sensors

Article
Novel Electromagnetic Sensors Embedded in
Reinforced Concrete Beams for Crack Detection
Michaela Gkantou 1 , Magomed Muradov 1,2 , George S. Kamaris 1 , Khalid Hashim 1 ,
William Atherton 1 and Patryk Kot 1, *
1 Built Environment and Sustainable Technologies (BEST) Research Institute, Liverpool John Moores
University, Liverpool L3 3AF, UK
2 Sensor City Liverpool Limited, 31 Russell Street, Liverpool L3 5LJ, UK
* Correspondence: p.kot@ljmu.ac.uk; Tel.: +44-0151-231-2420




Received: 30 September 2019; Accepted: 23 November 2019; Published: 26 November 2019


Abstract: This paper investigates the possibility of applying novel microwave sensors for crack
detection in reinforced concrete structures. Initially, a microstrip patch antenna with a split
ring resonator (SRR) structure was designed, simulated and fabricated. To evaluate the sensor’s
performance, a series of structural tests were carried out and the sensor responses were monitored.
Four reinforced concrete (RC) beam specimens, designed according to the European Standards, were
tested under three-point bending. The load was applied incrementally to the beams and the static
responses were monitored via the use of a load cell, displacement transducers and crack width gauges
(Demec studs). In parallel, signal readings from the microwave sensors, which were employed prior
to the casting of the concrete and located along the neutral axis at the mid-span of the beam, were
recorded at various load increments. The microwave measurements were analysed and compared
with those from crack width gauges. A strong linear relationship between the crack propagation
and the electromagnetic signal across the full captured spectrum was found, demonstrating the
technique’s capability and its potential for further research, offering a reliable, low-cost option for
structural health monitoring (SHM).

Keywords: concrete beams; crack detection; electromagnetics; non-destructive; sensor

1. Introduction
The maintenance of structures is essential in various industries, including civil, offshore, marine
and transportation engineering. The inspection of structures in order to detect structural damage and to
assess the condition of materials is an important part of the maintenance process. Crack formation and
propagation due to fatigue or earthquake loading, corrosion in metals and delamination in composites
can be addressed via effective maintenance [1]. Effective maintenance not only improves the health and
safety standards of a structure, but can also allow for cost savings. Often costs related to continuous
monitoring, inspection and maintenance can affect civil infrastructure systems and potentially result in
decision-making under uncertain environments. This is particularly common in highways and railway
bridges, where an over-stretched transport network is self-evident [2]. Highways England alone has
an annual budget of over £1 billion allocated to the maintenance of UK’s strategic highways, which
make up approximately 10% of the overall UK road network [3].

Sensors 2019, 19, 5175; doi:10.3390/s19235175 www.mdpi.com/journal/sensors

Sensors 2019, 19, 5175 2 of 14

Structural health monitoring (SHM) provides continuous feedback regarding the “state” of the
constituent materials and structural members. Continuous monitoring over a period of time offers a
full history database for a structure, which can be used to provide a prognosis (evaluation of damage,
residual life, etc.) [4]. Deterioration owing to usage, environmental conditions and accidents can be
monitored and utilised in order to assess whether the condition of a structure is in accordance with
design standard limits and requirements.
In recent years, different SHM technologies have been investigated aiming to minimise the
costs related to damage assessment and maintenance works, while securing the reliability of
the structural performance. Traditional sensor technologies (e.g., strain sensors, displacement
transducers, accelerometers, etc.) can be utilized, along with traditional visual inspection and
damage detection, for structures subjected to various load conditions, such as earthquakes, wind and
fatigue loading. In addition to traditional sensors, new measuring approaches (e.g., GPS, terrestrial laser
scanning, vision-based systems, etc.) have been applied for structural health monitoring purposes [5].
Non-destructive evaluation (NDE) can be time-consuming and expensive, and access is not always
possible [6]. Both wired and wireless sensor technologies are available, each of them having their own
advantages and limitations, while the main concern should always be obtaining high accuracy data.
Existing wired sensor techniques can be expensive in terms of additional costs and the space required
for cable installation, while wireless sensors require due consideration for the power supply.
Aiming to complement existing sensing technologies for SHM, the present study investigates
the potential of using microwave sensors for crack detection in concrete structures. Recently,
electromagnetic sensors have received interest as a method to monitor and predict the excess moisture
content in concrete structures due to exposure [7–9]. The use of microwave technology offers low
cost, low profile flexibility in the design and the adaptation of sensors. This study will demonstrate a
potential sensor solution for the detection and prediction of crack propagation in reinforced concrete
(RC) beams. Note that wired sensors are examined herein, whereas wireless microwave sensor
technologies are planned to be studied in a future research programme, in which energy harvesting
systems will be considered as a power source for the sensors.

2. Design of the Experiments

An experimental programme was designed and carried out in order to investigate the accuracy
of the microwave sensing technology in predicting the crack formation in reinforced concrete beams.
The current section provides details regarding the design of the RC beams (Section 2.1) and of the
microwave sensors (Section 2.2).

2.1. Design of the Beams

Four RC beams were designed in accordance with Eurocode 2 [10] to be used for the experimental
set-up of this study. The total length of the beams was 1400 mm and their cross-section was rectangular
with dimensions of 200 mm × 200 mm. The beams were subjected to a vertical design load equal to a
maximum of 100 kN based on the capacity of the actuator and associated load cell. Grade C40/50 was
used for the concrete and steel grade S500 (Scimitar Steels, Liverpool, UK) for the reinforcement. The
cover to the reinforcement for bond and durability was 30 mm. Steel bars with diameters of 20 mm and 8
mm were used for the longitudinal reinforcement and the hanger bars of the beams, respectively. Finally,
shear links (Scimitar Steels, Liverpool, UK) with diameter 8 mm with 100 mm spacing’s were used for the
shear reinforcement of the beams. The design of the beams is shown in Figure 1. The microwave sensor
(Liverpool John Moores University, Liverpool, UK) was placed 100 mm from the bottom of the concrete
specimen in order to capture the crack propagation in the tensile zone just below the neutral axis.
diameters of 20 mm and 8 mm were used for the longitudinal reinforcement and the hanger bars of
the beams, respectively. Finally, shear links (Scimitar Steels, Liverpool, UK) with diameter 8 mm with
100 mm spacing’s were used for the shear reinforcement of the beams. The design of the beams is
shown in Figure 1. The microwave sensor (Liverpool John Moores University, Liverpool, UK) was
placed
Sensors 100
2019, mm from the bottom of the concrete specimen in order to capture the crack propagation
19, 5175 3 of 14
in the tensile zone just below the neutral axis.

Sensors 2019, 19, x FOR PEER REVIEW 3 of 14

(a)
(a)

(b)
Figure1.1.Design
Figure Designdetails
detailsofofthe
thebeams
beamsused
usedfor
forthe
theexperimental
experimentalset-up
set-up
inin this
this study:(a)(a)Longitudinal
study: Longitudinal
andshear
and shearreinforcement;
reinforcement; (b)(b) Cross-section
Cross-section ofof
thethe beam.
beam.

2.2.
2.2.Design
Designofofthe
theMicrowave
MicrowaveSensors
Sensors
InInthis study,
this a microstrip
study, a microstrip patchpatch
antenna with a with
antenna split ring resonator
a split (SRR) structure
ring resonator (SRR) was designed,
structure was
simulated and fabricated as a potential microwave sensing technique.
designed, simulated and fabricated as a potential microwave sensing technique. SRR structures SRR structures have been
utilised
have beento realise microwave
utilised to realiseresonators,
microwave filters and microwave
resonators, filters and antennas
microwave[11], and have been
antennas [11], used for
and have
various applications, namely chemical sensors [12], gas sensors [13], crack
been used for various applications, namely chemical sensors [12], gas sensors [13], crack detection detection sensors [14], angle
sensors
sensors[14],
[14],alignment and position
angle sensors sensors and
[14], alignment [15], position
displacementsensors sensors [16] and microwave
[15], displacement sensorsimaging
[16] and
sensors [17].
microwave imaging sensors [17].
Figure
Figure2 2presents
presents(a) (a)a atraditional
traditionalcircular
circularpatch
patchantenna
antennaresonating
resonatingatat10.810.8GHzGHz(30 (30mmmminin
diameter) and (b) a traditional circular patch antenna with an additional
diameter) and (b) a traditional circular patch antenna with an additional SRR structure, which were SRR structure, which
were modelled
modelled andand simulated
simulated using
using thetheHigh
HighFrequency
FrequencyStructural
Structural Simulation
Simulation software
software tool
tool(HFSS)
(HFSS)
(ANSYS Inc., Canonsburg, PA, USA) [18]. SRR structures offer an improved
(ANSYS Inc., Canonsburg, PA, USA) [18]. SRR structures offer an improved performance compared performance compared to
traditionally designed microstrip patch antennas as they allow for
to traditionally designed microstrip patch antennas as they allow for miniaturised antenna miniaturised antenna dimensions
and/or multiband
dimensions operations,
and/or multiband i.e.,operations,
multiple resonant frequencies
i.e., multiple resonant [19]. In this study,
frequencies [19].the
InSRR
this structure
study, the
was
SRR structure was used to provide multiple resonant frequencies between 1 to 10 GHz,details
used to provide multiple resonant frequencies between 1 to 10 GHz, which provided more which
about the concrete.
provided In addition,
more details about thethe use of SRR
concrete. shifted the
In addition, the use
resonant
of SRRfrequencies to lower frequency
shifted the resonant frequencies
ranges,
to lowerwhich in turn increased
frequency ranges, which the penetration capability
in turn increased theofpenetration
the antenna,capability
while preventing disturbances
of the antenna, while
inpreventing
the concrete structure by keeping the dimensions of the antenna small.
disturbances in the concrete structure by keeping the dimensions of the antenna small. Figure 2c demonstrates
the comparison
Figure between the
2c demonstrates the comparison
resonant frequency betweenresponse
the resonant of thefrequency
traditional circularofpatch
response antenna
the traditional
(Figure
circular2a)patch
and the proposed
antenna SRR2a)
(Figure structure
and the (Figure 2b), which
proposed resonated(Figure
SRR structure at multiple frequencies
2b), which belowat
resonated
10.8 GHz, namely 3.4 GHz, 4.7 GHz, 5.7 GHz, 8.7 GHz and 10.6 GHz
multiple frequencies below 10.8 GHz, namely 3.4 GHz, 4.7 GHz, 5.7 GHz, 8.7 GHz and 10.6 GHz withwith a return loss below −10 dB.
The dimensions
a return (in mm)
loss below –10of dB.
the proposed
The dimensions antenna(in aremm)presented
of theinproposed
Figure 2d.antenna are presented in
Figure 2d.
 Sensors 2019, 19, 5175 4 of 14
Sensors 2019, 19, x FOR PEER REVIEW 4 of 14

(a) (b)

(d)
(c)

Figure 2. Simulated (a) original antenna, (b) antenna with split ring resonators (SRRs), (c) the S
Figure 2. Simulated (a) original antenna, (b) antenna with split ring resonators (SRRs), (c) the S1111
simulation results and (d) the design of the proposed antenna with dimensions (mm).
simulation results and (d) the design of the proposed antenna with dimensions (mm).

As the sensor was aimed to be embedded into the concrete, a concrete sample was modelled and
As the sensor was aimed to be embedded into the concrete, a concrete sample was modelled and
simulated using HFSS. According to Piladaeng [20], the concrete dielectric constant for a 7-day-old
simulated using HFSS. According to Piladaeng [20], the concrete dielectric constant for a 7-day-old
specimen is between ε0 = 4 and ε0 = 5.5, conductivity (S/m) in the range between 0.18 and 0.25 and
specimen is between ε’= 4 and ε’= 5.5, conductivity (S/m) in the range between 0.18 and 0.25 and
dielectric loss factor between ε” = 0.45 and ε” = 0.65. Therefore, in this study, the concrete specimen
dielectric loss factor between ε’’= 0.45 and ε’’= 0.65. Therefore, in this study, the concrete specimen
was modelled using ε0 = 5.5, conductivity (S/m) of 0.25 and a loss factor of ε” = 0.5. The model of the
was modelled using ε’= 5.5, conductivity (S/m) of 0.25 and a loss factor of ε’’= 0.5. The model of the
concrete sample was simulated with various crack widths (0.0 mm, 0.1 mm, 0.2 mm, 0.3 mm, 0.4 mm,
concrete sample was simulated with various crack widths (0.0 mm, 0.1 mm, 0.2 mm, 0.3 mm, 0.4 mm,
0.5 mm and 0.6 mm) and with the SRR antenna underneath the sample (see Figure 3). The simulation
0.5 mm and 0.6 mm) and with the SRR antenna underneath the sample (see Figure 3). The simulation
was carried out seven times, including the case of an uncracked concrete sample. The simulation
was carried out seven times, including the case of an uncracked concrete sample. The simulation
Sensors 2019, 19, 5175 5 of 14
Sensors2019,
Sensors 2019,19,
19,x xFOR
FORPEER
PEERREVIEW
REVIEW 5 5ofof1414

measurements (S11 , reflection coefficient) were realised in the frequency range of 1 to 10 GHz and with
measurements(S(S1111, ,reflection
measurements reflectioncoefficient)
coefficient)were
wererealised
realisedininthe
thefrequency
frequencyrange
rangeofof11toto10
10GHz
GHzand
and
4000 sweep points.
with4000
with 4000sweep
sweeppoints.
points.

(a)
(a) (b)
(b)
Figure3.3.HFSS
Figure HFSSsimulation
simulation
simulationofofof
the
the proposed
proposed
the antenna
antenna
proposed andand
and
antenna theconcrete
the concrete
the sample:
sample:
concrete (a)3D
(a)
sample: 3D view
view
(a) and
3Dand (b)
(b)
view top
top
and
view.
view.
(b) top view.

The
Thesimulation
simulationresults resultsare areshown
shownin
shown ininFigure
Figure4a,
Figure 4a,which
4a, whichmainly
which mainlydemonstrates
mainly demonstratesa achange
demonstrates changeininthe the
the
amplitude
amplitudeof
amplitude ofofthe
thereflected
the reflectedsignal
reflected signal(S
signal (S(S
11 The
).).The
1111 full
Thefull simulated
fullsimulated frequency
simulatedfrequency
frequencyspectrumspectrum(1–10 (1–10GHz,GHz,4000 4000sweep
sweep
points)
points)was
points) was
was scanned
scanned
scanned across
acrossusing
across a bespoke
using
using aabespoke LabVIEW
bespoke LabVIEW
LabVIEW (v2013) program
(v2013)
(v2013) (National
program
program Instruments,
(National
(National Austin,
Instruments,
Instruments,
TX, USA)
Austin,
Austin, TX,aiming
TX, USA)to
USA) identify
aiming
aiming aidentify
lineararelationship
totoidentify alinear
linearrelationshipbetween
relationship the sensor
between
between response
thesensor
the (S11 ) and
sensorresponse
response (S(S11the
) )and
11 crack
and the
the
width
crack (mm).
crackwidth Figure
width(mm). 4b
(mm).Figure presents
Figure 4b the linear
4bpresents
presents fitlinear
thelinear
the across the
fitfitacrossfullthe
across simulated
the spectrum
fullsimulated
full simulated and theand
spectrum
spectrum crack
and the width
the crack
crack
(mm).
width The
width(mm). strongest
(mm). The linear correlation
Thestrongest
strongest can be can
linearcorrelation
linear correlation observed
canbe aroundaround
beobserved
observed 6.2 GHz.
around 6.2Figure
6.2 4c
GHz.Figure
GHz. shows
Figure the
4c4cshows linear
shows the
the
relationship
linear between
linearrelationship
relationship S11 andSS11
between
between the crack
and
11and thewidth
the crack(mm)
crack widthchanged
width (mm)changed
(mm) at 6.2 GHz,
changed atat6.2 with
6.2 GHz,
GHz, 2 =
R with
with0.91.
RR =The
2 2 =0.91.figure
0.91. The
The
demonstrates
figure that S11that
figuredemonstrates
demonstrates increased
that as the crack
SS1111increased
increased asasthe
the width grew,
crackwidth
crack widthi.e., lessi.e.,
grew,
grew, energy/power
i.e., lessenergy/power
less was absorbed
energy/power wasabsorbed
was by the
absorbed
concrete
bybythe sample
theconcrete as
concretesample there was
sampleasasthere an opening
therewas wasan air gap,
anopening which
openingair changed
airgap,
gap,which the penetration
whichchanged
changedthe of the electromagnetic
thepenetration
penetrationofofthe the
(EM) signal, as well
electromagnetic
electromagnetic (EM)
(EM) as changing
signal, asasthe
signal, wellreflection
well asas changingof the the
changing signal. Based on
the reflection
reflection ofofthese simulation
the signal.
the signal. Based
Based results
on and
on these
these
previous
simulation
simulation work
results[8],and
results the investigation
andprevious
previouswork work was taken
[8],the
[8], forward, i.e.,
theinvestigation
investigation was
was totaken
the fabrication
taken forward,i.e.,
forward, ofto
i.e., the
tothe sensors
the and
fabrication
fabrication
subsequently
ofofthe
thesensors
sensors to thesubsequently
and
and experimentaltotesting.
subsequently tothetheexperimental
experimentaltesting. testing.

(a)
(a)

Figure 4. Cont.
Sensors 2019, 19, x FOR PEER REVIEW 6 of 14
Sensors 2019, 19, x FOR PEER REVIEW 6 of 14
Sensors 2019, 19, 5175 6 of 14

(b)
(b) (c)
(c)

Figure Simulated (a) raw (b) R222 values

raw data, (b) values acrossthe
the full spectrum
spectrum (1–10 GHz)
GHz) and (c)
(c) linear fit
fit
Figure 4.
Figure 4. Simulated
4. Simulated (a) (a) raw data,
data, (b) R
R values across
across the full
full spectrum (1–10
(1–10 GHz) and
and (c) linear
linear fit
between the
between the SS1111change
changeand
andcrack
crackwidth
widthat at6.2
6.2GHz, withRR222==0.91.
GHz,with 0.91.
between the S11 change and crack width at 6.2 GHz, with R = 0.91.
The proposed antennas
The were fabricated on a double-sided 1.6 mm mmFR4 PCB (Printed Circuit Board)
The proposed
proposed antennas
antennas were
were fabricated
fabricated on
on aa double-sided
double-sided 1.61.6 mm FR4FR4 PCB
PCB (Printed
(Printed Circuit
Circuit
sheet
Board) using an LPKF ProtoMat D104 unit (see Figure 5). Then the SMA connectors were inserted
Board) sheet
sheet using
using anan LPKF
LPKF ProtoMat
ProtoMat D104
D104 unit
unit (see
(see Figure
Figure 5).
5). Then
Then the
the SMA
SMA connectors
connectors were
were
through the probe-fed
inserted hole of the antennas and soldered on the patchon(top layer) and thelayer)
ground plane
inserted through
through the
the probe-fed
probe-fed hole
hole of
of the
the antennas
antennas and
and soldered
soldered on the
the patch
patch (top
(top layer) and
and the
the
(bottomplane
ground layer)(bottom
of the antennas.
layer) of the antennas.
ground plane (bottom layer) of the antennas.

Figure 5. Fabrication of the antennas on a LPKF ProtoMat D104.

Figure
Figure 5.
5. Fabrication
Fabrication of
of the
the antennas
antennas on
on aa LPKF
LPKF ProtoMat
ProtoMat D104.
D104.
3. Experimental Programme
3.
3. Experimental
Experimental Programme
Programme
In order to evaluate the capability of the microwave sensors to predict the crack formation,
four In
In order
order to
reinforced to evaluate
concrete the
evaluate capability
beams,
the in lineof
capability the
the microwave
ofwith the sensors
sensors to
design described
microwave predict
toin Section
predict the crack
the2.1, were
crack formation, four
tested under
formation, four
reinforced concrete
three-pointconcrete
reinforced beams,
loading.beams, in line
The experimental with the
in line with programme design described
was carried
the design described in Section 2.1,
out at the
in Section were
2.1,Henry tested under
Cottonunder
were tested three-
Laboratory
three-
point loading.
of Liverpool
point loading.John The experimental
The Moores programme
University.
experimental programme was
was carried
carriedofout
The preparation theat
out the
the Henry
atspecimens, Cotton
Cotton Laboratory
Henryincluding the attached
Laboratory of
of
Liverpool
Liverpool John
instrumentation, Moores
is described
John Moores University.
in Section
University. The
The preparation
3.1,preparation of the
while Sectionof3.2the specimens,
provides including
information
specimens, includingon thethe attached
thetest set-up
attached
instrumentation, is
is described
and the process utilised.
instrumentation, described in in Section
Section 3.1,3.1, while
while Section
Section 3.2
3.2 provides
provides information
information on on the
the test
test set-up
set-up
and the process utilised.
and the process utilised.
3.1. Specimens’ Preparation and Instrumentation
3.1.
3.1. Specimens’
Specimens’ Preparation
Preparation and Instrumentation
and for
Instrumentation
Initially, the formwork the concrete beams was constructed and the longitudinal steel
Initially,
reinforcement the
bars formwork
and the
Initially, the formwork for the for
shear the
links concrete beams
beams was
were fabricated.
concrete wasUponconstructed and
and the
careful treatment
constructed longitudinal
the of the surfaces,
longitudinal steel
six
steel
reinforcement
strain gauges (Techni
reinforcement bars and
bars and the
Measure shear
the shear links
- Sensors were
links were fabricated.
& Transducers, Upon
fabricated. Doncaster, careful
Upon careful treatment
UK) of
were attached
treatment the
of the atsurfaces, six
the tensile
surfaces, six
strain gauges
reinforcement (Techni
bars of Measure
the - Sensors
specimens. In & Transducers,
particular, the Doncaster,
strain gauges UK)
were
strain gauges (Techni Measure - Sensors & Transducers, Doncaster, UK) were attached at the tensilewere attached
adhered to at
the the tensile
bottom and
reinforcement
the top of a tensile
reinforcement bars
bars of the
the specimens.
reinforcement
of specimens. In
In particular,
bars and at the the
particular, the strain
mid gauges
gauges were
and quarter
strain were adhered
span. to
to the
A microwave
adhered the bottom
sensor and
bottom was
and
the
the top
located of
top of a
at thetensile reinforcement
neutral
a tensile axis (NA) ofbars
reinforcement bars and
the and at the mid
cross-section
at the mid and
mid-span quarter
and quarter span.
of each A microwave
specimen,
span. A microwave sensor
wheresensor was
negligible
was
located at
at the
longitudinal
located neutral
thestresses
neutralandaxis (NA)
axisstrains of
(NA) wereof the
the cross-section
cross-section mid-span
expected. mid-span of of each
each specimen,
specimen, where
where negligible
negligible
longitudinal
Following stresses
this, and
the strains
concrete were
longitudinal stresses and strains were expected.mix expected.
was prepared in accordance with the BRE (Building Research
Establishment) concrete mix design [21], which gave a 28-day compressive strength equal to 57 N/mm2 .
 Sensors 2019, 19, x FOR PEER REVIEW 7 of 14
Sensors 2019, 19, 5175 7 of 14
Following this, the concrete mix was prepared in accordance with the BRE (Building Research
Establishment) concrete mix design [21], which gave a 28-day compressive strength equal to 57
Crushed
N/mmlimestone
2. Crushedaggregates (Beers Timber
limestone aggregates andTimber
(Beers Building
and Suppliers,
Building Liverpool, UK) (20 mm
Suppliers, Liverpool, UK)down),
(20 mm
sand (Mersey Grit) (Beers Timber and Building Suppliers, Liverpool, UK), OPC
down), sand (Mersey Grit) (Beers Timber and Building Suppliers, Liverpool, UK), OPC Type II Type II (Travis Perkins,
Liverpool, UK) andLiverpool,
(Travis Perkins, a water-to-cement
UK) andratio equal to 0.45 were
a water-to-cement ratioused.
equal A toslump test used.
0.45 were was also performed
A slump test was
to ensure that the mix was in accordance with the design guidelines. The
also performed to ensure that the mix was in accordance with the design guidelines. The beamsbeams were cast and the were
concrete was
cast and thevibrated
concreteusing a poker using
was vibrated vibrator according
a poker to the
vibrator designto
according guidelines
the design[21]. Note that
guidelines [21].the
Note
microwave sensor was sensor
that the microwave embedded was into concrete
embedded and
into the accuracy
concrete and the of accuracy
its location
of was regularly
its location was checked
regularly
during casting.
checked during casting.
Concrete
Concretecubes
cubeswith dimensions
with dimensionsof 150 × 150
of 150 × 150
× 150 mm
× 150 mm were
werealso prepared
also prepared in order to check
in order to checkthethe
compressive strength of the concrete at 7 and
compressive strength of the concrete at 7 and 28 days.28 days.
TheThe
specimens’
specimens’ preparation,
preparation,including
including thethe
details of the
details attached
of the attachedsensors andand
sensors gauges along
gauges alongwithwith
thethe
concrete casting, are shown in Figure
concrete casting, are shown in Figure 6. 6.
UponUpondemoulding,
demoulding, thethe
beams
beamswerewerepainted
paintedon on
bothbothlateral sides
lateral in order
sides in order to make
to makethethe
crackcrack
formation and propagation detectable. Demec studs were applied at the central
formation and propagation detectable. Demec studs were applied at the central tension region of tension region ofthe
thebeams
beamsininorder
orderto tomonitor
monitorthethecrack
crackopening
openingwithin
withinthethemost
mostheavily
heavilystressed
stressedareas.
areas.Four
Fourstudstudsets
setswere
werebonded
bondedon onthethe concrete’s
concrete’s surface
surface at at a set horizontal distance of 50 mm on either side the
a set horizontal distance of 50 mm on either side of of the
centreline andand
centreline at vertical distances
at vertical equal
distances to 80
equal to mm,
80 mm,60 mm,
60 mm, 40 mm
40 mmandand20 mm
20 mmfrom thethe
from beam’s
beam’sneutral
neutral
axisaxis
(NA).
(NA).

Concrete poker
vibrator

Microwave sensor at the

neutral axis (NA)
Cables for the strain
gauges

Figure 6. Specimens’
Figure preparation.
6. Specimens’ preparation.
3.2. Experimental Set-Up
3.2. Experimental Set-Up
Having attached the required instrumentation (i.e., strain gauges attached on the tensile
Having attached the required instrumentation (i.e., strain gauges attached on the tensile
reinforcement bars, microwave sensors embedded in the concrete beam and Demec studs glued
reinforcement bars, microwave sensors embedded in the concrete beam and Demec studs glued on
on the concrete beam’s surface), the specimens were cured for 7 days and were then located at the test
the concrete beam’s surface), the specimens were cured for 7 days and were then located at the test
rig for the experimental investigation. The beams spanned 1200 mm with an overhang of 100 mm from
rig for the experimental investigation. The beams spanned 1200 mm with an overhang of 100 mm
both sides of the supports. Simply-supported conditions were achieved with the use of roller supports.
from both sides of the supports. Simply-supported conditions were achieved with the use of roller
A calibrated hydraulic 100 kN load actuator was used for the load application. A constant slow loading
supports. A calibrated hydraulic 100 kN load actuator was used for the load application. A constant
rate, excluding any dynamic effects, was implemented, while a load cell was used for the recording
slow loading rate, excluding any dynamic effects, was implemented, while a load cell was used for
of the load increments. Two strain gauge type displacement transducers (DTs) were placed at the
the recording of the load increments. Two strain gauge type displacement transducers (DTs) were
mid-span in order to measure the vertical deflections of the specimens during testing. Synchronised
placed at the mid-span in order to measure the vertical deflections of the specimens during testing.
data acquisition systems were used for the structural response measurements. The experimental set-up,
Synchronised data acquisition systems were used for the structural response measurements. The
along with the employed instrumentation, are shown in Figures 7 and 8. The cables shown in Figure 8
experimental set-up, along with the employed instrumentation, are shown in Figures 7 and 8. The
present the current available instrumentation for the structural response monitoring via strain gauges,
cables shown in Figure 8 present the current available instrumentation for the structural response
Sensors 2019, 19, x FOR PEER REVIEW 8 of 14
Sensors 2019, 19, 5175 8 of 14
Sensors 2019, 19, x FOR PEER REVIEW 8 of 14
monitoring via strain gauges, Linear Variable Displacement Transducers (LVDTs) and the load
cell.monitoring via strain
The microwave gauges,
sensor Linear
in this Variable
study Displacement
was (LVDTs)
connected Transducers
using a single (LVDTs) and which
the loadwill be
Linear Variable Displacement Transducers and the load cell.coaxial cable,
The microwave sensor in this
cell. The microwave sensor in this study was connected using a single coaxial cable, which will be
eliminated
study was inconnected
the future.using a single coaxial cable, which will be eliminated in the future.
eliminated in the future.

Figure
Figure 7. Experimental
7. Experimental set-up.set-up.
Figure 7. Experimental set-up.

Load cell
Crack width gauge
LVDTs Load cell
Crack width gauge Cable of
LVDTs
microwave

Pinned support
Cable of Cables of strain
sensor Pinned support
gauges
microwave
sensor Cables of strain
Pinned support Pinned support
gauges
Load, displacements and
Hydraulic jack
Signal data acquisition strain acquisition

Load, displacements and

Hydraulic jack
Signal data acquisition strain acquisition

Figure
Figure8.8.Photos
Photosfrom thethe
from experimental set-up
experimental and instrumentation.
set-up LVDTs: LVDTs:
and instrumentation. ??? ???

4.4.Results
Results

4.1.Results
4.1. Results on
on the
theStructural
StructuralResponse
Response
Figure 8. Photos from the experimental set-up and instrumentation. LVDTs: ???
The structural response for all specimens is shown in Figure 9, where the applied load is plotted
4. Results
against the average of the two DTs for the mid-span displacement. As shown, the load versus vertical
displacement response of the beams were linear, and very similar for all specimens. The crack width
4.1. Results on the Structural Response
recordings for the specimens are presented in Figure 10, where in all cases, the crack opening increased
for higher loads. Additionally, the observed crack width openings were more pronounced for a vertical
 Thestructural
The structuralresponse
responsefor forall
allspecimens
specimensisisshownshownin inFigure
Figure9,9,where
wherethe theapplied
appliedload
loadisisplotted
plotted
againstthe
against theaverage
averageof ofthe
thetwo
twoDTs DTsfor forthe
themid-span
mid-spandisplacement.
displacement.As Asshown,
shown,the theload
loadversus
versusvertical
vertical
displacementresponse
displacement responseof ofthe
thebeams
beamswere werelinear,
linear,and
andveryverysimilar
similarforforall allspecimens.
specimens.The Thecrack
crackwidth
width
recordings
recordings for
for
Sensors 2019, 19, 5175
the
the specimens
specimens are
are presented
presented in
in Figure
Figure 10,
10, where
where in
in all
all cases,
cases, the
the crack
crack opening
opening
9 of 14
increasedfor
increased forhigher
higherloads.
loads.Additionally,
Additionally,the theobserved
observedcrack crackwidth
widthopenings
openingswere weremoremorepronounced
pronounced
for aa vertical
for vertical distance
distance of of 80
80 mmmm from from thethe neutral
neutral axis,
axis, where
where the the tensile
tensile stresses
stresses were
were higher,
higher,
distance
compared
compared of 80to mm
to fromof
those
those thesmaller
of neutral axis,
smaller where
vertical
vertical the tensile
distances
distances stresses
from
from the were
the neutral
neutral higher, compared
axis.
axis. to those
The strain
The strain of
gauges
gauges
smaller vertical
measurementswere
measurements distances
wereonly from
onlyused
usedtothe neutral
tofurther axis.
furthervalidate The
validatethe strain gauges
thestructural measurements
structuralresponse
responseof ofthe were
thebeam
beamand only
andwill used
willnot to
notbe be
further
further
further validate
utilisedthe
utilised structural
inthis
in paper.response of the beam and will not be further utilised in this paper.
thispaper.
ForFor
For each
each
each of of
the specimens,
of the
the specimens,
specimens, a concrete cube compressive
aa concrete
concrete cube compressive
cube strength
compressive test wastest
strength
strength performed.
test The average
was performed.
was performed. The
The
compressive strength was 49.92 N/mm 2 and 57.15 N/mm 2 for the 7- and 28-day tests, respectively. In
average compressive
average compressive strengthstrength was was 49.92
49.92 N/mmN/mm and
2 2 and 57.15
57.15 N/mm
N/mm for 2 2 for thethe 7-7- and
and 28-day
28-day tests,
tests,
Figure 11, a typical
respectively.
respectively. InFigure
In cracked
Figure 11,aaspecimen
11, typical at the end
typicalcracked
cracked of the at
specimen
specimen testing
atthe
theendis depicted.
end ofofthe
thetesting Theiscrack
testing evolution
isdepicted.
depicted. and
Thecrack
The crack
propagation
evolutionand
evolution was
and marked during
propagation
propagation testing.
markedA
wasmarked
was symmetrical
during
during testing.crack
testing. pattern, ascrack
AAsymmetrical
symmetrical anticipated
crack pattern,for
pattern, asreinforced
as anticipated
anticipated
concrete
for beamsconcrete
forreinforced
reinforced under
concrete pure
beamsflexure,
beams under
under was observed
pure
pure flexure,in
flexure, most
was
was cases. It
observed
observed inismost
in noteworthy
most cases.ItItis
cases. that even though
isnoteworthy
noteworthy that
that
the
evenspecimens
even though the
though did not reach their
the specimens
specimens did failure
did notnot reachload,
reach thefailure
their
their recorded
failure response
load,
load, up to 100
the recorded
the recorded kN allowed
response
response up to
up tofor
100the
100 kN
kN
comparison
allowedfor
allowed between
forthe the
thecomparison Demec
comparisonbetween studs
betweenthe and the
theDemec new
Demecstuds sensor
studsand signal
andthethenewvalues,
newsensor thus
sensorsignal serving the
signalvalues, scope
values,thus of this
thusserving
serving
study,
the as will
thescope
scope of be
ofthis discussed
this study,as
study, asin Section
will
will 4.2.
bediscussed
be discussed inSection
in Section4.2.4.2.

Figure
Figure
Figure 9.9.9. Load
Load
Load versus
versus
versus average
average
average mid-span
mid-span
mid-span displacement
displacement
displacement graph
graph
graph of
ofof all
all
all test
test
test specimens.
specimens.
specimens.

(a)Specimen
(a) Specimen11 (b)Specimen
(b) Specimen22

(c)Specimen
(c) Specimen33 (d)Specimen
(d) Specimen44

Figure 10. Load versus crack width opening.

 Sensors 2019, 19, x FOR PEER REVIEW 10 of 14
Sensors 2019, 19, 5175 10 of 14
Figure 10. Load versus crack width opening.

Figure
Figure 11.Typical
11. Typical cracked
cracked specimen
specimenatatthe maximum
the applied
maximum load.load.
applied

4.2. 4.2. Results

Results for for
thethe MicrowaveSensor
Microwave Sensor

The The microwave

microwave measurements
measurements were
were provided
provided by by
thethe S-parameter
S-parameter (i.e.,
(i.e., , reflectioncoefficient),
S11S,11reflection coefficient),namely
namely complex data (real and imaginary data of the signal) and magnitude
complex data (real and imaginary data of the signal) and magnitude from the Vector Network from the Vector NetworkAnalyser
Analyser (VNA) (Rohde & Schwarz, Munich, Germany). The data was captured via a bespoke
(VNA) (Rohde & Schwarz, Munich, Germany). The data was captured via a bespoke LabVIEW program
LabVIEW program on a laptop connected to the VNA. The data acquisition took place at each load
on a laptop connected to the VNA. The data acquisition took place at each load increment. The recorded
increment. The recorded measurements were in the 1 to 10 GHz frequency range using the maximum
measurements
number of were
sweepinpoints
the 1 to 10 GHz(4000).
available frequency range using
The captured the maximum
results were analysednumberagainstof sweep points
the crack available
width
(4000). Thesamples
of the captured resultstowere
in order analysed
determine against the
a relationship crack width
between of the in
the changes samples
the EMin order toand
response determine
the a
relationship between
crack width. Figure the
12changes in theaEM
demonstrates response
linear and the
relationship crack width.
between the crack Figure 12 demonstrates
propagation and the EM a linear
relationship between
signal across the crack
the fully propagation
captured spectrumand theGHz).
(1–10 EM signal across the
The strongest fullyrelationship
linear captured spectrum (1–10 GHz).
was identified
between the 2.8 GHz and 3 GHz frequency range.
The strongest linear relationship was identified between the 2.8 GHz and 3 GHz frequency range.
Sensors 2019, 19, x FOR PEER REVIEW 11 of 14

Figure 12. Linear

Figure fit between
12. Linear the crack
fit between widthwidth
the crack and electromagnetic (EM)
and electromagnetic response
(EM) responsefor
forallallfour
fourspecimens
across the 1–10 GHz
specimens acrossfrequency spectrum.
the 1–10 GHz frequency spectrum.

The EM response from all four specimens demonstrated a strong linear relationship against the
crack width with R2 above 0.90 for specimens 1, 2, 3 and 4 (Figure 13). The slopes of the linear
regression lines for the four specimens at 20 mm, 40 mm, 60 mm and 80 mm distances from NA
(Figure 14) demonstrate a similar response for specimens 1, 2 and 4, namely the slope increased with
the increase of the distance from the NA. For specimen 3, the crack width at 20 mm and 60 mm
Sensors 2019, 19, 5175 11 of 14

The EM response from all four specimens demonstrated a strong linear relationship against the crack
width with R2 above 0.90 for specimens 1, 2, 3 and 4 (Figure 13). The slopes of the linear regression lines
for the four specimens at 20 mm, 40 mm, 60 mm and 80 mm distances from NA (Figure 14) demonstrate a
similar response for specimens 1, 2 and 4, namely the slope increased with the increase of the distance
from the NA. For specimen 3, the crack width at 20 mm and 60 mm distances from NA were found to be
similar (Figure 13c), leading to a slope decrease from 40 mm to 60 mm, as shown in Figure 14. In addition,
the sensor response varied between the specimens owing to the inconsistency of the crack patterns on
each specimen. Note also that only small changes (crack widths less than 1 mm) were considered for
this study. Nevertheless, the sensor was able to determine these small changes in the structure, i.e., it
demonstrated a potential to detect anomalies in RC structures, which could be used as an indicator of
a structural failure. Further study is required for the proposed technique, namely more samples with
various crack
Sensors 2019, 19, xwidths andREVIEW
FOR PEER patterns in order to obtain necessary parameters/data for the analysis12orofthe
14
potential implementation of machine learning algorithms.

(a) Specimen 1 (b) Specimen 2

(c) Specimen 3 (d) Specimen 4

Figure
Figure 13. Linear correlation
13. Linear correlation between
between the
the crack
crack width
width and
and sensor
sensor response
response at
at 2.9
2.9 GHz.
GHz.

Figure 14. Slopes of linear regression lines for four specimens at 20 mm, 40 mm, 60 mm and 80 mm
 (c) Specimen 3 (d) Specimen 4
Sensors 2019, 19, 5175 12 of 14
Figure 13. Linear correlation between the crack width and sensor response at 2.9 GHz.

Figure
Figure 14. Slopes
Slopes of
of linear
linear regression
regression lines
lines for
for four
four specimens
specimens at
at 20
20 mm,
mm, 40
40 mm,
mm, 60
60 mm
mm and
and 80
80 mm
mm
distances from the NA.
distances

5. Conclusions
5. Conclusions
Aiming to
Aiming to assess
assess microwave
microwave sensing
sensing technology
technology asas aa means
means of
of structural
structural health
health monitoring,
monitoring, four
four
reinforced concrete beams were tested under three-point loading in the Henry Cotton
reinforced concrete beams were tested under three-point loading in the Henry Cotton Laboratory of Laboratory of
Liverpool John
Liverpool John Moores
Moores University.
University. In
In addition
addition to
to the
the commonly implemented instrumentation
commonly implemented instrumentation inin
structural testing (i.e., load cells, displacement transducers, Demec studs, etc.), a novel
structural testing (i.e., load cells, displacement transducers, Demec studs, etc.), a novel microwave microwave
sensor, specifically
sensor, specifically designed
designed for
for this
this application,
application, was
was inserted
inserted into
into each
each beam’s
beam’s reinforcement
reinforcement cage
cage
before the concrete’s casting. In order to allow for a reduced size of the sensor and to shift
before the concrete’s casting. In order to allow for a reduced size of the sensor and to shift the resonant the
resonant frequencies to lower frequency ranges, hence increasing the antenna’s penetration capability,
a microstrip patch antenna with an SRR structure was designed and subsequently fabricated.
During the load application, both the structural performance and the signal measurements were
incrementally recorded. As anticipated, a linear load–displacement structural response was found
to be very similar for all specimens. The microwave data sets, which were captured via a bespoke
LabVIEW program, were provided by the S11 reflection coefficient, and were within the 1–10 GHz
frequency range. For the largest anticipated tensile crack at the mid-span of the beam, the readings for
the crack width gauges and the sensors were processed and compared. A strong linear relationship
with R2 values very close to unity were observed. The variance on the sensor response was related to
the different crack patterns observed in each specimen. The present study considered crack widths
below 1 mm and thus only small changes in the concrete structure. It is noteworthy that the sensor
was able to detect these small changes in the concrete’s structure, demonstrating that the proposed
technique could be further studied in order to detect cracking in RC structures. Further investigation is
required to obtain more data for various crack patterns and widths, thus enhancing the measurements’
repeatability and reliability and allowing for the technique’s validation.
Future studies could also investigate the possibility of insulating/sealing the antennas to avoid
direct contact with the concrete, which might change the properties of the antenna owing to the physical
and chemical properties of the concrete mixture. The extra layer could protect antennas from possible
corrosion/oxidation of the copper part of the antenna. Finally, additional sensors (i.e., moisture and
temperature) could be added to provide additional variables for the sensor algorithm, which could
further improve crack detection.

Author Contributions: M.G., M.M., G.S.K., W.A. and P.K. organised the conceptualisation of the idea and the
methodology employed in this paper. Following that, P.K., M.G. and M.M. worked on the software, data validation
and formal analysis for the investigation. K.H., G.S.K., W.A. and M.G. worked on the concrete samples’ preparation.
M.M. and M.G. undertook data acquisition. The original writing and draft preparation were carried out by M.G.,
M.M. and P.K., and W.A. carried out the review and editing. The project was supervised by P.K.
Sensors 2019, 19, 5175 13 of 14

Funding: This research was funded by Liverpool John Moores University under the Pump Priming Award
Scheme 2019.
Acknowledgments: The authors of this paper would like to thank Sensor City Liverpool Limited for providing
access to their lab facilities to print the electromagnetic sensors.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Staszewski, W. Structural Health Monitoring Using Guided Ultrasonic Waves. In Advances in Smart
Technologies in Structural Engineering; Springer: Berlin/Heidelberg, Germany, 2004; pp. 117–162.
2. Rafiq, M.I. Structural health monitoring for maintenance management of deteriorating structures: Current
practice and state of the art. In Proceedings of the 2nd International and 6th National Conference on
Earthquake & Structures, Kerman, Iran, 14–15 October 2015; pp. 1–19.
3. Highways Agency. Highways Agency Business Plan 2014-15; Highways Agency: London, UK, 2014.
4. Balageas, D.; Fritzen, C.P.; Güemes, A. Structural Health Monitoring; ISTE: Newport Beach, CA, USA, 2006.
5. Park, H.S.; Shin, Y.; Choi, S.W.; Kim, Y. An Integrative Structural Health Monitoring System for the
Local/Global Responses of a Large-Scale Irregular Building under Construction. Sensors 2013, 13, 9085–9103.
[CrossRef]
6. Chang, P.C.; Flatau, A.; Liu, S.C. Review Paper: Health Monitoring of Civil Infrastructure. Struct. Heal.
Monit. 2003, 2, 257–267. [CrossRef]
7. Kot, P.; Ali, A.S.; Shaw, A.; Riley, M.; Alias, A. The application of electromagnetic waves in monitoring
water infiltration on concrete flat roof: The case of Malaysia. Constr. Build. Mater. 2016, 122, 435–445.
[CrossRef]
8. Teng, K.H.; Kot, P.; Muradov, M.; Shaw, A.; Hashim, K.; Gkantou, M.; Al-Shamma’a, A. Embedded Smart
Antenna for Non-Destructive Testing and Evaluation (NDT & E) of Moisture Content and Deterioration in
Concrete. Sensors 2019, 19, 547.
9. Kot, P.; Shaw, A.; Riley, M.; Ali, A.S.; Cotgrave, A. The Feasibility of Using Electromagnetic Waves in
Determining Membrane Failure through Concrete. Int. J. Civ. Eng. 2017, 15, 355–362. [CrossRef]
10. Eurocode 2: Design of Concrete Structures–Part 1-1: General Rules and Rules for Buildings; EN 1992-1-1; European
Committee for Standardisation (CEN): Brussels, Belgium, 2004.
11. Salim, A.; Sungjoon, L. Complementary split-ring resonator-loaded microfluidic ethanol chemical sensor.
Sensors 2016, 16, 1802. [CrossRef] [PubMed]
12. Rawat, V.; Dhobale, S.; Kale, S.N. Ultra-fast selective sensing of ethanol and petrol using microwave-range
metamaterial complementary split-ring resonators. J. Appl. Phys. 2014, 116, 164106. [CrossRef]
13. Barochi, G.; Rossignol, J.; Bouvet, M. Development of microwave gas sensors. Sens. Actuators B Chem. 2011,
157, 374–379. [CrossRef]
14. Albishi, A.; Ramahi, O. Detection of surface and subsurface cracks in metallic and non-metallic materials
using a complementary split-ring resonator. Sensors 2014, 14, 19354–19370. [CrossRef] [PubMed]
15. Naqui, J.; Durán-Sindreu, M.; Martín, F. Novel sensors based on the symmetry properties of split ring
resonators (SRRs). Sensors 2011, 11, 7545–7553. [CrossRef] [PubMed]
16. Horestani, A.K.; Fumeaux, C.; Al-Sarawi, S.F.; Abbott, D. Displacement sensor based on diamond-shaped
tapered split ring resonator. IEEE Sens. J. 2014, 13, 1153–1160. [CrossRef]
17. Islam, M.; Islam, M.; Samsuzzaman, M.; Faruque, M.; Misran, N. A negative index metamaterial-inspired
UWB antenna with an integration of complementary SRR and CLS unit cells for microwave imaging sensor
applications. Sensors 2015, 15, 11601–11627. [CrossRef] [PubMed]
18. Ansys. 3D Electromagnetic Field Simulator for RF and Wireless Design. 2019. Available online: https:
//www.ansys.com/products/electronics/ansys-hfss (accessed on 1 November 2019).
19. Kaushal, D.; Shanmuganantham, T. Parametric enhancement of a novel microstrip patch antenna using
Circular SRR Loaded Fractal Geometry. Alex. Eng. J. 2018, 57, 2551–2557. [CrossRef]
Sensors 2019, 19, 5175 14 of 14

20. Piladaeng, N.; Angkawisittpan, N.; Homwuttiwong, S. Determination of relationship between dielectric
properties, compressive strength, and age of concrete with rice husk ash using planar coaxial probe. Meas.
Sci. Rev. 2016, 16, 14–20. [CrossRef]
21. Teychenné, D.C.; Franklin, R.E.; Erntroy, H.C.; Hobbs, D.W.; Marsh, B.K. Design of Normal Concrete Mixes;
H.M.S.O. [for the] Building Research Establishment [and the] Transport and Road Research Laboratory:
London, UK, 1975.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).