Quantitative Determination of Porosity by Active Thermography

Quantitative Determination of Porosity by Active ThermographyG. Hendorfer, G. Mayr, G. Zauner, M. Haslhofer, and R. Pree Citation: AIP Conference Proceedings 894, 702 (2007); doi: 10.1063/1.2718039 View online: http://dx.doi.org/10.1063/1.2718039 View Table of Contents: http://scitation.aip.org/content/aip/proceeding/aipcp/894?ver=pdfcov Published by the AIP Publishing

This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions.

Downloaded to IP: 193.52.108.46 On: Wed, 04 Dec 2013 15:10:23

QUANTITATIVE DETERMINATION OF POROSITY BY ACTIVE THERMOGRAPHY

G. Hendorfer, G. Mayr, G. Zauner, M. Haslhofer and R. Pree

University of Applied Science, StelzhamerstraBe 23, A-4600 Wels, Austria

ABSTRACT. We present a new approach based on Active Thermography by which it is possible to produce images of porosity in Carbon Fibre Reinforced Plastics that correspond striking well with results obtained by ultrasonic testing. We applied the method of Pulsed Thermography by using flashes of light for the generation of heat. The evolution of surface temperatures which depends on the diffusivity of the sample and on the sample's geometry could be well fitted by means of a heat conduction model. Images of porosity are generated by means of an evaluation by which the influence of the light intensity distribution on the data is eliminated. Thus corrections with respect to the lateral distribution of the heat generation are not necessary, nor are emissivity corrections. Corrections due to the influence of geometry, however, had to be taken into account. The quantitative evaluation of the porosity is based on its linear relation to the diffusivity. Images of porosity obtained thermographically are compared with corresponding images as obtained by state of the art ultrasonic testing. We show that the thermographic method exhibits a better sensitivity and resolution of porosity. Measurements have been performed on samples with average porosities between 1 and 5 %. With some modifications the method can be applied for the quantitative characterization of delaminations of multi-layered samples as well.

Keywords: active thermography, porosity, CFRP PACS: 87.63.Hg

INTRODUCTION

In the aircraft industry Carbon Fibre Reinforced Plastics (CFRP) are increasingly used for interior as well as external components, due to the necessity to reduce material and weight [1-3]. Additionally, the desired mechanical properties such as stiffness and compression strength have to be guaranteed, thus giving rise to the testing of material quality as of uttermost importance. A critical quantity in CFRP-materials is their porosity, which may occur as undesired side effect of the manufacturing process. Porosity, as well as other material properties, should be tested by so called 'in-line-methods' in order to be able to adjust critical process conditions without appreciable time delay. State of the art technology has up to now measured porosity by means of ultrasonic testing [4]. Recently attempts have been undertaken, however, to measure porosity by thermographic methods as well [5-10]. Although several authors have shown that this can be done in principle, there is a common estimation of experts in this field that a quantitative evaluation is difficult and the results of thermographic experiments may be influenced by the special conditions of the heat excitation as well as by surface properties. Especially for an extended range of porosity levels ultrasonic methods are considered more reliable.

CP894, Review of Quantitative Nondestructive Evaluation Vol. 26, ed. by D. O. Thompson and D. E. Chimenti © 2007 American Institute of Physics 978-0-7354-0399-4/07/S23.00

702

This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions.

Downloaded to IP: 193.52.108.46 On: Wed, 04 Dec 2013 15:10:23

In this paper, we present a procedure to generate images of porosity in CFRP plates. Our method is based on Active Thermography. The duration from the light flash, from the evaluation to the completion of the porosity image does not last longer than about 5 seconds. Thus a very rapid testing seems possible, which is interesting especially in automated production cycles projects and promises a significant reduction in cost. The accuracy and speed of this method has been achieved because the influence of the heat absorbed by the samples has been eliminated. Thus the results are independent of the lateral distribution of the light intensity and consequently no corrections with regard to that are necessary.

The samples provided by Austrian aircraft components manufacturer FACC are rectangularly shaped with a diameter of about 600 mm and a thickness of 2 mm and 4 mm, respectively. Five samples with 2 mm thickness are presented in this paper. The nominal porosities of these samples are 1%, 2%, 3%, 4% and 5% respectively. All samples have been investigated by ultrasonic-testing, too. Porosity images deduced are given in fig. 4 and discussed below.

EXPERIMENTAL SETUP

The experiments were setup, as illustrated in Fig. 1. The samples are illuminated from one side by a flash lamp which delivers about 1500 J per pulse. The duration of illumination is about 10 ms. The distance of the flash lamp to the samples is about 400 mm. The light is normally incidental to the sample surface. On the back side the surface temperatures are measured by means of an infrared camera FLIR Thermacam PM695. Its temperature resolution is 60 mK. The time resolution per pixel is 40 ms. According to the geometrical conditions of our experimental setup our images exhibit a spatial resolution of 1.5 mm.

MEASUREMENTS AND DISCUSSION

Each pixel in the IR-detector-array correlates to a well defined area of 2.25 mm2 at the sample backside, at which the evolution of the surface temperature after excitation is evaluated. A typical time-dependence of the rear surface temperature is given in Fig.2.

Heat source

FIGURE 1. Experimental setup.

t o , p, c

703

This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions.

Downloaded to IP: 193.52.108.46 On: Wed, 04 Dec 2013 15:10:23

FIGURE 2. Temperature versus time after excitation.

The zero of the time axis represents the excitation of the sample. After a rapid increase in temperature a maximum is reached, followed by a slow decrease in temperature. The latter is dependent mainly on the transfer of heat from the sample to the surrounding air, which depends on the specific heat capacity and the density of the sample, involved, in addition to the heat transfer coefficient. The increase in temperature, however, is determined by the diffusivity a and the sample thickness, a is given by

a = "klp.c, (1)

where "k represents the thermal conductivity, p the density and c the specific heat capacity. Porosity obviously has an effect on the diffusivity. This can be seen in Fig.2, where different samples lead to different rates of the temperature increase. The decrease in temperature, however does not exhibit a clear dependence on porosity levels.

Although in Fig.2 there are different curves due to different porosity or diffusivity levels, respectively, it is not possible to determine these quantities unequivocally. Instead we apply a physical model of one-dimensional time dependent heat conduction, given by

dlldi = a.32T/32x. (2)

The solution of this equation at position x is given by

T = T0 + [(Q/A)/(p.c.(4:ra.t)1/2)] .exp(-x2/4a.t), (3)

where T0 denotes the ambient temperature, Q the amount of heat deposited on the front side of the samples and A the illuminated sample area, respectively. Q/A can be understood as intensity of excitation, a quantity which will exhibit a distinct decrease in lateral direction from the centre of the illumination.

704

This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions.

Downloaded to IP: 193.52.108.46 On: Wed, 04 Dec 2013 15:10:23

FIGURE 3. In AT versus 1/t.

We plotted ln(T-To) versus 1/t, substituting for x the sample thickness. These plots are shown in Fig. 3. In this way we separate the effects due to excitation from the effects due to diffusivity. By evaluating the slope of these plots between times, corresponding to 10 and 90 % of the signal maximum, respectively, we are able to determine the diffusivity with a resolution of 0,05%. This kind of evaluation is done for each pixel and the diffusivity determined quantitatively is plotted as an image by means of a false-colour representation. In Fig. 4 a comparison between the ultrasonic and the thermographic method is given for the samples with 2 mm thickness. Above images obtained by ultrasonic-testing and below images obtained thermographically are given.

The results of these two methods obviously correspond very well as far as the spatial dependence of porosity is concerned. The thermographic images exhibit even more details than ultrasonic ones. This can be seen even more distinctly in samples with low porosity values. A significant difference between the methods exists with respect to the testing speed. While thermographic images are produced within 5 seconds, it requires several minutes to produce comparable images by ultrasonic methods.

A great advantage of this approach is the fact that the effects of the light intensity and its decrease in a direction normal to the light propagation are excluded. This is demonstrated in fig. 5, where four segments of equal area have been illuminated in such a way that the maximum of the light intensity was positioned in the centre of each segment, respectively. Another image has been produced where the centre of the light intensity was coinciding with the centre of the plate itself. The image of fig. 5 obtained by composing the four segments to one picture is identical to the image obtained by investigating the whole plate at once.

*wr* m

• few--,-...

FIGURE 4. Comparison of thermographic and ultrasonic images.

705

This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions.

Downloaded to IP: 193.52.108.46 On: Wed, 04 Dec 2013 15:10:23

FIGURE 5. Themographic image obtained by composing four segments of a sample to one picture.

In Fig. 6 the correlation between nominal porosity and the diffusivity, deduced from the thermographic method is shown. The nominal porosity values are obtained from ultrasonic testing by averaging all values available per sample. The diffusivity values in Fig. 6 are obtained accordingly, that is, by averaging all individual diffusivity values per sample. The resulting linear relation between nominal porosity and average diffusivity can serve as calibration for the quantitative evaluation of the inhomogeneous spread of porosity.

From Fig. 6 we deduce that the higher the porosity, the lower is the corresponding diffusivity, in accordance with the 'Dethermalization Theory' [10]. With x denoting the porosity as fraction of the void volume to the plate volume, the sample density can be approximated by

p = p0/(l+x) (4)

Due to the linear correlation of Fig. 6 we can deduce that the thermal conductivity exhibits a linear decrease with x, where, obviously the equation

-d^/dx>^(x=0) (5)

must be satisfied, since c does not depend on x essentially.

„ 1

" 1

>

b

' 1 2 3 4 5 porosity[%]

FIGURE 6. Correlation between porosity and diffusivity.

2

0

8

e

4

706

This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions.

Downloaded to IP: 193.52.108.46 On: Wed, 04 Dec 2013 15:10:23

Thermal images

Diffusivity images

FIGURE 7. Bonding defects between CFRP-panels - thermal and diffusivity images.

Evaluating Bonding Defects

The method described above is very well suited also for indicating and evaluating bonding defects between CFRP-plates, as can be seen in Fig.7. The images above are thermal ones with the drawback of artifacts like reflections, hot spots or emissivity effects which may conceal real defects. The images below are diffusivity images indicating bonding defects without disturbing artifacts.

SUMMARY AND OUTLOOK

Summarizing this paper outlines an approach for producing images of porosity in CFRP materials. The method is based on Active Thermography. All samples investigated have been investigated by means of ultrasonic testing, as well, and images of the porosity have been produced. The ultrasonic method is based on ultrasonic attenuation, which has, up to now represented the state of the art testing method in the aircraft industry. We have shown that images produced applying the thermographic method bring about the same results on porosity as the ultrasonic method. As a matter of fact the thermographic approach turns out to give much more detailed information with regard to the spatial distribution of porosity, especially in samples with low porosity values. An experimentally obtained linear relation between porosity and diffusivity can be used as calibration in quantitative analysis. The most striking advantage of that approach, however, is the increase in the speed of testing. This advantage alone makes the method of thermographic testing a serious alternative to ultrasonic testing, at least for a sample thickness up to 4 mm. The increase of testing speed is due to the fact that corrections to compensate for the intensity distribution of the exciting light and for emissivity effects are not necessary. For more complex geometries the temperature data would have to be combined with geometric data in order to yield spatially resolved porosities. Although the evaluation is more complex than it is with plates, current research indicates that it may be possible to produce images of porosity as well.

707

This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions.

Downloaded to IP: 193.52.108.46 On: Wed, 04 Dec 2013 15:10:23

ACKNOWLEDGEMENTS

This work is supported by FACC (Fischer Advanced Composite Components), UAR (Upper Austrian Research) and FFG (Forschungsforderungsgesellschaft, Austria) to which we owe our thanks.

REFERENCES

1. X. P.V. Maldague, Theory and Practice of Infrared Technology for Nondestructive Testing, Wiley Series in Microwave and Optical Engineering, Kai Chang series editor, John Wiley & Sons, Inc., New York, 2001, p. 250

2. Nondestructive Testing Handbook, Volume 3 Infrared and Thermal Testing, Editors: X. P.V. Maldague and P.O. Moore, American Society for Nondestructive Testing, 2001, p. 490

3. A. Kiick, R. Heinrich, Int. Conf, on Joining Ceramics, Glass and Metal, Munich, 2002 4. E. A. Birt, R.A. Smith, Insight, Vol 46, No 11, 2004 5. Nondestructive Testing Handbook, Volume 3 Infrared and Thermal Testing,

Editors: X. P.V. Maldague and P.O. Moore, American Society for Nondestructive Testing, 2001, p. 495

6. J.N. Zalameda, Journal of Composites Technology & Research Vol. 21, No 2, 1999 7. P.H. Johnson, W.P. Wmfree, E.R. Long, S.M. Kullerd, N. Nathan, R.D. Partos, Review

of Progress inQNDE, 11B, 1555 (1992) 8. A. Ciliberto, G. Cavaccini, O. Salvetti, M. Chimenti, L. Azzarali, P.G. Bison, S.

Marinetti, A. Freda, E. Grinzato, Infared Physics & Technology Vol. 43, 2002 9. X. P.V. Maldague, Theory and Practice of Infrared Technology for Nondestructive

Testing, Wiley Series in Microwave and Optical Engineering, Kai Chang series editor, John Wiley & Sons, Inc., New York, 2001, p.527

10. H.I. Ringermacher, D.R. Howard, R. S, Gilmore, Review of Quantitave Nondestructive Evaluation Vol.21, ed. by D.O. Thompson and D.E. Chimenti, 528-535, 2002

708

This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions.

Downloaded to IP: 193.52.108.46 On: Wed, 04 Dec 2013 15:10:23