sensors Article

Development of a Distributed Crack Sensor Using Coaxial Cable Zhi Zhou 1,2, *, Tong Jiao 1,2 , Peng Zhao 1,2 , Jia Liu 1,2 and Hai Xiao 3 1 2 3

*

School of Civil Engineering, Dalian University of Technology, Dalian 116024, China; [email protected] (T.J.); [email protected] (P.Z.); [email protected] (J.L.) State Key Laboratory of Coastal and Offshore Engineering, Dalian University of Technology, Dalian 116024, China Electrical and Computer Engineering Department, Clemson University, Clemson, SC 29634, USA; [email protected] Correspondence: [email protected]; Tel.: +86-411-8470-9716

Academic Editors: Ferran Martín and Jordi Naqui Received: 21 June 2016; Accepted: 26 July 2016; Published: 29 July 2016

Abstract: Cracks, the important factor of structure failure, reflect structural damage directly. Thus, it is significant to realize distributed, real-time crack monitoring. To overcome the shortages of traditional crack detectors, such as the inconvenience of installation, vulnerability, and low measurement range, etc., an improved topology-based cable sensor with a shallow helical groove on the outside surface of a coaxial cable is proposed in this paper. The sensing mechanism, fabrication method, and performances are investigated both numerically and experimentally. Crack monitoring experiments of the reinforced beams are also presented in this paper, illustrating the utility of this sensor in practical applications. These studies show that the sensor can identify a minimum crack width of 0.02 mm and can measure multiple cracks with a spatial resolution of 3 mm. In addition, it is also proved that the sensor performs well to detect the initiation and development of cracks until structure failure. Keywords: coaxial cable; crack; distributed sensor

1. Introduction Cracking is a major sign of structural aging and damage, which shortens the integrity and service life of structures. While the width of cracks for most ordinary reinforced concrete (RC) structures in civil engineering are over 0.2–0.4 mm, the service life of the structure can be severely shortened due to the erosion of inner steel rebar under harsh environments. Moreover, structures may be damaged if the cracks reach 1–2 mm. Therefore, crack monitoring is essential to guarantee the safety of structures. Currently, many approaches have been developed for crack monitoring, such as ultrasonic methods [1], acoustic emission [2], infrared thermography [3], impact-echo [4], large area electronics [5], etc. All of the techniques mentioned above show good performance of crack detecting, however, they struggled to be applied in practical engineering due to the installation difficulties and vulnerability in the long-term harsh environment. Recently, the optical fiber sensing technology, such as the Bragg grating, the Michelson white light interferometer, and Brillouin scattering have been widely investigated for crack detecting [6–8]. However, all of the fibers in the above approaches are silica fibers which are brittle (their elongation rates are only about 1%) [9], which make it easy to break in real applications, and only small cracks can be monitored. Therefore, the brittleness of silica fibers limits the application of the approaches above. To overcome this disadvantage, coaxial cable, which has a high elongation rate (more than 20%) and works similarly as an optical fiber since they share the same fundamental physics governed by the

Sensors 2016, 16, 1198; doi:10.3390/s16081198

www.mdpi.com/journal/sensors

Sensors 2016, 16, 1198

2 of 13

same electromagnetic (EM) theory, was attempted to monitor the cracks. Lin et al. [10,11] investigated the feasibility of using an embedded commercial coaxial electrical time domain reflectometry (EDTR) cable to detect crack damage in a reinforced concrete beam subjected to a three-point bending load. Although the ETDR signal response waveforms of the sensing cable were able to capture the location of the crack damage site and indicate the relative magnitude of the crack opening, the commercial ETDR sensing cable has low signal-to-noise ratio and low sensitivity. Afterward, a prototype cable was designed and fabricated by Lin et al. [12,13], using rubber as an insulator of the cable instead of polyethylene or Teflon materials employed in commercial cables. According to the calibration tests conducted by Lin et al. [12,13], the prototype cable is approximately ten times more sensitive than commercial cables. The commercial and the prototype ETDR sensing cable are all designed according to the geometric change of the cables, which is low in sensitivity due to the lower sensitivity of the cable impedance to the cross-sectional change. To improve the sensitivity, Chen et al. developed two designs of coaxial cable sensors based on the change in topology of the cable’s outer conductor under strain conditions. One was designed by wrapping a dielectric rubber tube spirally with copper tape, which is the outer conductor of the cable [14]. The other was reconstructed by using Teflon as the dielectric layer and a steel spiral as the outer conductor. The steel spiral can slide along the Teflon surface under strain conditions [15]. These sensors are claimed to be more sensitive than commercial products under applied loads. The uniformity of the sensors, which are mainly associated with their fabrication process, is significantly improved by the plasma-spray coating technique [16]. The topology-based cable sensors have shown many advantages in practical applications [17,18]. However, the spirally-wrapped outer conductor of the sensors can easily lead to loose contact and substantial signal attenuation, which make it easy to break during the installation process and cannot apply to long distance monitoring. In this study, an improved topology-based cable sensor with a shallow helical groove on its external surface of the outer conductor is developed. The inter-relationship among various design parameters is explored through numerical simulations. The following calibration tests investigate the sensor performance for crack detecting. Finally, a feasibility study is also conducted to demonstrate the utility of the sensor in practical applications. 2. Sensing Mechanism of ETDR Coaxial cable is a microwave transmission line. It consists of an inner and outer conductor sandwiched by a tubular insulating layer with a high dielectric constant. Based on the transmission line theory, the characteristic impedance of a cable is the ratio of the voltage to the current of the guided wave. It is related to the distributed parameters, including inductance, capacitance, resistance, and conductance of the cable. Using an equivalent circuit model, their quantitative description can be derived as Equation: d Z0 “

U` pZq U´ pZq “´ “ I` pZq I´ pZq

R ` jωL G ` jωC

(1)

where Z0 is the characteristic impedance, U` pZq and U´ pZq are the incident and reflected voltage at point Z, respectively, I` pZq and I´ pZq are the incident and reflected current at point Z, respectively, L, C, R, G represent the inductance, capacitance, resistance, and conductance, respectively, and ω is the angular frequency of EM wave. In general, the propagating signal in the coaxial cable is high frequency EM waves (>100 kHz), so R ăă ωL, G ăă ωC and, thus, the characteristic impedance can be simplified as: c Z0 “

L C

(2)

Equation indicates that the characteristic impedance of the sensing line has a distributed feature. For an ideal coaxial cable, the EM wave propagating along the sensing line will not be reflected since

reflection at the point of discontinuity. As shown in Figure 1, the reflected step signal carries the information about the location and severity of the discontinuity. The position of the discontinuity, D measured from the monitoring point can be determined by: Sensors 2016, 16, 1198

D

1 vt 2

(3)13 3 of

where v is the propagation velocity of the signal along the sensing cable, and t is the transit time the characteristic impedance is constant from and uniform. If therepoint is a load-induced local deformation of the signal propagating forth-and-back the monitoring to the discontinuity point. of theThe cable’s cross-section, or if a change in topology of the cable’s outer conductor happens, an severity of the discontinuity is related to the reflection coefficient, which is defined as the impedance discontinuity willvoltage be present along the sensing cable, in turn, results in ρthe signal ratio between the reflected to the incident voltage. Thewhich, reflection coefficient, , can be reflection at the point of discontinuity. As shown in Figure 1, the reflected step signal carries the expressed as: information about the location and severity of the discontinuity. The position of the discontinuity, Ur Z'  Z0 by: D measured from the monitoring point canρbe  determined  (4) Ui Z'  Z0 1 D “ v¨t (3) 2 incident voltage, respectively. Z ' and Z0 are where Ur and Ui are the reflected voltage and the the characteristic impedances at ofthe and alongcable, the and uniform where v is the propagation velocity thediscontinuity signal along the sensing t is thetransmission transit time ofline, the respectively. signal propagating forth-and-back from the monitoring point to the discontinuity point.

Figure 1. Schematic illustration illustration of of the the waveform Figure 1. Schematic waveform of of aa coaxial coaxial sensing sensing cable cable under under loading. loading.

According mentioned above, cable can be developed to detect the The severitytoofthe theprinciple discontinuity is related to thecoaxial reflection coefficient, which is defined as both the ratio location and width of voltage a crack to at the anyincident point in voltage. a structure thecoefficient, ETDR technique. between the reflected Thethrough reflection ρ, can be expressed as: Ur Z1 ´ Z0 3. Design and Fabrication of the Prototype Sensor ρ“ “ 1 (4) Ui Z ` Z0 A new type of coaxial cable sensor for crack detecting is designed based on the ETDR sensing principle this As shown in voltage Figure 2a, the prototype is respectively. a topology-based sensing cable where Ur in and Uipart. are the reflected and the incident sensor voltage, Z1 and Z0 are the with a shallow helical groove ondiscontinuity the external and surface of the the uniform outer conductor of a semi-rigid coaxial characteristic impedances at the along transmission line, respectively. cable,According whose outer conductor is mentioned a seamless copper tube. Tocable precisely control the shape and depth of to the principle above, coaxial can be developed to detect both the the helical grooves, a computer numerical controlled (CNC) milling machine capable of pulling, location and width of a crack at any point in a structure through the ETDR technique. rotating, and notching grooves is used to fabricate the sensor. In the process of fabrication, a semi3. Design and Fabrication of on thethe Prototype Sensor rigid coaxial cable is mounted CNC milling machine with the milling depth set properly. Then the grooves continuously the outer conductor by a milling cut while cablesensing rotates A new are typeproduced of coaxial cable sensoron for crack detecting is designed based on thethe ETDR and moves along its axis. The key factor in the fabrication is to ensure that the shallow helical grooves principle in this part. As shown in Figure 2a, the prototype sensor is a topology-based sensing cable cannot cut through thegroove outer on conductor, but surface could beofseparated under (Figure 2b). with a shallow helical the external the outer easily conductor of loading a semi-rigid coaxial Figurewhose 3 shows theconductor schematicisofa the sensorcopper fabrication cable, outer seamless tube. process. To precisely control the shape and depth of the helical grooves, a computer numerical controlled (CNC) milling machine capable of pulling, rotating, and notching grooves is used to fabricate the sensor. In the process of fabrication, a semi-rigid coaxial cable is mounted on the CNC milling machine with the milling depth set properly. Then the grooves are produced continuously on the outer conductor by a milling cut while the cable rotates and moves along its axis. The key factor in the fabrication is to ensure that the shallow helical grooves cannot cut through the outer conductor, but could be separated easily under loading (Figure 2b). Figure 3 shows the schematic of the sensor fabrication process.

Sensors 2016, 16, 1198 Sensors 2016, 16, 1198 Sensors 2016, 16, 1198

4 of 13 4 of 13 4 of 13

(a)(a)

(b) (b)

Figure Sketch map: Prototype crack sensor; details the inner surface the sensor. Figure 2. 2. Sketch map: (a)(a) Prototype crack sensor; (b)(b) details ofof the inner surface ofof the sensor. Figure 2. Sketch map: (a) Prototype crack sensor; (b) details of the inner surface of the sensor.

Sensors 2016, 16, 1198

Figure Prototype sensor fabrication process. 3. 3. fabrication process. Figure Prototype sensor

5 of 13

Forthe theprototype prototypesensor sensorapplied appliedtotocrack crackmonitoring, monitoring,the theouter outerconductor conductoris isclosely closelyconnected connected For with its inner surface before crack generation. When a crack on the structure forms and grows, the with its inner surface before crack generation. When a crack on the structure forms and grows, the shallow spiral grooves will easily separate under mechanical loading, resulting in the change of the shallow spiral grooves will easily separate under mechanical loading, resulting in the change of the currentflow flowpath pathalong alongthe theouter outerconductor. conductor.Based Basedononthe theequivalent equivalentcircuit circuitmodel, model,the thedistributed distributed current inductance is related with the voltage and current, in accordance with Equation. Thus, the change inductance is related with the voltage and current, in accordance with Equation. Thus, the change inin current flow path will introduce a lumped inductor (Figure 4). According to Equation, an “impedance current flow path will introduce a lumped inductor (Figure 4). According to Equation, an “impedance discontinuity” will formed because the impedance the location the crack changes from L /LC/ C discontinuity” will bebe formed because the impedance atat the location ofof the crack changes from Thus,a aportion portionofofthe theincident incidentwave wavewill willbebereflected reflectedwhen whenit itencounters encountersthis this toto ( L (LL' ) L/ )C/ C. .Thus, '

discontinuity.Following Followingthe theETDR ETDRsensing sensingprinciple, principle,the thelocations locationsofofcracks cracksare aredetermined determinedbybythe the discontinuity. arrivaltime timeofofthe thereflected reflectedsignal signalwhile whileitsitsamplitude amplituderepresents representsthe thedegree degreeofofthe thecracks. cracks.Thus, Thus,the the arrival proposed sensor embedded in the structure can detect both location and width of the crack. Equation proposed sensor embedded in the structure can detect both location and width of the crack. Equation indicatesthat thatthe thespatial spatialresolution resolutionfor forcrack crackdetecting detectingis isdetermined determinedbybythe therise risetime timeofofthe theincident incident indicates impulse signal and the bandwidth of the sensor. For a certain rise time of the incident impulse signal, impulse signal and the bandwidth of the sensor. For a certain rise time of the incident impulse signal, Figure 4. Equivalent circuit model of the prototype sensor. the larger the transmission bandwidth of the sensor corresponds to a higher spatial resolution. Figure 4. Equivalent circuit model of the prototype sensor. the larger the transmission bandwidth of the sensor corresponds to a higher spatial resolution.

t )dt 4. Numerical Simulations =U U(t)(dt For the prototype sensor applied to crack monitoring, the outer conductor is closely connected (5)(5) L=L I (the tI)(t ) sensor with its surface before generation. When a crackperformances on the structure and grows, To inner fully understand thecrack correlation between and forms their major design the shallow spiral grooves easily separate under mechanical in the change parameters including crackwill width, spiral grooves interval and riseloading, time of resulting incident impulse signal, 1 1 v T 2 2  (0.35 / BW2 )2 L  Δ min ofthethe currentsensors flow path the outer the equivalent ΔLmin / BW ) on numerical  v2conductor. Tstepstep  (0.35 (6)(6) proposed were along numerically simulated using a Based full-wave solver circuit (Ansoftmodel, HFSS). 2 the distributed inductance is related with the voltage and current, in outer accordance withdiameter Equation. In the simulation, several 100 mm-long sensors were modeled with an conductor of Thus, theand change in current flow path will introduce lumpedpermittivity inductor (Figure According to 0.5 mm the dielectric layer diameter of 1.7 mm. Thea relative of the 4). dielectric material Equation, will be formed the impedance location of the was set toanbe“impedance 2.1, and characteristic impedance wasbecause 50 Ω. When removingatathe portion of several a thediscontinuity” a 1 q{C. Thus, a portion of the incident wave will be reflected when crack changes from L{C to pL ` L helix wires, an air gap can be created to represent a partial separation of spirals caused by the cracks itapplied encounters discontinuity. Followingwaveforms the ETDR under sensing principle, the of pulse crackswere are on a this sensor. The ETDR response the excitation oflocations a Gaussian determined the arrivalthe time of thewaveform reflected signal whileHerein, its amplitude represents the degree of obtained byby integrating voltage over time. crack width is characterized bythe the cracks. Thus, the proposed sensor embedded in the structure can detect both location and width of separation width of the spiral grooves. the crack. Equation indicates that the spatial resolution for detecting the rise Figure 5a shows the simulated ETDR waveforms of crack the sensor withisadetermined fixed spiralby separation length of one turn and a spiral groove interval of 6 mm, and a different crack width from 0.25 mm to 2.5 mm with an increment of 0.25 mm. Figure 5b illustrates the relationship between the peak values of the reflection coefficient and the crack width. Simulation results demonstrate that the peak values of the reflection coefficient increase almost linearly within the crack width. For the simulated sensors with a fixed spiral separation length of one turn and a crack width of 1.0 mm, when the spiral groove

Sensors 2016, 16, 1198

5 of 13

time of the incident impulse signal and the bandwidth of the sensor. For a certain rise time of the incident impulse signal, the larger the transmission bandwidth of the sensor corresponds to a higher spatial resolution. r Uptqdt L“ (5) Iptq b 1 (6) ∆Lmin “ v Tstep 2 ` p0.35{BWq2 2 4. Numerical Simulations To fully understand the correlation between the sensor performances and their major design parameters including crack width, spiral grooves interval and rise time of incident impulse signal, the proposed sensors were numerically simulated using a full-wave numerical solver (Ansoft HFSS). In the simulation, several 100 mm-long sensors were modeled with an outer conductor diameter of 0.5 mm and the dielectric layer diameter of 1.7 mm. The relative permittivity of the dielectric material was set to be 2.1, and the characteristic impedance was 50 Ω. When removing a portion of several helix wires, an air gap can be created to represent a partial separation of spirals caused by the cracks applied on a sensor. The ETDR response waveforms under the excitation of a Gaussian pulse were obtained by integrating the voltage waveform over time. Herein, crack width is characterized by the separation width of the spiral grooves. Figure 5a shows the simulated ETDR waveforms of the sensor with a fixed spiral separation length of one turn and a spiral groove interval of 6 mm, and a different crack width from 0.25 mm to 2.5 mm with an increment of 0.25 mm. Figure 5b illustrates the relationship between the peak values of the reflection coefficient and the crack width. Simulation results demonstrate that the peak values of the reflection coefficient increase almost linearly within the crack width. For the simulated sensors with a fixed spiral separation length of one turn and a crack width of 1.0 mm, when the spiral groove interval changes from 2.0 mm to 7.0 mm with an increment of 1.0 mm, the peak values of the reflection coefficient gradually decrease as shown in Figure 6. It indicates that the sensitivity of the proposed sensor is related to the spiral groove interval, and the higher peak value of the reflection coefficient corresponds to a smaller interval of the spiral groove. Therefore, the proposed sensor with a smaller interval of spiral grooves is expected to perform with better sensitivity. To investigate the effect of the rise time of the incident impulse signal on the spatial resolution, a sensor with a fixed spiral separation length of two turns and a spiral groove interval of 6 mm, as well as a crack width of 0.5 mm was simulated. According to Figure 7, there are two clear peaks in the response waveforms when the rise time is 40 ps, however, the two response peaks become gradually indistinct with the increase of the rise time. It means that the spatial resolution is related to the rise time of incident impulse signal, and a smaller rise Sensors 2016, 16, 1198time can lead to a higher spatial resolution. 6 of 13

(a)

(b)

Figure 5. Results of simulations: (a) reflection waveform; and (b) the correlation between reflection Figure 5. Results of simulations: (a) reflection waveform; and (b) the correlation between reflection coefficient and crack width. coefficient and crack width.

(a) (a)

(b) (b)

Figure 5. Results of simulations: (a) reflection waveform; and (b) the correlation between reflection coefficient and crack width. coefficient and crack width.

SensorsFigure 2016, 16, 5. 1198 Results of simulations: (a) reflection waveform; and (b) the correlation between reflection6 of 13

Figure 6. Responses Responses of the the sensor under under various intervals intervals of spiral spiral grooves. Figure Figure 6. 6. Responses of of the sensor sensor under various various intervals of of spiral grooves. grooves.

Figure 7. Correlation between the spatial resolution and the system rise time. Figure 7. 7. Correlation between the the spatial spatial resolution resolution and and the the system system rise rise time. time. Figure Correlation between

5. Sensor Calibration 5. Sensor 5. SensorCalibration Calibration In this part, the crack detecting capability of the prototype sensor was investigated using a test In this part, the detecting capability ofof thethe prototype sensor waswas investigated using a test In this part, thecrack crack detecting capability prototype sensor investigated using a system, which consists of a vector network analyzer (VNA), a tensile testing device and a prototype system, which consists of a vector network analyzer (VNA), a tensile testing device and a prototype test system, which consists of a vector network analyzer (VNA), a tensile testing device and a sensor. The overall setup of the experiment and their schematic representations are shown in sensor. The overall of the experiment and their schematic representations are are shown in prototype sensor. Thesetup overall the experiment and their representations shown Figure 8a,b, respectively. Thesetup VNAof used in the system has a 40schematic GHz measurement bandwidth and Figure 8a,b, respectively. The VNA used in the system has a 40 GHz measurement bandwidth and in Figure 8a,b, respectively. VNA used inof the system a 40testing GHz measurement bandwidth and launches the Gaussian pulseThe with a rise time 6 ps. The has tensile device, capable of measuring launches the the Gaussian Gaussian pulse pulse with with aa rise rise time time of of 66 ps. ps. The The tensile tensile testing testing device, device, capable capable of of measuring measuring launches micro-displacements, can elongate the cable with the minimum stepping movement of 0.01 mm. The micro-displacements, can elongate the with thethe minimum stepping movement of 0.01 mm.mm. The micro-displacements, elongate thecable cable minimum stepping movement of 0.01 commercial semi-rigidcan SFT-50-3 coaxial cablewith was selected to fabricate the sensor following the commercial semi-rigid SFT-50-3 coaxial cable was selected to fabricate the sensor following the The commercial semi-rigid SFT-50-3 coaxial cable was selected to fabricate the sensor following the method of the above third part. Table 1 lists the geometric configuration and electrical properties method of of the above above third part. part. Table lists the the geometric geometric configuration configuration and and electrical electrical properties properties method Tableof11the lists relevant to the the sensingthird characteristics sensor. relevant to to the the sensing sensing characteristics characteristics of of the the sensor. sensor. relevant Sensors 2016, 16, 1198 7 of 13

(a)

(b)

Figure 8. Setup of sensor test system: (a) schematic representation; and (b) overview. Figure 8. Setup of sensor test system: (a) schematic representation; and (b) overview. Table 1. Parameters of SFT-50-3 coaxial cable. Diameter of Inner

Diameter of Outer

Thickness of Outer

Characteristic

Dielectric

Conductor

Conductor

Conductor

Impedance

Constant

0.92 mm

3.60 mm

0.30 mm

50 Ω

2.10

Sensors 2016, 16, 1198

7 of 13

Table 1. Parameters of SFT-50-3 coaxial cable. Diameter of Inner Conductor

Diameter of Outer Conductor

Thickness of Outer Conductor

Characteristic Impedance

Dielectric Constant

0.92 mm

3.60 mm

0.30 mm

50 Ω

2.10

5.1. Sensitivity Test In theory, the proposed sensor has the distributed detecting ability of cracks because any separations along the spiral groove resulting from cracks will introduce the reflected signal on the EDTR waveform. Since the objective of this test is to investigate the influences of crack width and the spiral groove interval to the reflected coefficient, the sensors tested were made of only one turn of the spiral separation. Thus, two types of prototype sensors were considered in the sensitivity test. The main difference between the two sensors is their spiral groove interval as presented in Table 2. Table 2. Prototype sensors for a sensitivity test. Sensor

Length of Sensor

Turns of Spiral Separation

Initial Position of the Spiral Groove

Spiral Groove Interval

Sensor-I Sensor-II

150 cm 150 cm

1.0 1.0

118.5 cm 99.5 cm

3.0 mm 5.0 mm

Each prototype sensor mounted on the test system was elongated at a step of 0.02 mm, and the response waveforms were captured by the VNA. It was observed during the experiment that the shallow spiral groove was gradually separated from 0 mm up to 1.52 mm of Sensor-I and 2.62 mm of Sensor-II with the increase of the load level. Figure 9a,c presents the resulting waveforms in terms of the reflection coefficient distribution of the two tested sensors. For the convenience of quantitative analysis, their reflection coefficient variations with crack width are also shown in Figure 9b,d. It is apparent that the separating process of the spiral groove goes through three stages under tensile loads. In stage I, the spiral groove starts to separate when an initial strain occurs. Then, the crack develops along the helical direction of spiral groove with gradual increases from 0, up to 0.5 turns. As shown in Figure 9b,d, the peak values of the reflection coefficient increase almost linearly with the applied strain during this process. Stage II presents a sharp jump in the relationship curve between the reflection coefficient and the crack width since the crack of the spiral groove stops developing along its helical direction and experiences a sudden extension along the axial orientation of the cable. Starting from the stage III, the crack width linearly increases with the applied displacement. Consequently, the peak values of the reflection coefficient are linearly related to crack width in this stage. It could be seen that the sensitivity in stage III is higher than that in stage I for both Sensor-I and Sensor-II. The comparison between Sensor-I and Sensor-II also show that Sensor-I is more sensitive to crack than Sensor-II. It means that a higher sensitivity can be obtained if the spiral groove interval is designed to be smaller. To verify the resolution of the prototype sensor, response waveforms of Sensor-II are partially enlarged and presented in Figure 10, under different crack widths that vary from 2.40 mm to 2.60 mm, with an increment of 0.02 mm. It can be concluded that the prototype sensor can distinguish the crack width of 0.02 mm.

is is more more sensitive sensitive to to crack crack than than Sensor-II. Sensor-II. It It means means that that aa higher higher sensitivity sensitivity can can be be obtained obtained if if the the spiral spiral groove interval is designed to be smaller. groove interval is designed to be smaller. To To verify verify the the resolution resolution of of the the prototype prototype sensor, sensor, response response waveforms waveforms of of Sensor-II Sensor-II are are partially partially enlarged and presented in Figure 10, under different crack widths that vary from 2.40 mm enlarged and presented in Figure 10, under different crack widths that vary from 2.40 mm to to 2.60 2.60 mm, with an increment of 0.02 mm. It can be concluded that the prototype sensor can distinguish mm, with an 1198 increment of 0.02 mm. It can be concluded that the prototype sensor can distinguish the Sensors 2016, 16, 8 ofthe 13 crack crack width width of of 0.02 0.02 mm. mm.

(a) (a)

(b) (b)

(c) (c)

(d) (d)

Figure 9. Test results: (a) reflection coefficient distribution along Sensor-I; (b) reflection coefficient Figure 9. 9. Test Test results: results: (a) (a) reflection reflection coefficient coefficient distribution distribution along along Sensor-I; Sensor-I; (b) (b) reflection reflection coefficient coefficient Figure variations with crack width of Sensor-I; (c) reflection coefficient distribution along Sensor-II; and variations with crack width of Sensor-I; (c) reflection coefficient distribution along Sensor-II; and (d) (d) variations with crack width of Sensor-I; (c) reflection coefficient distribution along Sensor-II; reflection coefficient variations with crack width of Sensor-II. reflection coefficient variations with crack Sensor-II. and (d) reflection coefficient variations withwidth crackof width of Sensor-II.

Figure 10. Relationship the reflection coefficient with crack width (varying from 2.40 mm to 2.60 Figure10. 10.Relationship Relationshipofof ofthe the reflection coefficient with crack width (varying 2.40to mm 2.60 Figure reflection coefficient with crack width (varying fromfrom 2.40 mm 2.60tomm) mm) of Sensor-II. mm) of Sensor-II. of Sensor-II.

5.2. Multiple Crack Sensing Test Based on transmission line theory, any separation of the spiral groove on the designed sensor will cause EM wave reflection. The sensing cable, thus, can function as a group of many sensing elements when it suffers from multiple cracks. Therefore, it is important to validate the spatial resolution of the

Sensors 2016, 16, 1198

9 of 13

5.2. Multiple Crack Sensing Test Sensors 2016, 16, 1198

9 of 13

Based on transmission line theory, any separation of the spiral groove on the designed sensor will cause EM wave reflection. The sensing cable, thus, can function as a group of many sensing elements it cracks suffersdetecting. from multiple it is important to validate the spatial sensor for when multiple Table 3cracks. lists theTherefore, spatial resolution and parameters, obtained from resolution of kinds the sensor for multiple cracks detecting. Table 3 lists the spatial resolution and the test of two of sensors with multi-spiral grooves fabricated. parameters, obtained from the test of two kinds of sensors with multi-spiral grooves fabricated.

Sensor Sensor Sensor-III Sensor-III Sensor-IV Sensor-IV

Table 3. Types of prototype sensors for the spatial resolution test. Table 3. Types of prototype sensors for the spatial resolution test. Turns of Spiral Initial Position of Length of Sensor Turns of Spiral Initial Position of the Separation the Spiral Groove Length of Sensor Separation Spiral Groove 150 cm 134.0 cm 150 cm 4.04.0 134.0 cm 150 cm 2.0 79.5 cm 150 cm 2.0 79.5 cm

Spiral Groove Spiral Groove Interval Interval 5.0 5.0mm mm 3.0 mm 3.0 mm

In this test, of the incident impulse signal launched by theby VNA 6 ps, is and In test, the therise risetime time of the incident impulse signal launched the isVNA 6 the ps, bandwidth of the tested sensor is 12 GHz.isSensor-III Sensor-IV onmounted the test system and the bandwidth of the tested sensor 12 GHz. and Sensor-III andmounted Sensor-IV on thewere test elongated at elongated a step of 0.02 mm of respectively, and their reflection distributions along the system were at a step 0.02 mm respectively, and their coefficient reflection coefficient distributions sensorthe captured the VNA Figure 11 is seen11. in It Figure 11a four11a peaks the along sensor by captured byare thepresented VNA are in presented in. ItFigure is seen inthat Figure thatoffour waveform corresponding to the spiral grooves of Sensor-III areofclearly distinguished. the spiral peaks of the waveform corresponding to the spiral grooves Sensor-III are clearlyWhen distinguished. groovethe interval is 3.0 mm, as in itsSensor-IV, reflection its peaks can also be can distinguished but not as When spiral groove interval is Sensor-IV, 3.0 mm, as in reflection peaks also be distinguished distinct asdistinct Sensor-III. From Figure 11b,Figure the adjacent willpeaks overlap if overlap the spiral interval is but not as as Sensor-III. From 11b, thepeaks adjacent will if groove the spiral groove less thanis3.0 mm, resulting in resulting a minimum resolution of resolution 3.0 mm. Consequently, the theoretical interval less than 3.0 mm, in spatial a minimum spatial of 3.0 mm. Consequently, spatial resolution (2.98 mm) calculated Equation well with well the experiment result. result. the theoretical spatial resolution (2.98 mm)by calculated byagrees Equation agrees with the experiment

(a)

(b)

Figure 11. 11. Reflection Reflection coefficient coefficient distribution distribution along along the the sensor: sensor: (a) (a) Sensor-III; Sensor-III; and and (b) (b) Sensor-IV. Sensor-IV. Figure

6. Experiment Validation on RC Members 6. Experiment Validation on RC Members To confirm the effectiveness of the prototype sensor in practical applications, a three-point To confirm the effectiveness of the prototype sensor in practical applications, a three-point bending test and a four-point bending test were conducted separately on two identical small-scale bending test and a four-point bending test were conducted separately on two identical small-scale RC RC beams with an embedded coaxial prototype crack-detecting sensor. The three-point bending test beams with an embedded coaxial prototype crack-detecting sensor. The three-point bending test is is expected to explore the single crack detecting ability of the prototype sensor, and the main objective expected to explore the single crack detecting ability of the prototype sensor, and the main objective of the four-point bending test is to validate the multiple cracks sensing ability. The concrete beam of the four-point bending test is to validate the multiple cracks sensing ability. The concrete beam was 500 mm long with a 150 mm × 250 mm rectangular cross-section. Two steel rods of 8.00 mm was 500 mm long with a 150 mm ˆ 250 mm rectangular cross-section. Two steel rods of 8.00 mm diameter were used to reinforce the beam and were placed at a distance of 15 mm from the bottom diameter were used to reinforce the beam and were placed at a distance of 15 mm from the bottom surface and 60 mm apart symmetrically about the centerline of the cross-section. The sensor made of surface and 60 mm apart symmetrically about the centerline of the cross-section. The sensor made semi-rigid SFT-50-3 coaxial cable has a diameter of 3.60 mm, then placed on the middle-line of the of semi-rigid SFT-50-3 coaxial cable has a diameter of 3.60 mm, then placed on the middle-line of the specimen at the same height of the steel rods. The geometric configuration of the cross-section of the specimen at the same height of the steel rods. The geometric configuration of the cross-section of the beam is shown in Figure 12. beam is shown in Figure 12.

Sensors 2016, 16, 1198 Sensors 2016, 16, 1198 Sensors 2016, 16, 1198

10 of 13 10 of 13 10 of 13

Figure 12. Geometric configuration of the specimen. Figure 12. Geometric configuration of the specimen.

The concrete beam specimens were cast using the commercial C25 concrete mix. A wooden mold The concrete concrete beam beam specimens specimens were were cast cast using using the the commercial commercial C25 C25 concrete concrete mix. mix. A wooden mold was used for casting the sample. During the casting, the sensing cable was held in place by two spring was was used used for for casting casting the the sample. sample. During During the casting, casting, the sensing sensing cable cable was was held held in place by two spring clamps in which a taut tension was exerted to prevent the cable from sagging. The specimens were clamps in which a taut taut tension tension was was exerted exerted to to prevent prevent the the cable cable from from sagging. sagging. The specimens specimens were cured for 28 days before testing. Then step loading on the RC beam was conducted by a reaction cured testing. Then step loading on the was conducted by a reaction frame cured for for28 28days daysbefore before testing. Then step loading on RC the beam RC beam was conducted by a reaction frame and a pressure transducer, as well as a jack. A dial indicator was also instrumented at the midand a pressure transducer, as well as a jack. A dial indicator was also instrumented at the mid-span frame and a pressure transducer, as well as a jack. A dial indicator was also instrumented at the midspan of the beam to measure deflections. A strip of narrow steel plate was added in between the of theof beam measure deflections. A stripAofstrip narrow steel plate was added between the loading span the to beam to measure deflections. of narrow steel plate wasinadded in between the loading wedge and the specimen, to prevent a local cracking of the specimen at the point of the load wedge the specimen, to preventtoa prevent local cracking the specimen the pointatofthe thepoint load application loadingand wedge and the specimen, a localofcracking of theatspecimen of the load application in the three-point bending test. Photographs showing the test setups of three-point in the three-point test. Photographs the test setups of three-point bending test and application in thebending three-point bending test.showing Photographs showing the test setups of three-point bending test and four-point bending test are given in Figure 13. A quasi-static monotonic bending four-point bending test are given in Figure 13. A quasi-static monotonic bending load with a step load bending test and four-point bending test are given in Figure 13. A quasi-static monotonic bending load with a step load increment of 0.2 kN was applied up to the failure of the specimen. At every load increment 0.2 load kN was appliedofup the failure of the Atof every load step,At theevery response load with aofstep increment 0.2to kN was applied up specimen. to the failure the specimen. load step, the response ETDR signal waveforms of the embedded sensor were acquired using a VNA, and ETDR signal waveforms the embedded sensor acquired usingwere a VNA, and a crack gauge step, the response ETDRof signal waveforms of thewere embedded sensor acquired usingwidth a VNA, and a crack width gauge measures the crack width. measures the crack a crack width gaugewidth. measures the crack width.

(a) (a)

(b) (b)

Figure 13. Test setup: (a) three-point bending test; and (b) four-point bending test. Figure 13. Test setup: (a) (a) three-point three-point bending bending test; test; and and (b) (b) four-point four-point bending bending test. test. Figure 13. Test setup:

6.1. Results Analysis of a Three-point Bending Test 6.1. Results Analysis of a Three-point Bending Test 6.1. Results Analysis of a Three-point Bending Test An initial crack with 0.4 mm width occurs at 3.6 kN of applied load during the test; then the An initial crack with 0.4 mm width occurs at 3.6 kN of applied load during the test; then the initial with 0.4 mmThe width occurs at 3.6 width kN of recorded applied load during the test;isthen the crackAn grows as crack the load increases. maximum crack at specimen failure 1.7 mm. crack grows as the load increases. The maximum crack width recorded at specimen failure is 1.7 mm. crack grows as the load increases. The maximum crack width recorded at specimen failure is 1.7 mm. A photograph of the failed specimen and the corresponding ETDR signal responses of the embedded A photograph of the failed specimen and the corresponding ETDR signal responses of the embedded A photograph of the failed specimen and the corresponding signal14responses of the embedded sensing cable regarding reflection coefficient are presentedETDR in Figures and 15, respectively. The sensing cable regarding reflection coefficient are presented in Figures 14 and 15, respectively. The sensing cable regarding reflection coefficient are presented in Figures 14 and 15, respectively. The crack crack damage induced on the specimen near the mid-span of the beam (Figure 14), is clearly indicated crack damage induced on the specimen near the mid-span of the beam (Figure 14), is clearly indicated damage induced on the specimen near the mid-span theNote beam (Figure 14), is clearly indicated by by a spike in the response waveform shown in Figure of 15a. that the reflection coefficient outside by a spike in the response waveform shown in Figure 15a. Note that the reflection coefficient outside athe spike in the region response waveform shown in Figure 15a. Notetothat theelongation reflectionin coefficient outside mid-span is all within 3~10 milli-rho, which is due small the non-cracking the mid-span region is all within 3~10 milli-rho, which is due to small elongation in the non-cracking the mid-span region is all within 3~10 milli-rho, due to small elongation in the non-cracking area. Figure 15b indicates that the peak values which of the is reflection coefficient increase almost linearly area. Figure 15b indicates that the peak values of the reflection coefficient increase almost linearly area. Figure 15b indicates that the peak the reflection coefficientsensor increase almost with with respect to the crack width. The testvalues resultsofverify that the prototype can detectlinearly the location with respect to the crack width. The test results verify that the prototype sensor can detect the location respect to themagnitude crack width. test results verify the prototype sensor can detect the location and and relative of The the crack damage of athat structure after calibration. and relative magnitude of the crack damage of a structure after calibration. relative magnitude of the crack damage of a structure after calibration.

Sensors 2016, 16, 1198

11 of 13

Sensors 2016, 16, 1198 Sensors 2016, 16, 1198

11 of 13 11 of 13

Sensors 2016, 16, 1198

11 of 13

Figure 14. Photograph of the specimen failed with single crack damage.

Figure 14.14. Photograph failedwith with single crack damage. Figure Photographof of the the specimen specimen failed single crack damage. Figure 14. Photograph of the specimen failed with single crack damage.

(a) (a)

(b) (b)

Figure 15. Test results of the three-point bending test: (a) experimental reflection waveform; and (b)

(b)waveform; and (b) Figure 15. Test results of(a) the reflection three-point bending and test: (a) experimental reflection the15. correlation between coefficient Figure Test results ofthethe three-point bendingcrack test:width. (a) experimental reflection waveform; the correlation between reflection coefficient andtest: crack Figure 15. Test results the of the three-point bending (a)width. experimental reflection waveform; and (b) and (b) the correlation between the reflection coefficient and crack width. 6.2. Results Analysisbetween of a Four-Point Bending Test and crack width. the correlation the reflection coefficient 6.2. Results Analysis of a Four-Point Bending Test Under progressive loading,Bending two dominant cracks appeared at two loading points in the four6.2. Results Analysis of a Four-Point Test 6.2.Under Results Analysis of aloading, Four-Point Bending Test cracks appeared at two loading points in the fourprogressive two dominant point bending test as presented in Figure 16a. Two spikes corresponding to the cracks’ locations point bending test asloading, presented Figure 16a.cracks Two spikes corresponding the cracks’ locations Under progressive two dominant appeared at two loading points Under loading, dominant cracks appeared at two loading points inthe thefour-point fourcapture and progressive clearly identify theintwo cracks from the reflection waveform astoshown in in Figure 16b. capture and clearly identify the cracks from the reflection waveform as shown in Figure 16b. point bending test as presented Figure Twocorresponding spikes corresponding to the 17 cracks’ locations bending test asthe presented in Figure 16a. Two16a. spikes to theFigure cracks’ locations capture Therefore, embedded sensor in is capable of detecting multiple cracks. indicates the the embedded capable ofwidth detecting multiple Figure indicates the captureidentify and clearly identify theis cracks from the waveform reflection waveform Figure 16b. waveform of crack1 an sensor example. The grows gradually fromas 0.4 mm17toin 1.6 mmTherefore, with and Therefore, clearly theascracks from the crack reflection as cracks. shown inshown Figure 16b. waveform ofthe crack1 as an example. The crack 17a, width gradually from 0.4 mm to mm with Therefore, embedded sensor is capable of detecting multiple Figure 171.6 indicates the of the increase of applied loading. Figure wegrows can observe thatcracks. reflection coefficient changes the embedded sensor is capable ofFrom detecting multiple cracks. Figure 17 indicates the waveform the increase of applied loading. From Figure 17a, we can observe that reflection coefficient changes waveform of crack1 as an example. The crack width grows gradually from 0.4 mm to 1.6 mm with generally as the crack develops. The peak values of the reflection coefficient induced by crack1 are crack1 as an example. The crack width grows gradually from 0.4 mm to 1.6 mm with the increase of generally the peak values ofwe the reflection coefficient by crack1 are the increase of crack applied loading. Figure can observe thatthat reflection coefficient changes plotted inasFigure 17b,develops. acting asThe aFrom function of 17a, crack width. It shows theinduced peak values increase applied loading. From Figure 17a, we can observe that reflection coefficient changes generally as the plotted in Figure 17b, acting as a function of crack width. It shows that the peak values increase generally crack develops. TheThese peak results values imply of the that reflection coefficient induced by crack1 linearly as as thethe crack width enlarges. the prototype sensor has the ability are of cracklinearly develops. The peak values of the reflection coefficient induced by crack1 are plotted in Figure the crack width enlarges. These results imply that the prototype hasvalues the ability of 17b, plottedasin Figure 17b, acting as a function of crack width. It shows that sensor the peak increase multiple crack monitoring. acting as a function of crack width. It shows theimply peakthat values increase linearly as the the ability crack width multiple crack linearly as themonitoring. crack width enlarges. These that results the prototype sensor has of enlarges. These results imply that the prototype sensor has the ability of multiple crack monitoring. multiple crack monitoring.

(a) (a)

(b) (b)

Figure 16. Test results: (a) crack patterns of the tested beam; and (b) the corresponding reflection (b) Figure 16. Test results:(a) (a) crack patterns of the tested beam; and (b) the corresponding reflection waveform. waveform. Figure 16. Test results: (a) crack patterns of the tested beam; and (b) the corresponding reflection

Figure 16. Test results: (a) crack patterns of the tested beam; and (b) the corresponding reflection waveform. waveform.

Sensors 2016, 16, 1198 Sensors 2016, 16, 1198

12 of 13 12 of 13

(a)

(b)

Figure 17. results of aoffour-point bending test: (a) reflection waveform; and (b) Figure 17. Test Test results a four-point bending test:experimental (a) experimental reflection waveform; correlation between the reflection coefficient and the crack width. and (b) correlation between the reflection coefficient and the crack width.

7. Conclusions 7. Conclusions This paper reports a distributed crack sensor using a coaxial cable based on the ETDR sensing This paper reports a distributed crack sensor using a coaxial cable based on the ETDR sensing principle. The prototype sensor designed with a shallow helical groove on the outer surface of the principle. The prototype sensor designed with a shallow helical groove on the outer surface of the coaxial cable is successfully fabricated by a CNC milling machine. The sensor is expected to have coaxial cable is successfully fabricated by a CNC milling machine. The sensor is expected to have lower lower signal attenuation than the state-of-the-art sensors, but a comparison test needs to be signal attenuation than the state-of-the-art sensors, but a comparison test needs to be implemented in implemented in future work. The numerical simulations explored the correlation between the sensor future work. The numerical simulations explored the correlation between the sensor performances performances and their major design parameters. It concludes that a higher peak value of the and their major design parameters. It concludes that a higher peak value of the reflection coefficient reflection coefficient corresponds to a smaller interval of the spiral groove, and a higher spatial corresponds to a smaller interval of the spiral groove, and a higher spatial resolution might be obtained resolution might be obtained by reducing the rise time of the incident impulse signal. The sensor by reducing the rise time of the incident impulse signal. The sensor performance was investigated performance was investigated through calibration tests. In the test, the reflection waveform measured through calibration tests. In the test, the reflection waveform measured with a cable sensor correlates with a cable sensor correlates well with the location and width of a crack. Results demonstrate that well with the location and width of a crack. Results demonstrate that the prototype sensor is capable the prototype sensor is capable of detecting cracks of 0.02 mm and sensing multiple cracks along the of detecting cracks of 0.02 mm and sensing multiple cracks along the cable with a minimum spatial cable with a minimum spatial resolution is 3.0 mm. Finally, a feasibility study was also conducted by resolution is 3.0 mm. Finally, a feasibility study was also conducted by using the prototype sensor using the prototype sensor to detect the presence of load-induced crack damages and identify their to detect the presence of load-induced crack damages and identify their locations in a RC member. locations in a RC member. It confirms that the prototype sensor can successfully be used to monitor It confirms that the prototype sensor can successfully be used to monitor the initiation and development the initiation and development of cracks on a structure up to its failure. of cracks on a structure up to its failure. Acknowledgments: This This work work is is supported Acknowledgments: supported by by the the National National Key Key Basic Basic Research Research Program Program of of China China (973 (973 Program) Program) under Grant Grant No. No. 2011CB013705, and also also by by the of China under under 2011CB013705, and the National National High-tech High-tech R&D R&D Program Program of China (863 (863 Program) Program) under Grant 2014AA110401. The results, findings, Grant No. No. 2014AA110401. findings, and and opinions opinions are are those those of the authors only and do not necessarily represent represent those those of of the the sponsors. sponsors. Zhi Zhou Zhou and and Hai Hai Xiao Author Author Contributions: Contributions: For For the the work work reported reported in in this this article. article. Zhi Xiao proposed proposed the the idea idea of of distributed coaxial cable sensor for crack detecting. Zhi Zhou conceived and designed the study. Tong Jiao, distributed coaxial cable sensor for crack detecting. Zhi Zhou conceived and designed the study. Tong Jiao, Peng Peng Zhao and Jia Liu designed and performed the experiments, and analyzed the data. Tong Jiao wrote and Zhao and Liu designed and performed the experiments, and analyzed the data. Tong Jiao wrote and edited edited the Jia manuscript. the manuscript. Conflicts of Interest: The authors declare no conflict of interest. Conflicts of Interest: The authors declare no conflict of interest.

References References 1. 1. 2. 2. 3. 3.

4.

Her, S.C.; Lin, S.T. Non-destructive evaluation of depth of surface cracks using ultrasonic frequency analysis. Sensors 2014, 14,S.T. 17146–17158. [CrossRef] [PubMed] Her, S.C.; Lin, Non-destructive evaluation of depth of surface cracks using ultrasonic frequency Maria, S.; Sensors Dimitrios, G.A.; B.; Patrick, G.; Danny, H. Evaluation of SHM System Produced by Additive analysis. 2014, 14, Dieter, 17146–17158. Manufacturing via Acoustic Emission OtherG.; NDT Methods. Sensors 2015, 26709–26725. Maria, S.; Dimitrios, G.A.; Dieter, B.;and Patrick, Danny, H. Evaluation of 15, SHM System Produced by Cheng, Cheng, T.M.; Chiang, C.H. Defect of Methods. concrete structures using both infrared AdditiveC.C.; Manufacturing via Acoustic Emission anddetection Other NDT Sensors 2015, 15, 26709–26725. thermography and elastic waves. Autom. Constr. 2008, 18, 87–92. Cheng, C.C.; Cheng, T.M.; Chiang, C.H. Defect detection of [CrossRef] concrete structures using both infrared thermography and elastic waves. Autom. Constr. 2008, 18, 87–92. Mori, K.; Spagnoli, A.; Murakami, Y.; Kondo, G.; Torigoe, I. A new non-contacting non-destructive testing method for detection in concrete. NDT&E Int. 2002, 35, 399–406.

Sensors 2016, 16, 1198

4. 5. 6. 7. 8. 9. 10.

11.

12.

13. 14. 15.

16.

17.

18.

13 of 13

Mori, K.; Spagnoli, A.; Murakami, Y.; Kondo, G.; Torigoe, I. A new non-contacting non-destructive testing method for detection in concrete. NDT&E Int. 2002, 35, 399–406. Yao, Y.; Glisic, B. Detection of steel fatigue cracks with strain sensing sheets based on large area electronics. Sensors 2015, 15, 8088–8108. [CrossRef] [PubMed] Li, Q.B.; Li, G.; Wang, G.L.; Yuan, L.B. CTOD measurement for cracks in concrete by fiber optic sensors. Opt. Lasers Eng. 2004, 42, 377–388. [CrossRef] Mao, J.H.; Chen, J.Y.; Cui, L.; Jin, W.L.; Xu, C.; He, Y. Monitoring the corrosion process of reinforced concrete using BOTDA and FBG sensors. Sensors 2015, 15, 8866–8883. [CrossRef] [PubMed] Sahay, P.; Kaya, M.; Wang, C.J. Fiber loop ringdown sensor for potential real-time monitoring of cracks in concrete structures: An exploratory study. Sensors 2013, 13, 39–57. [CrossRef] [PubMed] Ball, G.A.; Morey, W.W. Compression-tuned single-frequency Bragg grating fiber laser. Opt. Lett. 1994, 19, 1979–1981. [CrossRef] [PubMed] Lin, M.W.; Abatan, A.O.; Zhang, W.M. Crack damage detection of structures using distributed electrical time domain reflectometry (ETDR) sensors. In Proceedings of the SPIE Conference on Smart Structures and Materials 1998: Smart Systems for Bridges, Structures, and Highways, Newport Beach, CA, USA, 1 March 1998; Volume 3325, pp. 173–180. Lin, M.W.; Abatan, A.O.; Zhang, W.M. Crack damage detection of concrete structures using distributed electrical time domain reflectometry (ETDR) sensors. In Proceedings of the SPIE Conference on Smart Structures and Materials 1999: Smart Systems for Bridges, Structures, and Highways, Newport Beach, CA, USA, 18 May 1999; Volume 3671, pp. 297–304. Lin, M.W.; Abatan, A.O.; Zhou, Y.M. High sensitivity electric TDR distributed strain sensor. In Proceedings of the SPIE Conference on Smart Structures and Materials 2000: Sensory Phenomena and Measurement Instrumentation for Smart Structures and Materials, Newport Beach, CA, USA, 12 June 2000; Volume 3986, pp. 463–471. Lin, M.W.; Thaduri, J.; Abatan, A.O. Development of an electrical time domain reflectometry (ETDR) distributed strain sensor. Meas. Sci. Technol. 2005, 16, 1495–1505. [CrossRef] Chen, G.D.; Mu, H.M.; Pommerenke, D.J.; Drewniak, J.L. Damage detection of reinforced concrete beams with novel distributed crack/strain sensors. Struct. Health Monit. 2004, 3, 225–243. [CrossRef] Chen, G.D.; McDaniel, R.D.; Sun, S.S.; Mu, H.M.; Pommerenke, D.J.; Drewniak, J.L. Distributed crack sensors featuring unique memory capability for post-earthquake condition assessment of RC structures. Smart Struct. Syst. 2005, 1, 141–158. [CrossRef] Sun, S.S.; Pommerenke, D.J.; Drewniak, J.L.; Chen, G.D.; Xue, L.; Brower, M.A.; Koledintseva, M.Y. A novel TDR-based coaxial cable sensor for crack/strain sensing in reinforced concrete structures. IEEE Trans. Instrum. Meas. 2009, 58, 2714–2725. Chen, G.D.; Sun, S.S.; Pommerenke, D.J.; Drewniak, J.L.; Greene, G.; McDaniel, R.D. Crack detection of a full-scale reinforced concrete girder with a distributed cable sensor. Smart Mater. Struct. 2005, 14, S88–S97. [CrossRef] Chen, G.D.; McDaniel, R.D.; Brower, M.A.; Pommerenke, D.J. Crack detectability and durability of coaxial cable sensors in reinforced concrete bridge applications. Transp. Res. Rec. 2010, 2172, 151–156. [CrossRef] © 2016 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/).