Instantaneous Delamination Detection in a Composite Plate using a Dual Piezoelectric Transducer Network

Transcription

1 Instantaneous Delamination Detection in a Composite Plate using a Dual Piezoelectric Transducer Network Chulmin, Yeum Department of Civil and Environmental Engineering Korea dvanced Institute of Science and Technology Daejeon, Republic of Korea June 4th, 2 Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory

2 Outline Introduction 2 Instantaneous Damage Detection lgorithm 3 Extraction of the Lamb Wave Mode 4 Damage Classification 5 Experimental Results 6 Conclusion Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 2

3 Motivation Greater use of super strengthened plastics in the 787 raises concerns about detecting damage- now done using a quarter- but company says visual inspection will be enough.. Like Boeing, irbus says it also has designed the 38 so that invisible damage cannot produce a significant subsurface flaw, and that ultrasounds and other imaging methods are needed only if there's visible damage... Chicago Tribune September 2, 27 Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 3

4 Non-destructive Inspection Materials for Boeing % 5% Composit e luminum 5% Visual Inspection Non-destruction Testing Structural Health Monitoring Themography Phased array UT Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 4

5 The Objective of This Study Development of a delamination detection technique on a composite plate Robustness of the proposed technique against environmental effects such as temperature variation pplication of the Lamb wave mode decomposition technique using concentric ring and circular PZTs Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 5

6 The Uniqueness of This Study Identification of delamination without using prior measured baseline data Propose the instantaneous delamination detection algorithm using the time delay of the Lamb wave mode Extraction of the Lamb wave mode without changing the PZT size and/or spacing configuration Journal and Patent Publication Chul Min Yeum, Hoon Sohn and Jeong Beom Ihn, Lamb wave mode decomposition using concentric ring and circular PZT Transducers, Submitted to Journal of coustical Society of merica, 2. (Impact factor: 2.8) Hoon Sohn, Chul Min Yeum and Jeong Beom Ihn, lamb wave mode decomposition technique using amplitude matching, Submitted to the Us patent office, Docket No Chul Min Yeum, Hoon Sohn and Jeong Beom Ihn, Reference-free delamination detection and localization in a composite plate using a dual piezoelectric transducer network, In preparation for Composites Structures, 2. (Impact factor: 2.53) Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 6

7 Characteristics of Laminate Composites Laminate composites Laminate composites involve two or more layers of the same or different materials. The layers can be arranged in different directions to give strength. dvantage : Lightweight, superior specific strength Disadvantage : Invisible impact damage, expensive dvantages of Lamb wave based damage detection technique Inspection of large areas with little attenuation Excellent sensitivity to multiple defects Te lack of need for complicated and expersive insertion/radiation devices Difficulties of Lamb wave based damage detection technique on the laminate composite nisotropic nature of a laminate composite very fast wave velocity ( The S mode is four or five times faster than the mode) Highly damping coefficients Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 7

8 Literature Reviews Incident wave Delamination Transmitted wave Excitation/Sensing Pulse-echo Delamination Reflection Converted mode Excitation Pitch-catch Sensing The effect of dealmination on Lamb wave propagation Delamination. Measuring the group velocity and/or energy of the reflected modes [Valdes(22);Ip(24) ] 2. Image construction using the cross-correlation of the scatter signal envelop [Ng (29)] 3. Damage quantification based on the changes in energy contents of scatter waves [Ihn(28)] 4. Compute the probability of damage using correlation coefficients of measured signals [Wang(28)] 5. Group delay measurements using modally selective Lamb wave [Petculescu(28)] 6. Delamination size detection using time of flight of the converted mode [Ramadas(2)] Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 8

9 Outline Introduction 2 Instantaneous Damage Detection lgorithm 3 Extraction of the Lamb Wave Mode 4 Damage Classification 5 Experimental Results 6 Conclusion Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 9

10 The Effect of Delamination on the Mode mode delay Incident wave Delamination If Lamb waves propagating along a thin plate encounter delamination, some portion of waves are scattered or reflected at the boundary and others are transmitted Delamination through it. Especially, the transmitted mode is more delayed and attenuated than mode attenuation the transmitted S mode. [Petculescu at all (28)] Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory

11 Why Should We pply the Reference-free Damage Detection Technique? Output(V) Output(V) Excitation Sensing. Intact(2 o C) Inatact(6 o C) Time(ms). Intact(2 o C) Damage(2 o C).5 mode mode Time(ms) Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory

12 Reference-free Damage Detection Techniques. Damage identification using instantaneously measured Lamb wave signals obtained from only one pair of PZTs B Tested path PZT 2. Damage identification using instantaneously measured Lamb wave signals obtained from other undamaged paths (Independent of path lengths and directions) B Tested path Reference paths 3. Damage identification using instantaneously measured Lamb wave signals obtained from other undamaged paths having same directions and lengths B Tested path Reference path Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 2

13 Output Voltage(V) (V) Determination of the Damage Sensitive Feature Experimental configuration Undamaged condition 7-8 path 6-7 path Time(ms) Multimodal characteristic Complex boundary conditions Very fast S mode velocity Sensitive to the bonding condition mode Time delay Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 3

14 Outline Introduction 2 Instantaneous Damage Detection lgorithm 3 Extraction of the Lamb Wave Mode 4 Damage Classification 5 Experimental Results 5 Conclusion Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 4

15 Voltage What is the Lamb Wave Decomposition? Excitation Sensing 2 S + modes 2 How (?) mode Time S mode Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 5

16 Conventional Techniques for the Lamb Wave Decomposition Wedge transducer [Wilcox (22)] Collocated PZTs on both surfaces [Kim (27)] D PZT Plate (Intact) B C B S C B C Comb transducer [Rose ( 998)] n array of PZTs with time delays [Gao (27)] Tuning of the driving frequency [Giurgitiu (23)] Lamb wave PZT Structure Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 6

17 Conventional Techniques for the Lamb Wave Decomposition - Wedge transducer - Wedge transducer [ Wilcox (22) ] Collocated PZTs [ Kim and Sohn (27) ] D PZT Plate (Intact) S D D Comb Transducer [ Rose ( n array of PZTs with time delays [ Gao (27) ] Problems Tuning of the driving frequency [ Giugitiu (25) ] Difficulty of setting the angle of incidence with appreciable accuracy Consideration of time delay due to block. PZT Lamb wave Significant signal attenuation before impinging the inspection material Generation of additional reflected waves from interfaces Structure Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 7

18 Conventional Techniques for the Lamb Wave Decomposition - Collocated PZTs - S mode Wedge transducer [ Wilcox (22) ] Collocated PZTs [ Kim and Sohn (27) ] mode D PZT Plate (Intact) S D D Problems Comb Transducer [ Rose ( n array of PZTs with time delays [ Gao (27) ] Tuning of the driving frequency [ Giugitiu (25) ] sensors Lamb wave Entrance PZT Structure Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 8

19 Conventional Techniques for the Lamb Wave Decomposition - Comb transducer - Problems Wedge transducer [ Wilcox (22) ] Decomposition of Lamb waves at a specific frequency Collocated PZTs [ Kim and Sohn (27) ] Needs for a multi channel data acquisition system Sensitive to prescribed time delay profiles or wavelength PZT S D Plate (Intact) D D Comb transducer [ Rose (998) ] n array of PZTs with time delays [ Gao (27) ] Tuning of the driving frequency [ Giugitiu (25) ] Lamb wave PZT Structure Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 9

20 Conventional Techniques for the Lamb Wave Decomposition - Tuning of the driving frequency - Wedge transducer [ Wilcox (22) ] Collocated PZTs [ Kim and Sohn (27) ] PZT S D Plate (Intact) luminum 224-T3.7 mm luminum 224-T3 7 mm D D Comb Transducer [ Rose ( n array of PZTs with time delays [ Gao (27) ] Problems Tuning of the driving frequency [ Giurgitiu (25) ] Decomposition of Lamb waves at a specific frequency Needs for a baseline tuning curve Lamb wave PZT Structure Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 2

21 Overview of the Proposed Technique V rc Dual PZT (excitation) Dual PZT (sensing) Input S mode mode S mode (reflection) V rc * Subscript s r and c : ring and circular PZTs * S is an ratio of S mode scaling at a specific frequency Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 2

22 Overview of the Proposed Technique V rr Dual PZT (excitation) Dual PZT (sensing) mode V rc V rr X S Input S mode mode S mode (reflection) V rc V rr Input S mode S mode (reflection) * Subscript s r and c : ring and circular PZTs * S is an arbitrary scaling at a specific frequency Input mode Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory V rc - V rr X S 22

23 Theoretical Response Model for 2D PZTs y x PZT (Excitation) a r s PZT B (Sensing) c ux V t sin a sin asinc c Displacement at x from the PZT [ Giurgiutiu (23)] u t i e e S S sin a NS S ( ) sin i x t a N i( xt) x () S ' S ' DS D Voltage Response at PZT B [Giurgiutiu (23)] Eshsg3 sin a sin c sin a sin c V ( t) e e S S S N S S N ( ) S i r ( st i rst ) S 2c ' S 2c ' DS DS Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 23

24 Theoretical Response Model for 3D Circular PZTs r 2a z r s 2c Infinite isotropic plate 2h u () t aj a r V t aj a? PZT transducer (Excitation) PZT transducer (Sensing) Displacement at x from the PZT [jay(24)] N S a it S S (2) S (2) ur ( r, z b) i e J a H ' S r J a H r ' DS D Voltage Response at PZT B [Lee and Sohn (2)] S rs c 2 2 Eshs g3a NS it S S (2) S 4r rs V ( t) i e J 2 a rh ' r 2tan dr S c DS rs c r rs c N Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 24

25 Two Noticeable Factors of Theoretical Equations for 3D Circular PZTs r 2a z r s 2c Infinite isotropic plate 2h Vt () 2 E h g a 2 S S S ' a J c S S S S ' S i t r s ( ) 4 S s s 3 i c r ( s N i t rs ) 4 N J a J c D J S D e e PZT transducer (Excitation) PZT transducer (Sensing) V t aj a J c c The amplitudes of the S and modes are fucntions of the excitation and sensing PZT sizes (a and c) In the fixed distance between the sensing and Input S mode mode S mode (reflection) V rc V rr excitation PZTs, signal phases does not change with respect to the variations of the PZT size Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 25

26 mplitude Voltage(V) Voltage(V) Voltage(V) Example of the Mode Extraction from Raw Signals ) Signals measured different sensing size Input S mode mode S mode (reflection) V rc V rr F22(3-4path) V rr F23(3-4path) V rc 2) Matching of the amplitude of the S mode Input S mode mode S mode (reflection) V rc V rr X S Time(ms) F22(3-4path) V rr F23(3-4path) V rc Time(ms) 3) Extraction of the mode mode(3-4path) Input mode V rc - V rr X S Time(ms) Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 26

27 Outline Introduction 2 Instantaneous Damage Detection lgorithm 3 Extraction of the Lamb Wave Mode 4 Damage Classification 5 Experimental Results 6 Conclusion Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 27

28 Summary of the Proposed Delamination Classification Delamination classification procedure Technical details ) ttach dual PZTs to the host composite plate 2) Decide an input frequency range using experimental tuning curves 3) Measure responses from all actuator and sensor pairs 4) Extract the Lamb wave mode. Determination of the time range of the mode using experimental group velocity 2. Calculation of correlation coefficients using instantaneous measured signals having same directions and lengths 3. Computation of the damage index from all paths 4. Setting up the threshold values based on ) Generalized extreme value distribution 2) K-mean clustering 3) Outlier analysis 4) Beta distribution 5. Decision making 5) Establish the damage classification Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 28

29 Cross Correlation based a Damage Detection Technique Cross correlation Cross-correlation analysis is used to determine the degree to which two signals are linearly related. This algorithm is sensitive to signal shape changes, but insensitive to amplitude changes. Damage index ( DI ) where corr is the cross correlation n d DI ( i) ( corr( ai, a j)) d, 45,9,35 2 n a i is the mode obtained from the path i. d is the direction of the path i. n d is the number of paths of the d direction a j is the mode obtained from the same direction of path i. d j i 2 DI Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 29

30 /2 x (- DI corr coeff) Determination of a Threshold Value using the Beta Distribution Beta distribution The beta distribution is a family of continuous probability distributions defined on the interval (, ) parameterized by two positive shape parameters, typically denoted by α and β. The damage index is bounded on (, ) The double bounded distribution should be used Threshold 99.7 % Driving Frequency(kHz) frequency Threshold Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 3

31 Outline Introduction 2 Instantaneous Damage Detection lgorithm 3 Extraction of the Lamb Wave Mode 4 Damage Classification 5 Experimental Results 6 Conclusion Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 3

32 Experimental Setup Impact head Impact tester Data acquisition system Temperature chamber - The dimension of each PZT : Data cquisition System Specimen * 9 packaged dual PZTs 4 rbitrary Waveform Generator : Sending signal to Multiplexer Multiplexer 4: Sending signal to Digitizer 2 PZT ctuator ctuating 3 Sensing * PSI-54E type - Input signal : tone-burst signal with ± peak-to-peak voltage frequency range 8 khz to 2 khz with an increment of khz - Sampling rate : 2MS/s Digitizer PZT Sensor - Power amplifier gain : 5 - Data averaging : 2 times - Temperature : -, 2, 5 o C Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 32

33 Sensor Configuration Dual PZT Impact location 2 Thickness =.5 Unit : mm Dual PZT Impact on the back side Path directions o : o : 4 Unit: mm 5 2 Thickness : o : 6 35 o : 4 Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 33

34 Output Voltage(V) Output Voltage(V) Comparison of the Raw Signals. Undamage Time(ms)). Damage Time(ms)) Raw signals obtained from o paths Damage path Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 34

35 mplitude Voltage(V) mplitude Voltage(V) Comparison of the Extracted mode Undamage Time(ms)) Damage Time(ms)) Extracted mode from o paths Damage path Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 35

36 DI /2 x (- corr coeff) DI /2 x (- corr coeff) /2 x (- corr coeff) DI DI /2 x (- corr coeff) /2 x (- corr DI coeff) DI /2 x (- corr coeff) Comparison of Correlation Coefficients Undamage Damage - o C - o C Driving Frequency(kHz) frequency (khz) 2 o C Driving Frequency(kHz) frequency (khz) 2 o C Driving Frequency(kHz) frequency (khz) 5 o C Driving Frequency(kHz) frequency (khz) 5 o C Driving Frequency(kHz) frequency (khz) Driving Frequency(kHz) frequency (khz) Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 36

37 Outline Introduction 2 Instantaneous Damage Detection lgorithm 3 Extraction of the Lamb Wave Mode 4 Damage Classification 5 Experimental Results 6 Conclusion Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 37

38 Limitations of the Proposed Technique. Enough paths with same direction and length are needed 2. Majority of the collected data will be recorded over undamaged sections 3. The group velocities do not change with respect to the input energy or propagation distance 4. Contributions of converted modes or scatter waves to measured signals are relatively small rather than those of the first arrival of the mode 5. The first arrival mode should not be overlapped with its reflection wave from structural boundaries Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 38

39 Concluding Remarks Summary. Development of the reference-free damage detection technique 2. Propose the mode extraction technique 3. Investigation of environmental effects on the proposed technique Future study. Development of a reference-free damage detection technique without using other reference paths 2. Investigation of sensor installation techniques to improve bonding condition between sensors and a structure 3. Works on the more reliable damage classifier Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 39

40 Reference S. H. Diaz Valdes and C. Soutis, "Real-time non-destructive evaluation of fibre composite laminates using low-frequency Lamb waves," J. coust. Soc. m., (22). [ISI] K. H. Ip and Y. W. Mai, Delamination detection in smart composite beams using Lamb waves, Smart Mater. Struct (24). C. T. Ng and M. Veidt, Lamb-wave-based technique for damage detection in composite laminates, Smart. Mater. Struct. 8, 746 (29). J. B. Ihn and F. K. Chang, Pitch-catch active sensing methods in structural health monitoring, Smart Mater. Struct (28). D. Wang, L. Ye and Z. Su, Probability of the presence of damage estimated from an active sensor network in a composite panel of multiple stiffeners, Compos. Sci. Technol (29) G. Petculescu, S. Krishnaswamy and J. D. chenbach, Group delay measurements using modally selective Lamb wave transducers for detection and sizing of delaminations in composites, Smart Mater. Struct (28). C. Ramadas, M. J. Padiyar, K. Balasubramanjam, M. Joshi and C. V. Krishnamurthy, Delamination size detection using time of flight of ani-symmetric() and mode converted mode of guided Lamb waves, J. Intell. Mater. Syst. Struct. 2, (2). P. D. Wilcox, M. J. S. Lowe, and P. Cawley, "Mode and transducer selection for long range Lamb wave inspection," J. Intell. Mater. Syst. Struct. 2, (2). S. B. Kim, and H. Sohn, Instantaneous reference-free crack detection based on polarization characteristics of piezoelectric materials, Smart Mater. Struct. 6, (27). J. L. Rose, S. P. Pelts, and M. J. Quarry, " comb transducer for mode control in guided wave NDE," IEEE Ultras. Symp. Proc (996). Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 4

41 Reference H. Gao, J. Rose and C. Lissenden, Ultrasonic guided wave mode selection and tuning in composites using piezoelectric phased arra ys, International workshop on Structural health monitoring (27) V. Giurgiutiu., "Lamb wave generation with piezoelectric wafer active sensors for structural health monitoring," SPIE. 556, -22 ( 23). Raghavan and C. E. S. Cesnik, Modeling of piezoelectric-based Lamb-wave generation and sensing for structural health monitorin g, SPIE. 539 (24) H. Sohn, S. J. Lee, Lamb wave tuning curve calibration for surface-bonded piezoelectric transducers, Smart Mater. Struct. 9, 5 7 (2). S. R. nton and D. I. Inman DJ, Reference-free damage detection using instantaneous baseline measurements, I. 47, (29). S. S. Kessler, S. M. Spearing, and C. Soutis, Damage detection in composite materials using Lamb wave methods, Smart Mater. Str uct., (22). I.. Viktorov, Rayleigh and Lamb Waves (Plenum, New York, 967). S. C. Wooh and Y. J. Shi, "Synthetic phase tuning of guided waves," IEEE Trans. Ultrason. Ferroelectr. Freq. Control 48, (2 ). Z. Su, L. Ye, and Y. Lu, Guided Lamb waves for identification of damage in composite structures: review, J. Sound Vib. 295, (26). Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 4

42 Do You Have ny Questions? I would be happy to help Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 42

43 Back up Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 43

44 DI DI DI /2 x (- corr coeff) DI /2 x (- corr coeff) /2 x (- corr coeff) /2 x (- corr coeff) /2 x (- corr coeff) /2 x (- corr coeff) Comparison of Correlation Coefficients Obtained from Undamage and Damage Conditions Undamage Damage Frequency(kHz).5 DI DI Frequency(kHz) Frequency(kHz) Frequency(kHz) Frequency(kHz) Frequency(kHz) Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 44

45 Comparison of Correlation Coefficients Obtained from same path direction and lengths /2 x (- corr coeff) /2 x (- corr coeff) /2 x (- corr coeff) 2 o C Frequency(kHz) 5 o C Frequency(kHz) - o C Frequency(kHz) Civil & Environmental Engineering Department, Smart Structures and Systems Laboratory 45