sensors Article
A High-Sensitivity Flexible Eddy Current Array Sensor for Crack Monitoring of Welded Structures under Varying Environment Tao Chen *, Yuting He and Jinqiang Du Aeronautic Engineering College, Air Force Engineering University, 1 Baling Road, Xi’an 710038, China;
[email protected] (Y.H.);
[email protected] (J.D.) * Correspondence:
[email protected]; Tel.: +86-029-84787254 Received: 4 May 2018; Accepted: 1 June 2018; Published: 1 June 2018
Abstract: This paper develops a high-sensitivity flexible eddy current array (HS-FECA) sensor for crack monitoring of welded structures under varying environment. Firstly, effects of stress, temperature and crack on output signals of the traditional flexible eddy current array (FECA) sensor were investigated by experiments that show both stress and temperature have great influences on the crack monitoring performance of the sensor. A 3-D finite element model was established using Comsol AC/DC module to analyze the perturbation effects of crack on eddy currents and output signals of the sensor, which showed perturbation effect of cracks on eddy currents is reduced by the current loop when crack propagates. Then, the HS-FECA sensor was proposed to boost the sensitivity to cracks. Simulation results show that perturbation effect of cracks on eddy currents excited by the HS-FECA sensor gradually grows stronger when the crack propagates, resulting in much higher sensitivity to cracks. Experimental result further shows that the sensitivity of the new sensor is at least 19 times that of the original one. In addition, both stress and temperature variations have little effect on signals of the new sensor. Keywords: flexible eddy current array sensor; structural health monitoring; sensitivity; crack detection; effects of stress and temperature variations
1. Introduction Crack damage has become the main factor to the safety of structures, and connection parts of metal structures are generally stress-concentration areas where cracks are most likely to occur. As the most important structural connection in various equipment and facilities, welds are also vulnerable to cracks. Many equipment and facilities, such as cranes, power generation equipment, pressure vessels, pipelines, bridges, ships, etc., have all experienced catastrophic accidents due to the fracture of welded structures. According to a statistic of pressure vessels, more than 70% of accidents are caused by cracks [1]. In the inspection statistics for many harbor cranes, the potential threat to the safe use of crane metal structures is mainly cracks at weld seams, and the resulting failures account for approximately 80% of all failures [2]. Despite all kinds of cracks in these cranes, these equipment are still in service due to various reasons, which brings huge safety risks. Structural health monitoring (SHM) technology can promptly detect defects and improve the safety and maintainability of equipment and facilities. Current SHM technologies include optical fiber [3–6], CVM sensor [7], acoustic emission [8,9], ultrasonic [10,11], and other sensor technologies [12,13]. Flexible eddy current sensors are manufactured by the flexible printed circuit technology, which can be used for crack monitoring of various curved structures. Alexi Rakow and Fukuo Chang [14] developed a flexible eddy current sensor integrated with the bolt for monitoring crack growth of structures around bolt
Sensors 2018, 18, 1780; doi:10.3390/s18061780
www.mdpi.com/journal/sensors
Sensors 2018, 18, 1780
2 of 16
holes. The signals of the sensor are affected by stress changes during the experiment, resulting in a lower sensitivity to cracks. JENTEK Sensors [15–19] has developed a variety of Meandering Winding Sensors 2017, 17, x FOR PEER REVIEW 2 of 16 Magnetometer (MWM) sensors for damage detection and quality evaluation of metal structures, including blades, turbine pressure vessels, oil pipelines other[15–19] structures during aeroengine the experiment, resulting in disks, a lower sensitivity to cracks. JENTEKand Sensors has with developed a variety of Meandering Winding Magnetometer for damage of detection complex curved surfaces. In addition, the sensor has been (MWM) used to sensors crack monitoring bolt-jointed and quality evaluation metal structures, including aeroengineon blades, turbine disks, signal pressure structures. However, the of influence of the service environment the sensor output has not vessels, oil pipelines and other structures with complexofcurved surfaces.changes In addition, the sensor been studied. In a previous paper [20], the influence temperature on signals of has a rosette been used to crackismonitoring of bolt-jointed structures. However, the influence service eddy current sensor studied. Results show that the ambient temperature hasofa the great influence environment on the sensor output signal has not been studied. In a previous paper [20], the on the output signal of the sensor by affecting the conductivity and permeability of the structures influence of temperature changes on signals of a rosette eddy current sensor is studied. Results show under testing. that the ambient temperature has a great influence on the output signal of the sensor by affecting the Therefore, the influence of stress, temperature, and crack on the output signal of a traditional conductivity and permeability of the structures under testing. FECA sensor for crack monitoring of welded structures wason studied. A finite model was Therefore, the influence of stress, temperature, and crack the output signal element of a traditional established to study the effect of crack perturbation on eddy currents distribution and output signals. FECA sensor for crack monitoring of welded structures was studied. A finite element model was According to the obtained bycrack the finite elementon analysis, the HS-FECA sensor was proposed. established to results study the effect of perturbation eddy currents distribution and output According the results obtained by the finite element analysis, simulation the HS-FECA sensor was Thesignals. sensitivity of the to sensor to cracks was studied using finite element and experiments. proposed. The sensitivity of the sensor to cracks was studied using finite element simulation and
2. Experiments experiments. of Traditional FECA Sensor 2.1.2.FECA Sensor of Traditional FECA Sensor Experiments The schematic of a traditional FECA sensor is shown in Figure 1. The meandering exciting coil 2.1. FECA Sensor forms several rectangular loops. Each sensing coil lies in the middle of each rectangular loop. When the schematic of a traditionalcurrent FECA sensor is shown Figure 1. The meandering exciting sensor isThe working, a time-varying I is applied toin the exciting coil to generate eddy coil currents forms several rectangular loops. Each sensing coil lies in the middle of each rectangular loop. When on the surface of the structure under monitoring, and the varying eddy currents causes each sensing the sensor is working, a time-varying current I is applied to the exciting coil to generate eddy coil to generate an induced voltage V. When the crack propagates below the corresponding sensing currents on the surface of the structure under monitoring, and the varying eddy currents causes each channel, the currents the structure will be changed, resulting below in variations of the induced sensing coileddy to generate an of induced voltage V. When the crack propagates the corresponding voltage. This paper defines the sensor output signal A as: R sensing channel, the eddy currents of the structure will be changed, resulting in variations of the induced voltage. This paper defines the sensor output signal AR as:
V A R =𝑉 𝐴𝑅 = I
(1)
𝐼
(1)
where V isV the voltage sensingchannel channel and is the magnitude ofexciting the exciting current. where is the voltagemagnitude magnitude of of aa sensing and I isI the magnitude of the current.
Structures under testing D
Crack
Channel 1 Channel 2 Channel 3 Channel 4 Channel 5
V1 V2 V3 Exciting coil
V4
V5 Sensing coil
Figure 1. 1. The The schematic FECA sensor. Figure schematicofofa atraditional traditional FECA sensor.
Sensors 2017, 17, x FOR PEER REVIEW Sensors2017, 2018,17, 18,x1780 Sensors FOR PEER REVIEW
3 of 16 16 3 3ofof16
The FECA sensor is manufactured using a flexible printed circuit board manufacturing process, The FECA sensor is manufactured using a flexible printed circuit board manufacturing process, whichThe is comfortable various complex curved structures. dangerous part of the welded FECA sensorfor is manufactured using a flexible printed The circuit board manufacturing process, which is comfortable for various complex curved structures. The dangerous part of the welded structure iscomfortable often located atvarious the weldcomplex seam. The FECAstructures. sensor can be used for crackpart monitoring at this which is for curved The dangerous of the structure is often located at the weld seam. The FECA sensor can be used for crack monitoringwelded at this curved and rough location. structure is often located at the weld seam. The FECA sensor can be used for crack monitoring at this curved and rough location. curved and rough location. 2.2. Effects of Stress Variations on the Output Signal of the Sensor 2.2. Effects of Stress Variations on the Output Signal of the Sensor 2.2. Effects of Stress Variations on the Output Signal of the Sensor The experimental system used is shown in Figure 2. The system includes a measurement The experimental system used is shown in Figure 2. The system includes a measurement The experimental system usedsystem, is shown 2. The systemwith includes a measurement instrument, MTS810 material testing andin theFigure specimen integrated the sensor (Figure 3). instrument, MTS810 material testing system, and the specimen integrated with the sensor (Figure 3). instrument, MTS810 instrument material testing system,anandembedded the specimen integrateda with the sensorsource, (Figure 3). The measurement includes computer, stimulation a The measurement instrument includes an embedded computer, a stimulation source, a The measurement instrument includes an embedded computer, a stimulation source, conditioning conditioning module and an AD converter. The frequency of the exciting current is 1a MHz in this conditioning module and an AD converter. The frequency of the exciting current is 1 MHz in this module and an AD converter. The frequency of the exciting 1 MHz inisthis paper, paper, the sampling rate is 50 MHz and the resolution is 16 current bit. Theisspecimen welded bythe a paper, the sampling rate is 50 MHz and the resolution is 16 bit. The specimen is welded by a sampling rate is 50 MHz and the resolution is 16 bit. The specimen is welded by a single-side welding single-side welding and double-side forming process. The sensor is directly mounted on the single-side welding and double-side forming process. The sensor is directly mounted on the and double-side forming process. sensor is directly mounted the specimen sealants without specimen by sealants without any The other surface treatment and theonsense channels by cover the welding specimen by sealants without any other surface treatment and the sense channels cover the welding any other surface treatment and the sense channels cover the welding seam. seam. seam.
MTS 810 Material Testing System MTS 810 Material Testing System
Specimen Specimen
Measurement instrument Measurement instrument
Figure The Figure2. Theexperimental experimentalsystem systemfor foreffects effectsof ofstress stressand andcracks. cracks. Figure 2.2.The experimental system for effects of stress and cracks. Sensor Sensor
Sealant Sealant
Specimen Specimen
Figure 3. The specimen integrated with the sensor. Figure Figure3.3.The Thespecimen specimenintegrated integratedwith withthe thesensor. sensor.
Due to the magnetoelastic effect of the ferromagnetic material, output signals of the sensor are Due to the magnetoelastic effect of the ferromagnetic material, output signals of the sensor are Due thestress. magnetoelastic the ferromagnetic material, output signals of sensor are affected bytothe Therefore,effect beforeofcarrying out crack monitoring experiments, thethe influence of affected by the stress. Therefore, before carrying out crack monitoring experiments, the influence of affected theoutput stress. signal Therefore, before carrying crack monitoring the influence of stress onby the of the sensor was out studied. During the experiments, test, a constant–amplitude stress on the output signal of the sensor was studied. During the test, a constant–amplitude stress on the output signal the=sensor was studied. During test, a constant–amplitude spectrum spectrum (maximum stressofSmax 220 MPa, stress ratio R = the 0, loading frequency f = 0.02 Hz) was spectrum (maximum stress Smax = 220 MPa, stress ratio R = 0, loading frequency f = 0.02 Hz) was (maximum stress Smax =while 220 MPa, stress ratio R = 0, loading frequency f = 0.02 Hz) was applied to applied to the specimen, signals of the sensor were collected. As shown in Figure 4, the output applied to the specimen, while signals of the sensor were collected. As shown in Figure 4, the output the specimen, while signals of the sensor were collected. As shown in Figure 4, the output signal of signal of channel 1 changes with the change of stress, and the varying amplitude is about 1.5%. This signal of channel 1 changes with the change of stress, and the varying amplitude is about 1.5%. This channel 1 changes with the change of varying amplitude is crack aboutmonitoring. 1.5%. This would would reduce the sensor’s sensitivity tostress, cracksand andthe have a certain impact on would reduce the sensor’s sensitivity to cracks and have a certain impact on crack monitoring. reduce the sensor’s sensitivity to cracks and have a certain impact on crack monitoring.
Sensors 2018, 18, 1780
4 of 16
Sensors 2017, 17, x FOR PEER REVIEW Sensors 2017, 17, x FOR PEER REVIEW
4 of 16 4 of 16 Channel 1 Channel 1
230 230 190 190 150 150 110 110 70 70 30 30 -10 200 400 600 800 1000 1200 -10 Sample number 200 400 600 800 1000 1200 Sample number
Stress / MPa Stress / MPa
Normalized AR Normalized AR
1.005 1.005 1 1 0.995 0.995 0.99 0.99 0.985 0.985 0.98 0 0.98
Stress Stress
0
Figure4.4.The The output output signal stress. Figure signalvaries varieswith with stress.
Figure 4. The output signal varies with stress.
2.3. Experimental Research on Crack Monitoring of Welded Structures 2.3. Experimental Research on Crack Monitoring of Welded Structures 2.3. Experimental Research on Crack Monitoring of Welded Structures To perform the crack monitoring experiment, a constant–amplitude spectrum (maximum stress perform the crack monitoring experiment, af =constant–amplitude spectrum (maximum stress ToMPa, perform theratio crack experiment, a15Hz) constant–amplitude (maximum stress STo max = 220 stress R =monitoring 0, loading frequency was applied to spectrum the specimen. As shown Smaxin= S220 MPa, stress ratio R = 0, loading frequency f = 15Hz) was applied to the specimen. As shown max = 220 MPa, stress ratio R = 0, loading frequency f = 15Hz) was applied to the specimen. As shown Figure 5, output signal of each channel arrives at each turning point A–E when crack tip in Figure output of signal of eachcoil channel arrives at Therefore, each point turning point A–E when tip in Figure 5, output signal eachexciting channel arrives at each turning A–E when crack tip crack propagates propagates to5,corresponding of each channel. the loading cycles separately propagates to corresponding exciting coil of each channel. Therefore, the loading cycles separately correspondingexciting to crack coil length a of 5channel. mm, 9 mm, 13 mm,the 17 loading mm, andcycles 21 mm can be obtained, as to corresponding of each Therefore, separately corresponding corresponding crack length ofthat 5 mm, 9 mm, 17can mm, and 21 mm as can be obtained, shown in Figure should be noted because the13 sensor can bebe easily affected byshown stress and the as 6. to crack length a of 6.5toIt mm, 9 mm, 13a mm, 17 mm, and 21mm, mm obtained, in Figure shown inofFigure 6. It should beisnoted that because the sensor can be easily affected stress andinthe sensitivity the that sensor to cracks so low, the estimation errors areby larger. The maximum change It should be noted because the sensor can be easily affected stress and theby sensitivity of the sensitivity of theofsensor to cracks is so3,low, the estimation errors are larger. The maximum change in the output signal channel 1, channel and channel 5 is approximately 5%, while the maximum sensor to cracks is so low, the estimation errors are larger. The maximum change in the output signal the output signal of channel 1, channel 3, channel and channel 5 isapproximately approximately1.5%. 5%, while the maximum change in the output signal of channel 2 and 4 is just of channel 1, channel 3, and channel 5 is approximately 5%, while the maximum change in the output change in the output channel 2 and channel is just approximately 1.5%. In addition, it can signal be seenofin the figure that, when4 no crack exits in the structure, the output signal of channel 2 and channel 4 is just approximately 1.5%. Inthe addition, can be seen in the figure when no crack exits the structure, thetooutput signal of sensor itfluctuates greatly under thethat, effect of stress. When theincrack propagates the Insignal addition, can befluctuates seen in the the fluctuation figure that, when nostress. exits in the structure, of theitsensing sensor greatly under the effect of the crack propagates tooutput the corresponding channel, amplitude ofcrack the When signal increases first andthe then signal of the sensor fluctuates greatly under the effect of When the crack propagates corresponding fluctuation amplitude of the signal first and then to decreases. This issensing mainlychannel, due to the stress concentration in stress. the crack tip increases region during crack the corresponding sensing channel, the fluctuation of signal increases first then decreases. and This is mainly duevery to small stressstress concentration in between thethe crack region during crack propagation, there only exists inamplitude the region thetip fatigue source andand the propagation, and there only exists very small stressininthe thecrack regiontip between fatigue source and the decreases. This is mainly due to stress concentration regionthe during crack propagation, crack tip. crack tip. exists very small stress in the region between the fatigue source and the crack tip. and there only Normalized AR signal (normalized) Output
A
1.041.05 1.031.04
1.021.03
Crack Crack
1.011.02 11.01
0.99
1
0.980.99 80,000 80000 0.98 80,000 80000
90,000 100,000 90000 100000 Loading cycles 90,000 100,000 90000 100000
110,000 110000 110,000 110000
1.014 1.012 1.014 1.01 1.012 1.0081.01 1.006 1.008 1.004 1.006 1.002 1.004 1 1.002 0.998 1 0.996 0.998 0.994 0.996 80000 80,000 0.994 80000 80,000
Normalized AR signal (normalized) Output
A
1.051.06
Normalized AR signal (normalized) Output
Normalized AR signal (normalized) Output
1.06
Loading cycles
Normalized AR signal (normalized) Output
C
Crack
1.031.04
1.021.03 1.011.02 11.01
0.99
1
0.980.99 80,000 80000 0.98 80,000 80000
90,000 100,000 90000 100000 Loading cycles 90,000 100,000 90000 100000
90000 100000 90,000 100,000 Loading cycles 90000 100000 90,000 100,000
110000 110,000 110000 110,000
110,000 110000
D
1.02Crack 1.015
Normalized AR signal (normalized) Output
Normalized AR signal (normalized) Output
Normalized AR signal (normalized) Output
1.041.05
Crack Crack
(b) a = 9 mm, N = 102,320 (b) a = 9 mm, N = 102,320
1.02
C
1.051.06Crack
B
Loading cycles
(a) a = 5 mm, N = 97,580 (a) a = 5 mm, N = 97,580
1.06
B
D
Crack
1.015 1.01
1.01 1.005
1.005 1
1 0.995 80000 80,000 0.995
110,000 110000
Loading cycles
80000 80,000
90000 100000 90,000 100,000 Loading cycles 90000 100000 90,000 100,000
Loading cycles
(c) a = 13 mm, N = 104,580 (c) a = 13 mm, N = 104,580
110000 110,000
(d) a = 17 mm, N = 105,725 (d) a = 17 mm, N = 105,725
Figure 5. Cont.
110000 110,000
Sensors 2018, 18, 1780
5 of 16
Sensors 2017, 2017, 17, 17, xx FOR FOR PEER PEER REVIEW REVIEW Sensors
Normalized AR (normalized) signal Output Normalized AARR signal Output Normalized (normalized) signal(normalized) Output
Sensors 2017, 17, x FOR PEER REVIEW
of 16 16 55 of 1.07 1.07 1.06 1.06
1.05 1.07 1.05 1.04 1.06 1.04 1.03 1.05 1.03 1.02 1.04 1.02
5 of 16
EE Crack Crack E Crack
1.01 1.03 1.01 1.0211 0.99 1.01 0.99 80,000 80000 80,000 1 80000 0.99 80,000(e) 80000
90,000 100,000 90000 100000 90,000 100,000 90000 100000 Loading cycles cycles Loading
110,000 110000 110,000 110000
90,000 100,000 90000 100000 Loading cycles
21 mm, mm, N N == 106,640 106,640 (e) aa == 21
110,000 110000
Figure 5. Output signals vary with the growth of crack.
Figure Output vary the crack. Figure 5. 5. Output vary with thegrowth growthofof crack. (e) signals asignals = 21 mm, Nwith = 106,640 Figure 5. Output signals vary with the growth of crack.
95,000 95,000 95,000
100,000 100,000 100,000
105,000 105,000 105,000
110,000 110,000 110,000
Figure 6. The The relationbetween betweenthe the crack crack length length and loading cycles. Figure 6. relation between the crack cycles. Figure 6. The relation lengthand andloading loading cycles.
2.4. Effects Effects of of Temperature Temperature Variations on Output Output Signals of the the Sensor FigureVariations 6. The relation between the crack length and loading cycles. 2.4. Signals of 2.4. Effects of Temperature Variations ononOutput Signals of the Sensor Sensor Both stress stress and and temperature temperature have have influences influences on on the the electrical electrical conductivity conductivity and and permeability permeability of of Both 2.4. Effects of Temperature Variations oninfluences Output Signals of the Sensor conductivity and Both stress and temperature have on the electrical permeability of the the material, causing interference with output signals of the sensor. As shown in previous sections, the material, causing interference with output signals of the sensor. As shown in previous sections, material, causing withhave output signals ofthe theelectrical sensor. As shown previous sections, Both stress and temperature influences and permeability of the the change change of interference stress has aa greater greater impact on the theon output signal of ofconductivity the sensor,inbut but will not not cause the of stress has impact on output signal the sensor, itit will cause the material, causing interference with output signals of the sensor. As shown in previous sections, change of stress has a greater impact on the output signal of the sensor, but it will not cause excessive excessive interference to the monitoring process. Therefore, this section focuses on the effect of excessive interference to the monitoring process. Therefore, this section focuses on the effect of the change of stress has a greater impact on the output signal of the sensor, but it will not cause temperature changes on output output signalsTherefore, of the the sensor. sensor. interference to the monitoring process. this section focuses on the effect of temperature temperature changes on signals of excessive interference toofthe monitoring process. Therefore, this section focuses the effectwas of Asoutput shown in Figure Figure 7, during the experiment, experiment, the specimen specimen integrated withonthe the sensor was changes on signals the sensor. As shown in 7, during the the integrated with sensor temperature changes on output signals of the sensor. placed in the the environmental testthe chamber. As shown shown in Figure 8, 8,integrated the sensor’s sensor’swith output signal increases placed in test chamber. As Figure the output increases As shown inenvironmental Figure 7, during experiment, the in specimen thesignal sensor was placed As shown in Figure 7, during thevariation experiment, the When specimen integrated with the sensor was by about 6% in the whole temperature variation process. When the crack grows, variations of output by about 6% in the whole temperature process. the crack grows, variations of output in the environmental test chamber. As shown in Figure 8, the sensor’s output signal increases by about placed in the environmental test chamber. As shown in Figure 8, the sensor’s output signal increases signals are are less than than 5%. 5%. Therefore, Therefore, changes changes in in the the ambient ambient temperature would would cause cause significant significant 6% insignals the wholeless temperature variation process. When the cracktemperature grows, variations of output signals are by about 6% in the output whole variation When crack grows, of output interference with outputtemperature signal of of the the sensor. process. Therefore, the the sensitivity of the thevariations sensor needs needs to be be interference with signal sensor. Therefore, the sensitivity of sensor to less signals than 5%. Therefore, changes in the ambient temperature would cause significant interference with are less than 5%. Therefore, changes in the ambient temperature would cause significant improved by optimizing the design of the sensor. improved by optimizing the design of the sensor. output signal of the sensor. Therefore, the sensitivity of the sensor needs to be improved by optimizing interference with output signal of the sensor. Therefore, the sensitivity of the sensor needs to be the design of the sensor.The specimen Environment test chamber chamber Measurement instrument instrument improved by optimizing the design sensor.test integratedof the Environment Measurement The specimen integrated with the the sensor sensor with
The specimen integrated with the sensor
Environment test chamber
Measurement instrument
Figure 7. 7. Experimental Experimental research research on on effects effects of of temperature. temperature. Figure Figure 7. Experimental research on effects of temperature.
Figure 7. Experimental research on effects of temperature.
Sensors 2018, 18, 1780
6 of 16
Sensors 2017, 17, x FOR PEER REVIEW
6 of 16
1.07 Sensors 2017, 17, x FOR PEER REVIEW
Normalized AR / [V/A]AR Normalized AR / [V/A]AR
1.06 1.07 1.05 1.06 1.04 1.05 1.03 1.04 1.02 1.03 1.01 1.02 1 1.01 0.99 1 0
6 of 16
T=70℃ T=70℃ CH1 CH2 CH1 CH3 CH2 CH4
T=15℃
CH3 CH5 CH4
T=15℃
CH5 100
0.99
0
100
200 Sample number 200
300 300
400 400
Figure 8. Signals of the sensor when temperature changes. number Sample
Figure 8. Signals of the sensor when temperature changes. Figure 8. Signals the sensor when temperature changes. 3. The Finite Element Analyses of theofSensor
3. The Finite Element Analyses of the Sensor
this paper, a 3-D finite of element model is established in Comsol software [21], as shown in 3. TheInFinite Element Analyses the Sensor In this paper, a 3-D finite element model established in Comsol software [21], as that shown Figure 9. Figure 9. The coil thickness is so small andisthe exciting current frequency is so high the in skin In this paper, a 3-D element modelcurrent isTherefore, established Comsol software asto shown effect of eddy current in finite this model obvious. thisinpaper uses the that line [21], model simulate The coil thickness is so small and the is exciting frequency is so high the skin effectinof eddy Figure 9. The coil is so small and thesense exciting current frequency the is socrack highpropagation, that the skina exciting andthickness area model to simulate current in thiscoils model isthe obvious. Therefore, this paperchannels. uses the To linesimulate model to simulate exciting coils and effect of eddy current in this model is obvious. Therefore, this paper uses the line model to simulate 0.02model mm wide crack is established with a length from 0 the mmcrack to 23 mm. The distance between the left the area to simulate sense channels. To simulate propagation, a 0.02 mm wide crack is exciting coils and area to simulate sense channels. simulate crack propagation, a most exciting coilthe and themodel left edge of the structure is 1 mm.To The inducedthe voltage of each sensing established a crack lengthis from 0 mm with to 23amm. The distance between the left most exciting 0.02 mmwith wide established length from 0 mm to 23 mm. The betweencoil theand left the left channel can be obtained using Formula (2), and then the output signal Adistance R of each channel can be edgemost of the structure isand 1 mm. Theedge induced voltage of each sensing channel can be obtained using Formula exciting coil the left of the structure is 1 mm. The induced voltage of each sensing obtained by Formula (1). (2), and thencan thebe output signal AR Formula of each channel obtained by Formula (1). channel can be channel obtained using (2), and can thenbe the output signal AR of each 𝑑𝐵 obtained by Formula (1). (2) 𝑉=∬ x 𝑑𝑡dB𝑑𝑆 𝑑𝐵 V= dS (2) (2) 𝑉 = and ∬ Sdt 𝑑𝑆 where B is magnitude of magnetic flux density is the area of the sense element. 𝑑𝑡
where B isBmagnitude ofofmagnetic andS Sisis the area of the sense element. where is magnitude magneticflux flux density density and the area of the sense element.
(a) Isometric view
(b) Top view
sensor. (a) Isometric viewFigure 9. 3-D FE model of the(b) Top view Figure 9. 3-DisFE model as: of the sensor. The variation rate of the output signal defined Figure 9. 3-D FE model of the sensor. Δ𝐴𝑅 |𝐴𝑅 − 𝐴𝑅0 | The variation rate of the output signal is defined 𝑆𝐶 = = as: (3) 𝐴defined 𝐴𝑅0 𝑅0 The variation rate of the output signal is as: Δ𝐴𝑅 |𝐴𝑅 − 𝐴𝑅0 | 𝑆𝐶 = signal=and AR0 is the initial output signal of the sensor. (3) where SC is the variation rate of the output 𝐴𝑅0 𝐴 ∆A | A𝑅0 | As shown in Figure 10, when the crack propagates to A the exciting coil of a channel, R R− R0front-end SC = signal and = R0 is the initial output signal of the sensor. (3) where C is the variation rate of the output the SCSof the channel begins to increase rapidly; tip just reaches the rear-end exciting A R0 whenAthe Acrack R0 As shown in Figure 10, when the crack propagates to the front-end exciting coil of a channel, coil of the channel, the SC reaches its maximum value and then slowly decreases. Therefore, the the of thevalue channel to rapidly; when theAcrack tipcracks. just reaches the rear-end where SSCC is variation ratecharacterize ofincrease the output signal and the initial output signal ofexciting theshow sensor. R0 isto maximum of Sbegins C can the sensor’s sensitivity The simulation results coil of the channel, the S C reaches its maximum value and then slowly decreases. Therefore, the As in Figure 10, of when propagates front-end exciting coil propagates. of a channel, the thatshown the maximum value the Sthe C incrack each channel does to notthe exceed 5% when the crack value ofsmall SC can characterize the sensor’s cracks. The simulation results exciting show The is so that the sensor’s sensitivity to cracks istotip too low.reaches SC ofmaximum the change channel begins to increase rapidly; whensensitivity the crack just the rear-end coil that the maximum value of the SC in each channel does not exceed 5% when the crack propagates. of the channel, the SC reaches its maximum value and then slowly decreases. Therefore, the maximum The change is so small that the sensor’s sensitivity to cracks is too low.
value of SC can characterize the sensor’s sensitivity to cracks. The simulation results show that the maximum value of the SC in each channel does not exceed 5% when the crack propagates. The change is so small that the sensor’s sensitivity to cracks is too low.
Sensors 2018, 18, 1780
7 of 16
Sensors 2017, 17, x FOR PEER REVIEW
7 of 16
Sensors2017, 2017,17, 17,x xFOR FORPEER PEERREVIEW REVIEW Sensors 5.0%
7 7ofof1616
5.0% 4.5% 5.0% 4.5% 4.0% 4.5% 4.0% 3.5% 4.0%
SC SC SC
3.5% 3.0% 3.5%
CH1 CH2 CH1 CH1 CH3 CH2 CH2 CH4 CH3 CH3 CH5 CH4 CH4 CH5 CH5
3.0% 2.5% 3.0% 2.5% 2.0% 2.5% 2.0% 1.5% 2.0% 1.5% 1.0% 1.5% 1.0% 0.5% 1.0% 0.5% 0.0% 0.5% 0 0.0% 0.0% 00
5
10
55
15 a/mm 1515 1010 a/mm a/mm
20
25
2020
2525
Figure TheSSCof of each each channel thethe crack growth. Figure 10.10. The channelvaries varieswith with crack growth. C Figure10. 10.The TheSCSof C of eachchannel channelvaries varieswith withthe thecrack crackgrowth. growth. Figure each
As shown in Figures 11 and 12, changes of permeability and conductivity have a significant As shown in Figures 11of11 and 12,12, changes of and conductivity have a significant Asshown shown Figures 11 and 12, changes permeability andconductivity conductivity have significant effect on the output signal the sensor. Temperature and stress and variations can cause changes of the As ininFigures and changes ofofpermeability permeability have a asignificant effect on the output signal of the sensor. Temperature and stress variations can cause changes of the effect on the output signal of the sensor. Temperature and stress variations can cause changes the material’s conductivity and permeability, which is the reason the output signal of the sensor changes effect on the output signal of the sensor. Temperature and stress variations can cause changes ofofthe material’s conductivity and permeability, which is the reason the output signal of the sensor changes with temperature and stress variations. material’s conductivity andand permeability, theoutput outputsignal signal sensor changes material’s conductivity permeability,which whichisisthe the reason reason the of of thethe sensor changes withtemperature temperature andstress stress variations. variations. with with temperature andand stress variations. 2.50% 2.50% 2.50% 2.00% 2.00% 2.00%
SC SC SC
1.50% 1.50% 1.50%
CH1 CH2 CH1 CH1 CH3 CH2 CH2 CH4 CH3 CH3 CH5 CH4 CH4 CH5 CH5
1.00% 1.00% 1.00% 0.50% 0.50% 0.50% 0.00% 100 0.00% 0.00% 100 100
110
120
130
Relative120 permeability 110 120 130 110 130 Relativepermeability permeability Relative
140 140 140
Figure 11. The SC of each channel varies with permeability of the structure. Figure 11.The The Sof C of each channelvaries varieswith withpermeability permeabilityofof the structure. Figure C each channel varies the structure. Figure 11. 11. The S Sof each channel with permeability of the structure.
SC SC SC
C
11 9 1111 Conductivity (MS/m) 99 Conductivity(MS/m) (MS/m) Conductivity Figure 12. The SC of each channel varies with conductivity of the structure. Figure 12. 12. The SCSof channel varies with conductivity of the structure. Figure 12.The The SCofeach ofeach each channel varieswith withconductivity conductivityofof the structure. Figure C channel varies the structure. 7 77
Sensors 2017, 17, x FOR PEER REVIEW
8 of 16
Sensors 2017, 17, x FOR PEER REVIEW 8 of 16 Sensors 18, 1780 8 ofby 16 As2018, shown in Figure 13a, before the crack initiates, three large eddy current loops are induced
exciting coils oninthe structure surface toinitiates, the opposite flow of adjacent exciting coils, As shown Figure 13a, before thedue crack three current large eddy current loops are induced by corresponding to channel 1, channel 3 and channel 5, respectively. When the crack propagates exciting coils oninthe structure surface to the opposite flow current of adjacent exciting coils, As shown Figure 13a, before thedue crack initiates, threecurrent large eddy loops are induced through the coils monitoring area, thesurface three3 larger eddy current loops are split eddy current corresponding toonchannel 1, channel and channel 5, respectively. When the six crack propagates by exciting the structure due to the opposite current flow of into adjacent exciting coils, loops. through the monitoring area, the three larger eddy current loops are split into six eddy current corresponding to channel 1, channel 3 and channel 5, respectively. When the crack propagates through loops. the monitoring area, the three larger eddy current loops are split into six eddy current loops.
(a) a = 0 mm
(b) a = 23 mm
(a) a = 0 mm (b) apropagates = 23 mm through the Figure 13. Distributions of eddy currents before and after the crack monitoring Figure 13. area. Distributions of eddy currents before and after the crack propagates through the Figure 13. Distributions of eddy currents before and after the crack propagates through the monitoring area. monitoring area.
As shown in Figure 14, when the crack propagates to the front end of channel 1, eddy currents As shown inflows Figure 14, crack to the of eddy currents around crack thethe crack tip.propagates The crack to disturbs theend distribution Asthe shown in Figurearound 14, when when the crack propagates the front front end of channel channelof1, 1,eddy eddycurrents, currents around the crack flows around the crack tip. The crack disturbs the distribution of currents, and and S C of the channel 1 starts to increase. As the crack grows, more and more eddy currents are around the crack flows around the crack tip. The crack disturbs the distribution of eddy currents, S of the channel 1 starts to increase. As the crack grows, more and more eddy currents are disturbed, disturbed, and the S C continues to increase. When the crack propagates to the middle of the channel C and SC of the channel 1 starts to increase. As the crack grows, more and more eddy currents are continues to increase. theeddy crackcurrents propagates the middle of currents the channel 1 (a =by 3 mm), 1and (a =the3 Smm), itthe is Sobvious that When part of flowtotoward eddy excited the C and disturbed, C continues to increase. When the crack propagates to the middle of the channel it is obvious that of1eddy currents flow toward currents by thecurrents rear-end coilthe of rear-end and form loop. With the eddy growth of the excited crack, eddy 1 (a = 3 coil mm),of itchannel ispart obvious that parta of eddy currents flow toward eddy currents excitedflowing by channel 1 and form a loop. With the growth of the crack, eddy currents flowing around the crack tip around the crack tip are getting less, and the perturbation effect of the crack tip on eddy currents are rear-end coil of channel 1 and form a loop. With the growth of the crack, eddy currents flowing are getting and the effect ofperturbation the increasing crack tipeffect on eddy areon also getting also getting weaker. leads toless, the and fact that the rate ofthe Scurrents C is getting smaller as theweaker. crack around the less, crack tipThis are perturbation getting the of crack tip eddy currents are This leads to the fact that the increasing rate of S is getting smaller as the crack grows. grows. C also getting weaker. This leads to the fact that the increasing rate of SC is getting smaller as the crack grows.
Figure Figure14. 14.Distributions Distributionsof ofeddy eddycurrents currentswhen whenthe thecrack crackpropagates propagatesfrom from00mm mmto to55mm. mm. Figure 14. Distributions of eddy currents when the crack propagates from 0 mm to 5 mm.
Sensors 2018, 18, 1780 Sensors 2017, 17, x FOR PEER REVIEW
9 of 16 9 of 16
Figure 15shows shows eddy current distribution when crack tip in propagates in the Figure 15 thethe eddy current distribution when the crackthe tip propagates the corresponding corresponding of channel 2. Because directions of exciting at the upper end of opposite channel area of channelarea 2. Because directions of exciting currents at thecurrents upper end of channel 2 are 2and aretwo opposite and two exciting coils are so close, eddy currents in the region do not form loop exciting coils are so close, eddy currents in the region do not form a loop similar to thata of the similar to that of the channel 1, and the eddy current density in the region is smaller. When the crack channel 1, and the eddy current density in the region is smaller. When the crack propagates in this propagates incurrents this region, eddy currents crack tip less, resulting in a effect weaker region, eddy flowing around theflowing crack tiparound are less,the resulting in are a weaker perturbation on perturbation effect on eddy currents. Therefore, as shown in Figure 10, the maximum values of S C for eddy currents. Therefore, as shown in Figure 10, the maximum values of SC for channel 2 and channel channel 2 and channel 4 are only is less than that of channel channel 3, and 4 are only about 2.4%, which is lessabout than 2.4%, that ofwhich channel 1, channel 3, and channel 1, 5 (3.2%, 4.3%, and channel 5 (3.2%, 4.3%, and 4.2%, respectively). This phenomenon is verified in the crack monitoring 4.2%, respectively). This phenomenon is verified in the crack monitoring experiment (Figure 5): the experiment (Figureof5): values of SC for channel 1, channel and channel 5 are all maximum values SC the for maximum channel 1, channel 3, and channel 5 are all about3,5%, while the maximum about maximum values of4 Sare C for channel 2 and channel 4 are approximately 2%. values5%, of Swhile for the channel 2 and channel approximately 2%. C
Figure15. 15.Distributions Distributionsofofeddy eddycurrents currentswhen whenthe thecrack crackpropagates propagatesfrom from55mm mmto to99mm. mm. Figure
To the perturbation perturbationeffect effectofofcracks cracks eddy currents, current magnitude IS atcrack the To analyze the onon eddy currents, the the current magnitude IS at the crack tip surface is calculated as the crack propagation, as Figure shown16. in When Figurethe 16.crack When the crack tip surface is calculated as the crack propagation, as shown in propagates in propagates in the corresponding area of channel 1, I S first increases and then decreases. The increase the corresponding area of channel 1, IS first increases and then decreases. The increase is due to the ispresence due to the presence thecause crackeddy that currents cause eddy currents to flow around the tip. As the crack of the crack of that to flow around the crack tip. Ascrack the crack propagates, propagates, some of the front-end eddy currents start flowing towards the rear-end eddy currents some of the front-end eddy currents start flowing towards the rear-end eddy currents and form a and form loop, part of eddy currents continue flowing theIncrack tip. In the Imeantime, loop, andapart ofand eddy currents continue flowing around the around crack tip. the meantime, S begins to Idecrease, S begins to decrease, the perturbation of to theeddy crack tip to eddy currents to begin to decrease. causing the causing perturbation of the crack tip currents to begin to decrease. When the crack When the crack propagates in the area corresponding to channel 2 (channel 4), I S first decreases and propagates in the area corresponding to channel 2 (channel 4), IS first decreases and then increases. then This is the mainly because the regions corresponding channels 1, 3 and 5 form a This increases. is mainly because regions corresponding to channels 1, 3 andto5 form a relatively large current relatively large loop. density Thus, the current density of theofcorresponding regions of loop. Thus, the current eddy current of eddy the corresponding regions channel 2 and channel 4 ischannel smaller, 2and andthe channel 4 is smaller, the eddy of current at is the of the two channels is eddy current densityand at the middle the twodensity channels themiddle smallest. the smallest.
Sensors 2018, 18, 1780
10 of 16
Sensors 2017, 17, x FOR PEER REVIEW Sensors 2017, 17, x FOR PEER REVIEW
10 of 16 10 of 16
0.2 0.2
IS / A IS / A
0.15 0.15 0.1 0.1
0.05 0.05 0 0
1
5
9 13 17 21 9 a/mm 13 17 21 a/mm Figure 16. The current magnitude thecrack cracktip tip surface surface IISSvaries the crack growth. Figure 16. The current magnitude atatthe varieswith with the crack growth. 1
5
Figure 16. The current magnitude at the crack tip surface IS varies with the crack growth.
Therefore, adjacent exciting currents in the opposite direction forms current loops when the Therefore, adjacent exciting currents inin the opposite current loops the crack Therefore, adjacent exciting currents the opposite direction forms current loops when when of thethe crack propagates, making eddy currents at the crack tipdirection decrease,forms reducing the perturbation crack propagates, making eddy currents at the crack tip decrease, reducing the perturbation of the propagates, making eddy currents at the crack tip decrease, reducing the perturbation of the crack on crack on eddy currents, resulting in low sensitivity to crack detection. crack on eddy currents, resulting in low sensitivity to crack detection. eddy currents, resulting in low sensitivity to crack detection. 4. Design of the HS-FECA Sensor 4. Design HS-FECASensor Sensor 4. Design of of thethe HS-FECA
As shown in the previous section, the perturbation of the crack tip to the eddy currents plays a
As shown in previous section, perturbation thecrack cracktip tipthe theeddy eddy currents plays major role in in the thethe sensitivity to cracks. To improve the sensitivity of sensor tocurrents cracks, this paper As shown previous section, thethe perturbation ofof the totothe plays aa major major role in the sensitivity to cracks. To improve the sensitivity of the sensor to cracks, this paper proposes a new winding method in which the exciting currents are arranged in the same direction role in the sensitivity to cracks. To improve the sensitivity of the sensor to cracks, this paper proposes proposes a new winding method in which the exciting currents arranged in the same shown in Figure so that channels do notare form anyare eddy current whendirection the shown crack in a new(as winding method in17), which thethe exciting currents arranged in the sameloop direction (as (aspropagates. shown in Figure 17), so that the channels do not form any eddy current loop when the crack To verify the validity of this design method, a finite element model is established. Figure 17), so that the channels do not form any eddy current loop when the crack propagates. To verify propagates. verify the winding validity method, of this design a finite element established. In additionTo to the sensor the restmethod, of the parameters are the model same asisthose shown in the In validity of this design method, a method, finite element model isparameters established. Inthe addition tothose the sensor addition to the sensor winding the rest of the are same as shownwinding in Figure 9. method, Figurethe 9. rest of the parameters are the same as those shown in Figure 9.
Structures under testing Structures under testing D D Crack Crack
Channel 1 Channel 1
Channel 2
Channel 2
Channel 3
Channel 3
V1 V2 V3 V1 Exciting V2 coil V3 Exciting coil
Channel 4
Channel 4
Channel 5
Channel 5
V4 V5 V4 Sensing V5 coil Sensing coil
Figure 17. The schematic of the HS-FECA sensor. Figure 17. 17. The HS-FECA sensor. Figure Theschematic schematicofofthe the HS-FECA sensor. Figure 18 shows the eddy current distribution in the monitoring area during crack propagation. the eddy current distribution in the area during crack propagation. As Figure shown18 in shows the figure, the presence of a crack alters themonitoring flowing trajectory of eddy currents near the Figure 18 the the eddy current in the monitoring area during crack propagation. As shown inshows the figure, presence of distribution acurrents crack alters trajectory of eddy the crack. Because directions of exciting are the theflowing same, eddy currents alsocurrents flow in near the same As shown in the figure, thenear presence of adocrack alters the flowing trajectory of eddy currents near crack. Because directions exciting currents thea same, eddy currents flow the same direction. Eddy currents of the crack notare form loop, but always flow also around theincrack tip. As the direction. crack. Because directions exciting currents the but same, eddy currents also flowin inFigure the Eddy currents nearof the crack not currents formare a loop, always around the crack tip. As same the crack propagates, more and moredo eddy flow around theflow crack tip. As shown the crack propagates, more and more eddy currents flow around the crack tip. As shown in Figure direction. Eddy currents near the crack do not form a loop, but always flow around the crack tip. As the 19, the current magnitude at the crack tip surface IS increases rapidly with crack propagation, which 19,propagates, the current magnitude the crack tipcurrents surface ISflow increases rapidly crack more andatmore eddy around the with crackcrack tip. propagation, As shown inwhich Figure 19,
the current magnitude at the crack tip surface IS increases rapidly with crack propagation, which means the perturbation of the crack becomes stronger and stronger. As shown in Figure 20, SC of each
Sensors2018, 2017,18, 17,1780 x FOR PEER REVIEW Sensors
11 of of 16 16
means the perturbation of the crack becomes stronger and stronger. As shown in Figure 20, SC of channel in the new sensor reached more than 120%. Compared to 5% of the the each channel in the new has sensor has reached more than 120%. Compared to traditional 5% of the sensor, traditional sensitivity of the new sensor is at least 24 times that of the traditional one. sensor, the sensitivity of the new sensor is at least 24 times that of the traditional one.
(a) a = 0 mm
(b) a = 3 mm
(c) a = 7 mm
(d) a = 11 mm
(e) a = 15 mm
(f) a = 21 mm
Figure18. 18.The Theeddy eddycurrent currentdistribution distributionin inthe themonitoring monitoringregion. region. Figure
1212ofof 16 16 12 of of 16 16 12
IISS/I/SA A/ A
Sensors 2017, 17, x FOR PEER REVIEW Sensors 2018, 18, Sensors 2017, 2017, 17, 17, x1780 x FOR FOR PEER PEER REVIEW REVIEW Sensors 1.6 1.6 1.6 1.4 1.4 1.4 1.2 1.2 1.2 1 11 0.8 0.8 0.8 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 1 0.2 11
5 55
9 13 13 99 a/mm 13 a/mm a/mm
17 17 17
21 21 21
Figure 19. The current magnitude at the crack tip surface IS varies with the crack growth. Figure 19. 19. The The current current magnitude magnitude at at the the crack crack tip tip surface surface IISS varies varies with with the the crack crack growth. growth. Figure CH1 CH1 CH1 CH2 CH2 CH2 CH3 CH3 CH3 CH4 CH4 CH4 CH5 CH5 CH5
250% 250% 250%
SSCCSC
200% 200% 200% 150% 150% 150% 100% 100% 100% 50% 50% 50% 0% 0% 0 0% 00
5 55
10 10 10
a/mm a/mm a/mm
15 15 15
20 20 20
25 25 25
Figure Figure 20. 20. Output Outputsignals signals of of the the new new sensor sensor when when the the crack crack propagates. propagates. Figure 20. Output signals of the new sensor when the crack propagates.
Figures 21 and 22 show the influence of changes in conductivity and permeability on output Figures 21 and conductivity and permeability and 22 show the influence of changes in conductivity permeability on output output signals of the new sensor. Compared with effects of cracks on the output signal, effects of changes in signals of of the the new new sensor. sensor. Compared Compared with effects effects of cracks on the output output signal, signal, effects effects of changes changes in in signals conductivity and permeability are essentially negligible. conductivity and permeability are essentially negligible. conductivity CH1 CH1 CH1 CH2 CH2 CH2 CH3 CH3 CH3 CH4 CH4 CH4 CH5 CH5 CH5
SSCCSC
30% 30% 30% 25% 25% 25% 20% 20% 20% 15% 15% 15% 10% 10% 10% 5% 5% 5% 0% 0% 7.3 0% 7.3 7.3
8.3 9.3 10.3 8.3 9.3 10.3 8.3 9.3 10.3 Conductivity（MS/m） Conductivity（MS/m） Conductivity（MS/m）
Figure 21. The SC of the new sensor varies with conductivity of the structure. Figure 21. The SCCC of of the the new new sensor sensor varies varies with with conductivity conductivity of of the the structure. structure.
Sensors 2018, 18, 1780 Sensors 2017, 17, x FOR PEER REVIEW Sensors 2017, 17, x FOR PEER REVIEW
13 of 16 13 of 16 13 of 16
30.00% 30.00% 25.00% 25.00%
SC SC
20.00% 20.00%
CH1 CH1 CH2 CH2 CH3 CH3 CH4 CH4 CH5 CH5
15.00% 15.00% 10.00% 10.00% 5.00% 5.00% 0.00% 0.00% 100 100
120 140 120permeability 140 Relative Relative permeability
Figure 22. 22. The The SSC of the new sensor sensor varies varies with with permeability permeability of of the the structure. structure. Figure thenew new Figure 22. The SCCofofthe sensor varies with permeability of the structure.
5. Experimental Verification of the HS-FECA Sensor ExperimentalVerification Verification of of the HS-FECA Sensor 5.5.Experimental As shown in Figure 23, to verify the monitoring sensitivity of the HS-FECA sensor, As shown 23, to the monitoring sensitivity of the HS-FECA experimental As shown ininFigure Figure 23,verify to verify the monitoring sensitivity of the sensor, HS-FECA sensor, experimental investigations on effects of stress, temperature and crack are performed in this section. investigationsinvestigations on effects of stress, temperature crack are and performed in performed this section.inThe experimental on effects of stress,and temperature crack are thisspecimen section. The specimen is the same welded structure shown in Figure 3. is the same welded structure shown in Figure 3. in Figure 3. The specimen is the same welded structure shown
MTS 810 Material testing system MTS 810 Material testing system
MTS 651 Environmental chamber MTS 651 Environmental chamber
Specimen Specimen Measurement instrument Measurement instrument
Figure 23. Experimental verification of the HS-FECA sensor. Figure Figure23. 23.Experimental Experimentalverification verificationofofthe theHS-FECA HS-FECAsensor. sensor.
Firstly, effects of stress on output signals of the HS-FECA sensor were studied. A Firstly, effects of stress on output signals of the HS-FECA sensor were studied. A constant–amplitude stresssignals Smax = 220 ratio R sensor = 0, loading Firstly, effects spectrum of stress(maximum on output of MPa, the stress HS-FECA werefrequency studied.f constant–amplitude spectrum (maximum stress Smax = 220 MPa, stress ratio R = 0, loading frequency f = 0.02 Hz) was applied to the specimen, while signals were collected. As R shown in Figure 24, the A constant–amplitude spectrum (maximum stress Smax =of220 MPa, stress ratio = 0, loading frequency = 0.02 Hz) was applied to the specimen, while signals of were collected. As shown in Figure 24, the signal ofapplied channelto1the changes withwhile the change stress,collected. and the As varying of the the foutput = 0.02 Hz) was specimen, signals of were shownamplitude in Figure 24, output signal of channel 1 changes with the change of stress, and the varying amplitude of the output signal signalof is channel about 10%. output 1 changes with the change of stress, and the varying amplitude of the output output signal is about 10%. signal is about 10%.
14 of 16 14 of 16 14 of 16
1.02 1.02 1 1 0.98 0.98 0.96 0.96 0.94 0.94 0.92 0.92 0.9 0.9 0.88 0.88 0 0
CH1 CH1
Stress Stress 200 200 150 150 100 100
/ MPa SS / MPa
Normalized NormalizedARAR
Sensors 2018, 18, 1780 Sensors 2017, 17, x FOR PEER REVIEW Sensors 2017, 17, x FOR PEER REVIEW
50 50 200 200 Sample number Sample number
0 0
400 400
Figure 24. Output signal of the HS-FECA sensor varies with stress. stress. Figure 24. 24. Output Output signal signal of of the the HS-FECA HS-FECA sensor sensor varies varies with with Figure stress.
Then, the experiment of crack monitoring under temperature interference is carried out. A Then, the interference is carried out.out. A the experiment experiment ofofcrack crackmonitoring monitoringunder undertemperature temperature interference is carried constant–amplitude spectrum (maximum stress Smax = 220 MPa, stress ratio R = 0, loading frequency f constant–amplitude spectrum = 220 MPa, stress ratio R = 0, loading frequency f A constant–amplitude spectrum(maximum (maximumstress stressSSmax = 220 MPa, stress ratio R = 0, loading frequency max = 15 Hz) was applied to the specimen. As shown in Figure 25, when the loading cycles N = 755, the =f = 1515Hz) Hz)was wasapplied appliedtotothe thespecimen. specimen.As Asshown shownininFigure Figure25, 25,when whenthe the loading loading cycles cycles N == 755, 755, the temperature starts to rise from◦ 20 °C ◦to 80 °C. Then, the temperature decreases to the room temperature startstoto from 20to°C °C.the Then, the temperature decreases the room temperature starts riserise from 20 C 80 to C. 80 Then, temperature decreases to the roomto temperature. temperature. When the temperature reaches 80 °C, the maximum SC of each channel is 8%, 13%, 13%, temperature. When the temperature reaches 80 °C, theSmaximum SC of each 8%, 13%, When the temperature reaches 80 ◦ C, the maximum is channel 8%, 13%,is13%, 12%,13%, and C of each channel 12%, and 10%, respectively. 12%, respectively. and 10%, respectively. 10%, When the crack tip propagates to the rear end of each channel, the signal SC of each channel When the crack tip propagates to the rear end of each each channel, channel, the signal SSC of each channel reaches the corresponding inflection point (C1–C5). Accordingly, loading cycles NCcorresponding to reaches the corresponding inflection point (C1–C5). Accordingly, Accordingly, loading cycles N corresponding to crack lengths of 5 mm, 9 mm, 13 mm, 17 mm, and 21 mm can be obtained as 43,064, 49,735, 55,920, crack lengths of 5 mm, 9 mm, 13 mm, 17 mm, and 21 mm can be obtained as 43,064, 43,064, 49,735, 49,735, 55,920, 55,920, 62,669, and 68,777, respectively. Besides, the maximum SC of each channel is 122%, 172%, 150%, 62,669, and 68,777, respectively. Besides, Besides, the the maximum maximum SCC of of each each channel channel is is 122%, 122%, 172%, 172%, 150%, 150%, 125%, and 95%, respectively. The sensitivity of the HS-FECA sensor is at least 19 times that of the 125%, and 95%, respectively. The The sensitivity sensitivity of of the the HS-FECA HS-FECA sensor sensor is is at at least 19 times that of the original sensor. Compared with the variation of output signals resulting from the crack, the effect of original sensor. Compared with the variation of output signals resulting from the crack, the effect of temperature and stress on output signals of the sensor is so small that whether there is a crack temperature and and stress stresson onoutput outputsignals signalsofof sensor is small so small whether is a below crack thethe sensor is so that that whether there there is a crack below the sensor under the actual service environment can be easily judged. below the sensor under the service actual service environment bejudged. easily judged. the sensor under the actual environment can be can easily
Crack Crack C2 C2 C3 C3
C4 C4 C5 C5
SCSC ΔV/V ΔV/V
CH1 170% CH1 CH2 170% CH2 CH3 150% CH3 150% CH4 CH4 130% CH5 130% CH5 C1 C1 110% 110% 90% 90% 70% 70% 50% T1=20℃, N=775 50% T1=20℃, N=775 T2=80℃, N=4690 30% T2=80℃, N=4690 30% 10% 10% -10% 0 10000 20000 30000 40000 10,000 20,000 30,000 40,000 -10% 0 10000 20000 30000 40000 10,000 20,000 30,000 40,000 Loading cycles N Loading cycles N
50000 50,000 50000 50,000
60000 60,000 60000 60,000
70000 70,000 70000 70,000
Figure 25. S SC of the HS-FECA sensor varies with loading cycles. Figure of the the HS-FECA HS-FECA sensor sensor varies varies with with loading loading cycles. cycles. Figure 25. 25. SCC of
When the sensor is applied to monitor cracks of actual welded structures, the channel numbers When When the the sensor sensor is is applied applied to to monitor monitor cracks cracks of of actual actual welded welded structures, structures, the the channel channel numbers numbers of the sensor and sizes of the sensor can be redesigned according to the critical crack length of the of sensor can redesigned according of the the sensor sensor and and sizes sizes of of the the sensor can be be redesigned according to to the the critical critical crack crack length length of of the the structure under monitoring. Because eddy current testing is a non-contact detection method, the structure under monitoring. Because eddy current testing is a non-contact detection method, the structure under monitoring. Because eddy current testing is a non-contact detection method, the sensor sensor can be directly mounted on critical components of the structure using sealants without any sensor can be directly mounted on critical components of the structure using sealants without any
Sensors 2018, 18, 1780
15 of 16
can be directly mounted on critical components of the structure using sealants without any other surface treatment. What operators need to do is to make sure the sense channels cover the welding seam where cracks are most likely to initiate and grow. 6. Conclusions This paper mainly investigates FECA sensors for crack monitoring of welded structures under stress and temperature interference. Firstly, experimental research on effects of stress, temperature and crack on the output signal of the traditional sensor was carried out. Experimental results show that the sensitivity of the sensor to cracks is so low that stress and temperature variations have great influences on the crack monitoring process. Then, a 3-D FE model was established to analyze the perturbation of the crack to eddy currents. According to results, the reason the sensitivity of the sensor is so low is because adjacent exciting currents in the opposite direction forms current loops when the crack propagates, making eddy currents at the crack tip decrease, reducing the perturbation effect of the crack on eddy currents. Therefore, the HS-FECA sensor with a new winding method in which the exciting currents are arranged in the same direction was proposed. FE simulation reveals that the new winding method makes the perturbation effect of the crack on eddy currents stronger and stronger when the crack propagates, resulting in much higher sensitivity to cracks. The following experiments further prove that the sensitivity of the new sensor is at least 19 times that of the original one, both stress and temperature variations have little effects on output signals of the new sensor. Author Contributions: Conceptualization, T.C. and Y.H.; Methodology, T.C.; Software, T.C.; Validation, T.C. and Y.H.; Formal Analysis, T.C.; Investigation, T.C.; Resources, T.C. and J.D.; Data Curation, T.C.; Writing-Original Draft Preparation, T.C.; Writing-Review & Editing, T.C., Y.H. and J.D.; Visualization, J.D.; Supervision, Y.H.; Project Administration, Y.H.; Funding Acquisition, Y.H. Funding: This research is supported by the National Key Research and Development Program of China (Grant Nos. 2018YFF0214701 and 2018YFF0214702), the Natural Science Fund of China (Grant Nos. 50507186 and 51475470), Postdoctoral Science Found of China (project number: 2016M592967). Conflicts of Interest: The authors declare no conflict of interest.
References 1. 2. 3.
4.
5. 6.
7.
8.
Smith, T.A.; Warwick, R.G. A survey of defects in pressure vessels built to high standards of construction and its relevance to nuclear primary circuits. Int. J. Press. Vessels Pip. 1974, 2, 283–322. [CrossRef] Zhao, Z. Research and Application of Crack Diagnosis, Control and Maintenance Method for Mechanical Bearing Structures; Wuhan University of Technology: Wuhan, China, 2001; pp. 13–23. (In Chinese) Majumder, M.; Gangopadhyay, T.K.; Chakraborty, A.K.; Dasgupta, K.; Bhattacharya, D.K. Fibre Bragg gratings in structural health monitoring—Present status and applications. Sens. Actuators A Phys. 2008, 147, 150–164. [CrossRef] Kinet, D.; Mégret, P.; Goossen, K.W.; Qiu, L.; Heider, D.; Caucheteur, C. Fiber Bragg grating sensors toward structural health monitoring in composite materials: Challenges and solutions. Sensors 2014, 14, 7394–7419. [CrossRef] [PubMed] Rodríguez, G.; Casas J, R.; Villaba, S. Cracking assessment in concrete structures by distributed optical fiber. Smart Mater. Struct. 2015, 24, 035005. [CrossRef] Oromiehie, E.; Prusty, B.G.; Compston, P.; Rajan, G. Characterization of process-induced defects in automated fiber placement manufacturing of composites using fiber Bragg grating sensors. Struct. Health Monit. 2018, 17, 108–117. [CrossRef] Stehmeier, H.; Speckmann, H. Comparative Vacuum Monitoring (CVM) Monitoring of fatigue cracking in aircraft structures. In Proceedings of the 2nd European Workshop on Structural Health Monitoring, Munich, Germany, 7–9 July 2004; pp. 1–8. Racle, E.; Godin, N.; Reynaud, P.; Fantozzi, G. Fatigue Lifetime of Ceramic Matrix Composites at Intermediate Temperature by Acoustic Emission. Materials 2017, 10, 658. [CrossRef] [PubMed]
Sensors 2018, 18, 1780
9. 10. 11.
12. 13. 14. 15.
16. 17.
18.
19.
20.
21.
16 of 16
Farnam, Y.; Geiker, M.R.; Bentz, D.; Weiss, J. Acoustic emission waveform characterization of crack origin and mode in fractured and ASR damaged concrete. Cem. Concr. Compos. 2015, 60, 135–145. [CrossRef] Cho, H.; Lissenden, C.J. Structural health monitoring of fatigue crack growth in plate structures with ultrasonic guided waves. Struct. Health Monit. 2012, 11, 393–404. [CrossRef] Smithard, J.; Rajic, N.; van der Velden, S.; Norman, P.; Rosalie, C.; Galea, S.; Mei, H.; Lin, B.; Giurgiutiu, V. An Advanced Multi-Sensor Acousto-Ultrasonic Structural Health Monitoring System: Development and Aerospace Demonstration. Materials 2017, 10, 832. [CrossRef] [PubMed] Liu, Z.; Chen, K.; Li, Z.; Jiang, X. Crack Monitoring Method for an FRP-Strengthened Steel Structure Based on an Antenna Sensor. Sensors 2017, 17, 2394. [CrossRef] [PubMed] Roy, S.; Lonkar, K.; Janapati, V.; Chang, F.K. A novel physics-based temperature compensation model for structural health monitoring using ultrasonic guided waves. Struct. Health Monit. 2014, 13, 321–342. [CrossRef] Rakow, A.; Chang, F.K. A structural health monitoring fastener for tracking fatigue crack growth in bolted metallic joints. Struct. Health Monit. 2012, 11, 253–267. [CrossRef] Goldfine, N.; Grundy, D.; Craven, C.; Washabaugh, A.; Zilberstein, V.; Sheiretov, Y.; Weiss, V.; Davis, M.; Schaff, J.; Hullander, T.; et al. Damage and usage monitoring for vertical flight vehicles. In Proceedings of the Annual Forum Proceedings—American Helicopter Society, Virginia Beach, VA, USA, 1–3 May 2007; Volume 63. Sheiretov, Y.; Grundy, D.; Zilberstein, V.; Goldfine, N.; Maley, S. MWM-array sensors for in situ monitoring of high-temperature components in power plants. IEEE Sens. J. 2009, 9, 1527–1536. [CrossRef] Goldfine, N.; Denenberg, S.; Manning, B.; Thomas, Z.; Al Rushaid, R.; Haught, F. Modeling and Visualization for Imaging of Subsurface Damage. In Proceedings of the 7th Middle East NDT Conference & Exhibition, Manama, Kingdom of Bahrain, 13–16 September 2015. Goldfine, N.; Manning, B.; Thomas, Z.; Sheiretov, Y.; Denenberg, S.; Dunford, T.; Haque, S.; Al Rushaid, R.; Haught, F. Imaging Corrosion under Insulation and under Fireproofing, using MR-MWM-Arrays. In Proceedings of the 7th Middle East NDT Conference & Exhibition, Manama, Kingdom of Bahrain, 13–16 September 2015. Neil, G.; David, G.; Jennifer, M.; Manning, B.; Martin, C.; Root, C.; Nguyen, A.; Engstrand, C.; Kulowitch, P.; Barrett, A. Flight Testing of Permanently Installed Eddy Current Sensors for IVHM. In Proceedings of the 2014 Aircraft Airworthiness and Sustainment Conference, Baltimore, MD, USA, 14–17 April 2014. He, Y.; Chen, T.; Du, J.; Ding, H.; Jiao, S.; Li, P. Temperature-compensated rosette eddy current array sensor (TC-RECA) using a novel temperature compensation method for quantitative monitoring crack in aluminum alloys. Smart Mater. Struct. 2017, 26, 065019. [CrossRef] MULTIPHYSICS, COMSOL. v. 5.2; COMSOL AB: Stockholm, Sweden, 2015. © 2018 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/).
