·Home ·Table of Contents ·Materials Characterization and testing

Lock-in thermography for depth resolved defect characterisation A. Dillenz, D. Wu, K. Breitrück, and G. Busse University of Stuttgart Institute of Polymer Testing and Polymer Science (IKP) Department of Non-Destructive Testing (ZFP) Pfaffenwaldring 32 D-70569 Stuttgart, Germany e-mail address of lead author: dillenz@ikp.uni-stuttgart.de Contact

Abstract

Lock-In thermography allows for fast and reliable defect detection in components. Thermal waves are used as probes to reveal local material inhomogenities thereby indicating subsurface faults in the examined material.
However, there are basic limitations concerning the detection of hidden defects in thick materials with thermal waves: the penetration depth and thus the depth range of thermal waves is limited and depends on thermal diffusion length. This limitation can be turned into an advantage if several thermal waves with different wavelengths are used: one can achieve depth resolved imaging of a component.
In this paper defect-selective and depth resolved defect recognition is presented. Lock-In thermography using monofrequent sinusoidal excitation is compared to the transient excitation technique which is combined with phase evaluation.

INTRODUCTION

At present mainly two different procedures of active thermography are being used: Pulse and Lock-In Thermography. With pulse thermography [1], [2] the examined material is warmed up with a short energy pulse (light, eddy current, or ultrasonic pulse) and the heat response is recorded after a certain time. The result is an infrared image which indicates material defects in different depths, but usually it shows also local inhomogeneities of heating as well as of the material surface.
Remote material testing with monofrequent thermal waves by Lock-In Thermography [3-6] uses sinusoidally modulated excitation signals in order to derive information from the observed phase and magnitude of this wave. The phase angle has the advantage that it is independent of local variations of illumination or of surface emissivity [7]. However, if the thermal characteristics of the material are unknown or if defects in different depths have to be detected, then the phase images have to be taken at different frequencies which causes several measurements. A combination of pulse excitation and phase evaluation [8] can combine their advantages. The result is the long pulse thermography with Fourier phase evaluation or wavelet analysis [9] , as described in this paper. These procedures require an arithmetic performance which is provided by modern personal computers.

EVALUATION METHODS

In the following we present the Lock-In-method as well as the long-pulse method with different evaluation algorithms.

Lock-In Technique
The simplest evaluation algorithm is based on the four point procedure described previously [6]. This procedure is based on the fact that a sinusoidal thermal wave is generated on the whole sample. As the thermography signal also has a sinusoidal shape, the amplitude and phase can be calculated from four data points per modulation period. This evaluation is being used since several years in the AGEMA Thermovision 900 Lock-In system.

Transient Thermography with phase evaluation
Transient signals differ from sine signals by the fact that they contain several frequency components.If one heats a sample with a long pulse and records both the heating and the subsequent cooling process with the infrared camera, then phase and amplitude information can be derived from the Fourier transform of the local time sequence. As this transformation can be performed at various frequencies one obtains images that are similar to Lock-In thermography images taken at the same frequencies. Fig. 1 shows a typical temperature evolution on the surface of a steel sample that was heated up for 15 seconds, and the spectrum derived therefrom. In this case the frequency range from major peaks at 0.03 Hz up to the second local maximum at 0.11 Hz can be used for phase evaluation. Areas of higher frequencies are too noisy to be evaluated in phase and amplitude.

Fig 1: Time dependence of temperature(left) and spectrum(right)

Wavelet evaluation


Fig 2: Scalogram of temperature curve shown in Fig. 1. Red denotes high energy density, blue colour indicates low density.

The wavelet transformation uses a similar basic approach where the measuring signal is convoluted with a "wavelet". The difference to the Fourier transform is mainly that the frequency range is limited and additionally the time interval. From this the information about the defect depth can be determined not only by using the contained frequency proportions, but also directly over the time when defects in different depths show their contrast maximum.
The local time-frequency energy density can be displayed by the Scalogram [10] where the abscissa denotes the time and the ordinate the frequency. One point in the Scalogram determined by time and frequency is taken for an imaging wavelet transformation.

RESULTS AND COMPARISON

Depth profiling of a glass fibre composite sample with integrated heater foils


Sample configuration
The laminate consists of ten glass fibre laminate layers with epoxy resin matrix where heating foils are integrated at different depths to act as thermal wave transmitters (Fig. 3). The depths of the heating foils were selected in such a way that their distances to the rear side of the sample and to its front surface were different. This way a broad range of the phase angle is possible over the whole sample thickness. The heating foil next the surface is marked as number one, the deepest one with four. The sample was manufactured in the wet laminate procedure. The 50 x 50 mm sized heating foils consist of 0.15 mm thick polyamide with heating stripes on them.

Fig 3: Configuration of the heating foil sample. Foil 1 is closest to the front surface while foil 4 is deepest

Results
Fig. 4 shows the embedded heating foils in the phase image of Lock-In thermography. The difference between the heating foils in different depths is obvious. Similar results can be achieved with long pulse phase imaging as shown in Fig. 5.
The wavelet analysis does not show a usable signal in the phase image. The wavelet amplitude image permits a depth allocation, but without the advantages of the phase images as independence from surface inhomogeneity and local heating up variations.

Fig 4: Lock-in phase image f=0.03 Hz (front side). Measuring duration: 99 seconds. Excitation power: 1.5 W each heating foil.

Fig 5: Phase image long pulse excitation f=0.03 Hz (front side). Measuring duration: 30 seconds.

Fig 6: Wavelet phase (left) and amplitude image using a Morelet wavelet [8] within a time- frequency window located at (0.07 Hz / 3 seconds). The temperature data used was the same as evaluated with Fourier phase (results shown in Fig. 5). Due to the high memory consumption by the wavelet transform process on large images only a part of the sample is shown. The heating foils (denoted by numbers according to their depths) are displayed only half.

Depth profiling with ultrasound excitation
The purpose of the experiments described below was to investigate how well these results can be transferred to arrangements where acoustic excitation is used for hysteresis heating of defects at ultrasound frequencies. If the amplitude of ultrasound is modulated at a low frequency, one performs "Ultrasound Lock-In Thermography" (ULT) [11-13] where the local hysteresis acts as a thermal wave transmitter. Acoustic excitation is suited as well for long pulse heating.

Sample configuration


Fig 7: Configuration of the blind hole steel sample. Metal cylinders were bonded into the holes with a diameter of 36 mm and 18 mm to serve as ultrasonic absorbers. Front of sample was painted black.

We used a 20 mm thick steel sample provided with blind holes (diameter ranging from 9 to 36 mm, see Figure 5) at the bottom of which small steel cylinders were bonded with lossy polymer material which acted as local ultrasound absorber at different depths underneath the surface. This sample was excited using an ultrasound sonotrode at a frequency of 20 kHz in order to heat the bond between the cylinder and the plate due to their mechanical hysteresis. From the phase image one can determine the propagation time of the thermal wave that is emitted from the bottom of the blind hole where the polymer bond is located. The formation of standing ultrasound waves in the sample could influence both amplitude and phase of the thermal wave. Therefore a suitable excitation frequency was selected in order to avoid this effect.

Results
The long pulse measurement included a heating period of 15 seconds and afterwards a cooling period for another 15 seconds. In the phase images of both lock-in and long-pulse imaging the depth of the blind holes is detected. The phase contrast is better in lock-in phase image though excitation power was below the one used for the pulse phase image. The quality of the wavelet images corresponds to the one described above.

Fig 8: ULT phase image and profile of second row of holes. Lock-in frequency 0.03 Hz. Excitation frequency 19.4 kHz. Excitation power: 800W.	Fig 9: Ultrasound long pulse phase image and profile of second row of holes. Lock in frequency 0.03 Hz. Excitation frequency 19.4 kHz. Excitation power: 1200W.	Fig 10: Wavelet phase image at 0.06 Hz and 15 sec.
		Fig 11: Wavelet amplitude image at 0.06 Hz and 15 sec.

Depth resolved characterisation of an impact damage
After dealing with artificial samples the long pulse technique will now be applied to a realistic example.Fibre reinforced materials are sensitive to impact damage which is often only barely visible at the sample surface. However, inside the sample or at its rear surface a serious damage may occur. Damages are defective-selectively detected by sonic excitation [11-13]. Fig. 12 shows the phase analysis of a long pulse measurement with an evaluation frequency of 0.31 Hz where thermal diffusion length and depth range are small [14-16]. The damage appears only as delamination of the fibres near the surface. The development of the defect image points to a fibre orientation of 45 degrees. As typical for impact damages a butterfly pattern appears. In Fig. 13 a lower evaluation frequency was used. The damage pattern appears more diffuse, but the apparent damage remains the same. Fig. 14 shows also deeper-located damage areas at an analysis frequency of 0.05 Hz. Additionally to the 45 degrees oriented delamination there occurs another delamination perpendicularly to it. This corresponds to the next laminate layer with a fibre orientation of 135 degrees. Thus a depth profile of the damage area can be achieved within only one measurement.

Fig 12: Long-pulse phase, 0.31 Hz.

Fig 13: Long-pulse phase, 0.12 Hz.

Fig 14: Long-pulse phase, 0.05 Hz.

DISCUSSION

Depth-resolved measurements with different types of excitation were performed. Electrical, sonic, and light excitation were analysed in each case with Lock-In and with the pulse phase technique. Due to the strong attenuation of thermal waves, a depth resolved image of structures and defects can be obtained. The measuring time can be reduced clearly by the use of long pulse excitation, as several frequencies are contained in one measurement. The use of the wavelet analysis is limited by the low signal to noise ratio. However, this restriction might be less serious if one uses a FPA camera with a high thermal resolution.

ACKNOWLEDGEMENT

The authors are grateful to R. Aoki, DLR, Stuttgart for providing the sample used for Fig. 12 to Fig. 14.

REFERENCES

1. R. L. Thomas, L.D. Favro, P. K. Kuo, T. Ahmed, Xiaoyan Han, Li Wang, Xun Wang and S.M. Shepard, Pulse-Echo Thermal-Wave Imaging for Non-Destructive Evaluation,Proc. 15th International Congress on Acoustics, Trondheim, Norway, June 26-30, 1995, pp. 433- 436.
2. L. D. Favro, Xiaoyan Han, Zhong Ouyang, Gang Sun, Hua Sui, and R. L.Thomas, Infrared imaging of defects heated by a sonic pulse, Rev. Sci. Instr., 2000.
3. G. M. Carlomagno, P. G. Berardi, Unsteady thermotopography in non-destructive testing, Proc. 3rd Biannual Exchange, St. Louis/USA, 24.-26. August 1976, pp. 33-39.
4. J. L.; Merienne E.; Danjoux R.; Egee M.: Numerical system for infrared scanners and application to the subsurface control of materials by photothermal radiometry, Infrared Technology and Applications, SPIE Vol. 590 (1985) p. 287
5. P. K. Kuo, Z. J. Feng, T. Ahmed, L. D. Favro, R. L. Thomas, J. Hartikainen, Parallel thermal wave imaging using a vector lock-in video technique, Photoacoustic and Photothermal Phenomena, ed. P. Hess and J. Pelzl. Heidelberg: Springer-Verlag. (1987) pp. 415-418.
6. G. Busse, D. Wu, and W. Karpen, Thermal wave imaging with phase sensitive modulated thermography, J- Appl. Phys. 71, 1992, pp. 3962-3965.
7. A. Rosencwaig, G. Busse, High resolution photoacoustic thermal wave microscopy,Appl. Phys. Lett. 36, 1980, pp. 725-727.
8. X. Maldague, S. Marinetti, Pulse Phase Infrared Thermography, J. Appl. Phys. 79 [5], 1996, pp. 2694-2698.
9. F. Galmiche, X. Maldague, Pulsed Phased Thermography with the Wavelet Transform, QNDE, Montreal,(1999).
10. Stephane Mallat, A wavelet tour of signal processing, Academic Press, San Diego, 1998.
11. R. B. Mignogna, R. E. Green, J. Duke, E. G. Henneke, K. L. Reifsnider, Thermographic investigations of high-power ultrasonic heating in materials, Ultrasonics 7, 1981, pp. 159- 163.
12. K. F. Stärk, Temperaturmessungen an schwingend beanspruchten Werkstoffen, Werkstofftech. 13, 1982, pp. 333-338.
13. J. Rantala, D. Wu, and G. Busse, Amplitude modulated lockin vibrothermography for NDE of polymers and composites, Research in Nondestructive Evaluation 7, 1996, p. 215.
14. Rosenwaig, Photoacoustic microscopy, American Lab 11, 1979, pp. 39-49.
15. G. Busse, Optoacoustic phase angle measurement for probing a metal, Appl. Phys. Lett. 35, 1979, pp. 759-760.
16. A.Lehto, J. Jaarinen, T. Tiusanen, M. Jokinen, M. Luukkala, Magnitude and phase in thermal wave imaging, Electr. Lett. 17, 1981, pp. 364-365.

© AIPnD , created by NDT.net

|Home|    |Top|