The study reported in this paper yields useful information about the performance of various eddy current instruments with respect to detecting small cracks under fasteners. The detection of these cracks is a timely issue of great concern. This article illuminates the capabilities and demonstrates the detection limits of the various instruments investigated for various discontinuities pertinent to this problem. G.P. Singh Associate Contributing Editor

Figures 1-4
Figures 5-7
Tables 1-5

The nondestructive inspection (NDI) portion within the National Aging Aircraft Research Program predicates that improvements must be made to existing inspection techniques and devices so as to provide more reliable detection capabilities. The objectives for the NDI program are derived directly from concerns about the current system, as exemplified by factors which led up to the Aloha Airlines accident, which was caused by multiple origin fatigue cracks in the fuselage structure. These multiple origin fatigue cracks are generally short and thus hidden by the flush-head fasteners. Hence, there was a need to detect these short cracks before they could grow to an unstable size. The goal of this study was to detect cracks 1.25 mm (0.05 in.) or larger in aluminum fuselage skins with a thickness of approximately 1.6 mm (0.063 in.) or less.

This study was carried out under contract with the NASA Langley Research Center with William Winfree as the technical monitor. The results show that the Staveley EddyScan 30, the Northrop LFECA, the Hocking FastScan, and the GK Engineering surface scanning probe with the Elotest B1 mini rotor, can detect fatigue cracks 1 mm (0.04 in.) or larger, under flush-head aluminum rivets in aluminum aircraft skins with a thickness of 1.6 mm (0.063 in.) or less. Recorded results are presented in this paper.

---

There was a need to detect multiple origin fatigue cracks before they could grow to an unstable size.

---

Background
It is a known fact that aluminum aircraft fuselage skins can develop small fatigue cracks after an extended period of service due to the numerous pressurization cycles. Usually, the outer skin is countersunk to provide for assembly with flush-head fasteners. The sharp edge in the skin, caused by the countersink, provides a high stress concentration for fatigue crack generation. Numerous small cracks may be generated along a particular row of fasteners. When this happens, they are generally referred to as "multiple origin" fatigue cracks. The small cracks continue to grow with time and at some point they can join up, causing failure of the skin and rapid decompression of the aircraft cabin - hence the need to detect them at the smallest possible size.

Many high frequency eddy current inspection techniques have been developed to detect fatigue cracks under installed fasteners in aluminum aircraft fuselage skins. However, until recently, the shortest through-the-thickness crack that could be reliably detected was 2.54 mm (0.1 in.) in length. Damage tolerance (stress) engineers were unhappy with that figure and requested that improved eddy current techniques be developed to detect fatigue cracks on the order of 1.25 mm (0.05 in.) in length. Cracks of this length barely extend beyond the fastener head, which makes them difficult to detect. This particular discontinuity size was the goal of this evaluation.

In order to prove that certain improved eddy current equipment or probes could detect the discontinuity size goal, fatigue cracks and electrical discharge machined (EDM) notches were fabricated into typical countersunk fuselage skin panels. The cracks and notches ranged from 0.635 mm (0.025 in.) to 2.54 mm (0.1 in.) in length.

Test Procedure
Test Specimens. Eight plates were fabricated from 2024-T3/T4 bare aluminum sheet stock 1.6 mm (0.063 in.) thick. The plates were 101.6 mm (4 in.) wide with four holes drilled in each. Four of the upper plates were spot faced and one EDM notch was placed in one hole of each plate. The notches were 0.635 mm (0.025 in.), 1.27 mm (0.05 in.), 1.9 mm (0.075 in.), and 2.54 mm (0.1 in.) in length. Cracks were generated in one hole of the other four upper plates. The cracks were grown to the same sizes as indicated above for the notches. The four notched plates and the four cracked plates were stacked and drilled with a 1.6 mm (0.063 in.) thick 2024-T3/T4 bottom plate 813 mm (32 in.) long by 230 mm (9 in.) wide. After stack drilling, EDM notches 0.635, 1.27, 1.9, and 2.54 mm (0.025, 0.05, 0.075, and 0.1 in.) long were placed in the bottom panel to simulate second layer cracks. One panel was assembled with aluminum rivets and a second panel was assembled with steel flush-head Hilocks, washer, and nut (Figure 1).

Nortec-30 EddyScan. This system consists of a compact scanner and a portable computerized instrument. Pulsed eddy current techniques and a broadband Hall effect sensor allow complete inspection of the perimeter of the hole from the surface to the maximum depth capability of the system. During inspection, the scanner is centered over the fastener and the Hall effect sensor and driving coil are rotated by the scanner about the fastener perimeter. Eddy currents are induced by the coil into the part under test. The results are observed on a CRT. For further information about the EddyScan instrument, refer to Gibbs (1990).

The instrument gain was set at 32 dB to detect the 1.27 mm (0.05 in.) notch in plate N2 (Figure 1).The results for both the 1.27 mm (0.05 in.) EDM notch N2 and fatigue crack C2, beneath the flush-head aluminum rivets, are shown in Figure 2. When testing the panel containing the steel lock bolts, the 1.27 mm (0.05 in.) notch and crack yielded similar results at 24 dB. Attempts to raise the 1.27 mm (0.05 in.) discontinuity amplitude to 75 percent of screen height caused noise from the steel fasteners to activate the discontinuity gate. Typical noise obtained from holes without discontinuities is also shown in Figure 3. However, at 24 dB, the 1.9 mm (0.075 in.) notch and crack both yielded signal amplitudes of 100 percent screen height (not shown).

GK Engineering Probe and Elotest. The test panels were scanned with a variety of existing eddy current probes and the 1.27 mm (0.05 in.) notches and cracks could not be detected. A rotating absolute driver/receiver probe was fabricated but yielded excessive noise from the edge of the fasteners. A second probe was fabricated with an absolute driver coil and differential (two coils) receiver. Preliminary results with the GK 10148 manual scanner probe were excellent. Signal amplitudes of 80 percent were easily obtained from the 1.27 mm (0.05 in.) notch and crack. Based on these results, similar probes were fabricated for use with the Elotest B1 automated scanner.

The results obtained with the automated probe (10150-3) for the 1.27 mm (0.05 in.) crack and notch, at a gain of Y60 dB/X45 dB, are shown in Figure 4. The results, with aluminum rivets, were obtained at 50 kHz. The signal-to-noise ratio was poor when attempting to record the 1.27 mm (0.05 in.) notch and crack under the steel fasteners. The search for the 1.9 mm (0.075 in.) notch and crack under the steel fasteners yielded good results.

Northrop LFECA System. In 1989 Northrop identified a need for the improved detection of cracks under fasteners on the F/A-18 vertical stabilizers. Cracks under fasteners were occurring in the vertical stabilizer aluminum spars which were fastened under composite skins by titanium fasteners (Sheppard, 1994). A sensitive inspection system was needed to detect these cracks in the aluminum spars with the skins and fasteners installed. For this reason, Northrop designed and constructed an updated/modified low frequency eddy current array (LFECA) system in 1990.

The system uses a unique segmented (16 coil) outer receiver array with a center driver coil. The core configuration for steel fasteners is different from that for titanium or aluminum fasteners. The difference between the two cores is the design of the center pole. The center pole of the titanium fastener probe is a solid piece, but for the steel fastener probe the center pole is a hollow piece like a sleeve (Sheppard, 1985). For clarity, the probe designed for use with titanium fasteners is referred to as a solid center probe, and the probe designed for use with steel fasteners is referred to as a hollow center probe. The probe diameter varies with the diameter of the fastener to be inspected.

The array probe is placed against an unflawed hole or specimen as a reference standard. When the probe is centered over the standard by the operator with assistance from the electronic display, the computer triggers the function generator for a preprogrammed signal with the desired frequency and amplitude. The induced eddy current signals are picked up by the segmented coils, scanned by the multiplexer, digitized by the voltmeter, and stored in the computer as a numerical array. Each segment voltage is sampled several times to obtain an average reading.

The probe is then moved to the fastener in the structure or specimen to be inspected. The centering and signal processing sequences are repeated and the segment voltage readings are stored in the computer as a second numerical array. The computer calculates the difference between the two array values and displays the asymmetries in a graphical wave form.

When a probe is placed on a hole which is the same as the reference hole which has no discontinuities, the response will be a flat line (zero) amplitude if the probe is properly centered. Figure 5 shows a typical response from the aluminum rivet specimen containing the 1.27 mm (0.05 in.) notch and crack. The horizontal tick marks in the display indicate sense coil (1 to 16) and the vertical tick marks indicate response amplitude. The solid line is the real part (portion in phase with the probe drive signal) and the dashed line is the imaginary part (90 degrees out of phase with the probe’s drive signal) of the complex response. Figure 5 results were obtained at 2.8 kHz.

In order to determine the signal to noise ratio between a hole with discontinuities and one without, the LFECA instrument was operated in the Y/X mode and the amplitude of the signals were measured. Table 1 shows the results from fasteners 1 through 32 for the aluminum rivet panel. The 1.27 mm (0.05 in.) crack and notch signals are well above the noise level from the holes without discontinuities. The large difference in amplitude between the 2.54 mm (0.1 in.) notch and crack signals cannot be explained because the other test methods show similar separation. The crack length is correct but appears very tight. Table 2 shows the results obtained from the steel fastener panel. Again, the 1.27 mm (0.05 in.) crack and notch responses are well above the noise level from the holes without discontinuities.

KB Hocking FastScan. Applied to aluminum alloy multilayered structures, the FastScan technique is capable of detecting surface and subsurface cracks and corrosion around fasteners (Buckley and Calvert, 1995). Phase and magnitude analysis of the mixed signals can give an indication as to the depth and severity of discontinuities that are detected. The probe is connected to the FastScan socket on the instrument by means of the probe cable. The lower section is placed symmetrically over the fastener by means of the aperture and held firmly in place, resting on nonslip feet. The upper section is lowered into the bearing and rotated by 90-180 degrees. The probe is designed to minimize signals from the fastener itself by means of symmetrical rotation and balanced coils. Dual frequency techniques are used to reduce further the common mode signal profile, allowing very small discontinuity signals to be detected.

The frequency range used for FastScan is normally between 500 Hz and 10 kHz. A base frequency must be chosen that will allow sufficient penetration for the thickness of the structure where subsurface flaws are to be detected. By knowing the conductivity of the material and the thickness, determine the standard depth of penetration frequency. The standard depth of penetration frequency should be divided by a factor of 3 to give the base frequency. This will be close to the effective depth of penetration frequency for the structure.

The Phasec 3.4 instrument was used with the differential FastScan probe. The probe liftoff was in the horizontal direction on the CRT and the notch/crack signals were in a vertical direction. The instrument gain was set at approximately 50 dB. Channel 1 was set at a frequency of 4 kHz and channel 2 at 10 kHz. The upper signal displays in Figure 6 were obtained from the cracks under the aluminum rivets and the lower displays were obtained from cracks under steel fasteners. The phase angle is quite different between the cracks under the aluminum rivets and steel fasteners for the 0.635 and 1.27 mm (0.025 and 0.05 in.) cracks. For some unexplained reason, the signal amplitude from the 2.54 mm (0.1 in.) crack was less than that from the 1.9 mm (0.075 in.) crack under the steel fasteners. Similar results were obtained using the Elotest B1 and the GK Engineering probe.

Second Layer Notch Detection
The prior study was performed to determine the detectability of cracks under fasteners in the accessible first layer structure. However, EDM notches were added to the bottom plates in order to simulate second layer cracks (Figure 1). Unfortunately, the existing inspection methods used to detect second layer cracks under fasteners either require the removal of the fastener or are lacking in requisite sensitivity. Fastener removal is generally a difficult and costly procedure which can itself result in damage to the structure.

Both the upper and bottom plates were 1.6 mm (0.063 in.) thick 2024-T3/T4 aluminum. EDM notches were added to the lower plate, as indicated in Figure 1. The notch at N1 is 0.635 mm; at N2, 1.27 mm; at N3, 1.9 mm; and at N4, 2.54 mm in length (0.025, 0.05, 0.075, and 0.1 in., respectively). One panel contained flush-head aluminum rivets and the other contained steel flush-head fasteners.

The automated probes from GK Engineering could not detect the subsurface notches because the lower frequency limit of 100 kHz did not provide for sufficient penetration of the eddy currents. However, when using the manual probe (GK 10148) with the Elotest at 47 kHz, the 1.9 mm (0.075 in.) notch was detected under both the aluminum and steel fasteners. Also, the preamplifier gain was reduced from 18 to 12 dB for the steel fastener panel along with a reduction in test sensitivity from 43 to 32 dB for the same panel.

Figure 7 shows the second layer notch (1.27 mm [0.05 in.]) detection for the EddyScan 30 and LFECA instruments over the aluminum rivets and steel fasteners. The EddyScan 30 signal amplitude was about 80 percent over the aluminum rivets at a gain of 41 dB. However, for the steel fastener panel the signal amplitude was slightly above 36 percent at a gain of 24 dB. The instrument gain could not be increased for the steel fastener panel due to the increase in background noise. At 24 dB, the signal amplitude from the 1.9 mm (0.075 in.) second layer notch was saturated (twin peaks) on the EddyScan-30 screen.

Summary
The results obtained from these series of experiments indicate that all four instruments/probes; Hocking FastScan and Phasec 2.3, Nortec-30 EddyScan, Northrop LFECA, and the GK Engineering/Elotest are capable of detecting surface (first layer) cracks 1 mm (0.04 in.) in length under installed (flush-head) aluminum rivets. The EddyScan-30, the FastScan, and the GK Engineering/Elotest were capable of detecting the surface 0.635 mm (0.025 in.) notch/crack under the aluminum rivets. Although the 0.0635 mm (0.025 in.) notch/crack was detected, inspection at this sensitivity level is not recommended unless the noise level is very low.

The FastScan, the EddyScan-30, and the LFECA detected the surface 1.27 mm (0.05 in.) notch/crack under the steel fasteners. The GK Engineering automated probe detected the 1.9 mm (0.075 in.) notch/crack under the steel fasteners.

The GK Engineering Elotest automated probe could not detect the subsurface discontinuities due to a lower frequency limit of 100 kHz which did not provide for eddy current penetration through the first layer. However, when using the manual scanner probe (GK 10148) at 47 kHz, the 1.9 mm (0.075 in.) second layer notches were detected in both the aluiminum rivet and steel fastener panels.

The FastScan, the LFECA, and the EddyScan-30 detected the 1.27 mm (0.05 in.) second layer notches in both the aluminum rivet and steel fastener panels. Table 3 shows the LFECA signal response and signal to noise ratio for the steel fastener panel. Table 4 shows the LFECA signal response and signal to noise ratio for the aluminum rivet panel.

Table 5 shows the summary findings for the four instruments and for the four conditions, i.e., first layer cracks under aluminum rivets and steel fasteners and second layer notches under similar fasteners. The layers were 1.6 mm (0.063 in.) thick 2024-T3/T4 aluminum sheet.

Similar eddy current inspection reliability experiments have been performed (Spencer and Schurman, 1995) and are continuing at The Aging Aircraft NDI Development and Demonstration Center at the Sandia National Laboratories in Albuquerque, New Mexico. The author did perform the eddy current probability of detection experiment at that facility during the last week in August, 1995, using the GK Engineering/Elotest automated scanner.

Acknowledgment
The author expresses his sincerest thanks to the following persons who provided technical support to this experimental study: William Sheppard, Northrop Grumman Corporation, Hawthorne, California; Ernie Grimson, GK Engineering, Chatsworth, California (deceased); Dave Jankowski, Krautkramer Branson, Lewistown, Pennsylvania; John Calvert, Hocking, St. Albans, England; and Tom Reep, Staveley Instruments, Kennewick, Washington. Special thanks to Ed Generazio and William Winfee, of NASA Langley Research Center, for providing funds for this program.

References
Buckley, J., and J. Calvert, Improved Eddy Current Techniques for the Aerospace Industry, Hocking NDT Ltd., St. Albans, England, 1995.

Gibbs, M., Pulsed Eddy Current vs. Other Eddy Current Methods for Inspection of Aircraft Structure, Staveley Instruments, Inc., Kennewick,WA, Sep. 1990.

Sheppard, W., WL-TR-94-4006: Eddy Current for Detecting Second Layer Cracks Under Installed Fasteners, Wright Patterson Air Force Base, OH, 1994.

Sheppard, W., AFWAL-TR-85-4095: Manufacturing Technology for Advanced Second Layer NDE System Producibility, Wright Patterson Air Force Base, OH, 1985.

Spencer, F., and D. Schurman, Reliability Assessment at Airline Inspection Facilities, Vol. III, DOT/FAA/CT-92/12, III, FAA Technical Center, Atlantic City, NJ, 1995.

 

* McDonnell Douglas Aerospace, 2401 E. Wardlow Rd., Long Beach,

CA 90807-5309; (310) 593-7304; fax (310) 496-9244

Copyright © 1997 by the American Society for Nondestructive Testing, Inc. All rights reserved.
 

[ Materials Evaluation ]