Nondestructive Evaluation of Graphite/Epoxy Composite Damage

Ultrasonic and acoustic emission techniques were used to monitor and evaluate material damage in a graphite/epoxy laminate containing a machined hole as an initial flaw and subjected to fully reversed spectrum fatigue loading at room temperature. It was found that the flaw growth progressed radially around the initial hole at a uniform rate during cycling at the lower stress levels. At the higher levels, material damage accelerated dramatically, progressing faster in the transverse direction toward the free edges. By close examination of the A-scan and RF spectrum photographs, it was determined that damage modes could be defined as to their extent and relative location within the specimen. These conclusions were supported by photographs of the failed specimen. EXPERIMENTAL PROCEDURE S ecimen The specimen was a 16-ply graphite/epoxy AS3501-5A) coupon of [{0/±45/90)sl 2 layup. The specimen gage section was 5.08 em {2 in.) long and 3.81 em (1.5 in.) wide. The initial flaw was a 0.795 em (0.313 in.) diameter hole bored into a previously unflawed/undamaged specimen. The hole was drilled using a diamond tipped coring bit mounted on a variable speed drill press. Figure 1 shows the specimen after the hole was drilled and before fatigue loading was begun. The fiber damage around the hole was caused by the coring bit when it penetrated the back-side of laminate. An ultrasonic C-scan of the specimen, Fig. 2, and corresponding A-scans, showed the damage was superficial. Previous experience indicated that this type of superficial damage does not influence appreciably material damage caused by fatigue loading. Fig. Test Specimen with Initial Flaw (Hole) Before Fatigue Loading Loading The test specimen was preconditioned by oven-drying at 347 deg K (165°F) for five days. The specimen was tested under ambient environmental conditions because the acoustic emission transducers did not work well in humid and/or hot environments. The specimen was stress-cycled in a fiveposition chain, with load applied through an electrohydraulic closed-loop system. The load spectrum consisted of 127,500 cycles of tension-compression at various levels of peak stress per life-time. At a cycling frequency of three Hertz, this was accomplished in approximately twelve hours. The 69 Fig. 2 C-scan of Test Specimen with Initial Flaw (Hole) Before Fatigue Loading specimen studied was cycled to peak levels of 161 MPa {23 ksi), 223 MPa (32 ksi) and 276 MPa (40 ksi) applied over 510,000, 510,000, and 254,428 cycles, respectively, in an effort to accelerate and extend the material damage. NONDESTRUCTIVE MONITORING Acoustic Emission2 The acoustic emission system consisted of 0.95 em {0.38 in.) diameter broadband transducers operational in the 0 to 1 MHz range. A transducer was attached to the specimen using a clip and an acoustic coupling medium to minimize the entrance of extraneous transient noise into the system, Fig. 3. The acoustic signal was fed into a databus after being processed by 180 to 210 KHz filters and then by preamplifiers. The databus operated at a 0.1 sec. sampling rate feeding an eight channel event recorder and chart recorder for hard copies of the data. Acoustic emission records consisted of both instantaneous rate and cumulative counts recorded for the entire fatigue spectrum. As mentioned earlier, filters were used between the acoustic emission transducer and recorders. The frequency band of these filters was carefully selected to allow only noise generated by the Fig. 3 Specimen with Compression Stabilization Plates and Acoustic Emission Transducer specimen to be recorded. The filter frequency band was determined by testing until all extraneous noises could be identified and blocked out. Also, to guarantee that no other sound sources existed the specimen was the only one present in the fa-' tig~e.chain duri~g testing. The remaining four pos1t10ns were f1lled with aluminum "dummy" specim~ns. Whenever possible the specimen was run at n1~ht when the chance of picking up background no1se from other electric equipment was at a minimum. Ultrasonic Monitoring3 At intervals corresponding to each half lifetime of loading, the specimens were rem?ved from the fatigue machine and inspected ultrason1cally. The ultrasound was transmitted and received with a single 5 MHz focused immersion transducer of 2.54 em (1.0 in.) diameter and 6.35 em (2.5 in.) focal length. These dimensions produce a f?cal spot' o~ ap~roximately 1.27 mm (0.05 in.) in d1ameter, wh1ch 1s comparable to the thickness of commonly inspected laminates. The transducer was operated by an ultrasonic analyzer operating in the pulse echo mode. The analyzer provides a time domain output for viewing the received pulse on an oscilloscope (A-scan) and a peak output detector for detecting and recording the local peak within the gated portion of the pulse. The.scan~ing system that was used to inspect the ~pec1~en 1s capable of automatically scanning and ~ndex1ng the transducer so specimens up to app~oxlmately 38 em (15 in.) square can be inspected w1th a single positioning. The transducer is linked to an X-Y recorder via displacement transducers so that a hard copy of the data can be generated. Figure 4 shows the various types.of ~cans w~ich can ~e generated by the system at th1s t1me. F1gure 4a 1s a C-scan representation ?f.a.[(0/±45/~0lsl2 graphite/epoxy specimen with an 1n1t1al flaw 1n the form of a circular hole. In this mode an alarm circuit with two limits is used 70 which causes the pen to lift from the paper whenever the set limits are exceeded. These limits were determined using a standard specimen. Figure 4b is an analog scan of the same specimen. In this mode the gated peak voltage of the reflected pulse is recorded as a deflection of the pen normal to the scanning direction. The variations of this voltage are related to the presence of flaws. Figure 4c is an offset angle analog scan of the same specimen. In this mode a component of the X-axis signal is fed into the Y-axis signal resulting in a perspective view of the specimen. Flaws now appear as troughs and gray tones on a uniformly displaced background. When a series of C-scans for one specimen are superimposed upon one another a flaw map is generated. In this way the total flaw growth history for a specimen can be presented in one illustration. The ultrasonic scanning was rigidly controlled with all variable setting controls considered frozen throughout the test. A fixture was used to insure the relative position of the specimen to the scanning tr~nsducer for each inspection. Usually the C-scan 1s the only hard copy representation of a specimen under normal conditions, but this is only a small part of the data available by ultrasonic inspection. In addition to C-scans, such as Fig. 2, the ultrasonic wave was analyzed and recorded on a spot basis by photographing the ultrasonic pulse on the oscilloscope (A-scan) and the frequency spectrum of the pulse. GENERAL PROCEDURES To minimize the possibility of generating false or m~sleading data extreme care was taken during the test1ng procedure. The testing, including insertion and removal from the fatigue machine, attachment and removal of the acoustic transducer, and ultrasonic scanning, was ~ond~cted by one person. Photographs, such as shown 1n F1g. 1 were taken at regular intervals to supplement the acoustic emission and ultra-. sonic data. Due to its operation characteristics the ultrasonic scanner may not pick up surface irregularities which can be seen if the specimen is inspected visually. . To avoid damage to the specimen during insertlon or removal from the chain a special torquing sequence was developed for the grip bolts. This seguence required the use of a preset torque wrench wh1ch would assure satisfactory gripping forces on the tabs of the specimen without causing bending or dama~e to the speci~en. Also, to avoid buckling or bend1ng of the spec1men during loading, stabilization plates, visible in Fig. 3, were attached over the center of the specimen gage section. A teflon film was placed on the stabilization plate to minimize friction and extraneous noises. RESULTS Acoustic Emission2 Figure 5 shows flaw maps of the cumulative damage through 0, 4, 8, and just before 10 lifetimes, respectively. As was seen for other specimens with this initial flaw type,l the flawed area increased in a generally radial pattern around the hole at a nearly uniform rate. The damage indicated in Fig. 5a is associated with the fiber breakage caused when the hole was machined. The rapid