TIME-GATINGOF PULSED EDDY CURRENT SIGNALS FOR DEFECT CHARACTERIZATION AND DISCRIMINATION IN AIRCRAFT LAP-JOINTS

Transcription

1 TIME-GATINGOF PULSED EDDY CURRENT SIGNALS FOR DEFECT CHARACTERIZATION AND DISCRIMINATION IN AIRCRAFT LAP-JOINTS Jay A. Bieber, Sunil K. Shaligram, James H. Rose, and John C. Moulder Center for Nondestructive Evaluation Iowa State University Ames, la 511 INTRODUCTION Pulsed eddy current (PEC) nondestructive testing differs from conventional eddy current techniques in that the probe coil is excited by a pulse, rather than continuous excitation at a single frequency. Reviews of early work on pulsed eddy currents are given by Waidelich 1 and by Renkin. 2 Pulsed excitation causes the propagation of a highly attenuated traveling wave, which is govemed by the diffusion equation? The diffusive propagation of the eddy current pulse results in spatial broadening and a delay, or travel time, proportional to the square of the distance traveled. It was realized in early work on pulsed eddy current systems that this time dependence offered certain advantages over conventional eddy currents. 4 In the current study we demonstrate the ability of a prototype pulsed eddy current instrument, described elsewhere, 5 6 to take advantage ofthistime dependence to discrirninate flaws from such interfering signals as probe liftoff, air gaps, and fasteners. EXPERIMENT Figure 1 is a schematic of the sample and probe geometry for the pulsed eddy current experiments that are presented here. The coil that was used in our experiments was a 638-tum, right-cylindrical air-core coil of.22-inch ID,.47-inch OD, and.118-inch length. The probe was designed with a constant built-in wear surface, which gave a Iiftoff of approximately.7 inches. The geometry of the sample was chosen to simulate a twolayer aircraft lap-joint, with corrosion in the locations shown and with a certain amount of air gap to indicate plate separation. Theinputto the coil is a 5-volt, 1 khz. square wave, with a 5% duty cycle. This allows enough time for the coil current to rise to a steady state in 5 J.Ls, and decay back to zero before the next pulse. The software that was developed for the PEC instrument perrnits the drive pulse repetition rate and amplitude to be adjusted to allow probes of various inductance and resistance to be used. The resulting coil Review ofprogress in Quantitative Nondestructive Evaluation. Vol. 16 Edited by D.O. Thompson and D.E. Chimenti. Plenum Press, New York,

2 / Aircore Coil Flaw Locations Bottom of top layer Top of botiom layer Bottom of bottom layer Fig. 1. Geometry of pulsed eddy current experiments on hidden flaws. current is detected by sensing the voltage drop across a 1-ohm resistor in series with the probe. The coil response is then digitized with a 16-bit, 1 megasample per second analogto-digital converter. To obtain a pulsed eddy-current flaw signal, the probe coil is first p1aced on a reference area on the sample where there is no flaw. The coil response on this reference area is then stored in memory and is subtracted from subsequent incorning signals as the probe is scanned over the sample. The pulsed eddy current signal displayed on the instrument is thus the difference between the transient current in the coil over a flaw-free area and that over an area containing a flaw. Figure 2 shows a typical pulsed eddy current signal from a simulated aircraft lap splice containing an artificially thinned region. As shown in this figure, the pulsed eddy current waveform has two main features that are used in flaw characterization. The first is the peak: height, which is proportional to the amount of metalloss. The second is the zero crossover point of the waveform, which contains information about the depth of the flaw ir the structure. W aveforms are acquired continuously while the probe is scanned over an area of the sample using a portable two-axis scanner. Peak Am plitude Time-gate (All peak am plitudes outside this range are not plotted).. f ' Zero-Crossover / / I ~ ~ Fig. 2. Typical pulsed eddy current signal and illustration of time-gating. 1916

3 Data acquired from 2-D scans are presented in a C-scan pseudo-color image. To produce a C-scan image, the peak height of the PEC waveform is assigned a color corresponding to its amplitude at each x-y position of the probe. Peak amplitudes are assigned colors from a ten-color look-up table. The range of amplitudetobe colorized is user selectable in the control software. Using this method the sensitivity of the image can be increased or decreased, using the colors to represent any range of peak amplitudes desired. THEORETICAL BASIS FOR TIME DISCRIMINATION Theoretical models ofthe instrument have been developed in an earlier study, 7 which was based upon the analytical solutions of Cheng, Dodd, and Deeds 8 9. The plot on the left in Fig. 3 shows a set of theoretical inversion curves calculated for a modellap-joint consisting of two 1.5-mm aluminum plates. This plot shows the peak amplitude versus the time to zero crossover for a range of metalloss in the locations shown in the schematic below. The plot on the rignt shows a nurober of experimental PEC curves obtained on the same geometry for 3% metalloss in each layer. As is evident in the inversion graph, air gap, or simple plate separation, has the earliest time to zero crossover. This is because the total thickness of metal below the coil remains constant. This results in a lower inductance, and hence faster rise and decay tims for the probe coil, than occurs when there is metalloss under the coil. Slightly later in time are the zero crossover points for metal loss in the bottom of the top layer, followed by thinning in the top of the bottom layer, and, finally, in the bottom of the bottom layer. > iii c:...;; CO "" CD.. 8.DE < 224 Al um tnutn 1 S mm Ihle< 5'1. 1 3~ me1111cu - 1"CI 6.E < tor.ott~ ot Tlf 4 OE 2 OE < 1.\ Tu otlorr~" lort(i""oihll!ll"' > iii c:..." cii "" CD.. 3 OE 2 2 OE Alumtnum t S mm lhtck - Rrt l "o'* - L6.'\ct12' 1"' : lop;tbolb-. :" -- 11:11)1"11 ' I OE 2 -.-!.lcth"' oiioi:.q'it 2... ~. OE+O O.OE D r- :...---' 1 2E 4 I 6E < 2.E 2 4E 4 2 BE Time to Zero-Cross~ng, s I DE ~..---,- O.OE- 2 OE 4 OE 6 OE 8.E 1 OE 3 Tome to Zero-Cross~ng, s Flaw advance Fig. 3. Theoretical inversion chart for PEC signals (left) and experimental waveforms for a variety of simulated defects (right). The geometry of the test specimen is shown at bottom. 1917

4 The schematic at the bottom of Fig. 3 is useful for visualizing the relationship between flaw location and the slope of the inversion curves. For metalloss at the bottom of the top layer or the bottom of the bottom layer, deeper flaws result in a thinning of material between the coil and the metal-defect interface. As illustrated by the arrows in the schematic diagram, this results in slightly shorter eddy current diffusion times for deeper flaws as they advance toward the probe. For metalloss at the top of the bottom layer, the metal-defect interface remains at a constant distance from the probe as the amount of thinning increases, and therefore the inversion curve has a nearly vertical slope. The separation in time of flaw signals from different layers provides a means to discriminate flaws based on time-gating. Time-gating is accomplished in the PEC software, as illustrated by the vertical bars in the display of Fig. 2. After an initial C-scan image has been acquired, it may be redisplayed, plotting only those peak: heights which have a zero crossover within a user-selected time gate. The initial image contains the peak: heights from the entire spectrum of zero crossover times possible during the pulse duration, which, for a 1kHz square wave at 5% duty cycle, is to 5 f..ls. PULSED EDDY-CURRENT IMAGES To demonstrate the PEC instrument's ability to discriminate flaws in layered structures, a lap-joint calibration sample was constructed of two plates of.62-inch 224 aluminum, as shown in Fig. 4. Flat bottom holes.75 inch in diameter were machined into the surface of each layer, with depths of 5, 1, 2, and 3%. Holes were also drilled in the sample to simulate rivets on 1-inch centers. Figure 5 is the pulsed eddy current image obtained by scanning this sample. The top image represents the raw image, which displays peak: heights from every signal acquired during the scan. The lower image is the result obtained by setting the time gate minimum late enough to exclude the rivet holes which, because they are on the surface, have very early zero crossover. Figure 6 is the result of setting narrower time gates to include only the flaws from each individuallayer. With the time-gate set for the times shown, we are able to isolate the flaws in the bottom of the top layer, the top of the bottom layer, and the bottom of the bottom layer, as shown. Elaw d~l:lltl 3% 2% 1% 5% Top plate Bottom plate Flaw locations Bottom ol top layer Top of bottom layer Bottom ol bottarn layer Fig. 4. Lap-joint calibration standard (.62-inch 224 Al plates with.75-inch dia. FBH). 1918

5 3% 2% 1% 5% Raw data, amplitude versus xy position 3% 2% 1% 5% Time-gated to eliminate data from fastener holes Fig. 5. PEC image of.62-inch lap-joint calibration standard. Bottom of Top Layer Bottom of Bonom Layer Top of Bottom Layer (19m >l>22j.is) (215m >l>25j.1s) (246m >1>512J.IS) Fig. 6. PEC image of.62-inch lap-joint calibration standard, time-gated to discriminate the flaws from each layer. The upper left image in Fig. 7 is a PEC raw data image of a two layer lap-joint corrosion training sample made by Boeing. lt consists of a corroded frrst layer riveted to a corrosion free second layer. The top right image is an immersion ultrasound image of the same sample. Similar features are seen in both images revealing the corroded areas and rivets, the PEC scan being of poorer resolution due tothelarge.47-inch diameter probe. The lower left image shows the PEC data time-gated to eliminate the rivet signals, leaving only the corroded areas. Narrower time-gating produced the image on the lower right, showing only the deepest areas of this first layer corrosion. This is possible due to the slightly negative slope of the inversion curve for corrosion in the first layer as discussed above. It is worth noting one feature that was revealed in this image: the dark corrosion spot just below the center of the image. According to the low amplitude color value assigned, this area would not at first appear to be deep corrosion. Upon comparison to the ultrasonic image however, we see that this spot is caused by a deep corrosion pit, smaller in diameter than our probe. The low amplitude of the signal was the result of the flaw being smaller than the probe used to scan it. In this case, time-gating provides the additional information required to deterrnine that this spot is indeed one of the deepest spots of 1919

6 - - Fig. 7. PEC scan of Boeing lap-joint corrosion training sarnple containing corroded first layer, joined to a corrosion free second layer. corrosion in this sarnple. This demonstrates one of the principal advantages of the pulsed eddy current technique. Figure 8 is the result of an investigation of second layer crack detection by pulsed eddy current. The sample is an EDM notch lap-joint calibration sarnple, containing EDM slots in the second layer of a.65-inch, two layer lap-joint with rivets on 1-inch centers. The notch lengths and locations are shown in the figure to be emanating radially from the edge of the rivet holes. The PEC scan of the raw data is shown in the figure, where the.2 inch notches are just visible. After time-gating this image to eliminate the interference due to the rivets, the lower image was produced, revealing even the smallest notch in the sample,.6 inches in length. Cross section Raw data image Time-gated image Secondlayer Fig. 8. PEC image of.65-inch lap-joint time-gated to discriminate second layer EDM notches. 192

7 SUMMARY Wehave demonstrated the ability of a newly developed prototype scanning pulsed eddy current system to discrirninate flaws using time-gating. The instrument can detect and locate cracks and corrosion in multilayer metal aircraft structures. Using the time dependence of the pulsed eddy current flaw signal, we have demonstrated the ability to deterrnine the location of defects and to discrirninate against interference from such features as fasteners or surface flaws. According to theoretical calculations and preliminary experimental work not shown here, we are also able to discrirninate flaws in the presence of varying Iiftoff and air gap. The instrument thus provides an easy to interpret, quantitative nondestructive testing technique that can be calibrated, in the case of an air-core probe, by using a theoretical inversion chart. ACKNOWLEDGMENTS This work was supported in part by the FAA Center for Aviation Systems Rehability program at the Center for NDE at Iowa State University, andin part by AFOSR grant No. F DEF, and by the Federal Aviation Administration under Grant Nos. 95-G-25 and 95-G-32. The authors are grateful to Mr. Michael Hutehinsan of the Boeing Airplane Co. for supplying one ofthe corrosion samples used in this study. REFERENCES 1. D. L. Waidelich, in Research Techniques in Nondestructive Testing Val. 1, edited by R. S. Sharpe, (Academic Press, London, 197), pp C. J. Renken, "Theory and Some Applications of Pulsed Current Fields to the Problems of Non-Destructive Testing", in Progress in Applied Materials Research Val. 6, edited by E. G. Stanford, J. H. Fearon, and W. J. McGonnagle (Gordon & Breach, London, 1964) pp D. L. Waidelich, S. C. Huang, Mat. Eva!. 3, 2-24 (1972). 4. J. L Fisher, and R. E. Beissner, "Pulsed Eddy Current Crack Characterization Experiments," in Review of Progress in QuantitativeNDE Vol.5, edited by D.. Thompson and D. E. Chimenti, (Plenum, New York 1986) p J. C. Moulder, M. W. Kubovich, E. Uzal, and J. H. Rose, "Pulsed Eddy-Current Measurements of Corrosion-Induced Metal Loss: Theory and Experiment," in Review of Progress in QNDE Val. 14, edited by D.. Thompson and D. E. Chimenti, (Plenum, New York 1995) p J. C. Moulder, J. A. Bieber, W. W. Ward III, and J. H. Rose, "Scanned Pulsed-Eddy Current Instrument for Non-Destructive Inspection of Aging Aircraft," SPIE Proceedings Val (in press). 7. J. H. Rose, E. Uzal, and J. C. Moulder, "Pulsed Eddy-Current Characterization of Corrosion in Aircraft Lap-Splices: Quantitative Modeling," SPIE Proceedings Vol. 216, 164 (1994). 8. C. C. Cheng, C. V. Dodd, and W. E. Deeds, lnt. J. Nondestr. Test. 3, 19 (1971). 9. C.V. Dodd and W. E. Deeds, J. Appl. Phys. 39, 2829 (1968). 1921