ISSN 1061-8309, Russian Journal of Nondestructive Testing, 2009, Vol. 45, No. 2, pp. 103–108. © Pleiades Publishing, Ltd., 2009. Original Russian Text © L.N. Stepanova, E.Yu. Lebedev, S.I. Kabanov, V.N. Chaplygin, S.A. Katarushkin, I.S. Ramazanov, K.V. Kanifadin, 2009, published in Defektoskopiya, 2009, Vol. 45, No. 2, pp. 39–46.
ACOUSTIC METHODS
Study of Fracture of Specimens Made of Fiberglass Plastic Using Acoustic-Emission and Strain Measurements L. N. Stepanova, E. Yu. Lebedev, S. I. Kabanov, V. N. Chaplygin, S. A. Katarushkin, I. S. Ramazanov, and K. V. Kanifadin Chaplygin Siberian Research Institute of Aviation, ul. Polzunova 21, Novosibirsk, 630051 Russia Received September 9, 2008
Abstract—The crack resistance of specimens made of a fiberglass plastic composite material has been studied. All specimens had approximately the same geometric dimensions. At the center of the specimens, there were 25-mm-long notches on both sides. The specimens were loaded with a static or cyclic load at a frequency f = 5 Hz. The tests were performed using acoustic-emission (AE) and strain gage measurements. The AE technique allowed stable localization of a flaw at an early stage of growth and made it possible to automate the measurement process. The strain gage measuring system was used to find deformations in the zones of localization of AE signals. DOI: 10.1134/S1061830909020041
The imperfection of the contemporary technology of manufacture of various composite materials (CMs) and articles from them causes the formation of numerous flaws in a structure (unstuck layers, exfoliation, impurity inclusions, fracture of fibers and the matrix, etc.). The special features of interior and exterior flaws in a CM cause difficulties in choice of means and in development of methods of nondestructive testing. This necessitates studies of specimens made of CMs to find their crack resistance and the resistance of the material to static, cyclic, and dynamic stresses. For testing of CM articles, the acoustic-emission (AE) technique is used because it makes it possible to locate a flaw at an early stage of development and to detect nucleating cracks by registering acoustic noises. AE signals carry information about sizes of cracks, their propagation rates, stresses at the nucleation sites, and the elastic properties of the material. The problem of early detection of fracture of CMs is important and enables solution of a wide range of theoretical and practical problems [1–3]. The purpose of this study is to investigate the possibilities of early detection of growth of flaws using AE techniques and strain gage measurements in specimens made of fiberglass plastic under static and cyclic loading. The bearing capacity of the CM specimens was estimated during their testing without artificial stress concentrators. In order to find the critical loads that caused fracture of specimens, they were loaded with a linearly increased static load. A static load was imposed on ten of the studied specimens, and a cyclic load was imposed on eight specimens. Acoustic noises appearing during loading of the specimens reached acoustic-emission transducers (AETs), which transformed an acoustic signal into an electric signal. Information from the AET was registered by an ëñÄÑ-16.03 AE system (Certificate of the Federal Agency for Technical Regulation and Metrology RU.C.28.007 no. 19913/2, registered in the State Register of Means of Measurement under no. 18892-05) [3]. Deformation picked off the strain gages was measured by an ååíë-64.01 microprocessor multichannel strain gage system (Certificate of the Federal Agency for Technical Regulation and Metrology RU.C.28.007 A no. 10749, registered in the State Register of Means of Measurement under no. 21760-01) [3]. Information from the ååíë-64.01 strain gage system was translated according to a network protocol to the ëñÄÑ-16.03 system and the loads responsible for AE signals were registered during recording of the signals. The specimens were prepared from fiberglass plastic with the same geometric dimensions, 200 × 600 × 20 mm. To simulate a flaw, artificial concentrators, 25-mm-long notches, were made at the center on both ends of the specimens. Specimen nos. 4 and 7 were subjected to a static load. Figure 1a shows the results obtained during strain gage measurements of specimen no. 4. It is seen from the curves that the first plastic deformations were observed under a constant load P = 5 t, when the defor103
STEPANOVA et al. 10
8 (a) 7 6 5 5 4 4 3 3 2 2 1 1 0 6 (b)
9
Strain gage 1
8
Load 7 6
Strain gage 8
Strain gage 7 Strain gage 2 12
18 Time, min
10 9 8 7 6 5 4 3 2 1 0
Load, t
Strain × 10–3
104
24
30
36
3
0 1 2 3 4
2
1 AET 1
Strain gages 1–4
AET 2
Fig. 1. (a) Results of measuring strains dependent on time in specimen no. 4 and (b) localization of AE signals during loading of specimen no. 4.
12
Load
10
20000 7
15000
5
10000 5000 0
8
10
10
9
8
6
4
6 4
Total count
3
Load, t
Total count of AE signals
25000
2
1 2 4
8
12 16 Time, min
20
24
0 28
Fig. 2. Dependence of change in the total count of AE signals and load on time for specimen no. 4.
mation on measuring gage 8 was ε = 4 × 10–3. When the load was P = 7 t, strain gage 8 broke; specimen no. 4 stayed for ∆t = 14 min at P = 10 t and then failed. Figure 1b shows the localization of AE signals and the placement of AET 0–AET 3 and strain gages 1–4. Strain gages 5–8 are placed on the opposite side of the specimen symmetrically to strain gages 1–4. Registration of AE signals began at early stages of loading under loads P = 1–1.25 t and lasted up to fracture of a specimen. The majority of localized AE signals were detected from the artificial notches in the specimens. For specimen no. 4, the smoothed dependence of the change in the total count of AE signals on the load is shown in Fig. 2. The first plastic deformations during testing of specimen no. 7 appeared under a static load P = 10 t, while at P = 11 t it fractured in an avalanche-like manner. The appearance of plastic deformations in specimen no. 7 caused an increase in the total count of AE signals. Deformations registered at the place where strain gage 1 was stuck were about ε = 3 × 10–3. An increase in the static load applied to specimen no. 7 caused an increase in the total number of counted AE signals with high amplitudes (Fig. 3). RUSSIAN JOURNAL OF NONDESTRUCTIVE TESTING
Vol. 45
No. 2
2009
STUDY OF FRACTURE OF SPECIMENS MADE OF FIBERGLASS PLASTIC 4.0
18000 14000 12000 10000 8000 6000 4000
Total count of AE signals Strain at strain gage 1
3.338
3.5 3.0
3.022 2.688 2.334
2.004 1.668 1.294 0.945 0.62 0.311 1 2 3 4 5 6 7 Load, t
Strain × 10–3
Total count of AE signals
16000
2.5 2.0 1.5
105
A cyclic load at a frequency f = 5 Hz was applied to specimen no. 9, and the specimen was loaded up to P = 6–7 t; in the zone where the strain gages were stuck, deformations were about ε = (2–3) × 10–3. It endured N = 147.75 × 103 loading cycles and then failed at the place of the artificial notch. AE signals totaling about 140 × 103 in number were continually detected from the fracture zone of the specimen.
1.0
The localization of AE signals obtained during 3000 cycles of cyclic testing of the specimen up to fracture is shown in Fig. 4a. AE sig0 0 8 9 10 nals were mainly registered from the zone of the upper notch of the specimen, where it was later broken. The data of strain gage measureFig. 3. Dependence of total count of AE signals and strain on ments (Fig. 4b) also confirm the results the load for specimen no. 7. obtained with the ëñÄÑ-16.03 acoustic system. The greatest deformations were registered with strain gages 1 and 8, which were in the zone of the upper notch. In order to determine more precisely the localization zone of the fracture, 3D distributions of the total counted number of AE signals and the total energy deposited over the area of the specimen were calculated. The process of drawing these distributions is similar to drawing histograms in a plane. The localization zone is divided by a rectangular grid S into many meshes of a small area. For each AE event, the localization area L is calculated taking into account the error of measurements of the difference in the arrival times (DAT) of AE signals ∆T1 and ∆T2 at the strain gages of the piezoelectric antenna and the error of measurements of the 2000
0.5
(a) 3
0 1 2 3 4
2
1 AET 1
Strain gages 1–4
AET 2
8 (b)
Strain × 10–3
7
Strain gages 8
6 5
Strain gages 1
4 3 2
Strain gages 7
1 0
12
24
Strain gages 3
36 Time, min
48
60
Fig. 4. (a) Localization of AE signals in specimen no. 9 during 3000 cycles up to fracture; (b) readings of strain gages at peaks of load. RUSSIAN JOURNAL OF NONDESTRUCTIVE TESTING
Vol. 45
No. 2
2009
106
STEPANOVA et al. (a)
(b) 5E+008
15
15 10
AET 3
5 200
X,100 mm
AET 0
0 0 –50 –100 mm AET 1 0 –150 Y,
Distribution of the total count of AE signals
(c)
AET 2
200 X, 100 mm
AET 0 0
AET 0
1E+008 5E+007 0E+000
0 –50 –100 m 0 AET 1–150 Y, m
(d) 3E+008 2E+008
AET 3 AET 2 AET 0
200 X, 100 mm
(e)
1E+008 5E+007 0E+000
0 –50 m 1 –100 m 0 AET–150 Y,
Distribution of AE signal energy, mV2
AET 2
0 –50 –100 m 0 AET 1 –150 Y, m
2E+008
AET 2
5
3E+008
AET 3
10
AET 3
200 X, 100 mm
4E+008
(f) 3E+008
15
5 AET 2
200
X,100 mm
2E+008
10
AET 3 AET 0
0
0 –50 m 1 –100 , m 0 AET –150 Y
AET 3 1E+008
AET 2
200 X, 100 mm
AET 0
5E+007 0E+000
0 –50 –100 m m 0 AET 1–150 Y,
Fig. 5. Localization of AE signals taking into account errors in determining the arrival time at the strain gages of the piezoelectric antenna and errors in finding the sound velocity in specimen no. 9: (a, c, e) distribution of the total count of AE signals; (b, d, f) estimation of the energy distribution of AE signals.
sound velocity ∆C. The error of localization of an AE signal source is expressed as the sum of the contributions from ∆C, ∆T1, and ∆T2. In order to construct the distribution of the number of localized signals, the value 1/NL, where NL is the number of identical meshes, is added to the weight of each mesh S that has a nonempty intersection with the area L. In order to estimate the energy distribution, the following quantity was additionally calculated: i=K
Ef =
∑u , 2 i
i=0
where K is a fixed number of counts for estimating the energy (which was taken as corresponding to 50 µs); ui are AE signal counts. RUSSIAN JOURNAL OF NONDESTRUCTIVE TESTING
Vol. 45
No. 2
2009
STUDY OF FRACTURE OF SPECIMENS MADE OF FIBERGLASS PLASTIC Strain gage 6
8 7
(a) Strain gage 1
Strain × 10–3
6 5
Strain gage 4
4 3 2
Strain gage 5
1 0 Total count of AE signals
107
4000 3500 3000 2500 2000 1500 1000 500 0
500
1000 1500 2000 Number of loading cycles
2500
500
1000 1500 2000 Number of loading cycles
2500
(b)
Fig. 6. (a) Dependence of the maximum values of readings of strain gages on the number of loading cycles at peaks of load; (b) dependence of the total count of AE signals on the number of loading cycles for specimen no. 12.
The weight of S meshes corresponding to an AE signal has increased by Ef /NL. Figure 5 shows the distributions of the total number of counts and the energy of AE signals for the central part of specimen no. 9. The origin of coordinates coincides with AET 0. The initial stage of the tests corresponds to Figs. 5a and 5b. During the first cycles of loading, AE signals were received mostly from the zone of the end of the bottom notch of the specimen. Then, the zone of activity of AE signals shifted to the area of the upper notch. Figures 5c and 5d show the distributions of the total number of AE signals and their energy corresponding to the loading cycles during which the zone of activity of signals shifts. It is seen that the energy distribution has a distinct maximum in the area where AE signals will mostly be registered in the future. This means that estimation of the energy distribution of AE signals is more informative than the planar pattern of their localization. Figures 5e and 5f show the signals that correspond to the last stage of loading of specimen no. 9. For this case, the planar pattern of localization is shown in Fig. 4a. Thus, it follows from Fig. 5 that taking into account errors in finding the sound velocity ∆C and the DAT ∆Ti makes it possible to eliminate groups of signals detected with low reliability from the total pattern of signal localization and to obtain a clear and true pattern of localization of flaws. In Figs. 5e and 5f, the maxima of the distributions are seen to correspond to the area in the specimen where the most intense fracture process is observed. A cyclic load was applied to specimen no. 12 at a frequency f = 5 Hz, at which deformation was ε = 2 × 10–3. Then the load was increased to strains of (2.5–3.5) × 10–3 at the places where the strain gages were stuck. In this case, the specimen endured N = 2300 loading cycles, after which it was broken at the artificial notch. It follows from the readings of strain gage 1 (Fig. 6a) that, after N = 1500 loading cycles, it passed out of the range of acceptable strains. Its readings did not change during the following tests. This was due either to damage of the strain gage or to propagation of a fatigue crack under it. The strains of the other strain gages at load peaks were ε = (2.0–3.5) × 10–3. Figure 6b for specimen no. 12 shows the dependence of the total number of counted AE signals on the number of loading cycles. It should be noted that, in this specimen, there was a stable increase in the total number of AE signals as a result of growth of the fatigue crack. Signals registered by the ëñÄÑ-16.03 AE system were caused by the fatigue crack propagating from the artificial notch at the center of the specimen. RUSSIAN JOURNAL OF NONDESTRUCTIVE TESTING
Vol. 45
No. 2
2009
108
STEPANOVA et al.
Analysis of the front rise time of AE signals makes it possible to assert that they have some spread in this parameter. However, for the majority of AE signals, the duration of the leading edge did not exceed 40–100 µs. Figure 7 shows a photograph of the fracture zone of specimen no. 12. Studies of the fractures have shown that, in this specimen, the fiberglass fabric was exfoliated. Specimen no. 21 was loaded with a static load in two stages. At the first stage, the initial load applied to the specimen at the place where the strain gages were stuck caused strains ε = (1.7–2.3) × 10–3. The specimen endured this load for 17 h (segment AB in Fig. 8). During Fig. 7. Photographs of the fracture zone of specimen no. 12. this time, strains at the places where the strain gages were stuck almost did not change, and, at the end of this period, they were ε = (2–2.7) × Total count of AE signals 10–3. From the artificial notch, about 10 × 103 50000 AE signals were detected (Fig. 8). At the second stage of the studies, the load E was increased and strains at the strain gages 40000 were in the range ε = (2.5–3.5) × 10–3. At the Second loading 11th minute of the second stage, the load was 30000 decreased to zero (point C in Fig. 8), and the specimen stayed unloaded for 58 min (segment C D 20000 CD). It was found that residual strains remained at the strain gages. Then a load was First loading B applied to the specimen that provided deforma10000 tion of about ε = 3 × 10–3. During 18 min of loading (segment DE), 35 × 103 signals were 0 A registered, thus confirming the high counting 200 400 600 800 1000 1200 1400 Time, min rate of AE signals and intense fracture of specimen no. 21. Thus, fiberglass specimens were loaded with static and cyclic loads up to fracFig. 8. Time dependence of the total count of AE signals durture during the tests. The critical strain for the ing the first and second loadings of specimen no. 21. specimens was ε = (2–2.5) × 10–3. Residual strains most often appeared during cyclic tests. The studies performed using AE techniques and strain gage measurements made it possible to find that fatigue cracks in fiberglass specimens propagated in the region of the stress concentrator, namely, the artificial notch at the center of a specimen. However, in three specimens (nos. 10, 11, and 14), fractures were in the grip zones. Fractures of the specimens caused an intense flow of AE signals even in the case of an insignificant increase in the length of the fatigue crack. The propagation velocity of AE signals was 3000 m/s, and the leading edge was short and did not exceed several tens of microseconds; therefore, they were reliably localized. Taking into account errors in determining the sound velocity ∆C and the DAT ∆Ti makes it possible to eliminate the group of signals detected with low reliability from the localization pattern, which provides a clear and true pattern of the disposition of flaws. REFERENCES 1. Stepanova, L. N., Lebedev, E. Yu., Kareev, A. E., et al., Use the Acoustic Emission Method in Detecting the Fracture Process of Specimens Made of Composite Materials, Defektoskopiya, 2004, no. 7, pp. 34–41 [Rus. J. Nondestruct. Test. (Engl. Transl.), 2004, vol. 40, no. 7, pp. 455–461]. 2. Stepanova, L. N., Chaplygin, V. N., Lebedev, E. Yu., et al., Use of the Acoustic Emission during Cyclic Testing of Composite Elements of Aircrafts, Kontrol’. Diagnostika, 2004, no. 12, pp. 53–56. 3. Ser’eznov, A. N., Stepanova, L. N., Murav’ev, V. V., et al., Diagnostika ob’ektov transporta metodom akusticheskoi emissii (Diagnostics of Transport Objects by Acoustic-Emission Technique, Stepanova, L. N., Murav’ev, V. V, Eds., Moscow: Mashinostroenie, 2004. RUSSIAN JOURNAL OF NONDESTRUCTIVE TESTING
Vol. 45
No. 2
2009