Cement and Concrete Research 69 (2015) 10–18

Contents lists available at ScienceDirect

Cement and Concrete Research

journal homepage: http://ees.elsevier.com/CEMCON/default.asp

Electrical resistance tomography to monitor unsaturated moisture flow

in cementitious materials
Milad Hallaji a, Aku Seppänen b, Mohammad Pour-Ghaz a,⁎
a
Department of Civil Construction and Environmental Engineering, North Carolina State University, Raleigh, NC 27695, USA
b
Department of Applied Physics, University of Eastern Finland, Kuopio, Finland

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

Article history: Traditionally the electrically-based assessment of the moisture flow in cement-based materials relies on two- or
Received 18 April 2014 four-point measurements. In this paper, imaging of moisture distribution with electrical resistance tomog-
Accepted 12 November 2014 raphy (ERT) is considered. Especially, the aim is to study whether ERT could give information on unsaturat-
Available online 20 December 2014
ed moisture flows in cases where the flow is non-uniform. In the experiment, the specimens are monitored
with ERT during the water ingress. The ERT reconstructions are compared with neutron radiographs, which
Keywords:
Concrete (E)
provide high resolution information on the 2D distribution of the moisture. The results indicate that ERT is
Electrical Resistance Tomography (ERT) able to detect the moisture movement and to show approximately the shape and position of the water front
Image Analysis (B) even if the flow is nonuniform.
Transport Properties (C) © 2014 Elsevier Ltd. All rights reserved.

1. Introduction in cement-based materials. In the majority of previous studies, a set of

The durability of reinforced concrete structures is related to the abil-
ity of concrete to impede the ingress of water and aggressive agents
(e.g., [1,2]). Therefore, the water transport rate in concrete is often
used as a measure of the durability. Advanced methods have been de-
veloped to monitor the moisture movement in unsaturated cement-
based materials including nuclear magnetic resonance (NMR) spectros-
copy [3–5], methods based on attenuation of electromagnetic radiation
(e.g. gamma-ray [6–8], X-ray [9,7,10–13] and neutron [14–21] imag-
ing), and electrically-based methods [22–27].
Each of the aforementioned methods has advantages and limita-
tions. While NMR, gamma-ray, X-ray, and neutron imaging have a
high resolution, they are generally limited to small specimens (from a
few millimeters to a few centimeters, depending on the equipment
and source intensity) due to large energy requirements for imaging
large specimens. In addition, gamma-ray, X-ray and neutron imaging
are invasive methods and are mainly limited to laboratory due to the re-
quired facilities. For example, neutron imaging (radiography/tomogra-
phy) requires a neutron source, such as a nuclear reactor. Electrically-
based methods, in contrast, have a low spatial resolution, but they are
noninvasive, inexpensive and rapidly performed.
Various electrically-based methods, such as electrical impedance
spectroscopy (EIS) [26,28] and single frequency alternating current
measurements [28], have been used for monitoring unsaturated flow
⁎ Corresponding author. Tel.: +1 919 515 2235; fax: +1 919 515 7908.
E-mail address: mpourghaz@ncsu.edu (M. Pour-Ghaz).
electrode pairs was embedded in the cement-based material, and im-
pedances between pairs of electrodes were measured. For example,
McCarter et al. [22,24,27–29] monitored water ingress in concrete spec-
imens by measuring the impedance between electrode pairs embedded
at different depths. They showed that the impedance measured de-
creases significantly as the water front enters the “influence zone” of
the electrode pair. Due to the diffusive nature of the electric current,
however, the impedance between the electrodes changes even by the
moisture content change far from the height of the electrode pair. Con-
sequently, the inference of the water front location based on such mea-
surements is not a straightforward task.
To estimate the location of the water front based on pairwise imped-
ance measurements, both experimental and numerical calibration strat-
egies have been proposed. McCarter et al. [22,24,29] proposed a method
in which the rate of impedance change is monitored as a function of
time, and the arrival of the water front at the height of the electrode
pair is taken as the time at which the rate of impedance change is at
maximum. Rajabipour et al. [26] derived an analytical function based
on finite element simulations to relate the impedance between embed-
ded electrode pairs to the water front position. It is noted that the im-
pedance measurements from horizontally aligned electrode pairs can
provide estimates for the water front location only if the water front is
approximately horizontally aligned, i.e., the water flow is one-
dimensional (1D). This assumption is not always valid; the water flow
can be non-uniform for example due to the spatial variation of the
porosity.
Electrical resistance tomography (ERT) is an imaging modality
which could provide a robust tool for monitoring the moisture flow in
concrete. In ERT, the three-dimensional (3D) distribution of the

http://dx.doi.org/10.1016/j.cemconres.2014.11.007
0008-8846/© 2014 Elsevier Ltd. All rights reserved.
 M. Hallaji et al. / Cement and Concrete Research 69 (2015) 10–18
electrical resistivity is imaged using a set of current injections and elec-
trode potential measurements. ERT has been used for monitoring the
water infiltration in soil [30–34]. Buettner et al. [35,36] have also stud-
ied the feasibility of ERT for monitoring the water ingress in concrete.
Stacey [37] applied ERT for monitoring moisture movement in Berea
sandstone. While the studies of Buettner and Stacey showed the poten-
tial of ERT for monitoring the moisture flow in concrete and sandstone,
the results of ERT were not corroborated by other techniques. Recently,
duPlooy et al. [38] used ERT to monitor (essentially) 1D water ingress in
concrete slabs. One face of each slab was exposed to water for 23 days,
and ERT measurements were performed using electrodes attached to
the wetted side of the slab. The results were in agreement with destruc-
tive tests and the estimates from electrical capacimetry and ground
penetrating radar.
The objective of this study is to investigate whether ERT could give
feasible information on the distribution of moisture in cases where the
water flow is non-uniform. To this end, ERT imaging of moisture flow
is experimentally compared with high resolution neutron radiography.
Both a horizontally uniform (1D) and a nonuniform (2D) water source
are used, resulting in different shapes of the water front; in the 2D
case, the water front is not even approximately horizontally aligned.
2. Materials and methods
Two cement paste specimens were used in the experiment. A water
reservoir was mounted on top of each specimen. ERT and neutron radi-
ography measurements were simultaneously carried out during water
ingress. The entire top surface of the first specimen was exposed to
water, while in the second specimen the water reservoir covered only
about one third of the top surface. The aim of this setting was to induce
a 1D water flow in the first specimen and a 2D flow in the second one.
Photographs of the two specimens 60 min after the addition of water
are shown in Fig. 1. The dark surface areas indicate the wetted regions,
and qualitatively the difference between the two flows is clear: in the
first case (Fig. 1a) the water front is nearly horizontal, whereas in the
second case (Fig. 1b) water forms a curved plume which grows radially.
Although the photograph in Fig. 1a (and especially the neutron radio-
graphs in Section 3) shows that the water front in the first specimen is
not exactly horizontal, we refer to this test case as the 1D flow case
and the corresponding water source as the 1D source. Respectively,
the attributes of the second specimen (Fig. 1b) are referred to as the
2D flow and the 2D source.
Fig. 1. Photographs of the specimens 60 min after adding water: a) specimen with 1D water source; b) specimen with 2D water source.
11
2.1. Materials
To facilitate simultaneous ERT and neutron radiography measure-
ments, several constraints had to be addressed. First, the neutron imag-
ing requires a nuclear reactor laboratory, and the allowable time for
carrying out the experiment in this facility was limited. To reduce the
duration of the test, highly porous cement paste was used to facilitate
a rapid water ingress. The rapid moisture movement, on the other
hand, constrains the time allowed for acquiring a neutron image: if
the neutron exposure time is long, the moisture distribution changes
considerably during the exposure and causes artifacts in neutron im-
ages. To reduce the neutron exposure time, relatively thin samples
were used in the experiment. Furthermore, the initial moisture content
of the material had to be high enough; the electrical resistivity of a very
dry cementitious material is high, and this may cause problems in ERT
imaging [39]. It is also noted that enough contrast in conductivity, and
hence in moisture content, is necessary for the water movement to be
detectable by ERT. The exact threshold for this contrast depends on
many factors including the pore size distribution of the material, sensi-
tivity of and the noise in measurements, and sample and electrode size.
Establishing this contrast ratio requires experimental and computation-
al investigations. Finally, many setup accessories were made of Teflon or
aluminum, because they are almost “transparent” to neutrons and thus
do not cause artifacts in radiographs.
2.1.1. Cement paste material
Specimens used in this work were made of type I ordinary portland
cement (OPC, Type I) with a water-to-cement ratio (w/c) of 0.60. The
plain cement paste and high w/c were selected in order to produce a
highly porous material. ERT measurements can be performed on con-
crete and mortar as well [38–40]. However, the feasibility of ERT to
monitor moisture movement in concrete needs to be experimentally in-
vestigated in the future.
2.1.2. Sample preparation
The cement paste was mixed according to ASTM C305 and was cast
in a 7.6 × 7.6 × 30.5 cm steel mold. The steel mold rested on a 7.6 cm ×
30.5 cm face during casting. After casting, the mold was wrapped in a
wet cloth and a layer of plastic. The specimen was demolded after
18 h. To reduce the porosity gradient from top to bottom of the speci-
men (due to possible segregation in the high w/c cement paste) a
0.50 cm thick layer from the top of the prism was cut using wet saw
and discarded. Next, the cement paste prism was sliced into 1.37 cm
12 M. Hallaji et al. / Cement and Concrete Research 69 (2015) 10–18
thick square disks, resulting in 7.6 × 7.1 × 1.37 cm square disk speci-
mens. The accuracy of the cutting procedure was approximately ±
0.1 cm. The disks were moved to an oven at 50 °C for 5 h, and then
sealed in two layers of plastic bag and placed inside an environmental
chamber at 23 °C temperature for 21 days to achieve a relatively uni-
form moisture content throughout the specimen.
The initial moisture content was made approximately uniform to
investigate the feasibility of using ERT to monitor moisture transport
in controlled conditions, and to simplify the comparison of ERT images
and neutron radiographs. In a case of nonuniform initial moisture con-
tent, ERT imaging becomes a slightly more challenging task. However,
ERT is not limited to cases where the initial moisture content is uniform;
it is possible to account for the effect of the inhomogenous initial condi-
tion in ERT computations [41].
After conditioning, the two large surfaces of the specimens were
sealed with a thin transparent tape (packing tape) to avoid evaporation
during the test. Fig. 2a shows a photograph of the 2D specimen. The
exact dimensions of both 1D and 2D specimens are shown in Fig. 2b
and c respectively. Note that during the experiment the specimens
were placed with the longer dimension vertically. Therefore, the casting
direction – top to bottom during casting – was horizontal. The casting
direction for the 1D specimen was from right to left and for the 2D spec-
imen from left to right.
Fig. 2. (a) Photograph of the specimen with a stripe water source (2D flow), (b) schematic illustration of the specimen with uniform water source (1D flow), (c) schematic illustration of
the specimen with stripe water source (2D flow).
2.1.3. Water reservoirs
Water reservoirs were mounted on top of the specimens using a thin
layer of pure silicon caulk. The reservoirs were made of Teflon. In the 1D
source, the reservoir covered the entire top surface of the specimen,
while in the 2D source, the width of the reservoir was only 26 mm.
Fig. 2b and c schematically illustrate the locations and sizes of the
water reservoirs with dotted lines.
2.1.4. Electrodes
Surface mounted electrodes were placed at the perimeter surfaces of
the specimens. The electrodes were made by applying a thin layer of
colloidal silver paint directly to the surface of the cement paste, and
then applying a layer of conductive silver-filled epoxy on top of the
silver paint. The silver paint layer was intended to smoothen the cement
paste surface to get a better electrical contact. Electrodes were connect-
ed to the measurement unit by force contacting aluminum screws to the
surfaces of the mounted electrodes.
A total of 15 electrodes were mounted on the 1D flow specimen
(5 electrodes at each side except the top) as shown in Fig. 2b. In the
case of the 2D flow specimen, 19 electrodes were mounted, as shown
in Fig. 2c. The lengths of the electrodes were 8.0 mm in vertical direction
and covered the entire thickness of the specimen. Specimen dimensions
and the electrode numbering are shown in Figs. 2b and c. The
 M. Hallaji et al. / Cement and Concrete Research 69 (2015) 10–18
photograph in Fig. 2a shows the 2D flow specimen after the electrodes
were mounted. The spaces between the electrodes on the perimeter
surface were sealed with a transparent tape to avoid evaporation.
2.1.5. ERT measurement fixture
For simultaneous ERT and neutron radiography measurements, the
fixture shown in Fig. 3 was designed. The fixture and all connections
were made of aluminum, while the parts in contact with the aluminum
screws attached to electrodes were made of Teflon. A 5 mm thick alumi-
num sheet covered with a 2 mm thick Teflon sheet was installed at the
back of the fixture to improve the stability of the fixture and the speci-
mens. A steel mesh was secured to the back plate using aluminum
tape. The openings and the sizes of the wires provided a scale to cali-
brate the dimensions of the neutron radiographs. A triangular alumi-
num water reservoir was mounted on the fixture. The neutron image
of the water triangle was used for determining the dependence be-
tween the optical density values of the neutron images and the moisture
content (see Section 2.3).
In this work, the electrodes were placed all around the specimen.
However it is also possible to perform ERT measurements on samples
with only a single accessible side (e.g., slabs) [39]. In field applications
of ERT to such geometries the use of more advanced computational
methods such as domain truncation might become necessary.
2.2. Electrical resistance tomography imaging
In electrical resistance tomography (ERT), electric currents are
injected into an object through electrodes attached on the object's sur-
face, and corresponding to each current injection, a set of electrode po-
tentials are measured. Based on these measurements, the distribution of
Fig. 3. ERT measurement fixture. 1. Aluminum frame, 2. steel mesh, 3. teflon frame, 4. spec-
imen with 2D moisture flow, 5. water reservoirs, 6. back plate, 7. water triangle, and 8.
specimen with 1D moisture flow.
13
the electrical resistivity inside the object is reconstructed. In addition to
monitoring water ingress in concrete [35,36,38], ERT has been proposed
for detecting damage and other structural defects in concrete [39,40,
42–44]. In [41,45–48], surface breaking cracks in concrete were detect-
ed with ERT applied on conductive films on the surface of concrete
elements.
A variety of computational methods for the image reconstruction in
ERT has been developed [49]. In this paper, a standard difference imaging
approach [50] is adopted. In difference imaging, a temporally varying
object is imaged on the basis of the difference between ERT measure-
ments corresponding to two time instants. The difference data is used
for reconstructing the change of the resistivity between the two instants.
The difference reconstruction is conventionally based on a global linear-
ization of the non-linear observation model of ERT, and consequently,
the reconstructions are often qualitative in nature especially if the tem-
poral variation of the resistivity is large or if the linearization point is
poorly selected. On the other hand, difference imaging is relatively tol-
erant to modeling errors, and hence widely used in many ERT applica-
tions. In this study, the specimen before the start of the water ingress
was used as the reference state for difference imaging. The software
used in the ERT reconstructions was based on Matlab implementations
by Vauhkonen et al. [51,52].
Since in this study the electrodes on the perimeter surface covered
the entire thickness of a specimen, the ERT measurements did not
carry information on the (possible) variation of the electrical resistivity
in the direction perpendicular to the two large surfaces of the speci-
mens. In other words, the setting only allowed for 2D ERT imaging. On
the other hand, the neutron radiographs are also 2D images, and
hence, 2D ERT reconstructions were sufficient for the comparisons
aimed at in this study. It is also worth noticing that the thin sample ge-
ometry was determined by the neutron radiography; if the specimens
had been thicker, a longer neutron exposure would have been needed.
ERT, by contrast, is not limited to 2D geometries; in most applications
3D ERT is used [53].
For simultaneous ERT and neutron radiography imaging, the ERT fix-
ture (Fig. 3) was positioned inside a neutron chamber. The ERT mea-
surement equipment was connected to the fixture with 3-meter-long,
shielded, high quality, low resistance cables. An in-house developed
equipment set [48,41] was used in the ERT measurements. The equip-
ment consisted of a current supply, a voltmeter, a manual switch system
and a computer for the data collection. A sinusoidal alternating current
(AC) with an amplitude of 2.0 mA and a frequency of 1 kHz was used in
all measurements. The frequency was chosen based on electrical imped-
ance spectroscopy (EIS) measurements. The frequency was kept as low
as possible to avoid inductance effects due to the long cables connecting
electrodes to the ERT measurement unit, and on the other hand, the fre-
quency had to be high enough to avoid electrode polarization effects.
With the selected frequency, the imaginary component of the imped-
ance was negligible. The same frequency was used by McCarter [27] to
monitor moisture penetration in concrete.
For ERT, 31 current injections were applied in the 1D flow case, and
39 injections in the 2D flow case. The currents were injected between
electrodes 2–14, 3–13, 4–12, 5–11 and electrodes i − j, i = {1,15}, j =
{1,…,N}, i ≠ j, where N is the number of electrodes (N = 15 in the 1D
flow case, and N = 19 in the 2D flow case). Corresponding to each cur-
rent injection, the electric potentials on all electrodes were measured
against a common ground, which was connected to one of the current
carrying electrodes. This resulted in a total of 465 potential measure-
ments in the 1D flow case and 741 measurements in the 2D flow case.
Since the current injections and potential measurements were switched
manually, collecting one data set took approximately 4 min. It is noted,
however, that for an automatic ERT system carrying out the same mea-
surements would take less than 1 s.
For both samples, the first set of ERT measurements was carried out
before water was added to the reservoir. These data sets were used as
reference measurements in the difference imaging. After adding
14 M. Hallaji et al. / Cement and Concrete Research 69 (2015) 10–18
water, multiple ERT measurements were performed on each specimen.
For each specimen, a set of three ERT measurements are shown in this
paper. In the case of the specimen with the 1D water source, the times
of the measurements were 23 min, 32 min and 15 h 50 min. Here, the
reported times indicate the instants of completing an ERT measurement
cycle, counted from the instant of adding water to the reservoir. The cor-
responding times for the specimen with the 2D water source were
17 min, 28 min and 16 h 30 min. The measurement times were chosen
to match with the times of neutron radiography exposures, although an
exact match was not possible, due to long measurement times and some
preparations that were required between the measurements.
2.3. Neutron radiography
The experiment was carried out at the PULSTAR nuclear reactor of
North Carolina State University. The neutron beam is comprised of ther-
mal neutrons with a nominal kinetic energy of approximately 0.05 eV.
The diameter of the beam at the imaging plane is 48 cm. The collimation
ratio L/D is variable from 100 to 150 (where L is the distance between
aperture and imaging plane and D is the diameter of the entrance aper-
ture). The neutron flux at the image plane is approximately 1.8 ×
106 n cm−2 s−1. A conventional X-ray film using a Gadolinium convert-
er screen and a vacuum cassette was used. The dimensions of the film
were 35.5 cm × 43.2 cm, covering the entire EIT fixture. The maximum
resolution of the images using the film is approximately 33 μm [54].
Based on a series of trial measurements, 9 min was identified as a
feasible duration of the neutron exposure. The trial measurements
were also used for assessing the length of a transition period right
after adding water, during which the flow rate was high: after approxi-
mately 8 min the water movement was slow enough to obtain clear ra-
diographic images with 9-minute neutron exposure.
A total of four neutron radiographic images were taken. The first
image was taken before adding water to the reservoirs. This radiograph
was used as a reference for the rest of the neutron images (see below).
The second neutron exposure started 10 min after adding water and
was completed 9 min later. This radiograph is referred to as the
19 min image hereafter. Correspondingly, the neutron exposures for
the next two images were completed 35 min and 16 h after the addition
of water. To avoid water depleting during the experiment, more water
was added to the reservoirs after the second neutron exposure. Howev-
er, the water reservoirs of both specimens got empty between the third
and fourth neutron exposures.
To quantify the moisture movement, the radiographs were inter-
preted in terms of the moisture content based on the neutron images
of a triangular water reservoir (Section 2.1.5) as follows (cf. [55]). The
neutron images of the triangular water reservoir (Section 2.1.5) provid-
ed neutron data of a linearly increasing thickness t of water. The water
thickness t was converted to volumetric moisture content w (corre-
sponding to the thickness of the specimens ts) simply by division
w = t/ts. The pixel-wise optical density I in the neutron radiographs is
linearly proportional to the local intensity of the neutron flux. To ap-
proximate the dependence between w and I, a sixth order polynomial
w = f(I) was fitted to I vs. w values. This low order approximate
model provided an adequate fit to the data.
In the cement paste, the attenuation is not solely caused by the
water component, and consequently, the polynomial approximation
w = f(I) determined using the water triangle observations cannot be di-
rectly used for converting the pixel-wise optical density I to the volu-
metric moisture content w of the cement paste. However, the change
of the volumetric moisture content, Δw, can be estimated by subtraction
Δw ¼ w−winit , where winit ¼ f ðI init Þ and w ¼ f ðI Þ, respectively, are ob-
tained from the neutron data of the sample at the initial state and after
wetting. Here, winit ¼ winit þ wc and w ¼ w þ wc , where winit and w are
the actual water contents and wc can be interpreted as an equivalent
water content, i.e., a water thickness that causes neutron attenuation
equal with the attenuation caused by a completely dry specimen. Before
the conversions, the optical intensity images were smoothed by low
pass filtering, to reduce the effects of randomness in the radiation and
attenuation.
The above-described analysis of the neutron images differs slightly
from the analysis made in studies [12,55–58] where the neutron attenu-
ation coefficient of water, μw, was determined experimentally, and Beer's
Law was used for converting the neutron intensities to water contents. A
different analysis was used, because in the present experiment, the
water contents were high, up to 0.40. Due to the well-known beam
hardening effect, the Beer's law is applicable only for thin layers of
water, typically less than 1 mm total water thickness [59], which corre-
sponds to a water content of 0.073 in the present specimen geometry.
3. Results and discussion
The results of the 1D and 2D water flow cases are shown in Figs. 4
and 5, respectively. In both figures, the results of neutron radiography
are illustrated in the left column and the ERT reconstructions in the
right column. The (finishing) time of the neutron exposure/ERT mea-
surements is shown under each image. Note that due to the rather
long durations of the neutron radiography (9 min) and ERT measure-
ments (4 min), it was not possible to exactly match the times of the im-
ages. Since ERT measurements were faster than neutron radiography,
multiple ERT measurements were performed during the experiment
and the ERT measurements that most closely matched the time of neu-
tron radiography measurements were used for comparison. However,
since the water penetration slowed down approximately 8 min after
the exposure to water, the radiograph–ERT image pairs in each row in
Figs. 4 and 5 are comparable despite the time lags between the images.
All six neutron images in Figs. 4 and 5 are represented in a mutual
colorscale. The black and white color, respectively, correspond to the
maximum and minimum water content change Δw over all six neutron
images. The ERT images are shown in individual colorscales, i.e. the min-
imum and maximum value of each ERT reconstruction is taken as the
minimum and maximum limit of the colorscale in each ERT image, re-
spectively. This choice is made, because the difference in ERT recon-
structions considered in this paper are qualitative in nature. The use of
a common colorscale for all six ERT images would render the contrast
in some of the images extremely low, because the range of values varies
between the images. Hence, the ERT reconstructions in Figs. 4 and 5 can-
not be interpreted in terms of absolute moisture content. Instead, they
only provide qualitative information on how the moisture is spatially
distributed.
Qualitatively, a large negative resistivity change Δρ depicted with
red color in the ERT reconstructions should indicate a high water con-
tent change Δw depicted with black in the neutron images. Respective-
ly, a positive resistivity change should indicate the decrease of the water
content. Note, however, that the occurrences of the positive resistivity
change are most probably reconstruction artifacts — in a setup where
water ingresses to porous medium, the change of the resistivity with re-
spect to the initial state is expected to be non-positive. Correspondingly,
the negative values in Δw are probably artifacts — caused by imperfect
alignment of the neutron images corresponding to the initial state and
the subsequent states.
3.1. 1D source case
The water ingress into the 1D specimen is shown in the neutron im-
ages in Fig. 4 (left column). In the 19 min image, the water front has
moved approximately 2.0 cm from top of the specimen. After 35 min,
the distance is about 2.6 cm, as evidenced by the boundary between
the red and yellow color. These images also reveal a nonuniform pene-
tration of water. Clearly, water flows slightly faster on the right hand
side of the specimen. Moreover, the water content is higher (compared
to other wetted region) at the corners of the specimen, especially at the
top right corner. In the 16 h image, water has penetrated further. Again,
 M. Hallaji et al. / Cement and Concrete Research 69 (2015) 10–18
19 min
35 min
16 h
Fig. 4. Neutron radiography image (left column) and ERT images (right column) of the specimen with one-dimensional (1D) water source. The times of the neutron exposures/ERT mea-
surements are shown under the images.
water flow has been slightly faster on the right hand side of the speci-
men, and also this image shows the highest water content around the
top corners. As noted above, water in the reservoir depleted between
35 min and 16 h. In the absence of the external water source, the mois-
ture gradient within the specimen has decreased.
The nonuniform penetration of water and the high water content at
the corners of the specimen are probably caused by 1) a porosity gradi-
ent within the sample due to the high w/c and segregation, and 2) dam-
ages at the corners, induced by cutting the samples. Although measures
were taken to minimize damages, they may have occurred as the mate-
rial had a high w/c and the samples were sliced after only 18 h of curing.
The ERT reconstructions in Fig. 4 (right column) capture the mois-
ture movement relatively well. In the 23 min ERT image, the area of
the low resistivity (high negative resistivity change) corresponds well
to the wetted area in the 19 min neutron image (Fig. 4, first row). The
same applies to the pair of images in the second row (35 min
15
23 min
32 min
15 h 50 min
neutron–32 min ERT). Further, the 15 h 50 min ERT image clearly
shows that the moisture has moved downwards.
It is worth noticing that 23 min and 32 min ERT images include
wide green areas indicating a small negative change of the resistivity
in positions where Δw is zero and no change of the electrical resistivity
is expected. These errors in the reconstructions are partly due to
smoothness-promoting regularization used in the solution of the
ERT inverse problem and partly due to the linearization made in
the difference imaging approach — and also due to the ill-
posedness of the ERT inverse problem. Nevertheless, the movement
of the water front is very clearly indicated by ERT images in Fig. 4.
The ERT images also reveal the nonuniformity of the flow: ERT
shows correctly that the movement of the water front is fastest on
the right hand side of the specimen. Also, the highest changes in
the electrical resistivity are detected at the highest moisture content
region — the top corners of the specimen.
16 M. Hallaji et al. / Cement and Concrete Research 69 (2015) 10–18
19 min
35 min
16 h
Fig. 5. Neutron radiography image (left column) and ERT images (right column) of the specimen with two-dimensional (2D) water source. The times of the neutron exposures/ERT mea-
surements are shown under the images.
3.2. 2D source case
The neutron radiographs in Fig. 5 (left column) illustrate the mois-
ture movement in the specimen with 2D (strip) water source. The
shape of the plume in the 19 min and 35 min neutron images is similar
to that observed visually in the 60 min photograph (Fig. 1b). The size of
the plume has increased between 19 min and 35 min. Although the
water source was in the middle of the top surface of the specimen,
water has moved nonsymmetrically; the plume is off-centered and
slightly inclined towards the left. This is again probably due to the po-
rosity gradient within the specimen. The 16 h neutron image shows
that the depletion of the water reservoir has again resulted in the de-
crease of the moisture gradient by diffusion within the specimen. More-
over, the diameter of the plume has further increased with time to the
17 min
28 min
16 h 30 min
extent that at 16 h it covers the entire width of the specimen, and the
moisture is concentrated towards the top left corner of the specimen.
The ERT reconstructions are shown in the right column in Fig. 5. The
17 min and 28 min images illustrate the 2D flow well. The position and
shape of the plume is approximately tracked: The widths of the areas of
low resistivity in the 17 min and 28 min ERT images correspond well to
the widths of the wetted areas in the 19 min and 35 min radiographs.
The depth of the plume is more difficult to assess based on the ERT im-
ages, because of the smoothness of the images discussed above. Howev-
er, a slight growth of the plume can be detected based on the 17 min and
28 min ERT images. Moreover, these images show the non-symmetry of
the moisture distribution: resembling the corresponding neutron radio-
graphs, the low resistivity areas (depicted with red) are inclined to-
wards the left. Finally, the 16 h 30 min ERT reconstruction indicates
 M. Hallaji et al. / Cement and Concrete Research 69 (2015) 10–18
that the moisture has concentrated around the top left corner of the
specimen — a phenomenon shown also by the 16 h neutron image.
It is worth noticing that the three ERT images in Fig. 5 are represent-
ed in different colorscales. This choice was made because of the qualita-
tive nature of the difference reconstructions; in 17 min and 28 min ERT
images, the estimated resistivity change is much lower than in the 16 h
30 min image. The values of the moisture content cannot be quantita-
tively determined based on these ERT images. However, the series of
three ERT images gives a very good qualitative insight of the moisture
movement in the specimen.
As discussed above, most of the ERT reconstructions are over-
smoothed. This is related to the so-called smoothness promoting regu-
larization used in the image reconstruction [40]. With a smaller degree
of regularization the reconstructions become noisy. The spatial resolu-
tion of the ERT images could perhaps be improved by the absolute imag-
ing approach and application-specific prior models for the resistivity (cf.
[41]). Via development of the reconstruction methods, ERT might even
become a tool for assessing the moisture content quantitatively. In this
paper, however, we only adopted standard reconstruction methods to
study the capability of ERT for qualitatively detecting the moisture
flow in cement paste. The development of advanced computational
methods for this special application is left for future studies.
4. Summary and conclusions
In this paper, the feasibility of electrical resistance tomography
(ERT) for monitoring unsaturated moisture ingress within cement-
based materials was studied. An experiment was set up in which simul-
taneous ERT and neutron radiography measurements were performed
during the water ingress in cement paste. The neutron radiographs
served as quantitative high resolution 2D moisture distribution images,
against which the ERT reconstructions were compared. Two specimens
were prepared for the experiments: the first specimen with (nearly)
one-dimensional water penetration, and the second one with two-
dimensional flow.
In both 1D and 2D flow cases, ERT provided feasible information on
the distribution of the moisture within the specimen. Although the ERT
reconstructions were not quantitative, they indicated the movement of
the moisture and an approximate shape and position of the water front.
ERT also showed the nonuniformities and nonsymmetries of the flow,
which were probably caused by porosity gradients within the speci-
mens. A good agreement was observed between the ERT reconstruc-
tions and neutron radiographs in both 1D and 2D flow cases. The
results of this study strongly support the feasibility of ERT for imaging
moisture distributions in cementitious materials.
In this paper, the specimens were made of cement paste. However,
ERT is also applicable to imaging mortar and concrete, and it might pro-
vide a robust tool for monitoring the water ingress in these materials.
Acknowledgements
This work was performed in the Materials and Sensor Development
Laboratory (MSDL), Constructed Facilities Laboratory (CFL), the
PULSTAR nuclear reactor facility at North Carolina State University,
and in the Department of Applied Physics at University of Eastern
Finland. The authors would like to acknowledge the support which
has made these laboratories and this research possible. The authors
would like to acknowledge the contributions of the technical staff of
CFL, Dr. Gregory Lucier, Mr. Jerry Atkinson, and Mr. Johathan McEntire,
as well as the contributions of the technical staff of the PULSTAR nuclear
reactor facility at North Carolina State University, Mr. Scott Lassell, Mr.
Greg Gibson and Mr. Andrew Cook. The second author would like to ac-
knowledge the support provided by Academy of Finland (projects
140280, 270174 and 273536). The third author of this paper was in
part supported by National Science Foundation (Grant # 1230284)
which is greatly acknowledged. Any opinions, findings, and conclusions
17
or recommendations expressed in this paper are those of the author(s) and
do not necessarily reflect the views of the National Science Foundation.
References
[1] A. Neville, Properties of Concrete, 4th edition Wiley, New York, 1996.
[2] F. Ghasemzadeh, M. Pour-Ghaz, The effect of damage on moisture transport in con-
crete, J. Mater Civil, Eng, 2014, http://dx.doi.org/10.1061/(ASCE)MT.1943-5533.
0001211.
[3] R. Gummerson, C. Hall, W. Hoff, R. Hawkes, G. Holland, W. Moore, Unsaturated
water flow within porous materials observed by NMR imaging, Nature 281
(1979) 56–57.
[4] Q. Chen, M. Gingras, B. Balcom, A magnetic resonance study of pore filling processes
during spontaneous imbibition in Berea sandstone, J. Chem. Phys. 119 (18) (2003)
9609–9616.
[5] C. Hall, W. Hoff, Water Transport in Brick, Stone and Concrete, 2nd edition Taylor &
Francis, London and New York, 2012.
[6] A. Nielsen, Gamma-ray-attenuation used for measuring the moisture content and
homogeneity of porous concrete, Build. Sci. 7 (4) (1972) 257–263.
[7] S. Roels, J. Carmeliet, H. Hens, O. Adan, H. Brocken, R. Cerny, Z. Pavlik, A. Ellis, C. Hall,
K. Kumaran, A comparison of different techniques to quantify moisture content pro-
files in porous building materials, J. Therm. Envel. Build. Sci. 27 (4) (2004) 261–276.
[8] M. Nizovtsev, S. Stankus, A. Sterlyagov, V. Terekhov, K. R.A., Determination of mois-
ture diffusivity in porous materials using gamma-method, Int. J. Heat Mass Transfer
54 (17–18) (2008) 4161–4167.
[9] S. Roels, K. Vandersteen, J. Carmeliet, Measuring and simulating moisture uptake in
a fractured porous medium, Adv. Water Resour. 26 (3) (2003) 237–246.
[10] S. Roels, J. Carmeliet, Analysis of moisture flow in porous materials using microfocus
X-ray radiography, Int. J. Heat Mass Transfer 49 (25–26) (2006) 4762–4772.
[11] P. Baker, D. Bailly, M. Campbell, G. Galbraith, R. McLean, N. Poffa, C. Sanders, The
application of X-ray absorption to building moisture transport studies, Measure-
ment 40 (9–10) (2007) 951–959.
[12] M. Pour-Ghaz, F. Rajabipour, J. Couch, J. Weiss, Numerical and experimental assess-
ment of unsaturated fluid transport in saw-cut (notched) concrete elements, ACI
Spec. Publ. 266 (2009) 73–86.
[13] M. Zhang, Y. He, G. Ye, D. Lange, K. Breugel, Computational investigation on mass
diffusivity in Portland cement paste based on X-ray computed microtomography
(μct) image, Constr. Build. Mater. 27 (1) (2012) 472–481.
[14] H. Pleinert, H. Sadouki, F. Wittmann, Determination of moisture distributions in
porous building materials by neutron transmission analysis, Mater. Struct. 31 (4)
(1998) 218–224.
[15] V. Cnudde, M. Dierick, J. Vlassenbroeck, B. Masschaele, E. Lehmann, P. Jacobs, L. Van
Hoorebeke, High-speed neutron radiography for monitoring the water absorption
by capillarity in porous materials, Nucl. Inst. Methods Phys. Res. B 266 (1) (2008)
155–163.
[16] D. Brew, F. de Beer, M. Radebe, R. Nshimirimana, P. McGlinn, L. Aldridge, T. Payne,
Water transport through cement-based barriers — a preliminary study using neu-
tron radiography and tomography, Nucl. Inst. Methods Phys. Res. A 605 (1–2)
(2009) 163–166.
[17] A. El Abd, A. Czachor, J. Milczarek, Neutron radiography determination of water
diffusivity in fired clay brick, Appl. Radiat. Isot. 67 (4) (2009) 556–559.
[18] M. Kanematsu, I. Maruyama, T. Noguchi, H. Iikura, T. N, Quantification of water pen-
etration into concrete through cracks by neutron radiography, Nucl. Inst. Methods
Phys. Res. A 605 (1–2) (2009) 154–158.
[19] P. Zhang, F. Wittmann, T. Zhao, E. Lehmann, L. Tian, P. Vontobel, Observation and
quantification of water penetration into strain hardening cement-based composites
(SHCC) with multiple cracks by means of neutron radiography, Nucl. Inst. Methods
Phys. Res. A 620 (2–3) (2010) 414–420.
[20] P. Zhang, F. Wittmann, T. Zhao, L. E.H., Neutron imaging of water penetration into
cracked steel reinforced concrete, Phys. B Condens. Matter 405 (7) (2010)
1866–1871.
[21] P. Zhang, F. Wittmann, T. Zhao, L. E.H., P. Vontobel, Neutron radiography, a powerful
method to determine time-dependent moisture distributions in concrete, Nucl. Eng.
Des. 241 (12) (2011) 4758–4766.
[22] W. McCarter, M. Emerson, H. Ezirim, Properties of concrete in the cover zone: devel-
opments in monitoring techniques, Mag. Concr. Res. 47 (172) (1995) 243–251.
[23] W. McCarter, Monitoring the influence of water and ionic ingress on cover zone
concrete subjected to repeated absorption, Cement Concr. Aggr. 18 (1) (1996)
55–63.
[24] W. McCarter, H. Ezirim, M. Emerson, Properties of concrete in the cover zone: water
penetration, sorptivity, and ionic ingress, Mag. Concr. Res. 48 (176) (1996) 149–156.
[25] W. McCarter, D. Watson, Wetting and drying of coverzone concrete, Inst. Civ. Eng.
Struct. Build., vol. 112, 1997, pp. 227–236.
[26] F. Rajabipour, J. Weiss, J. Shane, T. Mason, P. Shah, Procedure to interpret electrical
conductivity measurements in cover concrete during rewetting, J. Mater. Civil Eng.
17 (5) (2005) 586–594.
[27] W. McCarter, T. Chrisp, G. Starrs, A. Adamson, E. Owens, P. Basheer, S. Nanukuttan, S.
Srinivasan, N. Holmes, Developments in performance monitoring of concrete ex-
posed to extreme environments, J. Infrastruct. Syst. 18 (3) (2012) 167–175.
[28] W. McCarter, G. Starrs, A. Adamson, T. Chrisp, P. Basheer, S. Nanukuttan, S.
Srinivasan, C. Green, Influence of different European cements on the hydration of
cover-zone concrete during the curing and postcuring periods, J. Mater. Civil Eng.
25 (9) (2013) 1335–1343.
[29] W. McCarter, M. Chrisp, Monitoring water and ionic penetration into cover-zone
concrete, ACI Mater. J. 97 (2000) 667–674.
18 M. Hallaji et al. / Cement and Concrete Research 69 (2015) 10–18
[30] W. Daily, A. Ramirez, D. LaBrecque, J. Nitano, Electrical resistivity tomography of
vadose water movement, Water Resour. Res. 28 (5) (1992) 1429–1442.
[31] A. Borsic, C. Comina, S. Foti, R. Lancellotta, G. Musso, Imaging heterogeneities with
electrical impedance tomography: laboratory results, Geotechnique 55 (7) (2005)
539–547.
[32] C. Comina, R. Cosentini, G. Della Vecchia, S. Foti, G. Musso, Hydrochemomechanical
processes in soil samples: monitoring through electrical resistivity tomography, EPJ
Web of Conferences, vol. 6, EDP Sciences, 2010, p. 22012.
[33] C. Comina, R.M. Cosentini, G. Della Vecchia, S. Foti, G. Musso, 3D-electrical resistivity
tomography monitoring of salt transport in homogeneous and layered soil samples,
Acta Geotech. 6 (4) (2011) 195–203.
[34] R.M. Cosentini, G. Della Vecchia, S. Foti, G. Musso, Estimation of the hydraulic pa-
rameters of unsaturated samples by electrical resistivity tomography, Geotechnique
62 (7) (2012) 583–594.
[35] M. Buettner, A. Ramirez, W. Daily, Electrical resistance tomography for imaging c-
oncrete structures, Structural Materials Technology an NDT Conference, San Diego,
California, 1996.
[36] M. Buettner, A. Ramirez, W. Daily, Electrical resistance tomography for imaging
the spatial distribution of moisture in pavement sections, Structural Materials
Technology an NDT Conference, San Diego, California, 1996.
[37] R. Stacey, Electrical impedance tomography, geothermal program interdisciplinary
research in engineering and earth sciences, Tech. Rep. SGP-TR-182, Stanford Univer-
sity, 2006.
[38] R. Du Plooy, G. Villain, S. Palma Lopes, A. Ihamouten, X. Dérobert, B. Thauvin, Electro-
magnetic non-destructive evaluation techniques for the monitoring of water and
chloride ingress into concrete: a comparative study, Mater. Struct. (2013), http://
dx.doi.org/10.1617/s11527-013-0189-z (Published online).
[39] K. Karhunen, A. Seppänen, A. Lehikoinen, J. Blunt, J. Kaipio, J. Monteiro, Electrical re-
sistance tomography for assessment of cracks in concrete, ACI Mater. J. 107 (5)
(2010) 523–531.
[40] K. Karhunen, A. Seppänen, A. Lehikoinen, P. Monteiro, J. Kaipio, Electrical resistance
tomography imaging of concrete, Cem. Concr. Res. 40 (1) (2010) 137–145.
[41] M. Hallaji, A. Seppänen, M. Pour-Ghaz, Electrical impedance tomography-based
sensing skin for quantitative imaging of damage in concrete, Smart Materials and
Structures 23 (8) 085001.
[42] T. Hou, J. Lynch, Tomographic imaging of crack damage in cementitious structural
components, 4th International Conference on Earthquake Engineering, Taipei,
Taiwan, 2006.
[43] T. Hou, J. Lynch, Electrical impedance tomographic methods for sensing strain fields
and crack damage in cementitious structures, J. Intell. Mater. Syst. Struct. 20 (2009)
1363–1379.
[44] K. Karhunen, A. Seppänen, J. Kaipio, Adaptive meshing approach to identification of cracks
with electrical impedance tomography, Inverse Probl. Imaging 8 (1) (2014) 127–148.
[45] K. Loh, T. Hou, J. Lynch, Nanotube-based sensing skins for crack detection and im-
pact monitoring of structures, 6th International Workshop on Structural Health
Monitoring, Stanford, California, 2007.
[46] K. Loh, T. Hou, J. Lynch, N. Kotov, Carbon nanotube sensing skins for spatial strain
and impact damage identification, J. Nondestruct. Eval. 28 (2009) 9–25.
[47] D. Tawil, D. Rye, M. Velonaki, Improved image reconstruction for an EIT based sen-
sitive skin with multiple internal electrodes, IEEE Trans. Robot. 27 (3) (2011)
425–435.
[48] M. Hallaji, M. Pour-Ghaz, A new sensing skin for qualitative damage detection in
concrete elements: rapid difference imaging with electrical resistance tomography,
NDT & E INT. 68 (2014) 13–21.
[49] L. Borcea, Electrical impedance tomography, Inverse Probl. 18 (2002) R99–R136.
[50] D. Barber, B. Brown, Applied potential tomography, J. Phys. E Sci. Instrum. 17 (1984)
723–733.
[51] M. Vauhkonen, D. Vadasz, J. Kaipio, E. Somersalo, P. Karjalainen, Tikhonov regulari-
zation and prior information in electrical impedance tomography, IEEE Trans. Med.
Imaging 17 (2) (1998) 285–293.
[52] M. Vauhkonen, W. Lionheart, L. Heikkinen, P. Vauhkonen, J. Kaipio, A MATLAB pack-
age for the EIDORS project to reconstruct two-dimensional EIT images, Physiol.
Meas. 22 (2001) 107–111.
[53] P. Vauhkonen, M. Vauhkonen, T. Savolainen, J. Kaipio, Three-dimensional electrical
impedance tomography based on the complete electrode model, IEEE Trans.
Biomed. Eng. 46 (9) (1999) 1150–1160.
[54] K. Kaushal, I. Hawari, V. Gillette, Design and performance of a thermal neutron im-
aging facility at the North Carolina State University PULSTAR reactor, IEEE Trans.
Nucl. Sci. 53 (6) (2006) 3904–3911.
[55] M. Pour-Ghaz, Detecting Damage in Concrete Using Electrical Methods and
Assessing Moisture Movement in Cracked Concrete(Ph.D. thesis) Purdue University,
Indiana, 2011.
[56] G. Sant, A. Eberhardt, D. Bentz, J. Weiss, Influence of shrinkage-reducing admixtures
on moisture absorption in cementitious materials at early ages, J. Mater. Civil Eng.
22 (3) (2010) 277–286.
[57] G. Sant, J. Weiss, Using X-ray absorption to assess moisture movement in cement-
based materials, J, ASTM Int. 6 (9) (2009).
[58] J. Weiss, M. Geiker, K. Hansen, Using X-ray Absorption to Detect Fluid Ingress in
Cracked Concrete, Purdue University, Tech. rep., 2005
[59] D. Hussey, K. Coakley, D. Vecchia, D. Jacobson, M. Arif, In situ fuel cell water metrol-
ogy at the NIST neutron imaging facility, J. Fuel Cell Sci. Technol. 7 (2) (2010)
021024.