时间：2022-10-12 11:11:12
( China South West Architectural Design and Research Institute, 610041, Chengdu )
Abstract. In civil engineering field, huge amount of structural health monitoring work is required throughout the service life of the structures. In pursuing an autonomous, real-time and online non-destructive evaluation (NDE) method, piezo-impedance transducer (PZT) has attracted more and more attention from researchers in recent years. Although up to now, the real application of Electro-Mechanical Impedance (EMI) method with PZT on structure health monitoring is not utilized, experiments results shows its impressive potential.
Keywords: Piezoceramic; PZT; Structural Health Monitoring; Fatigue Damage; Damage detection
The main objective of this research is to investigate the ability of PZT patch to detect and characterize the damage on a simple, lab size structure induced by graduate wearing (fatigue loading) produced by a 25 tons dynamic machine in laboratory.
Studies conducted for this thesis were mainly experimental based. Two aluminum beams with circular notches were tested in laboratory under fatigue loading until failure. With the surfaced bonded PZT, electrical admittances of this simple structure were recorded during the loading process.
The analyzed results from this study showed that EMI technique using PZT patch is capable of detecting damage and incipient crack induced by fatigue loading. The observed trend of electrical admittance shift with respect to loading cycle showed that EMI technique is good in detecting the progressive damage of the structure. A distinct turning point on the conductance shift versus load cycles plot showed a good indication of crack appearance. Possibility of calculating the remaining life of the structure was also discussed. Conclusion of this experimental based project was presented at the end.
Introduction
Structural health monitoring (SHM) needs to be done on a regular basis right after the structure has been constructed. According to US Federal reserve board, the failure of civil infrastructure systems to perform at the expected level may cause a reduction in the national gross domestic product (GDP) by 1% [1]. The non-destructive, intelligent heath monitoring methods especially those with smart materials are becoming more and more popular.
Piezoelectric material such as lead-zirconate-titanate (PZT) is one of them which has been intensively researched and exploited by researchers. The main advantage of this kind of material is that it can act as Actuator and Sensor simultaneously. PZT is able to actuate the targeted structure under an applied alternating current (A.C.) and simultaneously measure its response in terms of electrical admittance. This dual function of PZT better developed the feasibility of structural health monitoring in terms of accuracy, range of frequency and ease of use.
There are not many test conducted on fatigue damage detection. In this paper the main task will be focused on the study of feasibility of EMI technique employing PZT patch to detect damage caused by fatigue load.
EXPERIMENTAL PROCEDURE
Two aluminum (T6061 T6 Hand Forged) beams with dimensions (300x50x6 mm) were prepared with notches to undergo the fatigue tests. In order to reduce fatigue load cycles and also to get a reasonable result a notch on the specimen which introduces stress concentration on the specimen is a good option.
Although the presence of the notch will increases the stress concentration near its vicinity, and may still causes local yielding of that part. As long as no global yielding of the whole specimen happens, and imposes no significant effect on the PZT patch, it is considered to be acceptable.
Three different kind of notch were calculated. As shown in Fig 1.
Figure 1: 3 types of notch proposed (a) An Edge Notch, (b) A V Notch and (c) A Symmetric Circular Notch
As for notch a:
Material Property for aluminum T6061 T6 Hand Forged
Ultimate strength: Su = 340 MPa
Stress life curve intercept: SF’= 603 MPa
Slope of the stress life curve: b = -0.097
Fatigue limit for aluminum was set to be at 10E6 cycles
Fatigue limit stress: σFL= 603 x (10E6)-0.097 = 157.8 MPa
An edge notch with r = 20 mm, was calculated here.
Load factor: kL = 0.9 [2]
Surface finish factor: kSF = 0.98 [3] empirically chosen for this experiment
Stress concentration factor: kt = 2.17 [4]
Ultimate strength of the specimen: Su = 340 MPa
Stress life curve intercept: SF’= 603 MPa
Slope of the stress life curve: b = -0.097
Fatigue limit: σFL= 157.8 MPa
For this experiment, a mean load 24 KN with an alternating load of 12 KN was applied. The stresses on the specimen 300mm2 cross section were 80 MPa for mean stress (σmean) and 40MPa for alternating stress (σa). And the final life time was calculated.
If kT (σmean+σa) < 0.8 Su
Where n is the safety factor
kt (σmean+σa) = 2.17 x (80+40) = 226.8 < 0.8 x 340
Therefore
n -1 = (2.17 x 40) / (157 x 0.9 x 0.98) + (2.17 x 80) / 340
n = 0.88.
n < 1 indicated that the fatigue failure would occur.
The new slope of the S-N curve for the notched specimen was calculated
New slope = b log10 [ kf /(ksf*kl)]/6
New slope = -0.16216
Nf = {[σmin x (1 + σmax/σmean)]/ SF’}^slope-1
Therefore Nf = 2.2*105 cycles.
kf (σmean+σa) < 0.8 Su, the load applied in this test is quite small. It is good to avoid the yielding problem.
If the fatigue test machine operates at 10 Hz, it takes about 6 hours to finish this experiment without stopping, which is acceptable.
V notch and symmetric circular notch were calculated in the same way. The r, d and α for case b were set at 5mm, 20mm and 90 degree respectively , while for case c, the diameter of the hole was set at 20mm. All the other factors and the loading remained the same.
A summary of calculations for these three cases are presented in Table 1
Table 1: Results for 3 different types of notch.
Case c was chosen for fatigue loading test due to the reason that it was simple to be manufactured in the laboratory, and also easy for future finite element modeling.
The location of PZT patch and notch hole were determined as shown in Figure 2
Figure 2: Dimensions of fatigue test beam
Test for 1st specimen
The main instruments utilized in this experiment include Wayne Kerr precision impedance analyzer 6420, a laptop and the 25 ton dynamic test machine.
The PZT patches adopted in this study were PIC 151 [5] (PI Ceramics, 2006), a modified lead zirconate titanate (PZT) ceramic with high permittivity, coupling factor and charge constant. The 25 ton dynamic test machine was used to apply a constant amplitude tension force on the specimen to produce fatigue effect on the specimen. The impedance analyzer supplied the PZT patch with 1 volt A.C to excite the structure; it also measured the electrical admittance of the bonded PZT patch. The laptop not only recorded all the data measured by the impedance analyzer, but also input them into a Micro Soft Excel work sheet for future use.
The specimen together with the impedance analyzer were moved into the test room in the laboratory. After approximately 20 minutes for the specimen to adjust to room temperature, 2 baselines were recorded. The excitation voltage and frequency range for this test were set at 1 volt and 10 to 200 k Hz respectively. The frequency of the 25 ton dynamic machine was set at 15 Hz throughout this test. After the baselines were recorded, the specimen was mounted onto the 25 ton dynamic test machine. The specimen was always gripped at the top first, and the PZT patch side is facing up throughout the experiment so as to give a consistent result.
The control panel was then used to set the load to a predetermined value. A mean load of 24 KN and an alternative load of 12 KN were set to achieve a mean stress of 80 MPa and an alternative stress of 40 MPa on the 300mm2 cross section. The loading cycles were also set so the machine will stop at targeted cycles of loading and the specimen could be taken out from the machine for recording.
The electrical impedance was recorded twice after each loading operation. One recording was carried out right after the release of specimen, while the other one was conducted 10 minutes after the release. The specimen was expected to adapt to the room temperature before second recording. The specimen was put on a smooth and flat surface to achieve a free boundary condition when the electrical impedance was recorded.
The recording of electrical impedance of the first specimen was carried out from 10,000 to 80,000 cycles at a constant interval of 10,000 cycles. The specimen failed earlier than the calculated life cycles. No surface crack or crack propagation was observed.
Test for 2nd specimen
The setting for the test of the second specimen was the same as the first one. Larger number of cycles was input into the 25 ton dynamic machine at the early stage with all the excitation voltage, frequency range and load frequency remained unchanged. After 70,000 cycles of loading, the load interval was selected to be smaller and smaller. Crack was observed at 104,000 cycles with the aid of a special magnifying glass.
After crack was observed on the surface, not only the load cycles but also the load frequency was reduced so as to better trace the behavior of the structure with crack propagation.
Results at 30, 70, 75, 80, 85, 90, 95, 100, 104, 104.2, 104.8, 105.2 thousands of cycles and after failure were recorded along with the crack lengths. After the whole process, the PZT patch was still in good sharp and functional.
Another two aluminum beams with different dimensions were also tested in this project. Progressive damage was induced to simulate crack propagation which can be compared with the fatigue test.
Results and Discussion for Fatigue Test
1st fatigue test
No crack or crack propagation was observed. The test was unable to trace the crack propagation on the specimen.
However, from the results obtained, there were still some traceable fatigue damage effects on the aluminum beam. Figure 3 shows the effect of the constant amplitude fatigue load on this specimen. Note that, although electrical admittance consists of conductance and susceptance which are the real and imaginary part respectively, only the conductance is used here. This is because the conductance signatures are more sensitive to damage detection [6].
Figure 3: Conductance versus frequency plot for first fatigue test.
From Figure 3 it is apparent that after 30,000 cycles of loading, the conductance signature either peaked down a lot or shifted leftward, and after 60,000 cycles of loading the signature shifted left by 0.1-0.2 kHz. The reduced peak may be due to the slight shift of the resonance frequency, say in the order of 0.01 kHz, but the impedance analyzer utilized here only possess the ability to measure the electrical admittance to the step of 0.1 kHz. So instead of recording the exact peak, the smaller after peak value at frequency of 0.1 kHz stepwise was recorded. This result proved that the fatigue induced damage (micro cracks) before visible surface crack appeared was detected by the PZT patch.
Second fatigue test
15 sets of data were recorded including 2 sets of baseline. At around 80,000 cycles, minor scratches were observed around the circular notch. Its intensity increased as the applied loading cycles went up. At 104,000 cycles, a 5 mm crack was observed with the aid of magnifying glass. Data at 104,300, 104,800 and 105,200 cycles were recorded with crack lengths of 7, 8 and 10 mm respectively. The specimen failed when it was re-mounted on to the dynamic machine after the data acquisition for 105,200 cycles of loading. Although the crack propagated from only one side of the notch, the failure pattern was as expected, i.e., perpendicular across the notch center line. After failure, another set of data was recorded to ensure the PZT patch was still functional, while also serve as future reference.
As there were a lot of data recorded in this experiment, the analysis was divided into parts. Results are presented for the overall view, for before crack appearance and for crack propagation period.
Previous experience showed that for indentifying the peak shift, the frequency range selection should be 40-80 kHz. Data acquired from 80 kHz above subjected to relative large fluctuation. An overview of shift pattern from 40-60 kHz is presented in Figure 4. Data used here are intended to show the shifts between healthy state (baseline 2), fatigue load process (70,000 cycles) and crack appeared state (104,800 cycles).
Figure 4: Plot of conductance versus frequency for second fatigue test at healthy state, during loading process and with crack occurred
General left ward shift of the peak can be observed here. Figure 4 shows that the PZT patch worked well throughout the test, and it was able to detect the changes in specimen induced by fatigue damage.
Particularly for this test, the results from frequency range 56.5 to 57.5 kHz was found to be very good for damage characterization. Thurs, the following plots of conductance versus frequency are in this range.
Figure 5 is extracted from Figure 4, from the range 56.5 to 57.5 kHz, to show in detail how the peak shift happened.
Figure 5: Overall view of conductance versus frequency plot for specimen two (56.5-57.5 kHz)
Figure 6a presents the shift of the peak for data acquired before crack was observed. While Figure 4.6b shows the shift pattern after the crack was discovered. In both plots, conductance shift can be found, a good indication of induced by constant amplitude fatigue load. Although the shifts were about 0.1-0.3 kHz in both plots, the actual shifts per loading cycles are quite different. In plot a, the data sources have differences in multiple of ten thousands loading cycles while in plot b; the differences are only in the order of a few hundreds. This clearly indicates that after the crack was formed, deterioration of the structure became much faster.
Therefore a plot between loading cycles versus peak shift in that frequency range (56.5 - 57.5 kHz) which includes all the 15 sets of data were shown in Figure 7.
(a)Before crack detection
(b)After crack detection
Figure 6: Conductance versus frequency plot for specimen two from 56.5-57.5 kHz
Figure 7: Shift of conductance peak versus loading cycle plot for the second specimen. (56.5-57.5 kHz)
As shown in Figure 7, a sharp turning point was found at around 100,000 cycles. This is a good indication of the first presence of crack. Although the crack was observed at 104,000 cycles with a length of 5mm, crack with shorter length should have been developed before that. If more data were taken and the step interval of the impedance analyzer was smaller, a smoother curve would be plotted. Hence a more accurate turning point can be found which will serve as an indication of crack initiation.
Another issue regarding this test is that, with similar structure and more test results, this method can foretell the remaining life cycles of the structure. Assuming the crack started right at 100,000 cycles, shift of conductance peak versus crack length can also be plotted as shown in Figure 8.
Figure 8: Plot of shift of conductance peak versus crack length in the frequency range of 56.5-57.5 kHz
A slightly concave line is seen here which essentially means the disturbance on the structure’s stiffness become more severe when the crack is propagating.
Conclusion
The objective of this project has been achieved. Data for healthy state, progressive load up and crack appearance were recorded successfully for the fatigue test specimen. Shifts of conductance peak were observed in both fatigue test specimens. Incipient damage and visible crack were detected by the surface bonded PZT patch. The successful results from the second specimen proved that EMI technique employing PZT patch works well for fatigue damage detection. A plot of conductance peak shift versus loading cycles has a distinct turning point which is a good indication for crack appearance.
References
[1] Aktan, A. E., Helmicki, A. J. and Hunt, V. J., Issues in Health Monitoring for Intelligent Infrastructure, Smart Materials and Structures.[J] 1998, Vol. 7, No. 5, 674-692.
[2] Pook, L. Metal Fatigue, Springer, The Netherlands, 2007.
[3] Hertzberg, Richard W. Deformation and Fracture Mechanics of Engineering Materials, J. Wiley & Sons, 1996.
[4] Pilkey, Walter. D. Peterson's stress concentration factors, 3rd edition, John Wiley, 2008
[5] PI Ceramic , Product Information Catalogue, Lindenstrabe, Germany, 2010, www.piceramic.de.
[6] Sun, F. P., Chaudhry, Z., Rogers, C. A., Majmundar, M. and Liang, C. Automated Real-Time Structural Health Monitoring via Signature Pattern Recognition, edited by I. Chopra, Proceedings of SPIE Conference on Smart Structures and Materials, San Diego, California, 1995, Feb.27-Mar1, SPIE vol. 2443, 236-247.