HIGH-SPEED TIME AVERAGE DIGITAL HOLOGRAPHY FOR NDT OF CURVED SANDWICH STRUCTURES

Transcription

1 Proceedings of the National Seminar & Exhibition on Non-Destructive Evaluation NDE 2011, December 8-10, 2011 HIGH-SPEED TIME AVERAGE DIGITAL HOLOGRAPHY FOR NDT OF CURVED SANDWICH STRUCTURES Binu P. Thomas Engineer SF, EXMD/SDEG, STR Entity, Vikram Sarabhai Space Center, ISRO Post Thiruvananthapuram , Kerala, India. ABSTRACT Detection of debonds at sharply curved edges on sandwich structures is found difficult using conventional nondestructive testing (NDT) techniques. The advantages of digital holography and time average holographic interferometry were utilized to evolve a new approach to detect debonds on curved sandwich structures which is reported in this paper. Software developed for the high-speed recording, generation and storage of time average digital holograms within a fraction of a second per frame, made possible the capture of mode shape for every 1 Hz frequency increment during the frequency sweep. Thus the local resonance of debond region is easily and quickly visualized with zero chance of missing any such signatures and thus defects. The basic theory, procedure, software details and results are reported in this paper. Keywords: Digital holography, vibration modes, interferometry, time average 1. INTRODUCTION Honeycomb sandwich structures have a variety of engineering applications due to its high specific strength. It consists of metallic or composite skin sheet bonded on either side of metallic honeycomb core using film adhesive layers. A common defect found in these structures is the separation of adherents, known as debond. Skin-to-core debond affects the structural integrity of sandwich structure and reduces its overall stiffness and strength. Moreover, under repeated loading, debonding can propagate and trigger in-service failures. As many structures used in space applications are made of sandwich materials requiring high level of reliability, detection of debonds are very important. Many non-destructive testing techniques have been developed to detect and assess the debonds. Because of the complex geometry of honeycomb core, most of the conventional techniques like X-ray radiography and ultrasonic technique failed as practical NDT techniques. Thermography has the challenge to separate defect-related effects on the surface temperature from effects due to the complex sample shape like curvatures or edges [1,2]. Shop floor techniques like Fokker Bond and Wood pecker methods are time consuming point techniques and difficult to apply at sharp curved edges, which called for advanced NDT methods. Shearography is the latest candidate among the optical techniques, which has been successfully developed as a practical NDT tool in the recent years [3]. The high-speed time-average digital holography technique reported in this paper has lot of advantages over the shearography techniques. The high sensitivity to deformation, large coverage area and high quality fringes with low speckle noises are a few of the advantages, which guarantees fast and cent percentage debond detection. This paper gives details of the basic theory of the technique, experimental details, validation, vibration stressing method, holographic fringe signature analysis and application test results. 2. TIME AVERAGE DIGITAL HOLOGRAPHY The experimental setup for recording time-average digital hologram is same as that of double exposure digital holography [4] as shown in Figure 1. The laser beam is split into two using a beam splitter. One beam guided by an optic fiber is used to illuminate the object. The reference beam is guided through another fiber and is made to fall directly on the CCD sensor of the digital camera without lens. The object and reference waves arrive at the hologram plane and interfere with each other directly on the surface of the CCD sensor. The interference pattern recorded thus, called the digital hologram is reconstructed numerically in a computer. In time average digital holographic interferometry, the test object is allowed to vibrate while the camera is being exposed. The exposure time is made greater than the period of vibration so that the hologram records time averaged complex amplitude of light scattered by the object to the hologram plane.

2 NDE 2011, December 8-10, where T is the exposure time interval and M T (x,y) is known as the characteristic function. If the exposure time is long com pared to the period of vibration then M T is given by [6] as, Fig. 1 : Setup for recording hologram If d o (x, y) is the amplitude of mechanical vibration at location (x, y) and ω is the circular frequency of vibration then the displacement of a point on the vibrating object at time t is given by the relation [5] d(x, y, t) = d o (x, y) sin ωt. The phase shift ΔΦ of the light scattered from this point is then a function of time, which is given as [5] (ΔΦ(x, y, t)=k. d o (x, y) sin ωt where k is the sensitivity vector. Now let O (x, y) = O (x, y) e -iφ(x, y) represent complex amplitude of the scattering light when the object is stationary, where O (x, y) and Φ(x, y) are the amplitude and phase respectively of the scattering light at a point (x, y). Then complex amplitude of the scattering light of the vibrating object at any instant t, is given by [5] -i [Φ(x, y)+k. do (x, y) sin ωt ] O (x, y, t)= %O (x, y)%e The complex amplitude u (x, y) of the wave reconstructed by the hologram is given by [6] where J 0 is the zero order Bessel function of the first kind [6]. The intensity in the reconstructed image is then given by [6] as I(x,y) = u(x,y). u(x,y)* = O(x,y) M T (x,y) 2 where J 02 (Z) Z = k. d o (x, y). Hence the intensity of the time average holographic interferogram is proportional to the squared zero order Bessel function. The dark fringes, at which the intensity drops to zero, correspond to the zeros of the function J 02 (Z) and the bright fringes to its maxima [6]. The nodes of the vibrating object (for which the amplitude of oscillation is zero) yield the maximum possible value for the squared zero-order Bessel function, and therefore appear brighter than any other feature in the holographic interferogram. Hence the fringe pattern recorded through time average holography visually indicates where the vibrations are occurring and how strong they are at any given point on the surface of the object. 3. VIBRATION STRESSING Piezo shaker is used for vibrating the object. Piezo shaker system of make M/s isi-sys, Germany, used in the present study is shown in figures 2a and 2b. Function generator generates the excitation frequencies. The output from the function generator is fed to the dual channel bipolar amplifier which in turn connected to the piezo shaker. Piezo shakers are mounted on the object surface using a vacuum adaptor. Fig. 2 : (a) Power amplifier and vacuum pump unit Fig. 2 : (b) Piezo shaker with vacuum adaptor

3 120 Binu P. Thomas : Proceedings of the National Seminar & Exhibition on Non-Destructive Evaluation Fig. 3 : HDigitalRT environment with Time average module 4. NUMERICAL RECONSTRUCTION SOFTWARE The environment of the numerical reconstruction software HDigitalRT [7] is shown in figure 3. Time-average digital hologram is recorded while the object is vibrating and is reconstructed by taking the Fresnel transform of the hologram [7] whose intensity is proportional to the squared zero order Bessel function. The exposure time should be quite larger than the period of vibration. 5. EXPERIMENTATION VERDI 6W laser was used in the experiment at 1 W power level with PixeLINK CMOS camera and HDigitalRT for hologram recording and reconstruction. The laser was kept on a wheeled cart and bifurcated optical fiber was used for transmitting beams for illuminating the object and for reference beam as shown in figure Validation A few experiments were done first to validate the procedure and software to record and reconstruct time-average digital holograms. An aluminium plate with dimension 150 mm x 165 mm x 3 mm was clamped at one end as a cantilever and vibrated using the piezo shaker on the other end as shown in figure 5. The frequency of vibration was swept from 0 Hz to 20 k Hz. Frequency response analysis of a cantilever plate using NASTRAN finite element analysis (FEA) software was also done and obtained the contour plot of the deformation at frequencies with maximum vibration amplitude. The timeaverage holograms (left image) correspond to these frequencies were comparable with the numerical plots (right image) and were shown in figures 6a to 6c. 5.2 NDT of honeycomb sandwich panels Time average digital holographic interferometry technique was applied to detect the programmed debonds in honeycomb Fig. 4 : Laser on wheeled cart and experimental setup with bifurcated fiber

4 NDE 2011, December 8-10, Fig. 5 : Piezo shaker vibrating 3 mm thick plate fixed along one edge Fig. 7 : Panel under vibration stressing The panel with metallic skin sheets has dimension 300 mm x 70 mm and 25 mm thick core with programmed debond. Figure 8 shows time average holograms of the panel at different frequencies. (a) (b) 2 nd bending mode is shown in the figure 8a with the node indicated by the bright fringe at the center. Figure 8b corresponds to hologram at about 8 k Hz and, the debond between the skin and film locally resonates to give circular fringes. The fringe signature matches well with the numerical results in the paper by Vyacheslav et.al. [8] The panel with composite skin sheets has dimension of 600 mm x 95 mm and 40 mm thick core. Figure 9 shows time average holograms of panel having no defects and figure 10 shows time average holograms of the defective panel, with debond across its width, at different frequencies. The fringe patterns of figure 10 gives local resonant modes of debond region clearly indicating presence of defects. It is clear from (c) Fig. 6 : (a) At frequency 540 Hz (b) At frequency 820 Hz (c) At frequency 4040 Hz sandwich panels with metallic and composite skin sheets. The technique was used to obtain their local resonant frequency and the corresponding fringe signature. In both the cases the panels were vertically fixed at one end and other end was vibrated using a shaker with frequency swept upto 30 khz as shown in figure 7 Fig. 8 : Time average holograms at different frequencies of metallic sandwich panel (programmed debonds)

5 122 Binu P. Thomas : Proceedings of the National Seminar & Exhibition on Non-Destructive Evaluation Fig. 13 : Time average hologram of edge B at 800 Hz Fig. 9 : Time average holograms at different frequencies of non defective panel Fig. 12 : Time average hologram of edge A Fig. 10 : Time average holograms at different frequencies of defective panel Fig. 14 : Time average hologram of edge B at 8 k Hz Fig. 11 : Configuration of sandwich structure showing the three curved edges A, B and C figure 10a that the size of the debond matches well with that of its first local resonant mode. 6. NDT OF CURVED SANDWICH STRUCTURES Sandwich structures made of metallic honeycomb sandwich panels with folded type of joints making three curved edges as shown in the figure 11 were tested in this study. Vibration Fig. 15 : Curved edge A and shaker stressing was applied using a piezo shaker and the frequency of vibration swept from 1 Hz to 50 khz maximum. Time average holograms were recorded continuously in real time with a frequency increment of minimum 1 Hz. Figure 12 shows the mode shape of edge A at frequency of 800 Hz. Figures 13 and 14 show time average holograms of edge B at 800 Hz and 8 khz respectively and local vibration

6 NDE 2011, December 8-10, corresponding to metallic and composite skin sheet debond in honeycomb sandwich structures are presented and typical frequency range correspond to local resonance is found to be 8 khz. Sandwich structures with curved edges are tested and debond on the curved edge was detected through this technique and the fringe patterns for the corresponding frequencies obtained. The fringe signature was comparable with that of programmed debonds. Fig. 16 : Time average interferogram of edge A modes are not present in these holograms. Anomalous fringe patterns or local resonant modes are not observed throughout the frequency sweep at any of the three edges. Hence it is concluded that no debond exists at any of the three curved edges. Second structure was holographed for debond detection. No anomalous fringe patterns were observed along both the long slanted curved edges B and C. The edge A of the adaptor with shaker mounted on it is shown in figure 15. Figures 16a and 16b show the time average hologram of the curved edge A at a vibration frequency of 8 khz with vibration amplitude ±1 V and ±10 V respectively. Fringe pattern due to local resonance corresponding to debond is visible in the interferogram. The size of the fringe anomaly is measured about 15 mm 15 mm and hence the size of debond. It is observed that the local resonance of debonds are visible even with small vibration amplitude by time average holography. 7. CONCLUSION The principle, method and implementation of high-speed time average digital holography is explained in this paper. Time average digital holography gives the mode shape of a structure under vibration and is validated numerically. Reference hologram is not required for time average holography unlike double exposure holography and shearography. This minimizes the vibration isolation requirement and reduces rigid body motion effect on the fringe patterns. The fringe signature REFERENCE 1. G. Mayr, B. Dietermayr, et.al., Characterization of defects in curved CFRP samples using pulsed thermography and 3D finite element simulation, 9th International Conference on Quantitative InfraRed Thermography, Krakow - Poland, July 2-5, S. Sfarra1, C. Ibarra-Castanedo et.al., A comparative investigation for the nondestructive testing of honeycomb structures by holographic interferometry and infrared thermography, 15th International Conference on Photoacoustic and Photothermal Phenomena, Journal of physics: Conference Series, 214, Y. Y. Hung, Shearography for Non-destructive evaluation of composite structures, Optics and Lasers in Engineering 24 (1996) Binu P. Thomas and S. Annamala Pillai, Digital holographic interferometry for wholefield NDT applications, Journal of aerospace sciences and technologies, vol. 61, No. 2, pp , May Charles M. Vest, Holographic interferometry, John Wiley and Sons, New York, Hariharan P., Optical Holography Principles, techniques and applications, Cambridge University press, Cambridge, Binu P. Thomas and S. Annamala Pillai, High speed generation of digital holographic interferogram and shearogram for non-destructive testing, INSIGHT, Journal of The British Institute of NDT, vol. 51, no. 5, pp , May Vyacheslav N. Burlayenk and Tomasz Sadowski, Influence of skin/core debonding on free vibration behavior of foam and honeycomb cored sandwich plates International Journal of Non-Linear Mechanics 2009