sensors Review

A Review of Microwave Thermography Nondestructive Testing and Evaluation Hong Zhang 1, *, Ruizhen Yang 2 , Yunze He 3, *, Ali Foudazi 4 , Liang Cheng 5 and Guiyun Tian 5 1 2 3 4 5

*

School of Electronic and Information Engineering, Fuqing Branch of Fujian Normal University, Fuzhou 350300, China Department of Civil and Architecture Engineering, Changsha University, Changsha 410022, China; [email protected] College of Electrical and Information Engineering, Hunan University, Changsha 410082, China Electrical and Computer Engineering Department, Missouri University of Science and Technology, Rolla, MO 65409, USA; [email protected] School of Electrical and Electronic Engineering, Newcastle University, Newcastle upon Tyne NE1 7RU, UK; [email protected] (L.C.); [email protected] (G.T.) Correspondence: [email protected] (H.Z.); [email protected] (Y.H.); Tel.: +86-157-1591-8377 (H.Z.); +86-134-6769-8133 (Y.H.)

Received: 21 March 2017; Accepted: 10 May 2017; Published: 15 May 2017

Abstract: Microwave thermography (MWT) has many advantages including strong penetrability, selective heating, volumetric heating, significant energy savings, uniform heating, and good thermal efficiency. MWT has received growing interest due to its potential to overcome some of the limitations of microwave nondestructive testing (NDT) and thermal NDT. Moreover, during the last few decades MWT has attracted growing interest in materials assessment. In this paper, a comprehensive review of MWT techniques for materials evaluation is conducted based on a detailed literature survey. First, the basic principles of MWT are described. Different types of MWT, including microwave pulsed thermography, microwave step thermography, microwave pulsed phase thermography, and microwave lock-in thermography are defined and introduced. Then, MWT case studies are discussed. Next, comparisons with other thermography and NDT methods are conducted. Finally, the trends in MWT research are outlined, including new theoretical studies, simulations and modelling, signal processing algorithms, internal properties characterization, automatic separation and inspection systems. This work provides a summary of MWT, which can be utilized for material failures prevention and quality control. Keywords: infrared thermography; NDT; microwave thermography; volumetric heating; material

1. Introduction Infrared (IR) thermography plays an important role in structural health monitoring (SHM) [1] and non-destructive testing (NDT) [2]. IR thermography has great potential and advantages, including fast inspection time, high sensitivity and spatial resolution owing to commercial IR cameras’ ability to detect inner defects as a result of heat conduction. It can be split into two categories: passive and active. For the passive approach, the IR camera is used to measure the temperature of materials under test without any external excitation source. The passive thermography configuration is illustrated in Figure 1a. In many industrial processes, passive thermography has been used in production and predictive maintenance [3]. While passive thermography allows qualitative analyses to be performed, active thermography is both qualitative and quantitative [4]. Contrary to the passive approach, an external thermal excitation is required for active thermography. The known characteristics of this external excitation enable depth quantification

Sensors 2017, 17, 1123; doi:10.3390/s17051123

www.mdpi.com/journal/sensors

Sensors 2017, 17, x FOR PEER REVIEW Sensors 2017, 17, 1123

2 of 33 2 of 33

composites’ debonding detection [5]. As shown in Figure 1b, the configuration of the active infrared approach is similar to thatdetection of the passive except 1b, for the theconfiguration utilization of of anthe excitation source in composites’ debonding [5]. Asapproach, shown in Figure active infrared to generate distinctive thermal contrast.approach, As illustrated infor Figure 1b, the IR camera is situated on the approach is asimilar to that of the passive except the utilization of an excitation source to same side of the excitation source in reflection configuration. For transmission configuration, the IR generate a distinctive thermal contrast. As illustrated in Figure 1b, the IR camera is situated on the camera is situated on the opposite side of the excitation source. The IR camera is synchronizedthe with same side of the excitation source in reflection configuration. For transmission configuration, IR the excitation source by opposite a controlside unit.ofAthe computer is source. required toIR process display the obtained camera is situated on the excitation The cameraand is synchronized with the thermal images. contrast and quantify defects, is obtained often performed excitation source To by aimprove control unit. A computer is required to active processthermography and display the thermal with advanced signalcontrast processing the reflection mode isis often suitable for detecting images. To improve andmethods. quantify Normally, defects, active thermography performed with defects situated near the surface, while deeper defects can be detected in the transmission mode. advanced signal processing methods. Normally, the reflection mode is suitable for detecting defects However, thethe transmission approach becan used some cases the target is inaccessible situated near surface, while deepercannot defects be in detected in thewhere transmission mode. However, [6–8]. the transmission approach cannot be used in some cases where the target is inaccessible [6–8].

Figure Figure 1. 1. MWT setup for (a) the passive approach and (b) the active approach.

Depending on the external thermal excitation, different active thermography methods have been Depending on the external thermal excitation, different active thermography methods have been developed, such as pulsed thermography (PT) [9], step thermography (ST) [10] and modulated developed, such as pulsed thermography (PT) [9], step thermography (ST) [10] and modulated thermography (MT) or so-called lock-in thermography (LT) [11]. Finally, there is pulsed phase thermography (MT) or so-called lock-in thermography (LT) [11]. Finally, there is pulsed phase thermography (PPT) [12], developed by Maldague and Marinetti in 1996, which combines the thermography (PPT) [12], developed by Maldague and Marinetti in 1996, which combines the advantages of PT and MT [13,14]. advantages of PT and MT [13,14]. Various physical heating sources have been adopted as thermal stimulation sources, such as Various physical heating sources have been adopted as thermal stimulation sources, such as thermal thermal lamps, lasers, ultrasound devices, and electromagnetic waves. Accordingly, laser lamps, lasers, ultrasound devices, and electromagnetic waves. Accordingly, laser thermography [15], thermography [15], ultrasonic thermography and eddy current thermography [16,17] were ultrasonic thermography and eddy current thermography [16,17] were developed. Taking eddy current developed. Taking eddy current thermography as an example, it combines the advantages of IR thermography as an example, it combines the advantages of IR thermography and eddy current testing, thermography and eddy current testing, such as being fast and non-contact [18–21]. Eddy current such as being fast and non-contact [18–21]. Eddy current thermography can heat many materials thermography can heat many materials such as metals and carbon fiber reinforced polymer (CFRP) such as metals and carbon fiber reinforced polymer (CFRP) with eddy current heating. However, with eddy current heating. However, it only works for conductive materials [22–24], therefore, the it only works for conductive materials [22–24], therefore, the excitation source needs to be chosen excitation source needs to be chosen according to the specific problem. In the last decade, researchers according to the specific problem. In the last decade, researchers have shown an increased interest have shown an increased interest in microwave heating techniques. Microwave heating has been in microwave heating techniques. Microwave heating has been exhibited advantages of rapid heat exhibited advantages of rapid heat transfer (due to volumetric heating), efficiency, heating transfer (due to volumetric heating), efficiency, heating uniformity, compact equipment, and being uniformity, compact equipment, and being easy to control, etc. Meanwhile, microwave heating has easy to control, etc. Meanwhile, microwave heating has emerged as a powerful platform due to emerged as a powerful platform due to dielectric loss and eddy current heating with different dielectric loss and eddy current heating with different materials under test. So far, built microwave materials under test. So far, built microwave thermography devices have shown some unique

Sensors 2017, 17, 1123

3 of 33

thermography devices have shown some unique advantages such as: (1) microwaves will produce reflection, scattering, transmission at a discontinuous interface. With microwave signal reflection and scattering in the defect area, less microwave energy can be used for heating, and the temperature raise in a defect area is slower than in a non-defect area during microwave heating. IR cameras will capture this abnormal thermal image which will strengthen the effectiveness of defect detection; (2) the heating pattern of microwave heating is relatively uniform, volumetric and selective, and can be achieved in a short time; (3) microwave heating is easy to control, and it is easy to implement different heating function modulations. However, it is necessary to restrict microwave leakage as they are dangerous to human health, therefore, the leakage of the microwaves needs to be kept below a certain recommended level. Generally, in industry microwave heating is operated from 890 MHz to 2.45 GHz to minimize any possible interference with communication services [25]. Thermography has been associated with microwaves in numerous applications. MWT has been employed by some scientists to detect wet rotten wood [26], mines and surrogate signatures [27,28]. MWT has also been used to inspect and characterize various kinds of materials and phenomena, such as debonding and delamination in composite materials [29]. So far, although there are several review works [30–44], they have been limited to a specific field such as composite or renewable energy, so a review of MWT in the material detection field which includes the principles, advantages and disadvantages, developments and research trends is still needed. In this paper, a comprehensive review of MWT techniques for material evaluation has been provided, based on a detailed literature survey. The overall structure of this paper includes the following: the principle of MWT is presented in Section 2. Then, typical types of MWT applications are summarized in Section 3. Section 4 reviews the development of MWT with case studies. Then, a comparison and discussion are provided in Section 5. Trends are shown in Section 6. Finally, the conclusions are outlined in Section 7. 2. Principle of Microwave Thermography The principles of MWT mainly include microwave heating and 3D heat conduction. These are analyzed theoretically in the following subsections. 2.1. Microwave Based Heating The heating style of microwave thermography can be divided into volume heating and surface heating [45], therefore the heating process can be divided into volumetric heating (i.e., dielectric loss heating) and surface heating (i.e., eddy current heating). 2.1.1. Dielectric Loss Heating For dielectric materials, such as glass fiber composite materials, microwave heating is volumetric heating (i.e., dielectric loss heating). Considering glass fiber composites for instance, material dielectric loss in the microwave radiation field will generate heat. The dissipated power P per unit volume can be expressed as follows [46]: P = 2π f ε 0 ε00 E2 (1) where f is the frequency of an electric field, E is the RMS value of the electric field, ε0 is the permittivity of air, and ε” is the relative loss factor. Without considering the heat diffusion, the temperature change per unit at heating time t with a continuous microwave source is [46]: T (t) =

Pt ωε 0 ε00 E2 = t ρCp ρCp

(2)

where ρ is the density of the material and Cp is heat capacity. Obviously, with constant microwave parameters and constant properties of the material under test, the temperature increases linearly with time (during a short period of time).

Sensors 2017, 17, x FOR PEER REVIEW

4 of 33

The basic Sensorsprinciples 2017, 17, 1123 of MWT volumetric heating are shown in Figure 2a: firstly, 4 of a 33 microwave excitation module is used to generate a microwave radiation field; secondly, microwaves penetrate The basic MWT volumetric heatingwill are shown firstly, a microwave the material under test,principles and theofmedium molecules moveinatFigure the 2a: frequency of the electric field excitation module is used to generate a microwave radiation field; secondly, microwaves penetrate around the which generates heat and eventually is converted into Joule heat, which then will transfer the material under test, and the medium molecules will move at the frequency of the electric field material based on the diffusion equation; finally, an IR camera is utilized to obtain the temperature which generates heat and eventually is converted into Joule heat, which then will transfer around the variation inmaterial the material test. equation; Due to the differences inisdensity heat capacity between the based onunder the diffusion finally, an IR camera utilized toand obtain the temperature variation in the material underinformation test. Due to theabout differences in density heat capacity between materials under test and defects, surface andand internal defects canthebe obtained. materials under test and defects, information about surface and internal defects can be obtained. Thus, Thus, the processes of MWT for dielectric materials evaluation are based on microwave radiation, the processes of MWT for dielectric materials evaluation are based on microwave radiation, dielectric dielectric loss Joule heat transfer, and IR radiation. lossto to generate generate Joule heat,heat, heat transfer, and IR radiation.

Figure 2. Basic schematic of MWT: dielectric material (a) and conductive material (b). Figure 2. Basic schematic of MWT: dielectric material (a) and conductive material (b).

2.1.2. Eddy Current Heating

2.1.2. Eddy Current Heating

For conductive materials, like metals and carbon fiber composites, microwave heating is eddy current heating. Since the like material underand test iscarbon electrically conductive, microwaves cannot penetrate For conductive materials, metals fiber composites, microwave heating is eddy the conductor material. Thus, the principle of microwave heating is that the energy is radiated to the current heating. Since the material under test is electrically conductive, microwaves cannot penetrate conductive material surface by microwaves, an alternating electric field is generated, then, induced the conductor material. the from principle of microwave heating is that energy is radiated to the surface currentsThus, are excited the alternating electric field, resulting in an the alternating magnetic Next, surface a vortex electric field is generated this alternating magnetic field, vortex electric conductive field. material by microwaves, an by alternating electric field is the generated, then, induced field promotes the movement of electrons which will generate Joule heat. Finally, the conduct material surface currents are excited from the alternating electric field, resulting in an alternating magnetic is heated by Joule heat as shown in Figure 2b. Power P and heat Q generated by eddy current heating field. Next, can a vortex electric field is generated by this alternating magnetic field, the vortex electric be expressed as [47]: r field promotes the movement of electrons which will µ f generate Joule heat. Finally, the conduct 2 P ∼ Iinductor (3) σ material is heated by Joule heat as shown in Figure 2b. Power P and heat Q generated by eddy current r µf 2 heating can be expressed as [47]: Q = Pt ∼ Iinductor t (4) σ where, Iinductor is the current flowing through the inductor, σ is f the conductivity (S/m), µ is the magnetic 2 Pf is Ithe permeability of the material under test and frequency of the induced current. It is observed that inductor  with a stable induced current and induced current frequency, the generated heat is directly proportional to the microwave excitation time and it is inversely proportional to the square root of the conductivity. However, due to the presence of heat transfer and dissipation problems during actual applications, f 2 MWT must be corrected in order to minimize Q  Pt theImeasurement terror. inductor



(3)

(4)

where, Iinductor is the current flowing through the inductor, σ is the conductivity (S/m), μ is the magnetic permeability of the material under test and f is the frequency of the induced current. It is observed that with a stable induced current and induced current frequency, the generated heat is directly proportional to the microwave excitation time and it is inversely proportional to the square root of the conductivity. However, due to the presence of heat transfer and dissipation problems during

Sensors 2017, 17, 1123

5 of 33

Due to the skin effect, induced current depth (within the skin depth) in the conductive material is an extremely important factor. The skin depth can be obtained by the following equation [48]: δ= p

1 πµσ f

(5)

where f is the microwave’s frequency, σ is the conductivity (S/m), and µ is the magnetic permeability (H/m) of the material under test. A typical conductivity of CFRP is probably 1000 S/m, and the permeability is around 1. With 2.4 GHz microwave excitation, the skin depth is about 0.002 mm. Therefore, MWT belongs to the surface heating category as only the surface of the CFRP is heated. 2.2. 3D Heat Transfer and Temperature Field Heat Q generated by dielectric loss or the Joule heat will be conducted from inside to the surrounding material. The heat conduction equation is a time-dependent heat diffusion equation [49]: ∂T k = ∂t ρC p |



 ∂2 T ∂2 T ∂2 T 1 + + + Q( x, y, z, t) 2 2 2 ρC p ∂x ∂y ∂z {z } | {z }

Thermal diffusion

(6)

Microwave heating

where, T = T ( x, y, z, t) is the temperature distribution of the surface, k is the material thermal conductivity (W/m × K), C p is specific heat capacity (J/kg × K), ρ is the density (kg/m3 ), and Q( x, y, z, t) is the heat generation function with microwave heating (the dielectric loss heating or eddy current heating). A surface temperature distribution will eventually reflect disturbances of the electromagnetic and thermal fields. Therefore, MWT has the potential to characterize and track the property variations of the material, such as magnetic permeability, electrical conductivity, permittivity, thermal conductivity, thermal diffusivity, etc. In addition, the depth of defects can be quantified. The heat generated by Joule heat will propagate a certain distance within the material in the form of heat waves. The penetration depth δth of these heat waves is [50]:

√ δth ≈ 2 αt

(7)

α = (k/ρCp)

(8)

where, α is the thermal diffusivity, and t is the observation time. α can be expressed as a function of the density of the material ρ, heat capacity Cp and thermal conductivity k, as shown in Equation (8). It shows that the penetration depth of heat δth is proportional to the square root of t and α [11]. In the case of a modulated thermal wave, the length of thermal diffusion decides the penetration depth, which can be found from the following equation [11]: s µt =

2k = ωρCp

r

α πf

(9)

where, ρ is density, k is thermal conductivity, α is thermal diffusivity, Cp is heat capacity, and f is the frequency of the thermal wave. The penetration depth is proportional to the reciprocal of the square root of f and α. In other words, the detection depth varies according to the modulation frequency. In summary, the detection ability of microwave thermography is closely related to the electrical, dielectric and thermal properties of the material under test.

Sensors 2017, 17, 1123

6 of 33

3. Types of Microwave Thermography 3.1. Classification of Excitation Configuration According to the excitation configurations of microwave heating, MWT can be divided into microwave pulsed thermography (MPT),microwave pulsed phase thermography (MPPT), microwave lock-in thermography (MLT) [51], or microwave step thermography (MST), also known as microwave time-resolved thermography [52]. With MPT, the material under test is heated by a small period of microwave excitation as shown in Figure 3a. The variation of temperature is observed in the heating phase and the cooling phase. For MST, the sample is step heated by a long pulse as shown in Figure 3b, and the variation of temperature is observed in the heating phase. As shown in Figure 3c, the material under test is heated by a periodic amplitude modulated microwave with MLT and the periodic temperature change is captured. A square pulsed modulated excitation is used to derive phase Sensors 2017, 17, x FOR PEER REVIEW 6 of 33 information from multiple thermal waves with one inspection. The influence of non-uniform heating is been reduced and deeper displayed a displayed higher contrast. A pulse excitation uniform heating is beendefects reducedcan andbe deeper defectswith can be with a higher contrast. A pulsesignal is usedexcitation by MPPT. Phase analysis is carried outanalysis in the frequency domain Predictably, microwave signal is used by MPPT. Phase is carried out in the[13]. frequency domain [13]. Predictably, microwave frequency modulated thermography will employ a frequency modulated frequency modulated thermography will employ a frequency modulated microwave excitation in order microwave excitation in order to the derive phase information. From the above the that: conclusion to derive phase information. From above analysis, the conclusion can analysis, be reached MPT and can be reached that: MPT and MST analyze the temperature of thermal imaging in the time domain, MST analyze the temperature of thermal imaging in the time domain, which is affected by surface which is affected by surface emissivity variations and non-uniform heating; MLT obtains information emissivity variations and non-uniform heating; MLT obtains information in the frequency domain such in the frequency domain such as phase, which can suppress the influence of the surface emissivity as phase, which can suppress the influence of the surface emissivity variations and non-uniform heating. variations and non-uniform heating. However, the MLT inspection system requires a long However, the MLT time inspection requires a long measurement time and it is relatively complex. measurement and it issystem relatively complex.

Figure 3. Excitation functions of MWT: (a) MPT; (b) MST; (c) MLT.

Figure 3. Excitation functions of MWT: (a) MPT; (b) MST; (c) MLT. Comparisons among MPT, MST, MLT, and MPPT are listed in Table 1. Due to the use of an IR

Comparisons among MPT, MST, MLT, and are listed in Table 1. Due to thevisibility. use of an IR camera, all of them exhibit high sensitivity, highMPPT resolution, full-field detection and good camera, all of them exhibit high sensitivity, high resolution, full-field and good visibility. In addition, quantification information can be achieved based on the heatdetection conduction: In addition, quantification information can be achieved based on the heat conduction: 1. MPT can be fast and easily deployed. Surface temperature gradients will be introduced not only 1.

2.

3.

4.

defects, but easily also local variations in surface emissivitygradients and non-uniform heating. A long MPT from can be fast and deployed. Surface temperature will be introduced not only inspection time is required for a thick material. In addition, the material could be from defects, but also local variations in surface emissivity and non-uniformdamaged heating.due A long to the high heating energy. inspection time is required for a thick material. In addition, the material could be damaged due 2. MST is a time-resolved method and it can be used to quantify defect depth. However, the to the high heating energy. radiation from the heat source in the continuous heating process could deteriorate the MST temperature is a time-resolved method and it can used to quantify depth.emissivity However,variation the radiation measurements. Also, the be non-uniform heatingdefect and surface from have the adverse heat source continuous effectsin on the defect evaluation. heating process could deteriorate the temperature 3. MLT generally required less excitation energy MPT. MLTemissivity exhibits a higher sensitivity measurements. Also, the non-uniform heatingthan and surface variation havethan adverse MPT. The phase data can be extracted which is independent of surface emissivity and heating effects on defect evaluation. variations. Tests are repeated with various frequencies and it becomes a time-consuming process MLT generally required less excitation energy than MPT. MLT exhibits a higher sensitivity than MPT. to detect defects with various depths. However, MLT offer a compromise with a better depth The phase data can be extracted which is independent of surface emissivity and heating variations. resolution. Tests are repeated with various frequencies and it becomes a time-consuming process to detect 4. MPPT combines the advantages of MPT and MLT. MPPT is less sensitive to non-uniform heating defects with various depths. However, MLTphase offerinformation a compromise with a betterMoreover, depth resolution. and surface emissivity than MLT as only can be obtained. MPPT MPPTemployed combines the advantages of MPT and MLT. than MPPT is less sensitive to non-uniform a short excitation pulse which is faster MLT and wide frequency spectra can heating be obtained. However, with frequency, transferredcan energy is decreasedMoreover, with MPPT. and surface emissivity thanincreasing MLT as only phasethe information be obtained. MPPT With post processing algorithm, MPPT exhibits better detect ability and resolution than MPT for employed a short excitation pulse which is faster than MLT and wide frequency spectra can be deeper defects. with increasing frequency, the transferred energy is decreased with MPPT. obtained. However, With post processing algorithm, MPPT exhibits better detect ability and resolution than MPT for Table 1. Comparisons among MPT, MST, MLT and MPPT. deeper defects. Techniques

Strength

Microwave pulsed thermography [13]

Full-field, high resolution, high sensitivity, good visibility, quantification, fast, easy deployment

Microwave step thermography [52]

Full-field, high resolution, high sensitivity, good visibility, quantification, fast, easy deployment,

Limitation Small depth, long time for thick material, emissivity and non-uniform heating dependence, high power The radiation from heat source in the continuous heating process could have a negative influence on temperature

Sensors 2017, 17, 1123

7 of 33

Table 1. Comparisons among MPT, MST, MLT and MPPT. Techniques

Strength

Limitation

Microwave pulsed thermography [13]

Full-field, high resolution, high sensitivity, good visibility, quantification, fast, easy deployment

Small depth, long time for thick material, emissivity and non-uniform heating dependence, high power

Microwave step thermography [52]

Full-field, high resolution, high sensitivity, good visibility, quantification, fast, easy deployment, time-resolved

The radiation from heat source in the continuous heating process could have a negative influence on temperature measurements of MUT, emissivity, and non-uniform heating dependence

Microwave lock-in thermography [51]

Full-field, high resolution, higher sensitivity, good visibility, quantification, low power, emissivity independence, elimination of non-uniform heating

Compromise between depth and depth resolution, time-consuming,

Microwave pulsed phase thermography

Full-field, high resolution, high sensitivity, good visibility, quantification, fast, emissivity independence, elimination of non-uniform heating, greater depth and resolution, better detectability

Post signal processing, energy attenuation with frequency

Sensors 2017, 17, x FOR PEER REVIEW

Microwave pulsed Sensors 2017, 17, 1123 phase thermography

Full-field, high resolution, high sensitivity, good visibility, quantification, fast, emissivity independence, elimination of non-uniform heating, greater depth and resolution, better detectability

7 of 33

Post signal processing, energy attenuation 8 of 33 with frequency

3.2. Classification by Heating Style 3.2. Classification by Heating Style

MWT can be classified into surface heating thermography, volume heating thermography, and MWT can be classified into surface thermography, volume heating thermography, abnormal heating thermography [14,45].heating In Figure 4a, for eddy current heating, MWT can and be also abnormal heating thermography [14,45]. In Figure 4a, for eddy current heating, MWT can be also called surface heating thermography (SHT). Due to the great permeability, the skin depth is very called surface heating thermography (SHT). Dueoftothe theSHT greatfamily. permeability, the skin depth is the veryheat small [53–55]. Thus, it can be classified as part With reflection mode, small [53–55]. Thus, it can be classified as part of the SHT family. With reflection mode, the heat conduction from surface to inside is used to quantify the depth of defect. conduction from surface to inside is used to quantify the depth of defect. MWT exhibits volumetric heating for dielectric material inspection. MWT can be called volume MWT exhibits volumetric heating for dielectric material inspection. MWT can be called volume heating thermography (VHT), as illustrated in Figure 4b [45,55]. For the transmission and reflection heating thermography (VHT), as illustrated in Figure 4b [45,55]. For the transmission and reflection modes with VHT, thethe characterizations similar[54]. [54].Only Only interesting areas heated modes with VHT, characterizationsof ofdefects defects are are similar interesting areas are are heated without heating thethe host material Abnormalheat heatwill willcaused caused defects. Furthermore, without heating host materialininsome somecases. cases. Abnormal byby defects. Furthermore, this this abnormal heat is used to quantify depth information of the related defect. In Figure 4c, these abnormal heat is used to quantify depth information of the related defect. In Figure 4c, these methods are called abnormal heating thermography (AHT). For detecting water water in concrete structures, methods are called abnormal heating thermography (AHT). For detecting in concrete structures, MWT can be as a kind of AHT. MWT can be considered asconsidered a kind of AHT.

Figure4.4.Styles Stylesof ofheating: heating: (a) AHT. Figure (a)SHT; SHT;(b) (b)VHT; VHT;(c)(c) AHT.

4. Developments and Case Studies 4. Developments and Case Studies

4.1. 4.1. A History of MWT Development A History of MWT Development Developed countries have taken useof ofMWT MWTtotocarry carry related researches Developed countries have takenthe thelead lead on on the use outout related researches and and achieved some interesting results [56–58].Levesque Levesque and Ambrosio study on MWT in in achieved some interesting results [56–58]. Ambrosiodid dida apreliminary preliminary study on MWT the 1990s. Levesque et al. employedXXand andKu Ku bands bands (range toto 1818 GHz) horns andand parabolic the 1990s. Levesque et al. employed (rangefrom from88GHz GHz GHz) horns parabolic antennas to excite sampleunder undertest test[59]. [59]. The The excitation is provided by aby100 W W antennas to excite thethe sample excitationofofthe theantenna antenna is provided a 100 amplifier. Thermal images are produced by an AGEMA (model 782 LWB) IR camera with 8 μm–12 μm amplifier. Thermal images are produced by an AGEMA (model 782 LWB) IR camera with 8 µm–12 µm range. An area of 300 × 300 mm22 is been measured due to the limited excitation area. Several 10 mm range. An area of 300 × 300 mm is been measured due to the limited excitation area. Several 10 mm thick glass-epoxy composites have been tested to identify the inserted carbon particles area. The thick glass-epoxy composites have been tested to identify the inserted carbon particles area. The proposed proposed method were used to characterize artificial defects’ size and location under different depths method were used to characterize artificial defects’ size and location under different depths in composite in composite materials. Ambrosio et al. used a cavity with a minimum 600 W to detect artificial materials. et al. used a cavity with minimum 600×W to×detect artificial in non-metallic 3 whichdefects defectsAmbrosio in non-metallic composites [60]. Thea cavity was 300 200 280 mm is excited by a 2.45 3 which is excited by a 2.45 GHz magnetron and composites [60]. The cavity was 300 × 200 × 280 mm GHz magnetron and an Alter VPG1540 power supply unit (maximum power 1200 W). They used a an Alter VPG1540 powermeter supplytounit (maximum power and 1200reflected W). Theymicrowave used a Cober PM45 Two power meter to Cober PM45 power monitor the incident powers. cavity monitor the incident andcavity reflected microwave Two cavity applicators open cavity applicator applicators (an open applicator and a powers. large cavity applicator) have been(an proposed to avoid the edge effects and tohave improve field uniformity the sample. Toeffects estimate thetodefect and aperture a large cavity applicator) beenthe proposed to avoid over the aperture edge and improve permittivity andover depth surface temperature perturbation, theoretical modeling the field uniformity theinfluence sample. on To estimate the defect permittivity and depth influence onand surface numerical simulations were carried out. temperature perturbation, theoretical modeling and numerical simulations were carried out. In 1999, Takahide OsakaUniversity University applied surface breaking cracks in in In 1999, Takahide et et al.al.atatOsaka appliedMWT MWTtotodetect detect surface breaking cracks reinforced concrete structures and they pointed out that the tip of crack will produce more heat reinforced concrete structures and they pointed out that the tip of crack will produce more heat during during the test [61]. Heat was introduced by a time-gated microwave source into the homogeneously the test [61]. Heat was introduced by a time-gated microwave source into the homogeneously reinforced reinforced concrete structure. With the microwave penetration features, the subsurface of the concrete structure. With the microwave penetration features, the subsurface of the reinforced concrete reinforced concrete structure could be inspected. Moreover, wet cracks could be selectively heated. structure could be inspected. wet cracks could bewith selectively heated. Therefore, Therefore, subsurface cracksMoreover, will be immediately identified MWT. Analyses of time andsubsurface spatial cracks will be immediately identified with MWT. Analyses of time and spatial dependence features of measured thermal images are possible to quantify cracks’ depth and thermal diffusivity information [61]. In the 21st century, groups in the USA, UK, France, Poland, Italy, Korea, etc. have developed MWT for metal, dielectric material, cement/concrete, GFRP, CFRP, honeycomb and cement-based composites detection problems. These works have been simply summarized in Table 2 and are introduced in detail in the following Sections 4.2–4.9.

Sensors 2017, 17, 1123

9 of 33

Table 2. Summary of researches with MWT. Hardware Development

Software Development

Experimental Study

Operation Frequency

Antenna/Sensor

Power

IR Camera

Simulation Study

Sampling Software

Signal Processing

Material under Test

Defects

8 GHz to 18 GHz [59]

Horns and parabolic antennas [59]

100 W [59]

AGEMA 782 LWB [59]

Not available

CEDIP PTR-9010 system [59]

Normalized the infrared images pixel by pixel [59]

Glass-epoxy composites [59]

Delaminations

HP 8757 C network analyzer [60]

Open cavity applicator and large cavity applicator have been designed [60]

Kevlar or fiberglass slabs and sandwich samples [60]

Defects

Mortar block [61]

Cracks

Reinforced steel bars [62], Rebar in air [63]; Embedded in cement [63,65], CFRP [64]

Corrosion, delaminations, debonding and crack

Magnetron [60]

600 W–1700 W [60]

AGEMA AGA 880 [60]

Influence of defect permittivity and depth has been estimated [60].

Microwave oven [61]

1400 W [61]

Nikon LAIRD-3 [61]

Not available

Not available

Thermal image was taken at 20 s [61]

2 GHz to 3 GHz [62], 2.45 GHz [63,64]

TEM horn antenna [62,64]

50 W [62,64], 10 W [63]

DRS Tamarisk 320 [62,64], FLIR SC 500 [63]

CST Microwave Studio and MPHYSICS Studio [62].

Not available

The surface thermal profile was taken after 10 s of heating [62]. The surface thermal profile was taken after 5 s and 15 s of heating [63].

2.45 GHz [65]

Magnetron generated horn antenna [65]

600 W [65]

Not available

Not available

ALTAIR software

A contrast algorithm is used to analyze the thermogram series with 5 min of heating [65].

5 GHz to 10 GHz [66]

Horn antenna [66]

Mikron 6T61 and SantaBarbara IR camera [66]

Not available

Macintosh microcompute [66]

A 2.7 s microwave pulse was used [66]

Multilayered plexiglass-water-teflon specimen [66]

Debonding

Santa Barbara Focalplane [67]

Not available

LabVIEW

The surface thermal profile was taken after 8 s and 10 s of heating [67]

Carbon fibers in different epoxy structures [67]

Embedded fibers

Rohde & Schwarz SMF 100 A signal generator [57]

The surface thermal profile was taken after 10 min of heating [57]

2.45 GHz [60,61]

2.3 W [66,67] 9 GHz [67]

A single flare horn antenna [67]

18 GHz [57]

Flann 18 094-SF40 waveguide [57]

WR430 waveguide [57]

1 W [57]

FLIR SC7500 [57] COMSOL Multiphysics and CST Microwave Studio [57]

1000 W [57] Flir A325 [57,68]

2.45 GHz [56,68]

1 GHz [69]

The surface thermal profile was obtained after 10 s to 15 s of heating [57] Not available

Magnetron [68]

500 W [68]

Pyramidal horn antenna [56]

360 W [56]

Not available

Not available

Coaxial-type probes [69]

0.1 W [69]

FLIR T620 [69]

CST Microwave Studio [69]

Steel bar corrosion

Defects GFRP [57] Debonding and delamination

A sequence of 180 thermograms was obtained and processed by using normalized, standardized contrast and cosine transform [68]

Composite materials with adhesive bounded joints [68]

Defects

ALTAIR program [56]

The surface thermal profile was taken after 150 s of heating [56]

CFRP [56]

Defects

Spectrometer [69]

Normalized to the maximum value of field intensity [69]

Carbon-fiber composite materials [69]

Conductivity measurement

Not available

Sensors 2017, 17, 1123 x FOR PEER REVIEW

10 9 of 33

4.2. Metals and Corrosion etal. al.from from Missouri University of Science and Technology proposed the of Foudazi et thethe Missouri University of Science and Technology proposed the use ofuse MWT MWT the detection and characterization of corroded reinforced In Figure 5a, for thefor detection and characterization of corroded reinforced steel steel bars bars [62]. [62]. In Figure 5a, the 2 horn antenna to 2 horn the measurement setup for steel bars is illustrated. They employed a 14 24 cm measurement setup for steel bars is illustrated. They employed a 14 × 24 cm× antenna to irradiate irradiate with a 50 W microwave signal fordistance 10 s. The distance the steel and steel barssteel with bars a 50 W microwave signal for 10 s. The between thebetween steel bars and thebars antenna the antenna was 1the cm.experiment, During thefour experiment, piecesmaterial of corroded mounted on was 1 cm. During pieces offour corroded were material mountedwere on steel bars and steel bars and spaced 1 cmTamarisk apart. A320 DRS Tamarisk 320 was thermal camera was utilized to obtain spaced 1 cm apart. A DRS thermal camera utilized to obtain the thermal profilethe of thermal of the steel bars with aIn0.05 K sensitivity. In Figure 5b,profiles the surface thermal the steel profile bars with a 0.05 K sensitivity. Figure 5b, the surface thermal for steel bars profiles after 10 bars after s as detected GHz,are and 3 GHz energy are bars provided. steel sfor assteel detected with 10 2 GHz, 2.5 GHz,with and 23GHz, GHz 2.5 energy provided. The steel were aThe smooth bars were a smooth rebar without ribs.rebar The radius thiswhich rebar contained was 4.8 cma which containedarea. a 1 cm rebar without ribs. The radius of this was 4.8ofcm 1 cm corrosion In corrosion In Figure theare visible hot spots areas are theoncorroded areas on Because these steel Because Figure 5b,area. the visible hot5b, spots the corroded these steel bars. of bars. the relatively of the relatively low thermal conductivity corroded steel, heat quickly dissipatessteel. in uncorroded low thermal conductivity of corroded steel,ofheat quickly dissipates in uncorroded Moreover, steel. Moreover, these temperature differences betweenareas the corroded areas different these temperature differences between the corroded indicated thatindicated differentthat amounts of amounts ofabsorb corrosion absorbamounts different amounts of microwave a preliminary simulation corrosion different of microwave energy. energy. With aWith preliminary simulation and and experimental study ofMWT, the MWT, it demonstrated a higher excitation microwave frequency experimental study of the it demonstrated that that a higher excitation microwave frequency will will to a higher temperature for corrosion detection in bars. steel Moreover, bars. Moreover, increased corrosion lead lead to a higher temperature for corrosion detection in steel increased corrosion leads leads to absorption of microwave energy increasing results a greater difference in the measured to absorption of microwave energy increasing results in a in greater difference in the measured IR IR images. images.

Figure 5. (a) (a) MWT MWT measurement measurement setup setup for for corroded corroded rebar detection; detection; (b) (b) temperature temperature profile for corroded steel bar [62]. Reprinted/reproduced Reprinted/reproducedwith with permission permission from from IEEE. IEEE. ..

Pieper et al. demonstrated an active MWT for inspection of large corrosion areas on reinforcing Pieper et al. demonstrated an active MWT for inspection of large corrosion areas on reinforcing steel bars for cement-based structures [63]. They used CST Microwave studio and MultiPhysics studio steel bars for cement-based structures [63]. They used CST Microwave studio and MultiPhysics to construct a coupled microwave and thermal simulation model. The effects of steel bars have been studio to construct a coupled microwave and thermal simulation model. The effects of steel investigated in the air and in concrete. To detect the change in temperature with a thermal camera, the bars have been investigated in the air and in concrete. To detect the change in temperature with incident microwave power should be increased because some of the microwave energy is absorbed by a thermal camera, the incident microwave power should be increased because some of the microwave the concrete. The effect of different polarization has been studied with CST. With 10 W incident power, energy is absorbed by the concrete. The effect of different polarization has been studied with the parallel polarization generates the most uniform heating in that of orthogonal polarization and CST. With 10 W incident power, the parallel polarization generates the most uniform heating in circular polarization, but the temperature increase was the lowest (only 0.014 °C). The highest that of orthogonal polarization and circular polarization, but the temperature increase was the temperature increase was generated by orthogonal polarization (around 0.025 °C), therefore, lowest (only 0.014 ◦ C). The highest temperature increase was generated by orthogonal polarization orthogonal polarization is the best choice for thin corrosion detection. (around 0.025 ◦ C), therefore, orthogonal polarization is the best choice for thin corrosion detection. During the experimental study, two steel (AISI 1008) bars (each of length 150 mm and radius 4.8 During the experimental study, two steel (AISI 1008) bars (each of length 150 mm and radius 4.8 mm) mm) were measured, which have been embedded in parallel in a concrete block (170 × 150 × 50 mm33). were measured, which have been embedded in parallel in a concrete block (170 × 150 × 50 mm ). The sample was heated for 5 s by a microwave oven operating at 2.45 GHz. A FLIR Thermacam SC The sample was heated for 5 s by a microwave oven operating at 2.45 GHz. A FLIR Thermacam 500 was used to provide thermal images with a sensitivity of 0.1 °C. In Figure 6a, the two steel bars SC 500 was used to provide thermal images with a sensitivity of 0.1 ◦ C. In Figure 6a, the two have been heated by microwave energy for 15 s. One (top) with localized significant corrosion (on steel bars have been heated by microwave energy for 15 s. One (top) with localized significant the order of 1 mm–4 mm) on a portion of its length, the one below with light corrosion (on the order corrosion (on the order of 1 mm–4 mm) on a portion of its length, the one below with light corrosion 0.2 mm or less) along half of its length. As illustrated in Figure 6b and 6c, the corroded areas in the (on the order 0.2 mm or less) along half of its length. As illustrated in Figure 6b,c, the corroded areas two steel bars have been identified in the thermal images as indicated by the black circle. Moreover, in the two steel bars have been identified in the thermal images as indicated by the black circle. in Figure 6c, the uncorroded area in the steel bar is also visible, appearing as a relatively lower Moreover, in Figure 6c, the uncorroded area in the steel bar is also visible, appearing as a relatively temperature line (in the left of the black circle). In Figure 6e, the corroded steel bar is still detectable when embedded in a concrete block as highlighted by the black circle. These results show that steel Sensors 2017, 17, x; doi: FOR PEER REVIEW

www.mdpi.com/journal/sensors

Sensors 2017, 17, 1123

11 of 33

lower temperature line (in the left of the black circle). In Figure 6e, the corroded steel bar is still Sensors 2017, 17, x FOR PEER REVIEW 10 of 33 detectable when embedded in a concrete block as highlighted by the black circle. These results show that steel corrosion in concrete will to a higher loss tangent and the thermal properties are changed. corrosion in concrete will lead to alead higher loss tangent and the thermal properties are changed. With With microwave heating, the physical property changes due to corrosion become measurable microwave heating, the physical property changes due to corrosion become measurable in in the the thermograms. onon thethe orientation and thickness of corrosion, the polarization of the incident thermograms.Depending Depending orientation and thickness of corrosion, the polarization of the microwave signal can signal be optimized simulation. effect ofThe corrosion thicknesslayer on incident microwave can be through optimized through The simulation. effect layer of corrosion microwave heating has been investigated. thickness on microwave heating has been investigated.

Figure Figure6.6.(a) (a)Photograph Photographof oflocalized localizedcorrosion corrosion(top) (top)and andlight lightcorrosion corrosion(bottom); (bottom);(b) (b)thermal thermalimages images ofofaasteel steelbar barwith withlight lightcorrosion corrosionand and(c) (c)localized localizedcorrosion; corrosion;(d) (d)photograph photographofofclean clean(left) (left)and and corroded (right) steel bars; (e) thermal images of clean (left) and corroded (right) steel bars embedded corroded (right) steel bars; (e) thermal images of clean (left) and corroded (right) steel bars embedded ininthe theconcrete concreteblock blockafter aftermicrowave microwaveheating heating[63]. [63].Reprinted/Reproduced Reprinted/Reproducedwith withpermission permissionfrom fromAIP AIP Publishing PublishingLLC. LLC.

Keoetetal.al. presented a pyramidal antenna based toMWT detect corroded steel in a Keo presented a pyramidal horn horn antenna based MWT detecttocorroded steel in a reinforced reinforced concrete wall [65]. the experimental study, a commercial operating concrete wall [65]. During the During experimental study, a commercial magnetron magnetron operating at 2.45 GHzat 2.45 GHz was used to irradiate a maximum 800 W of microwave energy. The steel reinforcements was used to irradiate a maximum 800 W of microwave energy. The steel reinforcements (each one (eacha one with aofdiameter of 12 mm, a regular of 10acm) withconcrete a 38 mmcovering concrete with diameter 12 mm, located at alocated regularatspacing of spacing 10 cm) with 38 mm 3 covering were embedded in ablock concrete block (1000 mmspecimen ). The specimen was by heated by were embedded in a concrete (1000 × 1000 × ×651000 mm×3 ).65The was heated 600 W 600 W microwave energy for 5 min. The reflection mode was used as the thermal camera was placed microwave energy for 5 min. The reflection mode was used as the thermal camera was placed on on same the same as the excitation source. A data analysis method basedonona acontrast contrastalgorithm algorithmwas was the sideside as the excitation source. A data analysis method based used to analyze the thermogram series to reduce the effect of non-uniform excitation. As shown used to analyze the thermogram series to reduce the effect of non-uniform excitation. As shown inin Figure7a, 7a,the theinfrared infraredcamera camerawas wascomposed composedofofa a320 320××256 256matrix matrixdetector detectorofofindium indiumantimonide antimonide Figure ◦ (InSb)with with aa sensitivity sensitivity range antenna was placed in ain45° direction to heat (InSb) range of of 3–5 3–5μm. µm.The Theexcitation excitation antenna was placed a 45 direction to the largest surface area of the specimen. In addition, the infrared camera was placed at 2.32 m away heat the largest surface area of the specimen. In addition, the infrared camera was placed at 2.32 m ◦ direction from from the specimen in ain30° direction which can detect area of of the the away the specimen a 30 which can detectthe thelargest largestpossible possible surface surface area specimen.Thermograms Thermogramswere wererecorded recorded 1 image s by using ALTAIR software (developed specimen. atat 1 image perper s by using thethe ALTAIR software (developed by by FLIR for obtaining measurement data). Figure 7b shows the obtained thermograms at the 250 FLIR for obtaining measurement data). Figure 7b shows the obtained thermograms at the 250 s instant.s instant. The abnormal temperature areas correspond to the corrosion areasteel in the steel reinforcements. The abnormal temperature areas correspond to the corrosion area in the reinforcements. In the In the preliminary experimental study of the MWT, it was demonstrated that the detection depth of preliminary experimental study of the MWT, it was demonstrated that the detection depth of MWT MWT was about 3.8 cm in a thick concrete block. was about 3.8 cm in a thick concrete block.

Sensors 2017, 17, 1123 Sensors 2017, 17, x FOR PEER REVIEW

Sensors 2017, 17, x FOR PEER REVIEW

12 of 33 11 of 33

11 of 33

Figure 7. Photograph of microwave thermographymeasurement measurement setup setup (a) (a) and Figure 7. Photograph of microwave thermography and temperature temperatureprofile profile for for corroded steel bar (b) [65]. Reprinted/reproduced with permission from Elsevier. corroded steel bar (b) [65]. Reprinted/reproduced with permission from Elsevier.

7. Photograph 4.3. Figure Dielectric Materials of microwave thermography measurement setup (a) and temperature profile

4.3. Dielectric Materials for corroded steel bar (b) [65]. Reprinted/reproduced with permission from Elsevier.

Osiander et al. employed a time-resolved MWT system for surface and sub-surface inspection

Osiander et al. employed a time-resolved MWTatsystem for surface and sub-surface inspection on Plexiglass-water-Teflon specimens with absorbers different depths (0.15, 0.30 and 0.45 mm) [66]. 4.3. Dielectric Materials on Plexiglass-water-Teflon specimens with absorbers different depths (0.15,was 0.30 andto0.45 mm) [66]. Figure 8a shows a diagram of the experimental setup.atAn HP 6890B oscillator used generate al.with employed time-resolved for6890B surface and 1277 sub-surface inspection microwave a of frequency range fromMWT 5 GHzsystem to 10HP GHz. A Hughes X-band traveling Figure 8aOsiander showspulses aetdiagram thea experimental setup. An oscillator was used to generate on Plexiglass-water-Teflon specimens with absorbers at different depths (0.15, 0.30 and 0.45 mm) [66]. wave tube amplifier with 2.3 W maximum power was utilized as a power amplifier. An excitation microwave pulses with a frequency range from 5 GHz to 10 GHz. A Hughes 1277 X-band traveling Figure 8a shows a diagram of the experimental setup. An HP 6890B oscillator was used to generate horn antenna was placed in a 45° direction with respect to the sample surface. Two infrared cameras wave tube amplifier with 2.3 W maximum power was utilized as a power amplifier. An excitation pulses with the a frequency range from 5ofGHz to 10 GHz. A Hughes X-band traveling were usedwas to monitor temperature the sample test. One1277 was a Mikron 6T61 hornmicrowave antenna placed in a surface 45◦ direction with respect to the under sample surface. Two infrared cameras infrared withwith an HgCdTe detector. The second was aas SantaBarbara infraredAn camera with wave tubescanner amplifier 2.3 W maximum power wasone utilized a power amplifier. excitation were used to monitor the surface temperature of the sample under test. One was a Mikron 6T61 a 128antenna × 128 pixels InSb focal array. The resolution of the first camera wascameras 0.1 K. horn was placed in a plane 45° direction withtemperature respect to the sample surface. Two infrared infrared scanner with an HgCdTe detector. resolution The second one 0.003 was aK.SantaBarbara infrared camera For the second IR camera, the temperature is about The specimen under test is were used to monitor the surface temperature of the sample under test. One was a Mikron 6T61 withinfrared a 128 ×scanner 128 pixels InSb focal plane array. The temperature resolution of the first camera illustrated in Figure 8b. It is structured with a Teflon layer of varied thickness and a water layer with with an HgCdTe detector. The second one was a SantaBarbara infrared camera withwas 0.1 K. For the second IRbacked camera, the temperature resolution isthe about K. camera The specimen under constant Plexiglass layers. Considering power, thefirst microwave horn0.1 was aa128 × 128 thickness pixels InSb focal with plane array. The temperature resolution of0.003 the was K. test For isworking illustrated in Figure 8b. It is structured with a Teflon layer of varied thickness and a in the IR near field. The dataresolution was normalized to 0.003 the peak amplitude to eliminate thewater the second camera, the measured temperature is about K. The specimen under test is effects non-uniform distribution. Figure 8c shows IR images ofthe the test sample before, layerillustrated with aofconstant backed with layers. Considering the microwave in Figurethickness 8b. microwave It is structured withPlexiglass a Teflon layer of varied thickness andpower, a water layer with and afterinathe 2.7 near s microwave pulse. Three Considering water corresponding to the Teflon layer hornaduring was working field. The measured data layers was normalized tomicrowave the peak amplitude constant thickness backed with Plexiglass layers. the power, the horn was to thicknesses appear in the IRmeasured images indata a temporal order. With the potential of surface working the near The wasdistribution. normalized toFigure theMWT, peak eliminate the test eliminate theineffects offield. non-uniform microwave 8camplitude shows IRto images of the temperature versus time has been shown forFigure subsurface defects quantitative characterization. effects of non-uniform microwave distribution. 8c shows IR images of the test sample before, sample before, during and after a 2.7 s microwave pulse. Three water layers corresponding to the Compared with beam heating, dry epoxy coated steel samples exhibit a very small response during after laser a 2.7appear s microwave Three layers corresponding to the Teflon layer of Teflon layerand thicknesses in the pulse. IR images inwater a temporal order. With MWT, the potential with microwave heating. When a debonding region contains water, the whole structure of the thicknesses appear in thetime IR has images inshown a temporal order. With MWT, the potential of surface surface temperature versus been for subsurface defects quantitative characterization. debonding region can be illustrated. The wavelength independent resolution has been demonstrated temperature versus time has been shown for subsurface defects quantitative characterization. Compared with laser beam heating, dry epoxy coated steel samples exhibit a very small response to be 30 μm. Anlaser analytical hasdry been provided extract the time dependence of the surface with Compared with beam model heating, epoxy coatedtosteel samples exhibit a very small response microwave heating. When a debonding region contains water, the whole structure of the debonding temperature where quantitative such as theregion depth of the defect can be with microwave heating. Whendata a debonding contains water, theextracted. whole structure of the region can be illustrated. The wavelength independent resolution has been demonstrated to be 30 µm. debonding region can be illustrated. The wavelength independent resolution has been demonstrated An analytical model has beenmodel provided to extract thetotime dependence of the surface temperature to be 30 μm. An analytical has been provided extract the time dependence of the surface where quantitative data such as the depth of the defect can be extracted. temperature where quantitative data such as the depth of the defect can be extracted.

Figure 8. Schematic diagram of the experimental setup (a), the specimen under test (b) and IR images (c) [66]. Reprinted/reproduced with permission from SPIE.

Figure 8. Schematic diagram theexperimental experimental setup setup (a), under testtest (b)(b) andand IR images Figure 8. Schematic diagram ofofthe (a),the thespecimen specimen under IR images (c) [66]. Reprinted/reproduced with permission from SPIE. (c) [66]. Reprinted/reproduced with permission from SPIE.

Sensors 2017, 17, 1123 Sensors 2017, 17, x FOR PEER REVIEW

13 of 33 12 of 33

4.4. 4.4. Cement Cement and and Concrete Concrete Takahide Takahideet etal. al. proposed proposed the the use use of of MWT MWT for for detecting detecting surface-breaking surface-breaking cracks in concrete [61]. Surface-breaking cracks can be penetrated with water. The crack Surface-breaking cracks can be penetrated with water. The crack opening opening distance distance was was 0.2 0.2 mm mm and and 0.4 0.4 mm. mm. The size of the the cracks cracks was was 10 10 mm mm in in depth depth and and 40 40mm mmininwidth. width. Since Since the the microwave microwave absorptivity absorptivity of of the the concrete concrete is is much much smaller smaller than than that that of of water, water, cracks cracks containing containing water water can can be be selectively heated. By applying microwaves to the concrete structure, the thermal conduction of selectively By applying microwaves to the concrete structure, the thermal conduction of these these heated cracks generate localized high-temperature regions. experimental MWT setup heated cracks will will generate localized high-temperature regions. TheThe experimental MWT setup is is illustratedininFigure Figure9a. 9a.AAcommercially commerciallyavailable available2.45 2.45 GHz GHz magnetron magnetron was was used to illustrated to irradiate irradiate microwaves mortar block blockwith withaamaximum maximum1400 1400WW output power. specimen under microwaves into the mortar output power. TheThe specimen under test test placed in the microwave oven. NikonLAIRD-3 LAIRD-3with withPtSi PtSiarray arraywas was used used to measure waswas placed in the microwave oven. AA Nikon measure the the temperature distribution of the mortar block. In Figure 9b, four artificial cracks have been introduced temperature distribution of the mortar block. In Figure 9b, four artificial cracks have been introduced to hardening mortar blockblock specimen. By injecting water into crackinto C and applying microwaves toa arapid rapid hardening mortar specimen. By injecting water crack C and applying for 5 s, crack Cfor with selectively heated. Compared withCompared the non-crack area, temperature microwaves 5 s,water crackwas C with water was selectively heated. with thethe non-crack area, ◦ C to 8 ◦ C after microwave heating. As shown in Figure 9c, the position of of the crack rose about 6 the temperature of the crack rose about 6 °C to 8 °C after microwave heating. As shown in Figure 9c, the C can be crack easilyCdetected from detected the thermal taken immediately after the microwave the crack position of the can be easily fromimage the thermal image taken immediately after the excitation. authors found the temperature of the crackregion degraded aftercrack 20 s microwaveMeanwhile, excitation.the Meanwhile, the that authors found that region the temperature of the due to the thermal diffusion from the crack into the surrounding area. The relation between the crack degraded after 20 s due to the thermal diffusion from the crack into the surrounding area. The relation size and temperature rise should be determined for quantitative investigations. between the crack size and temperature rise should be determined for quantitative investigations.

Figure9. 9. Schematic Schematic diagram diagram of of MWT MWT (a), (a), mortar mortar block block specimen specimen with with artificial artificialcracks cracks(b) (b)and andthermal thermal Figure imagestaken takenafter aftermicrowave microwaveheating heating(c) (c)[61]. [61].Reprinted/reproduced Reprinted/reproduced with with permission permission from from SPIE. SPIE. images

4.5. Glass Glass Fiber Fiber Composite Composite Structures Structures 4.5. Bowenet etal. al. introduced introduced microwave-source microwave-source time-resolved time-resolvedinfrared infraredradiometry radiometryfor fordetecting detectingand and Bowen characterizing microwave absorption in dielectric materials [67]. An HP 6890B oscillator was used to characterizing microwave absorption in dielectric materials [67]. An HP 6890B oscillator was used excite a 9a GHz microwave signal. AA Hughes 1277 X-band traveling wave tube amplifier was used to to excite 9 GHz microwave signal. Hughes 1277 X-band traveling wave tube amplifier was used amplify this signal to a maximum 2.3 W power. A single flare horn antenna with about 50° beam ◦ to amplify this signal to a maximum 2.3 W power. A single flare horn antenna with about 50 beam width was was placed placed15 15cm cmfrom fromthe thesample sampleunder undertest. test.AASanta SantaBarbara Barbaracamera camerawith with128 128×× 128 128 InSb InSb width focal plane plane array array was was used used to to detect detect the the IR IR radiation. radiation. The The temperature temperature resolution resolution is is about about 0.003 0.003 K. K. focal With embedded fibers at different locations, a fiberglass-epoxy specimen was measured. The depth With embedded fibers at different locations, a fiberglass-epoxy specimen was measured. The depth of of the fibers was 0.25 mm and 0.75 mm. Due theloss losstangent tangentand andJoule Jouleheating, heating,there thereisisaavery veryhigh high the fibers was 0.25 mm and 0.75 mm. Due toto the contrast for defects in embedded epoxy materials. The size and orientation of these embedded fibers contrast for defects in embedded epoxy materials. The size and orientation of these embedded fibers in in the measured thermal images studied. The interaction is strongly dependent on the the measured thermal images havehave been been studied. The interaction is strongly dependent on the length length of the fiberto(930mm to and 30 mm) and orientations. The temperature showsdistribution a modal distribution of the fiber (9 mm mm) orientations. The temperature shows a modal along the along the fiber for fibers longer than 12 mm. fiber for fibers longer than 12 mm. Cheng et al. [57] developed a microwave pulsed thermography system for glass fiber composite measurement. Figure 10a illustrates the setup of the experimental system: (1) an adaptor connected with the waveguide; (2) GFRP wind turbine blade; (3) microwave generator linked with cable; (4) IR

Sensors 2017, 17, 1123

14 of 33

Cheng et al. [57] developed a microwave pulsed thermography system for glass fiber composite measurement. Figure 10a illustrates the setup of the experimental system: (1) an adaptor connected with the waveguide; (2) GFRP wind turbine blade; (3) microwave generator linked with cable; Sensors 2017, 17, x FOR PEER REVIEW 13 of 33 (4) IR camera. GFRP wind turbine blade with 4 holes (4.5 mm in radius) at the root was heated bycamera. an open-ended waveguide connected Flann Theheated adaptor linked GFRP wind turbine blade with 4 with holes a(4.5 mm18 in 094-SF40 radius) at adaptor. the root was by was an openwaveguide a Flann 18 094-SF40 adaptor. The adaptoroutput was linked to of a signal to ended a signal generatorconnected (Rohde &with Schwarz SMF 100 A) with the maximum power 30 dBm generator (Rohde Schwarz SMF A) with thethe maximum power aof45 30degree dBm (1illumination W) at 18 (1 W) at 18 GHz. The&waveguide was100 placed above sample output with roughly GHz. The was above sample withthe roughly a 45 degree illumination direction. direction. A waveguide FLIR SC7500 IRplaced camera was the used to obtain temperature distribution. Artificial holes A FLIR SC7500 IR camera was used to obtain the temperature distribution. Artificial holes with 4.5 with 4.5 mm radius were investigated. During the experiment, the signal generator provided 1W mm radius were investigated. the experiment, signal generator 1 W microwave microwave power, although onlyDuring 115.2 mW microwave the power was emitted provided from the waveguide due to although onlyto115.2 mW microwave was emitted waveguide due to loss the cable thepower, cable loss. In order maximize the actualpower illumination powerfrom andthe minimize the power during loss. In order to maximize the actual illumination power and minimize the power loss during the the transmission, 22.7 mm was selected as the standoff distance (distance between the microwave transmission, 22.7 mm was selected as the standoff distance (distance between the microwave antenna aperture and sample under test) due to the impedance matching (this distance should be antenna aperture and sample under test) due to the impedance matching (this distance should be around λ/4 to maximum power transfer, where λ is the wavelength of microwave). In their primary around λ/4 to maximum power transfer, where λ is the wavelength of microwave). In their primary experimental results, a discontinuous temperature was captured in the defect region (mainly at the experimental results, a discontinuous temperature was captured in the defect region (mainly at the edges). A high power heating system is required for a better contrast of heat patterns between defect edges). A high power heating system is required for a better contrast of heat patterns between defect and non-defect regions. and non-defect regions.

Figure 10. MWT setup in Newcastle University (a) and West Pomeranian University of Technology Figure 10. MWT setup in Newcastle University (a) and West Pomeranian University of Technology (b) [57]. Reprinted/reproduced with permission from IEEE. (b) [57]. Reprinted/reproduced with permission from IEEE.

A magnetron-based microwave pulsed thermography system has been developed by West A magnetron-based pulsed thermography systemdebonding has been and developed by West Pomeranian University microwave of Technology for glass fiber composite delamination Pomeranian University of Technology for(BLADES200W glass fiber composite debonding and detection [57]. A GFRP wind turbine blade from Navitron) with a 4.5 mmdelamination hole at the middle [57]. was the study wind object. Figureblade 10b illustrates system setup for high power pulsed detection A GFRP turbine (BLADES200W from Navitron) with microwave a 4.5 mm hole at the thermography. With aobject. maximum 1 kW the microwave excitation waspower a 2.45 microwave GHz magnetron. middle was the study Figure 10bpower, illustrates system setup for high pulsed An open-ended aperture was matched a WR-430 waveguide. 1000 thermography. With a maximum 1 kW impedance power, thewith microwave excitation wasDipolar a 2.45 Magdrive GHz magnetron. was used to control the output power. A Flir A325 device was used to obtain thermograms. A 0.238 An open-ended aperture was matched impedance with a WR-430 waveguide. Dipolar Magdrive 1000 °Cused maximum temperature raise can beAobtained 2 s heating. In to their primary experimentAresults, was to control the output power. Flir A325from device was used obtain thermograms. 0.238 ◦ C the location and the size information of the defect have been detected from the continuities in the line-scan results. Ryszard et al. also employed MWT for inspecting adhesive joints in composite materials [68]. Due to the dielectric property differences between defects (lack of adhesive or delamination) and the

Sensors 2017, 17, 1123

15 of 33

maximum temperature raise can be obtained from 2 s heating. In their primary experiment results, the location and the size information of the defect have been detected from the continuities in the line-scan results. Ryszard et al. also employed MWT for inspecting adhesive joints in composite materials [68]. Sensors 2017, 17, x FOR PEER REVIEW 14 of 33 Due to the dielectric property differences between defects (lack of adhesive or delamination) and the background, energy in the defect region and and the background in thein material under background,the theinduced inducedthermal thermal energy in the defect region the background the material test is different. The experimental setup is illustrated in Figure 11a: (1) an IR camera in a protective under test is different. The experimental setup is illustrated in Figure 11a: (1) an IR camera in a box to prevent damage caused by the highby power microwaves; (2) a 500 W (2) magnetron at protective box to prevent damage caused the high power microwaves; a 500 W working magnetron 2.45 GHz; (3) microwave absorbers; (4) an open-ended rectangular waveguide; (5) the sample under working at 2.45 GHz; (3) microwave absorbers; (4) an open-ended rectangular waveguide; (5) the test; (6) under a cooling for thesystem magnetron. MWT observation time was settime to 180 to obtain sample test;system (6) a cooling for theThe magnetron. The MWT observation wass set to 180 180 thermograms in one sequence. The measured results were obtained by using dedicated image s to obtain 180 thermograms in one sequence. The measured results were obtained by using dedicated processing algorithms. As shown Figure 11b–e, the original thermogram (b), cosine transform (c), image processing algorithms. Asinshown in Figure 11b–e, the original thermogram (b), cosine standardized contrast (d) and contrast enhancement (e) were used to obtain images of the defect. In the transform (c), standardized contrast (d) and contrast enhancement (e) were used to obtain images of primary experiment results,experiment approximate location and the sizelocation information about be the defect. In the primary results, approximate and the sizedelamination information can about obtained in the thermograms after processing. delamination can be obtained in the thermograms after processing.

Figure 11.11. Experimental setup for MWTfor (a) and measurement (b–e) [68]. Reprinted/reproduced Experimental setup MWT (a) andresults measurement results (b–e) [68]. with permission from authors. Reprinted/reproduced with permission from authors.

We also also investigated investigated MWT MWT for for defect defect detection detection in in aa glass glass fiber fiber wind wind turbine turbine blade, blade, as as shown shown in in We Figure 12. 12. A A horn horn antenna antenna (ETS (ETS Lindgren Lindgren 3115, 3115, frequency frequency range: range: 750 750 MHz–18 MHz–18 GHz.) GHz.) was was used used for for Figure microwave excitation. The antenna was connected to Rohde & Schwarz SMF 100 A signal generator microwave excitation. The antenna was connected to Rohde & Schwarz SMF 100 A signal generator with aa maximum maximum output output power power of of 11 W. W. The The waveguide waveguide was was placed placed above above the the sample sample with with roughly roughly with 90 degree degree illumination illumination direction. direction. A used to to obtain obtain the the temperature temperature 90 A FLIR FLIR SC7500 SC7500 IR IR camera camera was was used distribution of sample under test. During the experiment, the signal generator provided 1 W at 0.915 distribution of sample under test. During the experiment, the signal generator provided 1W at GHz and order detect during the experiment, the microwave heating duration 0.915 GHz2.45 andGHz. 2.45 In GHz. Intoorder todefects detect defects during the experiment, the microwave heating was selected 1 min as (as1 shown Figurein13a,c) 2 min shown in Figure 13b,d).13b,d). In our In initial duration was as selected min (asinshown Figureand 13a,c) and(as 2 min (as shown in Figure our experiment results, a discontinuous temperature was captured in the defect region. Furthermore, the initial experiment results, a discontinuous temperature was captured in the defect region. Furthermore, heating effect of 2.4 is not good as 0.915 GHz GHz due to thetopenetration abilityability difference. With the heating effect ofGHz 2.4 GHz is as not as good as 0.915 due the penetration difference. the captured IR images, defects around 1 mm radius with 0.2 mm in depth could be obtained in With the captured IR images, defects around 1 mm radius with 0.2 mm in depth could be obtainedthe in GFRP sample. With a longer heating duration, thethe number of defects became hard to identify due to the GFRP sample. With a longer heating duration, number of defects became hard to identify due thethe heat diffusion, therefore, thethe heating time must bebe optimized forfor different situations. to heat diffusion, therefore, heating time must optimized different situations.

Sensors 2017, 17, 1123

16 of 33

Sensors 2017, 17, x FOR PEER REVIEW Sensors 2017, 17, x FOR PEER REVIEW

15 of 33 15 of 33

Figure 12. Experimental setup for defect detection with MWT. Figure Figure 12. 12. Experimental Experimental setup setup for for defect defect detection detection with with MWT. MWT.

Figure 13. IR results with MWT for defect detection in glass fiber wind blade after 1 min microwave Figure 13. 13. IR IR results results with with MWT MWT for for defect defect detection detection in in glass glass fiber fiber wind wind blade blade after after 11 min min microwave microwave Figure excitation with 0.915 GHz (a), after 2 min microwave excitation with 0.915 GHz (b), after 1 min excitation with with 0.915 0.915 GHz GHz (a), (a), after after 22 min min microwave microwave excitation excitation with with 0.915 0.915 GHz GHz (b), (b), after after 11 min min excitation microwave excitation with 2.45 GHz (c) and after 2 min microwave excitation with 2.45 GHz (d). microwave excitation excitation with 2.45 GHz (c) and after 2 min microwave excitation with 2.45 GHz (d). microwave

4.6. Carbon Fiber Composite Materials 4.6. 4.6. Carbon Carbon Fiber Fiber Composite Composite Materials Materials In France, Keo et al. developed a microwave pulsed thermography for CFRP detection [56]. They In developed a microwave pulsed thermography for CFRP detection [56]. They In France, France,Keo Keoetetal.al. developed a microwave pulsed thermography for CFRP detection [56]. used a commercial 2.45 GHz magnetron to generate microwave signals. With a detector with a 320 × used commercial 2.45 GHz to generate microwave signals.signals. With a detector with a 320 × Theyaused a commercial 2.45magnetron GHz magnetron to generate microwave With a detector with 256 InSb matrix, the sensitivity of the IR camera is in a range of 3 μm–5 μm. To capture the whole 256 InSb matrix, sensitivity of the IRofcamera in a range 3 μm–5 Toµm. capture the whole a 320 × 256 InSbthe matrix, the sensitivity the IR is camera is in aofrange of 3μm. µm–5 To capture the inspection area, the sample was placed 1.5 m away from IR camera at 55°. The CFRP sample was 40 ◦ . Thesample inspection area, the sample was placed 1.5 m away camera at 55°. The was 40 whole inspection area, the sample was placed 1.5 mfrom awayIRfrom IR camera at 55CFRP CFRP sample 2 defect was created at the middle of the CFRP sample by the absence of × 40 × 4.5 cm33. A 10 × 10 cm 2 3 2 ×was 40 ×404.5 . A4.5 10cm × 10. cm was defect createdwas at the middle of the CFRPofsample by the absence of × cm 40 × A 10defect × 10 cm created at the middle the CFRP sample by the adhesive. The antenna was placed in the 45° direction as shown in Figure 14a. In Figure 14b, the adhesive. antenna placed inplaced the 45° as shown in Figure 14a. In14a. Figure 14b, 14b, the absence ofThe adhesive. Thewas antenna was in2direction the 45◦ direction as shown in Figure In Figure microwave beam was guided by a 59 × 56 cm2 pyramidal horn antenna onto the sample under test. microwave beam was was guided by aby 59a×59 56× cm56 pyramidal horn horn antenna onto onto the sample underunder test. the microwave beam guided cm2 pyramidal antenna the sample With the ALTAIR program provided by FLIR, a computer was used to record thermograms at 1 Hz. With the ALTAIR program provided by FLIR, a computer was used record thermograms at 1 Hz. test. With the ALTAIR program provided by FLIR, a computer was to used to record thermograms at A 360 W microwave was used to heat the specimen for 150 s. The same testing procedure was carried A 360 W microwave was used to heat the specimen for 150 s. The same testing procedure was carried 1 Hz. A 360 W microwave was used to heat the specimen for 150 s. The same testing procedure was out on samples with/without defect. As shown in Figure 15, the thermograms at the 100 s were out on samples with/without defect. As shown Figure the 15, thermograms at the at 100the s were carried out on samples with/without defect. As in shown in 15, Figure the thermograms 100 s obtained. In Figure 15a, a non-defect specimen has been shown in the thermogram at the 100 s instant. obtained. In Figure 15a, a non-defect specimen has beenhas shown the thermogram at the 100at s instant. were obtained. In Figure 15a, a non-defect specimen beeninshown in the thermogram the 100 In Figure 15b, a specimen with a defect has been shown in the thermogram. In Figure 15c, the In Figure In 15b, a specimen with a defect been in the in thermogram. In Figure 15c, 15c, the s instant. Figure 15b, a specimen with ahas defect hasshown been shown the thermogram. In Figure thermogram of the sample with a defect (Figure 15a) has been subtracted with the thermogram of the thermogram of the withwith a defect (Figure 15a)15a) has been subtracted withwith the thermogram of the the thermogram of sample the sample a defect (Figure has been subtracted the thermogram of sample without defect (Figure 15b) at the same instant. In Figure 15d, the thermogram in Figure 15c sample without defect (Figure 15b) at the same instant. InIn Figure the sample without defect (Figure 15b) at the same instant. Figure15d, 15d,the thethermogram thermogramin inFigure Figure 15c 15c is subtracted from its initial thermogram in order to highlight the defect. In the primary experiment is is subtracted subtracted from from its initial thermogram in order to highlight the defect. In In the the primary primary experiment experiment results, the CFRP defect area can be clearly found. It is hotter than the non-defect area due to the results, the CFRP defect area can be clearly found. It is hotter than the non-defect area due to the

Sensors 2017, 17, 1123

17 of 33

Sensors 2017, 17, x FOR PEER REVIEW Sensors 2017, 17, x FOR PEER REVIEW

16 of 33 16 of 33

results,ofthe CFRP defect area canarea be clearly found. It is directly hotter than thethe non-defect area due to the10absence absence adhesive. The defect cancan bebe estimated from thermograms (about × 10 absence of adhesive. The defect area estimated directly from the thermograms (about 10 × 10 of adhesive. The defect area can be estimated directly from the thermograms (about 10 × 10 cm2 ) [56]. 2) [56]. cmcm 2) [56].

(a)(a)

(b)(b)

Figure 14.14. MWT setup (a)(a) and Horn (b)(b) in in University Institute of of Technology of of Bethune [56]. Figure MWT setup (a) and Horn (b) University Institute of Technology Bethune [56]. Figure 14. MWT setup and Horn University Institute Reprinted/reproduced with permission from Scientific Research Publishing Inc. Reprinted/reproduced with permission permission from from Scientific Scientific Research Reprinted/reproduced with Research Publishing Publishing Inc. Inc.

Figure 15.15. MWT results forfor thethe absence of adhesive in CFRP: thethe thermogram of the sample without Figure MWT results absence of adhesive adhesive in CFRP: CFRP: thermogram of the the sample without Figure 15. MWT results for the absence of in the thermogram of sample without defect (a), the thermogram of the sample with the defect (b), the subtraction between the thermograms defect (a), the thermogram of the sample with the defect (b), the subtraction between the thermograms defect (a), the thermogram of the sample with the defect (b), the subtraction between the thermograms of the sample with and without thethe defect (c),(c), thermogram subtracted from its its initial thermogram (d)(d) of the the sample subtracted from initial thermogram of sample with with and andwithout without thedefect defect (c),thermogram thermogram subtracted from its initial thermogram [56]. Reprinted/reproduced with permission from Scientific Research Publishing Inc.Inc. [56]. Reprinted/reproduced with permission from Scientific Research Publishing (d) [56]. Reprinted/reproduced with permission from Scientific Research Publishing Inc.

In In 2014, Foudazi et et al.al. investigated MWT forfor thethe inspection of of rehabilitated cement-based 2014, Foudazi investigated MWT inspection rehabilitated cement-based In 2014, Foudazi et al. investigated MWTinfor the inspection of rehabilitated cement-based structures [64]. The experimental setup is shown Figure 16a,b, 2.4 GHz microwaves have been structures [64]. The experimental setup is shown in Figure 16a,b, 2.4 GHz microwaves have been structures [64]. The experimental setup is shown in Figure 16a,b, 2.4 GHz microwaves have been transmitted with 50 W power. A sweep oscillator (HP8690B) and a power amplifier (OphirRF transmitted with 50 W power. A sweep oscillator (HP8690B) and a power amplifier (OphirRF transmitted withto50generate W power. A sweep oscillator (HP8690B) and sample a powerunder amplifier (OphirRF 5303084) 5303084) areare used high power microwave signals. The test was illuminated 5303084) used to generate high power microwave signals. The sample under test was illuminated used to generateWith highapower microwave signals. The sample underoftest illuminated by awas horn byare a horn antenna. sensitivity of of 0.05 K, K, thethe thermal profile thewas sample’s surface by a horn antenna. With a sensitivity 0.05 thermal profile of the sample’s surface was 3 antenna. With a sensitivity of 0.05 K, the thermal profile of the sample’s surface was captured by captured by an IR camera (DRS Tamarisk 320). In Figure 16c, the sample is a 20 × 20 × 4 cm mortar 3 captured by an IR camera (DRS Tamarisk 320). In Figure 16c, the sample is a 20 × 20 × 4 cm mortar 3 mortar 2 2 an IR camera (DRS Tamarisk 320). In Figure 16c, the sample is a 20 × 20 × 4 cm sample sample covered with a 13 × 13 cmcmCFRP. There is ais 2a ×2 2×cm 2 mm depth delamination in in thethe 2 CFRP. 2 and sample covered with a 13 ×2 13 There 2 cmand 2 mm depth delamination covered with a 13 × 13 cm CFRP. There is a2× 2 cm2 and 2 mm depth delamination in standoff the center center of the sample [58]. With 5 s heating time, measurements were performed at a 6 cm center of the sample [58]. With 5 s heating time, measurements were performed at a 6 cm standoff of the sample [58]. With 5 s heating time, measurements were at a 6 cm standoff distance. distance. Meanwhile, measurements were also performed at at aperformed 45 cmcm standoff distance with 15 15 s s distance. Meanwhile, measurements were also performed a 45 standoff distance with Meanwhile, measurements were also performed at aand 45 cm standoff distance with16d, 15 sthe heating time. heating time. Thermal profiles were captured before after heating. In Figure thermal heating time. Thermal profiles were captured before and after heating. In Figure 16d, the thermal Thermal profiles were captured before and after heating. In Figure 16d, the 5thermal image before image before heating is is provided. Figure 16e16e shows thethe thermal image after In In their image before heating provided. Figure shows thermal image after s5 heating. s heating. their heating is provided. Figure 16e shows the thermal image after 5 s heating. In their primary experiment primary experiment results, thethe delamination in in CFRP cancan bebe easily identified. Both thethe level of of primary experiment results, delamination CFRP easily identified. Both level results, the delamination in CFRP can be easily identified. Both the level ofInpower andthey heating time power and heating time should be optimized to detect small size disbands. addition, found power and heating time should be optimized to detect small size disbands. In addition, they found should be optimized toisdetect smallaffected size disbands. Ineffects addition, they found that the measured result is that thethe measured result evidently byby edge with 45 45 cm standoff distance, as as shown that measured result is evidently affected edge effects with cm standoff distance, shown evidently affected by edge effects with 45 cm standoff distance, as shown in Figure 16f. in in Figure 16f. Figure 16f.

Sensors 2017, 17, 1123

18 of 33

Sensors 2017, 17, x FOR PEER REVIEW

17 of 33

Figure 16. 16. MWT cm (b)(b) standoff, photograph of Figure MWT measurement measurementsystem systemsetup setupused usedfor for4545cm cm(a) (a)and and6 6 cm standoff, photograph mortar sample with delamination (c). Measurements on mortar sample before heating (d), 6 cm of mortar sample with delamination (c). Measurements on mortar sample before heating (d), 6 cm standoff after after 55 ss heat heat (e) (e) and and 45 45cm cmstandoff standoffafter after15 15s sheat heat(f)(f)[58]. [58].Reprinted/reproduced Reprinted/reproduced with with standoff permission from IEEE. permission from IEEE.

Recently, Foudazi has shown that the MWT method can be used for detection of delaminations, Recently, Foudazi has shown that the MWT method can be used for detection of delaminations, debonding and cracks in rehabilitated cement-based materials [70]. The experimental setup is shown debonding and cracks in rehabilitated cement-based materials [70]. The experimental setup is shown in Figure 17a. Three CFRP-strengthened cement-based specimens have been measured (a unbonded in Figure 17a. Three CFRP-strengthened cement-based specimens have been measured (a unbonded defect was made by placing a 5 mm thin sheet of foam with dimensions of 6 cm × 8 cm in a sample defect was made by3 placing a 5 mm thin sheet of foam with dimensions of 6 cm × 8 cm in a sample with 52 × 38 × 9 cm , several delaminations (with thicknesses ranging from 1 mm to 3 mm and areas with 52 × 38 × 9 cm23 , several delaminations (with thicknesses ranging from 1 mm to 3 mm and areas ranging from 10 cm to 100 cm2) have been formed in a sample with dimensions of 52 × 38 × 7.8 cm3, ranging from 10 cm2 to 100 cm2 ) have been formed in3 a sample with dimensions of 52 × 38 × 7.8 cm3 , the crack sample had dimensions of 52 × 38 × 9 cm ), as shown in Figure 17b. Thermal profiles for the crack sample had dimensions of 52 × 38 × 9 cm3 ), as shown in Figure 17b. Thermal profiles these specimens are illustrated in Figure 17c. The approximate location and the size information for these specimens are illustrated in Figure 17c. The approximate location and the size information about defects in these specimens have been obtained. Meanwhile, the temperature of the defect will about defects in these specimens have been obtained. Meanwhile, the temperature of the defect be affected by the orientation of the carbon fibers (due to the electric field difference). In their initial will be affected by the orientation of the carbon fibers (due to the electric field difference). In their experimental results, for the case of unidirectional carbon fiber, if the electric field is along the fiber initial experimental results, for the case of unidirectional carbon fiber, if the electric field is along the direction, the temperature change for the defected area will be smaller compared to the case of fiber direction, the temperature change for the defected area will be smaller compared to the case of perpendicular polarization. This is due to the different electromagnetic response of the carbon fiber perpendicular polarization. This is due to the different electromagnetic response of the carbon fiber at different polarizations. In other words, if the wave is parallel to the fiber orientation, it will react at different polarizations. In other words, if the wave is parallel to the fiber orientation, it will react as an electric conductor, while for the perpendicular case it is similar to high loss dielectric materials. as an electric conductor, while for the perpendicular case it is similar to high loss dielectric materials. Additionally, the thermal contrast (TC) between healthy and defective areas is much greater. Additionally, the thermal contrast (TC) between healthy and defective areas is much greater. Moreover, Moreover, increasing defect dimensions also led to a greater TC. increasing defect dimensions also led to a greater TC. Furthermore, the authors used MWT to monitor debonding in CFRP by using different excitation antennas [64]. A horn antenna linked with a double-ridged waveguide operating at frequency ranged from 0.75 GHz to 18 GHz was employed. The aperture size of this antenna is 14 × 24 cm2 . In Figure 18a, the CFRP sample is 30 × 30 × 0.2 cm3 backed with 40 × 40 × 5 cm3 Al sheet which containing six debonds with dimensions of 6 × 6 cm2 , 5 × 5 cm2 , 4 × 4 cm2 , 3 × 3 cm2 , 2 × 2 cm2 , and 1 × 1 cm2 , respectively. Figure 18b shows top and side views of the CFRP backed by a layer of Al sheet. During experiment, the sample under test was illuminated at 2.4 GHz microwave signal of different power (50, 100, 150 and 200 W). In addition, the effect of the heating time (from 10 s to 30 s) has been investigated by using CST Microwave Studio and MultiPhysics Studio. As shown in Figure 18c,d, normalized thermal images of CFRP with different excitation power levels at 10 s and 30 s have been demonstrated. In the preliminary experimental results, they found that a higher excitation microwave power is needed for a smaller debond detection. In addition, the debonding becomes clearly visible

Sensors 2017, 17, 1123

19 of 33

with increasing excitation power due to the temperature difference increase. Therefore, increasing the incident power improves the detection of small disbonds. Meanwhile, an increase of the heating time leads to an increased temperature throughout the sample under test, thereby the possibility of detecting the disbonds is reduced, therefore, the heating time must be optimized according to the actual situation. Figure 17.x FOR MWT measurement setup (a), samples under test (b) and thermal profiles for samples Sensors 2017, 17, PEER REVIEW 18 of 33 under test (c) [70]. Reprinted/reproduced with permission from IEEE.

Furthermore, the authors used MWT to monitor debonding in CFRP by using different excitation antennas [64]. A horn antenna linked with a double-ridged waveguide operating at frequency ranged from 0.75 GHz to 18 GHz was employed. The aperture size of this antenna is 14 × 24 cm2. In Figure 18a, the CFRP sample is 30 × 30 × 0.2 cm3 backed with 40 × 40 × 5 cm3 Al sheet which containing six debonds with dimensions of 6 × 6 cm2, 5 × 5 cm2, 4 × 4 cm2, 3 × 3 cm2, 2 × 2 cm2, and 1 × 1 cm2, respectively. Figure 18b shows top and side views of the CFRP backed by a layer of Al sheet. During experiment, the sample under test was illuminated at 2.4 GHz microwave signal of different power (50, 100, 150 and 200 W). In addition, the effect of the heating time (from 10 s to 30 s) has been investigated by using CST Microwave Studio and MultiPhysics Studio. As shown in Figure 18c and 18d, normalized thermal images of CFRP with different excitation power levels at 10 s and 30 s have been demonstrated. In the preliminary experimental results, they found that a higher excitation microwave power is needed for a smaller debond detection. In addition, the debonding becomes clearly visible with increasing excitation power due to the temperature difference increase. Therefore, increasing the incident power improves the detection of small disbonds. Meanwhile, an increase of the heating time leads to an increased temperature throughout the sample under test, thereby the Figure 17. 17. MWTmeasurement measurementsetup setup (a), samples under (b) thermal and thermal profiles for samples Figure (a), under test test (b) and profiles samples under possibility of MWT detecting the disbonds is samples reduced, therefore, the heating timefor must be optimized under [70]. Reprinted/reproduced with permission from IEEE. test (c) test [70].(c) Reprinted/reproduced with permission from IEEE. according to the actual situation. Furthermore, the authors used MWT to monitor debonding in CFRP by using different excitation antennas [64]. A horn antenna linked with a double-ridged waveguide operating at frequency ranged from 0.75 GHz to 18 GHz was employed. The aperture size of this antenna is 14 × 24 cm2. In Figure 18a, the CFRP sample is 30 × 30 × 0.2 cm3 backed with 40 × 40 × 5 cm3 Al sheet which containing six debonds with dimensions of 6 × 6 cm2, 5 × 5 cm2, 4 × 4 cm2, 3 × 3 cm2, 2 × 2 cm2, and 1 × 1 cm2, respectively. Figure 18b shows top and side views of the CFRP backed by a layer of Al sheet. During experiment, the sample under test was illuminated at 2.4 GHz microwave signal of different power (50, 100, 150 and 200 W). In addition, the effect of the heating time (from 10 s to 30 s) has been investigated by using CST Microwave Studio and MultiPhysics Studio. As shown in Figure 18c and 18d, normalized thermal images of CFRP with different excitation power levels at 10 s and 30 s have been demonstrated. In the preliminary experimental results, they found that a higher excitation microwave power is needed for a smaller debond detection. In addition, the debonding becomes clearly visible with increasing excitation power due to the temperature difference increase. Therefore, increasing the incident power improves the detection of small disbonds. Meanwhile, an increase of the heating time leads to an increased temperature throughout the sample under test, thereby the possibility of detecting the disbonds is reduced, therefore, the heating time must be optimized according to the actual situation. Figure 18. MWT experimental setup (a), sample (b), measurement results for 10 s heating (c) and 30 s heating (d) [64]. Reprinted/reproduced with permission from IEEE.

A microwave time-resolved infrared radiometry system was proposed by scientists in The John Hopkins University [71]. The experimental setup is shown in Figure 19a. 9 GHz microwave signals were produced by an HP 6890B Oscillator. A Hughes 1277 X-band traveling wave tube amplifier was used to amplify these signals and then to feed them into a rectangular waveguide linked with a single flare horn antenna. The beam width of this horn antenna is about 50◦ and the sample was placed about 15 cm away. 2.3 W input power was transmitted to the antenna and a 20 mW/cm2 power

A microwave time-resolved infrared radiometry system was proposed by scientists in The John Hopkins University [71]. The experimental setup is shown in Figure 19a. 9 GHz microwave signals were produced by an HP 6890B Oscillator. A Hughes 1277 X-band traveling wave tube amplifier was used to amplify these signals and then to feed them into a rectangular waveguide linked with a20 single Sensors 2017, 17, 1123 of 33 flare horn antenna. The beam width of this horn antenna is about 50° and the sample was placed about 15 cm away. 2.3 W input power was transmitted to the antenna and a 20 mW/cm2 power density density was was created created for for the experimental experimental study. study. Both Both the the polarization polarization of of the the microwave microwave and and the the angle angle of incidence were controlled. By operating in the 3 µm–5 µm band, a 128 × 128 InSb focal plane array of incidence were controlled. By operating in the 3 μm–5 μm band, a 128 × infrared infrared camera camera was was used used to to monitor monitor the the surface surface temperature temperature of of the the sample sample under under test. test. Before, during during and after after the the application application of the microwave microwave step heating pulse, a series of frames were recorded for time-dependent initial experimental results, the authors measured carboncarbon fibers time-dependentmeasurements. measurements.InInthethe initial experimental results, the authors measured with different depthsdepths (0.25 mm and 0.75 mm) fiberglass-epoxy. The temperature at the center fiberstwo with two different (0.25 mm and 0.75inmm) in fiberglass-epoxy. The temperature at the of the fiber is shown in Figure 19b. Moreover, they found that the experimental data can be fitted center of the fiber is shown in Figure 19b. Moreover, they found that the experimental data can be with solidaline. determination of surface layer thickness and thermal diffusivity can be almost fittedawith solidThe line. The determination of surface layer thickness and thermal diffusivity can be independent from thefrom surface properties of the layer. almost independent the surface properties of the layer.

Figure Figure 19. 19. Schematic diagram of the experimental setup in The John Hopkins Hopkins University University (a) and and Surface as aafunction functionofofsquare squareroot roottime time a single point fibers in 0.25 Surface temperature as forfor a single point on on fibers in 0.25 mmmm and and 0.75 0.75 depth (b) [71]. Reprinted/reproduced permission mm mm depth (b) [71]. Reprinted/reproduced with with permission fromfrom SPIE.SPIE.

Lee et et al. al. proposed microwave probe probe pumping pumping technique technique to to characterize characterize the the anisotropic anisotropic Lee proposed aa microwave electrical conductivity in carbon-fiber composite materials [69]. They used CST Microwave Studio to to electrical conductivity in carbon-fiber composite materials [69]. They used CST Microwave Studio investigate the the electromagnetic electromagnetic field field distribution distribution with with different different anisotropic anisotropic conductivity. conductivity. Two Two 10 10 mm mm investigate coaxial probes were used to pump and scan the microwave field. The pumping probe was fixed on coaxial probes were used to pump and scan the microwave field. The pumping probe was fixed the backside of the sample under test. The near-field distribution was scanned by the scanning probe. on the backside of the sample under test. The near-field distribution was scanned by the scanning A spectrometer was used measure the microwave power. A network analyzer was was usedused as the probe. A spectrometer wasto used to measure the microwave power. A network analyzer as microwave source with a continuous mode. A FLIR T620 was used for measurement. A 1 GHz the microwave source with a continuous mode. A FLIR T620 was used for measurement. A 1 GHz microwave source source with with 20 20 dbm dbmpower powerwas wasused usedduring duringthe themeasurements. measurements.AA5050××50 50×× 0.1 0.1 mm mm33 microwave carbon-fiber/PEEK composite carbon-fiber/PEEK composite sheet sheet with with aa defect defect (characteristic (characteristic length length is is 55 mm) mm) has has been been measured. measured. In the initial experimental results, they obtained an intense area around the scanning probe. In the initial experimental results, they obtained an intense area around the scanning probe. Moreover, Moreover, found that the conductivity of the carbon-fiber/PEEK has an elliptical distribution. they found they that the conductivity of the carbon-fiber/PEEK has an elliptical distribution. 4.7. Honeycomb Structures Structures 4.7. Microwavepulsed pulsed thermography for water measurement in honeycomb materials was Microwave thermography for water measurement in honeycomb materials was developed ◦ beam developed in Poland They introduced 2 GHz antennas a 30° beamThe width. The by scientistsbyinscientists Poland [72]. They [72]. introduced 2 GHz antennas with a 30with width. surface 2 2 surface of a honeycomb was illuminated a 30 mW/cm power density. antenna of a honeycomb samplesample was illuminated with awith 30 mW/cm power density. TheThe antenna andand IR IR camera were arranged a reflection configuration.The Theproposed proposedMWT MWTdetection detectionsystem systemis is shown shown camera were arranged in in a reflection configuration. in Figure Figure 20a. 20a. The antenna was located at 1 m away from the sample sample under test and the IR IR camera camera 2 was placed at 0.7 m. The sample was a 290 × 215 mm sandwich panel with two 0.7 mm thickness Fibredux face skins. Different quantities of water (5.0, 2.5, 1.2 and