properties, and Materials,
McGraw-Hill, New York, 2006.
[28] K.L. Scrivener, K.M. Nemati, The percolation of pore space in the cement paste/
aggregate interfacial zone of concrete, Cem. Concr. Res. 26 (1) (1996) 35–40.
[29] G.W. Scherer, J.J. Valenza, G. Simmons, New methods to measure liquid
permeability in porous materials, Cem. Concr. Res. 37 (3) (2007) 386–397.
[30] H. Loosveldt, Z. Lafhaj, F. Skoczylas, Experimental study of gas and liquid
permeability of a mortar, Cem. Concr. Res. 32 (9) (2002) 1357–1363.
[31] V. Picandet, D. Rangeard, A. Perrot, T. Lecompte, Permeability measurement of
fresh cement paste, Cem. Concr. Res. 41 (3) (2011) 330–338.
[32] S. Shahbeyk, M. Hosseini, M. Yaghoobi, Mesoscale finite element prediction of
concrete failure, Comput. Mater. Sci. 50 (7) (2011) 1973–1990.
[33] H.F. Ma, Study on dynamic behaviors of fully-graded dam concrete based on
meso-mechanics, PhD thesis, China Institute of Water Resources and
Hydropower Research, Beijing, 2005, (in Chinese).
[34] H.F. Ma, H.Q. Chen, B.K. Li, Influence of meso-structure heterogeneity on
bending strength of concrete, J. Hydraulic Eng. 36 (2005) 846–852.
[35] S.G. Li, Q.B. Li, Method of meshing ITZ structure in 3D meso-level finite
element analysis for concrete, Finite Elem. Anal. Des. 93 (2015) 96–106.
[36] B.G. Chen, E.X. Song, X.H. Cheng, Plane-symmetrical simulation of flow and
heat transport in fractured geological media: a discrete fracture model with
Comsol, Multiphysical Testing of Soils and Shales, Springer Berlin Heidelberg,
2013, 149–154.
[37] COMSOL Multiphysics User’s Guide, Version 5.0. COMSOL Inc, Burlington,
Massachusetts, USA, www.comsol.com, 2014.
[38] S.D. Abyaneh, H.S. Wong, N.R. Buenfeld, Modelling the diffusivity of mortar
and concrete using a three-dimensional mesostructure with several aggregate
shapes, Comput. Mater. Sci. 78 (2013) 63–73.
[39] J.J. Zheng, X.Z. Zhou, X.F. Huang, C.Q. Fu, Experiment and modeling of the effect
of aggregate shape on the chloride diffusivity of concrete, J. Mater. Civ. Eng. 26
(9) (2013) 04014048.
[40] J.P. Ollivier, J.C. Maso, B. Bourdette, Interfacial transition zone in concrete, Adv.
Cem. Based Mater. 2 (1995) 30–38.
[41] K.L. Scrivener, Backscattered electron imaging of cementitious
microstructures: understanding and quantification, Cement Concr. Compos.
26 (8) (2004) 935–945.
[42] K.M. Nemati, P. Gardoni, Microstructural and statistical evaluation of
interfacial zone percolation in concrete, Strength Fract. Complexity 3 (2)
(2005) 191–197.
[43] C.C. Yang, J.K. Su, Approximate migration coefficient of interfacial transition
zone and the effect of the aggregate content on the migration coefficient of
mortar, Cem. Concr. Res. 32 (10) (2002) 1559–1565.
[44] Y.L. Li, J. Xia, S.S. Lin, A multi-phase model for predicting the effective diffusion
coefficient of chlorides in concrete, Constr. Build. Mater. 26 (2012) 295–301.