- Email: [email protected]
Analysis of defect detectability in polymeric composites using self-heating based vibrothermography
Composite Structures 201 (2018) 760–765
Contents lists available at ScienceDirect
Composite Structures journal homepage: www.elsevier.com/locate/compstruct
Analysis of defect detectability in polymeric composites using self-heating based vibrothermography
T
⁎
Andrzej Katunina, , Dominik Wachlaa a
Institute of Fundamentals of Machinery Design, Faculty of Mechanical Engineering, Silesian University of Technology, Konarskiego 18A, 44-100 Gliwice, Poland
A R T I C LE I N FO
A B S T R A C T
Keywords: Self-heating effect Vibrothermography Modal analysis Damage identification Vibrational excitation
The self-heating based vibrothermography (SHVT) is a new non-destructive testing method dedicated for testing of polymer matrix composite structures, where the excitation is performed by externally applied mechanical vibrations in a low frequency range. Exciting a structure with several resonant frequencies, the heating up of this structure is possible due to the occurrence of the thermoviscoleastic effect called the self-heating, whose nature originates from the mechanical energy dissipation. In order to examine an efficiency of the SHVT an analysis of defect detectability on composite specimens with milled artificial defects was performed. The post-processing of the series of resulting thermograms was performed in order to enhance defect detectability. The obtained results allow to conclude about high efficiency of SHVT NDT technique, which can be used especially in cases when a direct access to the testing structure in order to excite it externally is difficult or impossible.
1. Introduction Active infrared thermography (IRT) is a widespread non-destructive testing (NDT) method applied for testing and inspection of structural elements in various industrial branches. A special attention in IRT testing is paid to structures made of polymer matrix composites (PMCs) due to their wide applicability for manufacturing of elements of transport means. IRT NDT methods, considering its various modifications, allow for fast and non-contact inspection of such structures and detection of surface and subsurface manufacturing and operational defects and propagating damage resulting e.g. from fatigue processes. Numerous research studies prove the efficiency of application of IRT NDT for aircraft [1–3], aerospace [4–6], automotive [7,8], and other composite structures to detect and identify internal defects and damage. The difficulty in inspection of such structures with respect to homogeneous ones, is, first of all, its anisotropy, which influences on thermal wave speeds in various directions, and the directionality of thermomechanical properties of tested composite structures [9]. Moreover, the crucial influence on the defect detectability in such structures has an applied IRT method and the parameters of thermal excitation of a tested structure. Following this, it is essential to classify the methods and discuss possible excitation procedures. The classical IRT methods used for structural damage identification (SDI) in composites can be classified, in general, to pulsed IRT methods, transient IRT methods, and the methods with frequency-modulated excitation (lock-in IRT). The pulsed IRT methods are the simplest ones,
⁎
Corresponding author. E-mail address: [email protected] (A. Katunin).
https://doi.org/10.1016/j.compstruct.2018.06.105 Received 11 February 2018; Received in revised form 3 May 2018; Accepted 27 June 2018 Available online 30 June 2018 0263-8223/ © 2018 Elsevier Ltd. All rights reserved.
and are based on excitation by a thermal pulse and observation of a tested structure using infrared (IR) camera, while the delivered heat diffuses through the tested structure. A comprehensive review with the historical overview and theoretical background on the pulsed IRT methods is presented in [10]. The transient IRT methods are very similar to the pulsed ones with one significant difference: the duration of excitation in the case of transient IRT is much bigger, which allows using less powerful heating sources with respect to the pulsed IRT. The SDI ability of this group of methods is comparable with those of pulsed IRT. More details on this approach can be found in [11]. The application of lock-in IRT, where the excitation thermal wave is modulated at a fixed frequency and the thermal response is observed by IR camera, allows for significant improvement of defect detectability with respect to previously discussed methods (see [12,13] for more details). In all of the aforementioned types of IRT optical or external heating source excitation is used, usually in the form of flashes, halogen or IR lamps, lasers, fluid jets, heating blankets, etc. [14,15]. An alternative IRT approach with respect to already presented methods is vibrothermography (VT) – a group of methods which use mechanical or internal excitation, classified by the authors of [15] as an another group of methods with respect to all aforementioned IRT methods. The main difference of VT with respect to other IRT methods is a lack of external heating source, which is substituted by heating resulting from mechanical excitation of a tested structure. The concept of VT was proposed by the authors of [16,17] in the early 1980s. The heating of a tested specimen is usually performed in the form of
Composite Structures 201 (2018) 760–765
A. Katunin, D. Wachla
Fig. 1. The schemes of milled defects with characteristic dimensions.
damage identification in composite structures [29] show the potential of this approach in NDT practice. The preliminary studies presented in [29] allow for definition of a testing procedure based on the performed experiments, i.e. it was observed that the reasonable and the most effective approach is an excitation of a composite structure with a multiharmonic signal composed of several harmonics corresponding to the resonant frequencies of vibrations of a tested structure. An excitation with multiple resonant frequencies allows to excite the corresponding modal shapes, and thus, the regions of the highest stress for each modal shapes are different. Since in viscoelastic materials mechanical stress is directly related with the amount of energy dissipation (see [28] for instance), these regions correspond with regions of maximal heating up. This justifies the mechanical excitation with multiple resonant frequencies, assuming that the presence and location of eventual defect is not a priori known. The main goal of this paper is to investigate defect detectability efficiency, i.e. an estimation of depth at which a defect is still detectable. For this purpose the tests were performed on specimens with FBHs of various depths and shapes. It was shown that taking into consideration a series of thermograms instead of a single thermogram, and their further common analysis allow to enhance defect detectability. Additionally, the post-processing of thermograms in order to enhance defect detectability is discussed. The efficiency of the proposed approach is confirmed by the promising results obtained from self-heating based vibrothermography (SHVT) NDT experiments.
mechanical excitation by an elastic wave in a sonic or ultrasonic [18–20] frequency range, which results in energy dissipation in a tested structure. In the case of existence of a defect the bigger energy dissipation is observed due to the friction between the faces of a defect or stress concentration in the surrounding area of a defect [20]. The physics of this phenomenon is widely discussed in [21,22]. This leads to the situation, where an increase of temperature is observed only at the location of a defect, which can be captured by IR camera. The excitation is usually performed in the contact way using ultrasonic boosters or actuators placed on the surface or integrated with a tested structure. Besides of sonic and ultrasonic heating sources used in VT, other types of excitation are successfully developed. The authors of [23] based their approach of the thermal excitation on electromagnetic induction applied to the carbon fibre of a tested composite structure, which allows for local heating. Another approach of thermal excitation of a tested composite structure is using the thermoviscoelastic effect, known in a literature as the self-heating effect, accompanied by mechanical vibrations of such a structure. The nature of internal heat generation is different than for classical ultrasonic VT, since the heating in this case is coming from hysteresis resulting from phase lag of stress and strain amplitudes during mechanical vibrations, and originating from an internal structure of a polymeric matrix of a tested composite instead of frictional heating phenomena occurring during ultrasonic excitation of a tested structure [21]. The ability of heat generation of polymeric composites subjected to vibrations was successfully used in several previous studies. In particular, the self-heating effect was used by the authors of [24], where they presented results of investigation on viscoelastic response of flat-bottom holes (FBHs) filled with a viscous material, while the specimen was subjected to mechanical vibrations in the ultrasonic frequency range. The authors of [25,26] used mechanical excitation on resonant frequencies to localize the crack and monitor its propagation. The idea on excitation of tested structures with resonant frequencies was also used by the authors of [27], where they used absorptive viscoelastic coatings on the tested metallic specimens in order to observe the self-heating effect during excitation of specimens with resonant frequencies in the ultrasonic frequency range. Previous experimental studies of the authors’ team on a self-heating temperature distribution during mechanical excitation with resonant frequencies starting from the fundamental frequency of vibrations [28] as well as initial studies on application of the self-heating effect to
2. Specimens preparation The specimens made of the glass E-fabric-reinforced 14-layered epoxy composite material purchased from Izo-Erg S.A. (Gliwice, Poland) were cut from a sheet with the thickness of 2.5 mm into strips with the length of 250 mm and the width of 10 mm. For the investigation of defect detectability efficiency FBHs were milled in the specimens according to the schemes presented in Fig. 1. The milling was performed on the two-axis milling machine Roland Modela MDX-20 (Hamamatsu, Japan) using ø1 diamond milling cutter (see Fig. 2) with the following milling parameters: milling speed in the planar directions of specimens’ surface was of 10 mm/s, milling speed in the thickness direction was of 0.5 mm/s, and a spindle speed was of 6500 rpm.
Fig. 2. Milling process of artificial defects in specimens. 761
Composite Structures 201 (2018) 760–765
A. Katunin, D. Wachla
Fig. 3. The specimens with milled FBHs prepared for SHVT testing.
As it was shown in Fig. 1, four shapes of FBHs were considered: notch-type (L), circular (C), square (S), and rectangular (R). For each defined shape of the FBH six depths were milled, starting from 0.25 mm and ending at 1.5 mm with a step of 0.25 mm. Following this, 24 specimens in total were tested using the presented approach. For ensuring appropriate thermal emissivity the specimens were covered with a silicone black matt heat-resisting enamel manufactured by Dragon Poland Sp. z. o.o. (Cracow, Poland). The exemplary specimens after painting with all considered shapes of FBHs are presented in Fig. 3.
Table 1 Resonant frequencies of vibration of tested specimens with FBHs, Hz.
Depth of FBH, mm
Shape of FBH
L
C
S
R
0.25
191.01 528.91 1069.14 185.94 501.95 1031.64 190.23 520.70 1065.62 164.45 456.64 922.26 150.00 414.84 833.98 110.94 307.42 586.72 1005.08
189.06 533.05 1053.12 190.23 517.97 1075.39 191.80 526.17 1076.95 188.28 518.75 1048.44 187.50 516.02 1044.53 187.11 518.36 1042.97
189.06 524.22 1057.81 191.02 515.62 1072.66 190.62 524.61 1071.48 185.55 517.19 1037.50 185.94 508.59 1033.98 189.50 525.00 1069.00
189.45 523.44 1055.86 183.20 491.02 1008.98 187.89 99.22 1056.64 186.33 514.45 1040.23 184.37 506.64 1024.22 184.00 503.50 1025.50
0.5
0.75
3. Experimental testing procedure
1
The experiments were performed on the own-designed test rig presented in Fig. 4. The testing procedure consisted of two steps: performing the classical modal analysis, and, after acquisition of resonant frequencies of a tested specimen, excitation of this specimen with a multi-harmonic signal composed from the harmonics with acquired resonant frequencies and with simultaneous observation of the specimen’s top surface using IR camera in order to register its thermal response. In the first step, the modal analysis of a specimen 4, clamped in a steel holder 8 with polycarbonate inserts allowing for thermal insulation of a specimen 4 from the holder 8, was performed. The loading was performed using the TIRA® TV-51120 electrodynamic shaker 1 through the stinger 2, connected with a specimen holder 3 and the shaker 1 by screw connections. The pseudo-random signal generated from the dedicated measurement and control software on the PC 13, and amplified using the TIRA® BAA 500 shaker amplifier 10 was used for specimen excitation in order to obtain wide-band frequency response and identify natural frequencies. The measurements were performed using the Polytec® PSV-400 scanning laser Doppler vibrometer (LDV) 7 connected with the Polytec® PSV-W-400 data acquisition and control unit 9. To separate vibrations of a specimen 4 from vibrations of the whole system a reference signal was acquired from the Polytec® PDV100 single point LDV 6 focused on the surface of the holder 3. The surface of a specimen 4 subjected to testing and the holder 3 were covered by reflective tape to ensure appropriate reflection of laser beams of LDVs 6 and 7. The measurements of vibration velocity of tested specimens were performed in 45 equidistant points defined on the specimen’s surface across its length. The frequency response of the
1.25
1.5
specimens was examined in the frequency range of 0 ÷ 1250 Hz with a resolution of 0.390625 Hz (3200 data points in the defined frequency range). In every case the testing was performed on the undamaged surface, i.e. the FBHs were located on the bottom surfaces of specimens. In the considered frequency range three resonant frequencies were registered, except the specimen with the notch-type FBH with depth of 1.5 mm, for which four resonant frequencies were registered. The collected resonant frequencies for tested specimens are stored in Table 1. The exemplary frequency response function (FRF) and corresponding modal shapes for one of the tested specimens are presented in Fig. 5. In the second step, the collected resonant frequencies for a given specimen were used for a construction of a multi-harmonic signal, where the harmonics with resonant frequencies were the components of this multi-harmonic signal. This allows for excitation of a specimen on multiple resonant frequencies simultaneously. Since the maximal selfheating temperature values are observable in the regions of stress concentration, which was proven theoretically and experimentally in previous studies (see [28] for instance), for every modal shape the maximal heating is observed near the clamps and in the locations
(a)
(b)
Fig. 4. Experimental test rig: a) general view, b) scheme. 762
Composite Structures 201 (2018) 760–765
A. Katunin, D. Wachla
Fig. 5. Exemplary FRF a) and corresponding modal shapes b)-d) of tested specimens.
corresponding to maximal deflection in modal shapes. Considering that the location of a damage is not known a priori, it is essential to excite the whole surface of a tested structure, and therefore, the mechanical excitation with multiple resonant frequencies is necessary. It should be noticed that the selected resonant frequencies allow for the thermal excitation of the whole surface of a tested specimen in the presented case, however the selection process is geometry-dependent (due to the relation of local stress concentrations in modal shapes with areas of selfheating effect appearance), which means that the considered resonance frequencies in the excitation signal should be adjusted to a specific case. The signal constructed in the above-presented way was generated from the own-developed application, and after switching the switch 12 was delivered to the shaker 1 through the National Instruments™ cDAQ9174 compact DAQ chassis with 4-channel ± 10 V 16-bit analog output module NI-9263 11. The thermal response was registered by the InfraTec VarioCam® hr IR camera 5 with a sampling rate of 2 frames per second. The test duration in each case was equaled 5 min, i.e. 600 thermograms were collected for each specimen.
(a) L
(b) C
(c) S
(d) R
Fig. 6. Identified thermal signatures for various types of FBHs with the depth of 1 mm.
4. Analysis of thermograms and defect detectability The collected thermograms for various types of FBHs allowed for clear identification the notch-type FBHs only, while for other types of FBHs the thermograms were not sharp enough to clearly identify thermal signatures of FBHs (see Fig. 6). Moreover, it was observed that the sharpness and detectability depends on moment of capturing the thermogram, i.e. with increase of testing time the progressive blurring of thermograms was observed (see Fig. 7). Additionally, due to the temperature fluctuations during testing and considering low temperature increase with respect to the ambient temperature, the analysis of a single thermogram for identification of a thermal signature of defect can be inefficient. Therefore, it is essential to consider multiple thermograms during such an analysis. The results presented in Fig. 7 clearly show that the effective time period for damage identification is the period of self-heating temperature increase, i.e. the period between starting of rapid self-heating temperature increase and reaching the thermal equilibrium between the generated and released heat of a tested specimen. Considering that the self-heating temperature growth depends on material properties of a tested specimen as well as on boundary and loading conditions, the effective time period will be also dependent on these properties and conditions, and should be determined for a given tested structure individually. To unify the method of determination of the effective time
period the following approach is proposed. The maximal self-heating temperature evolution curve is fitted using the polynomial function in order to obtain the best fitting of approximation function to experimental data. It is especially important to select appropriate order of the applied approximation function which allow for correct reflection of the measurement data when the rapid self-heating temperature growth starts and when the thermal equilibrium is reached. The results of the performed empirical study show that the best results for such a character of a self-heating temperature evolution curve can be obtained using a polynomial function of 8th order. The example of such an approximation is presented in Fig. 8a. For clarity, the same self-heating temperature evolution curve as one presented in Fig. 7a was taken into consideration. Next, the obtained approximation curve is subjected to numerical differentiation and the absolute value is taken of the resulting data. This operation allows for identification of desirable boundaries of the effective time period (see the exemplary result in Fig. 8b). The boundaries of the effective time period is defined by first two minima of the obtained function starting from the origin. Having the effective time period defined by the above-presented approach, in order to enhance defect detectability, especially for the cases with unclear thermal signatures (see e.g. Fig. 6b-d), it is essential 763
Composite Structures 201 (2018) 760–765
A. Katunin, D. Wachla
Fig. 9. Exemplary results of the processed thermograms for considered types of FBHs. Table 2 Results of defects detectability analysis using SHVT method. Number of detected defects, –
Depth, mm
Fig. 7. Self-heating temperature evolution during testing (a) with accompanying thermograms for different moments of time (b) for the notch-type FBH with the depth of 1 mm.
Shape of FBH
L
C
S
R
0.25 0.5 0.75 1 1.25 1.5
undetectable 5 6 8 9 9
undetectable
undetectable 4 6 7 7 7
undetectable
3 6 6 6
3 4 4 4
signatures of FBHs are better detectable. In order to analyze detectability of defects in the form of FBHs with various shapes and depths the analysis according the above-presented approach was performed for all considered specimens. The results of defects detectability for particular cases are presented in Table 2. The resulting images for notch-type FBHs for all considered depths are presented in Fig. 10. The obtained results show that the thermal signatures of the introduced defects become detectable starting from the cases of the defects depth of 0.75 mm. One should note that the FBHs were introduced on the bottom surface of a specimen with respect to the surface subjected to testing. This means that the method reveals a sensitivity to damage located on the depth of 1.75 mm, which corresponds with 60% of the specimen thickness. According to the obtained results (see Table 2) not all the defects were well detectable (except the notch-type FBHs for the highest considered depths). There are two reasons that may explain this phenomenon. Firstly, harmonics in the excitation signal, in spite of initial equal magnitude of them during generation, have different influence on the resulting temperature distribution
to add up the acquired thermograms for the defined effective time period. The collected thermograms were exported to Matlab® environment in the form of matrices of temperature values, and then cut for specific dimensions of the specimen. The resulting matrices were added up for the defined effective time periods unique for each considered case, and the logarithm of the resulting matrix was taken for additional enhancement. The exemplary results of this operation are presented in Fig. 9. As the examples the same specimens as those presented in Fig. 6 were considered. The presented cases in Fig. 9 show that the introduced FBHs, except the notch-type ones, are barely detectable. The reason of such results is a difference in stress distribution between the notch-type FBHs and other shapes of FBHs considered in this study, and thus, the magnitude of resulting deflection of specimens with these FBHs. This allows for dissipation of more thermal energy in the case of the notch-type FBHs with respect to other considered cases, and accordingly, the thermal
Fig. 8. Maximal self-heating temperature evolution curve with the polynomial approximation (a), and absolute value of derivative of the approximation polynomial with defined boundaries of the effective time period (b). 764
Composite Structures 201 (2018) 760–765
A. Katunin, D. Wachla
defects of aircraft composites. NDT&E Int 2003;36(6):433–9. [2] Avdelidis NP, Almond DP, Dobbinson A, Ibarra-Castanedo C, Maldague X. Aircraft composite assessment by means of transient thermal NDT. Prog Aerosp Sci 2004;40:143–62. [3] Katunin A, Dragan K, Dziendzikowski M. Damage identification in aircraft composite structures: a case study using various non-destructive testing techniques. Compos Struct 2015;127:1–9. [4] Ibarra-Castanedo C, Genest M, Servais P, Maldague SPV, Bendada A. Qualitative and quantitative assessment of aerospace structures by pulsed thermography. Nondestr Testing Eval 2007;22(2–3):199–215. [5] Pawar S, Peters K, Transient infrared thermography for damage evaluation in aerospace composites. In: Proceedings of SPIE 7649, Nondestructive Characterization for Composite Materials, Aerospace Engineering, Civil Infrastructure, and Homeland Security, 2010, 76491A. [6] Meola C, Boccardi S, Carlomagno GM, editors. Infrared thermography in the evaluation of aerospace composite materials Woodhead Publishing; 2017. [7] Santulli C. IR thermography for impact damage evaluation of automotive composites and components. NDT.net J 2003;9(9). [8] Choi S-H, Park S-K, Kim J-Y. NDE of advanced automotive composite materials that apply ultrasound infrared thermography technique. Int J Modern Physics: Conf Ser 2012;6:515–20. [9] Liu D, Malvern LE. Matrix cracking in impacted glass/epoxy plates. J Compos Mater 1987;21:594–609. [10] Vavilov VP, Burleigh DD. Review of pulsed thermal NDT: physical principles, theory and data processing. NDT&E Int 2015;73:28–52. [11] Kretzmann JE, Venter G, Schreve K. Non-destructive testing with transient thermography on composite materials. R&D J South African Inst Mech Eng 2016;32:35–43. [12] Meola C, Carlomagno GM, Squillace A, Giorleo G. Non-destructive control of industrial materials by means of lock-in thermography. Meas Sci Technol 2002;13(10):1583–90. [13] Ullman T, Shi Y, Aoki R, Capabilities of lock-in thermography for non-destructive inspection of fibre reinforced composites. In: 11th International Conference on Quantitative InfraRed Thermography, Naples 2012, pp. 1–11. [14] Ibarra-Castanedo C, Tarpani JR, Maldague XPV. Nondestructive testing with thermography. Eur J Phys 2013;34(6):S91–109. [15] Ley O, Godinez V. Non-destructive evaluation (NDE) of aerospace composites: application of infrared (IR) thermography. Karbhari VM, editor. Non-destructive evaluation (NDE) of polymer matrix composites. Techniques and applications Oxford, UK: Woodhead Publishing Ltd.; 2013. p. 309–34. [16] Reifsnider KL, Henneke EG, Stinchcomb WW. The mechanics of vibrothermography. In: Stinchcomb WW, Duke JC, Henneke EG, Reifsnider KL, editors. Mechanics of Nondestructive Testing. Boston, MA, USA: Springer; 1980. p. 249–76. [17] Mignogna RB, Green RE, Duke JC, Henneke EG, Reifsnider KL. Thermographic investigation of high-power ultrasonic heating in materials. Ultrasonics 1981;19:159–63. [18] Zweschper T, Dillenz A, Riegert G, Scherling D, Busse G. Ultrasound excited thermography using frequency modulated elastic waves. Insight – Non-Destructive Testing Condition Monit 2003;45:178–82. [19] Pieczonka L, Aymerich F, Brozek G, Szwedo M, Staszewski WJ, Uhl T. Modelling and numerical simulations of vibrothermography for impact damage detection in composites structures. Struct Control Health Monit 2013;20:626–38. [20] Yang R, He Y. Optically and non-optically excited thermography for composites: a review. Infrared Phys Technol 2016;75:26–50. [21] Mabrouki F, Thomas M, Genest M, Fahr A. Frictional heating model for efficient use of vibrothermography. NDT&E Int 2009;42:345–52. [22] Renshaw J, Chen JC, Holland SD, Thompson RB. The sources of heat generation in vibrothermography. NDT&E Int 2011;44:736–9. [23] Yang R, He Y. Polymer-matrix composites carbon-fibre characterization and damage inspection using selectively heating thermography (SeHT) through electromagnetic induction. Compos Struct 2016;140:590–601. [24] Montanini R, Freni F. Correlation between vibrational mode shapes and viscoelastic heat generation in vibrothermography. NDT&E Int 2013;58:43–8. [25] Magi F, Di Maio D, Sever I. Damage initiation and structural degradation through resonance vibration: Application to composite laminates in fatigue. Compos Sci Technol 2016;132:47–56. [26] Di Maio D, Magi F, Sever IA, Damage monitoring of composite components under vibration fatigue using scanning laser Doppler vibrometer, Experimental Mechanics, 2018, in press, doi: https://doi.org//10.1007/s11340-017-0367-y. [27] Vaddi JS, Holland SD, Kessler MR. Absorptive viscoelastic coatings for full field vibration coverage measurement in vibrothermography. NDT&E Int 2016;82:56–61. [28] Katunin A, Fidali M. Self-heating of polymeric laminated composite plates under the resonant vibrations: theoretical and experimental study. Polym Compos 2012;33:138–46. [29] Katunin A. A concept of thermographic method for non-destructive testing of polymeric composite structures using self-heating effect. Sensors 2018;18:74.
Fig. 10. Results of defect detectability for notch-type FBHs for various depths considered in the study using SHVT method.
during excitation, i.e. the lower frequency of the harmonic the better thermal response is obtained. This explains difficulties with detectability of 2nd and 8th FBHs in all tested cases (see Figs. 9 and 10), which location corresponds mainly with maximal deflections of the third modal shape (cf. Fig. 5d). Secondly, the shape of FBHs has an influence on the stress distribution, and, considering that the amount of dissipated heat during excitation is proportional to stress (see [28] for instance), the differences in defects detectability for the same depths and various shapes of FBHs (see Table 2) can be explained. 5. Conclusions The efficiency of the proposed SHVT NDT method is analyzed basing on experiments performed on PMC specimens with FBHs of various depths. The FBHs were milled on the opposite surface with respect to the tested surface, which allows to simulate subsurface structural defects. The presented SHVT method together with the proposed postprocessing techniques of a collected sequences of thermograms based on estimation of a thermal gradient of the self-heating temperature evolution curves allows for enhancement of detectability of considered defects. The performed analysis of detectability shows satisfactory results, i.e. the defects become detectable starting from 1.75 mm of depth, which corresponds with 60% of the thickness of tested structures. Starting from 1.5 mm of depth (the distance from top surface of a tested specimen to the bottom of the milled FBH) the detectability was unchangeably fine, which proves the efficiency of the proposed approach. In many cases of FBHs their thermal signatures were ambiguous, which does not allow for their classification as a detected defect. However, the existence of such thermal signatures allows supposing that using advanced mathematical tools and image processing methods their detectability can be improved. In the next studies planned within the currently realized research project such an improvement is planned. Acknowledgements The results presented in this paper were obtained within the framework of research grant No. 2015/17/D/ST8/01294 financed by the National Science Centre, Poland. References [1] Avdelidis NP, Hawtin BC, Almond DP. Transient thermography in the assessment of
765
Contents lists available at ScienceDirect
Composite Structures journal homepage: www.elsevier.com/locate/compstruct
Analysis of defect detectability in polymeric composites using self-heating based vibrothermography
T
⁎
Andrzej Katunina, , Dominik Wachlaa a
Institute of Fundamentals of Machinery Design, Faculty of Mechanical Engineering, Silesian University of Technology, Konarskiego 18A, 44-100 Gliwice, Poland
A R T I C LE I N FO
A B S T R A C T
Keywords: Self-heating effect Vibrothermography Modal analysis Damage identification Vibrational excitation
The self-heating based vibrothermography (SHVT) is a new non-destructive testing method dedicated for testing of polymer matrix composite structures, where the excitation is performed by externally applied mechanical vibrations in a low frequency range. Exciting a structure with several resonant frequencies, the heating up of this structure is possible due to the occurrence of the thermoviscoleastic effect called the self-heating, whose nature originates from the mechanical energy dissipation. In order to examine an efficiency of the SHVT an analysis of defect detectability on composite specimens with milled artificial defects was performed. The post-processing of the series of resulting thermograms was performed in order to enhance defect detectability. The obtained results allow to conclude about high efficiency of SHVT NDT technique, which can be used especially in cases when a direct access to the testing structure in order to excite it externally is difficult or impossible.
1. Introduction Active infrared thermography (IRT) is a widespread non-destructive testing (NDT) method applied for testing and inspection of structural elements in various industrial branches. A special attention in IRT testing is paid to structures made of polymer matrix composites (PMCs) due to their wide applicability for manufacturing of elements of transport means. IRT NDT methods, considering its various modifications, allow for fast and non-contact inspection of such structures and detection of surface and subsurface manufacturing and operational defects and propagating damage resulting e.g. from fatigue processes. Numerous research studies prove the efficiency of application of IRT NDT for aircraft [1–3], aerospace [4–6], automotive [7,8], and other composite structures to detect and identify internal defects and damage. The difficulty in inspection of such structures with respect to homogeneous ones, is, first of all, its anisotropy, which influences on thermal wave speeds in various directions, and the directionality of thermomechanical properties of tested composite structures [9]. Moreover, the crucial influence on the defect detectability in such structures has an applied IRT method and the parameters of thermal excitation of a tested structure. Following this, it is essential to classify the methods and discuss possible excitation procedures. The classical IRT methods used for structural damage identification (SDI) in composites can be classified, in general, to pulsed IRT methods, transient IRT methods, and the methods with frequency-modulated excitation (lock-in IRT). The pulsed IRT methods are the simplest ones,
⁎
Corresponding author. E-mail address: [email protected] (A. Katunin).
https://doi.org/10.1016/j.compstruct.2018.06.105 Received 11 February 2018; Received in revised form 3 May 2018; Accepted 27 June 2018 Available online 30 June 2018 0263-8223/ © 2018 Elsevier Ltd. All rights reserved.
and are based on excitation by a thermal pulse and observation of a tested structure using infrared (IR) camera, while the delivered heat diffuses through the tested structure. A comprehensive review with the historical overview and theoretical background on the pulsed IRT methods is presented in [10]. The transient IRT methods are very similar to the pulsed ones with one significant difference: the duration of excitation in the case of transient IRT is much bigger, which allows using less powerful heating sources with respect to the pulsed IRT. The SDI ability of this group of methods is comparable with those of pulsed IRT. More details on this approach can be found in [11]. The application of lock-in IRT, where the excitation thermal wave is modulated at a fixed frequency and the thermal response is observed by IR camera, allows for significant improvement of defect detectability with respect to previously discussed methods (see [12,13] for more details). In all of the aforementioned types of IRT optical or external heating source excitation is used, usually in the form of flashes, halogen or IR lamps, lasers, fluid jets, heating blankets, etc. [14,15]. An alternative IRT approach with respect to already presented methods is vibrothermography (VT) – a group of methods which use mechanical or internal excitation, classified by the authors of [15] as an another group of methods with respect to all aforementioned IRT methods. The main difference of VT with respect to other IRT methods is a lack of external heating source, which is substituted by heating resulting from mechanical excitation of a tested structure. The concept of VT was proposed by the authors of [16,17] in the early 1980s. The heating of a tested specimen is usually performed in the form of
Composite Structures 201 (2018) 760–765
A. Katunin, D. Wachla
Fig. 1. The schemes of milled defects with characteristic dimensions.
damage identification in composite structures [29] show the potential of this approach in NDT practice. The preliminary studies presented in [29] allow for definition of a testing procedure based on the performed experiments, i.e. it was observed that the reasonable and the most effective approach is an excitation of a composite structure with a multiharmonic signal composed of several harmonics corresponding to the resonant frequencies of vibrations of a tested structure. An excitation with multiple resonant frequencies allows to excite the corresponding modal shapes, and thus, the regions of the highest stress for each modal shapes are different. Since in viscoelastic materials mechanical stress is directly related with the amount of energy dissipation (see [28] for instance), these regions correspond with regions of maximal heating up. This justifies the mechanical excitation with multiple resonant frequencies, assuming that the presence and location of eventual defect is not a priori known. The main goal of this paper is to investigate defect detectability efficiency, i.e. an estimation of depth at which a defect is still detectable. For this purpose the tests were performed on specimens with FBHs of various depths and shapes. It was shown that taking into consideration a series of thermograms instead of a single thermogram, and their further common analysis allow to enhance defect detectability. Additionally, the post-processing of thermograms in order to enhance defect detectability is discussed. The efficiency of the proposed approach is confirmed by the promising results obtained from self-heating based vibrothermography (SHVT) NDT experiments.
mechanical excitation by an elastic wave in a sonic or ultrasonic [18–20] frequency range, which results in energy dissipation in a tested structure. In the case of existence of a defect the bigger energy dissipation is observed due to the friction between the faces of a defect or stress concentration in the surrounding area of a defect [20]. The physics of this phenomenon is widely discussed in [21,22]. This leads to the situation, where an increase of temperature is observed only at the location of a defect, which can be captured by IR camera. The excitation is usually performed in the contact way using ultrasonic boosters or actuators placed on the surface or integrated with a tested structure. Besides of sonic and ultrasonic heating sources used in VT, other types of excitation are successfully developed. The authors of [23] based their approach of the thermal excitation on electromagnetic induction applied to the carbon fibre of a tested composite structure, which allows for local heating. Another approach of thermal excitation of a tested composite structure is using the thermoviscoelastic effect, known in a literature as the self-heating effect, accompanied by mechanical vibrations of such a structure. The nature of internal heat generation is different than for classical ultrasonic VT, since the heating in this case is coming from hysteresis resulting from phase lag of stress and strain amplitudes during mechanical vibrations, and originating from an internal structure of a polymeric matrix of a tested composite instead of frictional heating phenomena occurring during ultrasonic excitation of a tested structure [21]. The ability of heat generation of polymeric composites subjected to vibrations was successfully used in several previous studies. In particular, the self-heating effect was used by the authors of [24], where they presented results of investigation on viscoelastic response of flat-bottom holes (FBHs) filled with a viscous material, while the specimen was subjected to mechanical vibrations in the ultrasonic frequency range. The authors of [25,26] used mechanical excitation on resonant frequencies to localize the crack and monitor its propagation. The idea on excitation of tested structures with resonant frequencies was also used by the authors of [27], where they used absorptive viscoelastic coatings on the tested metallic specimens in order to observe the self-heating effect during excitation of specimens with resonant frequencies in the ultrasonic frequency range. Previous experimental studies of the authors’ team on a self-heating temperature distribution during mechanical excitation with resonant frequencies starting from the fundamental frequency of vibrations [28] as well as initial studies on application of the self-heating effect to
2. Specimens preparation The specimens made of the glass E-fabric-reinforced 14-layered epoxy composite material purchased from Izo-Erg S.A. (Gliwice, Poland) were cut from a sheet with the thickness of 2.5 mm into strips with the length of 250 mm and the width of 10 mm. For the investigation of defect detectability efficiency FBHs were milled in the specimens according to the schemes presented in Fig. 1. The milling was performed on the two-axis milling machine Roland Modela MDX-20 (Hamamatsu, Japan) using ø1 diamond milling cutter (see Fig. 2) with the following milling parameters: milling speed in the planar directions of specimens’ surface was of 10 mm/s, milling speed in the thickness direction was of 0.5 mm/s, and a spindle speed was of 6500 rpm.
Fig. 2. Milling process of artificial defects in specimens. 761
Composite Structures 201 (2018) 760–765
A. Katunin, D. Wachla
Fig. 3. The specimens with milled FBHs prepared for SHVT testing.
As it was shown in Fig. 1, four shapes of FBHs were considered: notch-type (L), circular (C), square (S), and rectangular (R). For each defined shape of the FBH six depths were milled, starting from 0.25 mm and ending at 1.5 mm with a step of 0.25 mm. Following this, 24 specimens in total were tested using the presented approach. For ensuring appropriate thermal emissivity the specimens were covered with a silicone black matt heat-resisting enamel manufactured by Dragon Poland Sp. z. o.o. (Cracow, Poland). The exemplary specimens after painting with all considered shapes of FBHs are presented in Fig. 3.
Table 1 Resonant frequencies of vibration of tested specimens with FBHs, Hz.
Depth of FBH, mm
Shape of FBH
L
C
S
R
0.25
191.01 528.91 1069.14 185.94 501.95 1031.64 190.23 520.70 1065.62 164.45 456.64 922.26 150.00 414.84 833.98 110.94 307.42 586.72 1005.08
189.06 533.05 1053.12 190.23 517.97 1075.39 191.80 526.17 1076.95 188.28 518.75 1048.44 187.50 516.02 1044.53 187.11 518.36 1042.97
189.06 524.22 1057.81 191.02 515.62 1072.66 190.62 524.61 1071.48 185.55 517.19 1037.50 185.94 508.59 1033.98 189.50 525.00 1069.00
189.45 523.44 1055.86 183.20 491.02 1008.98 187.89 99.22 1056.64 186.33 514.45 1040.23 184.37 506.64 1024.22 184.00 503.50 1025.50
0.5
0.75
3. Experimental testing procedure
1
The experiments were performed on the own-designed test rig presented in Fig. 4. The testing procedure consisted of two steps: performing the classical modal analysis, and, after acquisition of resonant frequencies of a tested specimen, excitation of this specimen with a multi-harmonic signal composed from the harmonics with acquired resonant frequencies and with simultaneous observation of the specimen’s top surface using IR camera in order to register its thermal response. In the first step, the modal analysis of a specimen 4, clamped in a steel holder 8 with polycarbonate inserts allowing for thermal insulation of a specimen 4 from the holder 8, was performed. The loading was performed using the TIRA® TV-51120 electrodynamic shaker 1 through the stinger 2, connected with a specimen holder 3 and the shaker 1 by screw connections. The pseudo-random signal generated from the dedicated measurement and control software on the PC 13, and amplified using the TIRA® BAA 500 shaker amplifier 10 was used for specimen excitation in order to obtain wide-band frequency response and identify natural frequencies. The measurements were performed using the Polytec® PSV-400 scanning laser Doppler vibrometer (LDV) 7 connected with the Polytec® PSV-W-400 data acquisition and control unit 9. To separate vibrations of a specimen 4 from vibrations of the whole system a reference signal was acquired from the Polytec® PDV100 single point LDV 6 focused on the surface of the holder 3. The surface of a specimen 4 subjected to testing and the holder 3 were covered by reflective tape to ensure appropriate reflection of laser beams of LDVs 6 and 7. The measurements of vibration velocity of tested specimens were performed in 45 equidistant points defined on the specimen’s surface across its length. The frequency response of the
1.25
1.5
specimens was examined in the frequency range of 0 ÷ 1250 Hz with a resolution of 0.390625 Hz (3200 data points in the defined frequency range). In every case the testing was performed on the undamaged surface, i.e. the FBHs were located on the bottom surfaces of specimens. In the considered frequency range three resonant frequencies were registered, except the specimen with the notch-type FBH with depth of 1.5 mm, for which four resonant frequencies were registered. The collected resonant frequencies for tested specimens are stored in Table 1. The exemplary frequency response function (FRF) and corresponding modal shapes for one of the tested specimens are presented in Fig. 5. In the second step, the collected resonant frequencies for a given specimen were used for a construction of a multi-harmonic signal, where the harmonics with resonant frequencies were the components of this multi-harmonic signal. This allows for excitation of a specimen on multiple resonant frequencies simultaneously. Since the maximal selfheating temperature values are observable in the regions of stress concentration, which was proven theoretically and experimentally in previous studies (see [28] for instance), for every modal shape the maximal heating is observed near the clamps and in the locations
(a)
(b)
Fig. 4. Experimental test rig: a) general view, b) scheme. 762
Composite Structures 201 (2018) 760–765
A. Katunin, D. Wachla
Fig. 5. Exemplary FRF a) and corresponding modal shapes b)-d) of tested specimens.
corresponding to maximal deflection in modal shapes. Considering that the location of a damage is not known a priori, it is essential to excite the whole surface of a tested structure, and therefore, the mechanical excitation with multiple resonant frequencies is necessary. It should be noticed that the selected resonant frequencies allow for the thermal excitation of the whole surface of a tested specimen in the presented case, however the selection process is geometry-dependent (due to the relation of local stress concentrations in modal shapes with areas of selfheating effect appearance), which means that the considered resonance frequencies in the excitation signal should be adjusted to a specific case. The signal constructed in the above-presented way was generated from the own-developed application, and after switching the switch 12 was delivered to the shaker 1 through the National Instruments™ cDAQ9174 compact DAQ chassis with 4-channel ± 10 V 16-bit analog output module NI-9263 11. The thermal response was registered by the InfraTec VarioCam® hr IR camera 5 with a sampling rate of 2 frames per second. The test duration in each case was equaled 5 min, i.e. 600 thermograms were collected for each specimen.
(a) L
(b) C
(c) S
(d) R
Fig. 6. Identified thermal signatures for various types of FBHs with the depth of 1 mm.
4. Analysis of thermograms and defect detectability The collected thermograms for various types of FBHs allowed for clear identification the notch-type FBHs only, while for other types of FBHs the thermograms were not sharp enough to clearly identify thermal signatures of FBHs (see Fig. 6). Moreover, it was observed that the sharpness and detectability depends on moment of capturing the thermogram, i.e. with increase of testing time the progressive blurring of thermograms was observed (see Fig. 7). Additionally, due to the temperature fluctuations during testing and considering low temperature increase with respect to the ambient temperature, the analysis of a single thermogram for identification of a thermal signature of defect can be inefficient. Therefore, it is essential to consider multiple thermograms during such an analysis. The results presented in Fig. 7 clearly show that the effective time period for damage identification is the period of self-heating temperature increase, i.e. the period between starting of rapid self-heating temperature increase and reaching the thermal equilibrium between the generated and released heat of a tested specimen. Considering that the self-heating temperature growth depends on material properties of a tested specimen as well as on boundary and loading conditions, the effective time period will be also dependent on these properties and conditions, and should be determined for a given tested structure individually. To unify the method of determination of the effective time
period the following approach is proposed. The maximal self-heating temperature evolution curve is fitted using the polynomial function in order to obtain the best fitting of approximation function to experimental data. It is especially important to select appropriate order of the applied approximation function which allow for correct reflection of the measurement data when the rapid self-heating temperature growth starts and when the thermal equilibrium is reached. The results of the performed empirical study show that the best results for such a character of a self-heating temperature evolution curve can be obtained using a polynomial function of 8th order. The example of such an approximation is presented in Fig. 8a. For clarity, the same self-heating temperature evolution curve as one presented in Fig. 7a was taken into consideration. Next, the obtained approximation curve is subjected to numerical differentiation and the absolute value is taken of the resulting data. This operation allows for identification of desirable boundaries of the effective time period (see the exemplary result in Fig. 8b). The boundaries of the effective time period is defined by first two minima of the obtained function starting from the origin. Having the effective time period defined by the above-presented approach, in order to enhance defect detectability, especially for the cases with unclear thermal signatures (see e.g. Fig. 6b-d), it is essential 763
Composite Structures 201 (2018) 760–765
A. Katunin, D. Wachla
Fig. 9. Exemplary results of the processed thermograms for considered types of FBHs. Table 2 Results of defects detectability analysis using SHVT method. Number of detected defects, –
Depth, mm
Fig. 7. Self-heating temperature evolution during testing (a) with accompanying thermograms for different moments of time (b) for the notch-type FBH with the depth of 1 mm.
Shape of FBH
L
C
S
R
0.25 0.5 0.75 1 1.25 1.5
undetectable 5 6 8 9 9
undetectable
undetectable 4 6 7 7 7
undetectable
3 6 6 6
3 4 4 4
signatures of FBHs are better detectable. In order to analyze detectability of defects in the form of FBHs with various shapes and depths the analysis according the above-presented approach was performed for all considered specimens. The results of defects detectability for particular cases are presented in Table 2. The resulting images for notch-type FBHs for all considered depths are presented in Fig. 10. The obtained results show that the thermal signatures of the introduced defects become detectable starting from the cases of the defects depth of 0.75 mm. One should note that the FBHs were introduced on the bottom surface of a specimen with respect to the surface subjected to testing. This means that the method reveals a sensitivity to damage located on the depth of 1.75 mm, which corresponds with 60% of the specimen thickness. According to the obtained results (see Table 2) not all the defects were well detectable (except the notch-type FBHs for the highest considered depths). There are two reasons that may explain this phenomenon. Firstly, harmonics in the excitation signal, in spite of initial equal magnitude of them during generation, have different influence on the resulting temperature distribution
to add up the acquired thermograms for the defined effective time period. The collected thermograms were exported to Matlab® environment in the form of matrices of temperature values, and then cut for specific dimensions of the specimen. The resulting matrices were added up for the defined effective time periods unique for each considered case, and the logarithm of the resulting matrix was taken for additional enhancement. The exemplary results of this operation are presented in Fig. 9. As the examples the same specimens as those presented in Fig. 6 were considered. The presented cases in Fig. 9 show that the introduced FBHs, except the notch-type ones, are barely detectable. The reason of such results is a difference in stress distribution between the notch-type FBHs and other shapes of FBHs considered in this study, and thus, the magnitude of resulting deflection of specimens with these FBHs. This allows for dissipation of more thermal energy in the case of the notch-type FBHs with respect to other considered cases, and accordingly, the thermal
Fig. 8. Maximal self-heating temperature evolution curve with the polynomial approximation (a), and absolute value of derivative of the approximation polynomial with defined boundaries of the effective time period (b). 764
Composite Structures 201 (2018) 760–765
A. Katunin, D. Wachla
defects of aircraft composites. NDT&E Int 2003;36(6):433–9. [2] Avdelidis NP, Almond DP, Dobbinson A, Ibarra-Castanedo C, Maldague X. Aircraft composite assessment by means of transient thermal NDT. Prog Aerosp Sci 2004;40:143–62. [3] Katunin A, Dragan K, Dziendzikowski M. Damage identification in aircraft composite structures: a case study using various non-destructive testing techniques. Compos Struct 2015;127:1–9. [4] Ibarra-Castanedo C, Genest M, Servais P, Maldague SPV, Bendada A. Qualitative and quantitative assessment of aerospace structures by pulsed thermography. Nondestr Testing Eval 2007;22(2–3):199–215. [5] Pawar S, Peters K, Transient infrared thermography for damage evaluation in aerospace composites. In: Proceedings of SPIE 7649, Nondestructive Characterization for Composite Materials, Aerospace Engineering, Civil Infrastructure, and Homeland Security, 2010, 76491A. [6] Meola C, Boccardi S, Carlomagno GM, editors. Infrared thermography in the evaluation of aerospace composite materials Woodhead Publishing; 2017. [7] Santulli C. IR thermography for impact damage evaluation of automotive composites and components. NDT.net J 2003;9(9). [8] Choi S-H, Park S-K, Kim J-Y. NDE of advanced automotive composite materials that apply ultrasound infrared thermography technique. Int J Modern Physics: Conf Ser 2012;6:515–20. [9] Liu D, Malvern LE. Matrix cracking in impacted glass/epoxy plates. J Compos Mater 1987;21:594–609. [10] Vavilov VP, Burleigh DD. Review of pulsed thermal NDT: physical principles, theory and data processing. NDT&E Int 2015;73:28–52. [11] Kretzmann JE, Venter G, Schreve K. Non-destructive testing with transient thermography on composite materials. R&D J South African Inst Mech Eng 2016;32:35–43. [12] Meola C, Carlomagno GM, Squillace A, Giorleo G. Non-destructive control of industrial materials by means of lock-in thermography. Meas Sci Technol 2002;13(10):1583–90. [13] Ullman T, Shi Y, Aoki R, Capabilities of lock-in thermography for non-destructive inspection of fibre reinforced composites. In: 11th International Conference on Quantitative InfraRed Thermography, Naples 2012, pp. 1–11. [14] Ibarra-Castanedo C, Tarpani JR, Maldague XPV. Nondestructive testing with thermography. Eur J Phys 2013;34(6):S91–109. [15] Ley O, Godinez V. Non-destructive evaluation (NDE) of aerospace composites: application of infrared (IR) thermography. Karbhari VM, editor. Non-destructive evaluation (NDE) of polymer matrix composites. Techniques and applications Oxford, UK: Woodhead Publishing Ltd.; 2013. p. 309–34. [16] Reifsnider KL, Henneke EG, Stinchcomb WW. The mechanics of vibrothermography. In: Stinchcomb WW, Duke JC, Henneke EG, Reifsnider KL, editors. Mechanics of Nondestructive Testing. Boston, MA, USA: Springer; 1980. p. 249–76. [17] Mignogna RB, Green RE, Duke JC, Henneke EG, Reifsnider KL. Thermographic investigation of high-power ultrasonic heating in materials. Ultrasonics 1981;19:159–63. [18] Zweschper T, Dillenz A, Riegert G, Scherling D, Busse G. Ultrasound excited thermography using frequency modulated elastic waves. Insight – Non-Destructive Testing Condition Monit 2003;45:178–82. [19] Pieczonka L, Aymerich F, Brozek G, Szwedo M, Staszewski WJ, Uhl T. Modelling and numerical simulations of vibrothermography for impact damage detection in composites structures. Struct Control Health Monit 2013;20:626–38. [20] Yang R, He Y. Optically and non-optically excited thermography for composites: a review. Infrared Phys Technol 2016;75:26–50. [21] Mabrouki F, Thomas M, Genest M, Fahr A. Frictional heating model for efficient use of vibrothermography. NDT&E Int 2009;42:345–52. [22] Renshaw J, Chen JC, Holland SD, Thompson RB. The sources of heat generation in vibrothermography. NDT&E Int 2011;44:736–9. [23] Yang R, He Y. Polymer-matrix composites carbon-fibre characterization and damage inspection using selectively heating thermography (SeHT) through electromagnetic induction. Compos Struct 2016;140:590–601. [24] Montanini R, Freni F. Correlation between vibrational mode shapes and viscoelastic heat generation in vibrothermography. NDT&E Int 2013;58:43–8. [25] Magi F, Di Maio D, Sever I. Damage initiation and structural degradation through resonance vibration: Application to composite laminates in fatigue. Compos Sci Technol 2016;132:47–56. [26] Di Maio D, Magi F, Sever IA, Damage monitoring of composite components under vibration fatigue using scanning laser Doppler vibrometer, Experimental Mechanics, 2018, in press, doi: https://doi.org//10.1007/s11340-017-0367-y. [27] Vaddi JS, Holland SD, Kessler MR. Absorptive viscoelastic coatings for full field vibration coverage measurement in vibrothermography. NDT&E Int 2016;82:56–61. [28] Katunin A, Fidali M. Self-heating of polymeric laminated composite plates under the resonant vibrations: theoretical and experimental study. Polym Compos 2012;33:138–46. [29] Katunin A. A concept of thermographic method for non-destructive testing of polymeric composite structures using self-heating effect. Sensors 2018;18:74.
Fig. 10. Results of defect detectability for notch-type FBHs for various depths considered in the study using SHVT method.
during excitation, i.e. the lower frequency of the harmonic the better thermal response is obtained. This explains difficulties with detectability of 2nd and 8th FBHs in all tested cases (see Figs. 9 and 10), which location corresponds mainly with maximal deflections of the third modal shape (cf. Fig. 5d). Secondly, the shape of FBHs has an influence on the stress distribution, and, considering that the amount of dissipated heat during excitation is proportional to stress (see [28] for instance), the differences in defects detectability for the same depths and various shapes of FBHs (see Table 2) can be explained. 5. Conclusions The efficiency of the proposed SHVT NDT method is analyzed basing on experiments performed on PMC specimens with FBHs of various depths. The FBHs were milled on the opposite surface with respect to the tested surface, which allows to simulate subsurface structural defects. The presented SHVT method together with the proposed postprocessing techniques of a collected sequences of thermograms based on estimation of a thermal gradient of the self-heating temperature evolution curves allows for enhancement of detectability of considered defects. The performed analysis of detectability shows satisfactory results, i.e. the defects become detectable starting from 1.75 mm of depth, which corresponds with 60% of the thickness of tested structures. Starting from 1.5 mm of depth (the distance from top surface of a tested specimen to the bottom of the milled FBH) the detectability was unchangeably fine, which proves the efficiency of the proposed approach. In many cases of FBHs their thermal signatures were ambiguous, which does not allow for their classification as a detected defect. However, the existence of such thermal signatures allows supposing that using advanced mathematical tools and image processing methods their detectability can be improved. In the next studies planned within the currently realized research project such an improvement is planned. Acknowledgements The results presented in this paper were obtained within the framework of research grant No. 2015/17/D/ST8/01294 financed by the National Science Centre, Poland. References [1] Avdelidis NP, Hawtin BC, Almond DP. Transient thermography in the assessment of
765