Guided waves for damage detection in rebar-reinforced concrete
beams
Ye Lu a ,⇑,Jianchun Li b ,Lin Ye c ,Dong Wang c ,d
a
Department of Civil Engineering,Monash University,Clayton,VIC 3800,Australia
b Centre for Built Infrastructure Rearch,School of Civil and Environmental Engineering,Faculty of Engineering and Information Technology,University of Technology,Sydney,NSW 2007,Australia c
Laboratory of Smart Materials and Structures (LSMS),Centre for Advanced Materials Technology (CAMT),School of Aerospace,Mechanical and Mechatronic Engineering,The University of Sydney,NSW 2006,Australia d
Aeronautical Science and Technology Rearch Institute of COMAC,Commercial Aircraft Corporation of China,Ltd.,Beijing 100083,People’s Republic of China
h i g h l i g h t s
Material property change due to cracking was assd by guided waves. Principal component analysis (PCA)was ud to classify corroded rebar.
Statistical parameters from wave signals were extracted as damage indices for PCA.
a r t i c l e i n f o Article history:
Received 28June 2012
Received in revid form 1May 2013Accepted 12May 2013
Available online 8June 2013Keywords:Guided waves
前言格式
Mechanical properties Crack detection Corrosion Concrete
a b s t r a c t
The propagation properties of ultrasonic waves in rebar-reinforced concrete beams were investigated for the purpo of damage detection.Two types of piezoelectric (PZT)elements were ud in experiments in which PZT disks were attached on the surfaces of concrete beams to obrve wave propagation in con-crete before and after a four-point bending test,while rectangular PZT patches were attached at the expod ends of the rebar to monitor wave transmission along the rebar with and without simulated cor-rosion in the form of partial material removal from the rebar.Experimental testing demonstrated that the surface-attached PZT disks were capable of detecting the change in material properties due to the exis-tence of cracking.In consideration of the inevitable discrepancies in different concrete beams due to specimen preparation and nsor installation,principal component analysis bad on statistical param-eters extracted from wave signals was applied to highlight the difference between benchmark and dam-aged rebar.The results
show the potential of the principal components as damage indices for quantifying integrity conditions of concrete structures.
Ó2013Elvier Ltd.All rights rerved.
1.Introduction
Rebar-reinforced concrete is currently widely ud in civil infrastructure including buildings,dams,power plants,bridges and roads,becau of its high load-carrying capacity and low maintenance.Although reinforced concrete is a relatively durable and robust constructional material,it can be verely weakened by poor manufacture or a hostile environment.Deterioration of reinforced concrete is generally attributable to either chemical degradation of the cementitious matrix,corrosion of the rebar,or physical damage (acking due to impact,fire,and ismic loads)[1].In the last decade or so there has been increasing awareness of the need for sustainable integrity surveillance for large reinforced concrete structures.Some novel damage identifi-cation techniques have emerged,bad on acoustic emission [2],impedance [3],and optical fiber [4]techniques,etc.However,be-cau the techniques offer only local measurement,den pop-ulations of nsors must be ud.More importantly,the approaches may lo their acuity with minu
te damage,for exam-ple,debonding between rebar and concrete,which is innsitive to static or low-frequency structural respons [5].In this respect,an identification technique bad on guided waves may be a promising solution,which has been validated for detecting diver defects in various structures [6].
In consideration of different wave modes propagating in con-crete and along embedded rebar,applications of guided waves for evaluation of the integrity of concrete structures have been reported recently to detect defects occurring in concrete and in reinforcement rebar [7–9].Generally speaking,bulk and surface (Rayleigh)waves can propagate into the concrete when a sur-face-attached transducer is excited [8].The bulk waves,which in-clude longitudinal (L)and shear (S)waves,propagate through the interior of the concrete,whereas Rayleigh (R)waves propagate
0950-0618/$-e front matter Ó2013Elvier Ltd.All rights rerved.dx.doi/10.buildmat.2013.05.016
Corresponding author.Tel.:+61399054995;fax:+61399054944.
E-mail address:ye.lu@monash.edu (Y.Lu).
mainly along the surface of the concrete,decreasing rapidly in magnitude with depth below the surface.On the other hand,L and S edge waves that also propagate along the surface are very weak and barely recognizable,although they propagate more quickly than L and S waves that propagate through the interior of the concrete[10].
Frequencies lower than100kHz are generally employed for identifying damage occurring in concrete[11],to avoid possible wave interactions between wave modes of smaller wavelengths at higher frequencies and aggregates in the concrete.Aggregate-scattered waves can significantly complicate the acquired wave signals,together with interference of wave reflection from the free surfaces of the concrete.Song et al.numerically and experimen-tally investigated Rayleigh wave propagation in concrete struc-tures using a surface-bonded piezoelectric actuator and nsor system[12].Sun et al.investigated the effect of uniaxial compres-sive stress and the resulting internal cracking of the concrete on the amplitude of the waveforms captured by piezoceramic nsors [8].Yang et al.numerically and experimentally evaluated the depth of surface-breaking cracks in concrete plates at low frequen-cies where surface waves evolve into Lamb waves with large wave-lengths[9].
On the other hand,for the ca of wave propagation through re-bar that is surrounded by concrete,th
e system should be consid-ered as a solid steel cylinder embedded in a solid medium with finite boundaries,showing more complex wave behaviors than in the ca of bare rebar[13].Substantial wave energy propagating in rebar may be lost or attenuated due to leakage into the sur-rounding concrete in the form of S and L bulk waves,parately or together[14].As a result,the inspection region in the longitudi-nal direction of rebar is limited becau of vere wave leakage.It is generally appreciated that,as the excitation frequency increas, the energy travelling in some wave modes becomes progressively more concentrated at the center of the rebar,with the velocity clo to the velocity of L bulk waves in steel,indicating little inter-action between rebar and surrounding concrete and therefore low-er energy leakage[15].
On the basis of the obrvations,Beard et al.succeeded in detecting steel bar deformation and an angled cut in a steel bar surrounded by mortar using L(0,12)mode at frequencies above 2MHz[15].Ervin and Reis monitored the accelerated corrosion of rebar embedded in mortar using L(0,9)mode at a frequency of 5.08MHz[13].However,it was reported that wave modes at high-er frequencies were particularly innsitive to surface defects and the medium surrounding the rebar[16],implying that they might be incapable of detecting damage such as delamination or debond-ing,which is generally simulated experimentally by resin/grea/ PVC coating.For this reaso
n,Wang et al.ud a spectral element method for the simulation of wave propagation along a steel rebar in concrete at50kHz,and evaluated the effect of different damage scenarios of debonding on wave propagation[5].Sharma and Muk-herjee termed the different properties of longitudinal guided wave modes at low and high frequencies as‘surface eking mode’and ‘core eking mode’respectively,and discusd their respective applications for detection of delamination and pitting caud by chloride corrosion[17].
In this study,the capability of ultrasonic waves for damage detection in rebar-reinforced concrete beams was investigated, with the ultimate aim to attach or embed slim and lightweight PZT elements into concrete structures to constitute‘smart con-crete’,in accordance with the concept of structural health monitor-ing[11,12,18,19].Two ries of experiments were conducted.PZT disks werefirst attached on the surfaces of rebar-reinforced beams to determine the change in dynamic properties of the concrete specimens after the occurrence of cracking damage,and rectangu-lar PZT patches were attached at the expod ends of the rebar to evaluate simulated corrosion in a form of partial removal of mate-rial from the rebar embedded in the concrete.Becau of differ-ences in the constitution of individual concrete beams and the installation of PZT elements,it was inappropriate for direct com-parison between the captured wave signals from different rebar specimens with and w
ithout damage.Principal component analy-sis was thus propod to highlight the exact difference in wave sig-nals due to the existence of damage,facilitating the identification of different conditions of the rebar.
可不可以不勇敢
2.Detection of change in elastic properties of concrete beams
找女生聊天With the assumption of concrete as an isotropic elastic medium and the lateral dimensions of the concrete beam being not small relative to the wavelength of the activated waves,the relationship between the velocities of elastic waves and the mechanical proper-ties of concrete can be given by the following equations[8,20]
c2
L
¼
kþ2l
qð1Þ超可爱的二次元头像
c2
S
¼
l
qð2Þwhere
l¼E
t;k¼
E t
t t
are known as Laméconstants,and c L and c S are the velocities of L waves and S waves,respectively.E,q and t are the Young’s modu-lus,material density and Poisson’s ratio,respectively.The velocity of R waves can be approximated as[20]
c R¼c S
0:87þ1:12t
1þt
ð3Þ
In this study,6PZT disks(PIÒPIC151,10mm in diameter and 1mm in thickness)were attached on the top and bottom surfaces of two rebar-reinforced concrete beams,respectively,with the dimensions detailed in Fig.1a,in which two pieces of rebar with a diameter of10mm were cast.The averaged mechanical proper-ties of the concrete tested from standard small and large cylinder samples are listed in Table1.A5-cycle Hanning-windowed tone-burst at different central frequencies was impod on P1with a peak-to-peak voltage of60V.Wave signals were acquired individ-ually by nsors P2–P6at a sampling frequency of20.48MHz to investigate the propagation properties of elastic waves in the con-crete beam.
Procesd with the assistance of a linear-pha bandpass signal filter[21],typical wave signals captured by P2at a central fre-quency of50kHz and200kHz,respectively,are shown in Fig.2a and b,w
here the signal at50kHz demonstrates a higher signal magnitude than that at200kHz.Thisfinding may be attributable to the greater energy dissipation of elastic waves at relatively high-er frequencies,induced by the complex interactions between the waves and the concrete aggregates.Typical wave signals captured by P3–P6at a central frequency of50kHz are shown in Fig.2c–f, respectively.Featuring the longest propagation distance,the wave signal captured by P6still displays good waveforms although the magnitude is much lower than that captured by nsor P2becau of wave attenuation.The results of P6at the frequency of200kHz are not shown here becau of the poor signal-to-noi ratio.
After ignoring L and S edge waves which propagate along the surface of concrete becau of their marginal magnitudes,two examples of propagating routes for L and S waves before they are captured by nsor P2are shown in Fig.1a,where the waves from actuator P1to nsor P2reflect from the bottom surface once(solid
关闭电灯Y.Lu et al./Construction and Building Materials47(2013)370–378371
line),as the shortest route,or twice (dotted line)[8].Thus the first wave packet in Fig.2a was recognized as L waves,and the cond wave packet was the combination of S waves and R waves a
s their velocities are very clo.It is also noteworthy that R waves exhib-ited stronger energy than both L and S waves,propagating mainly along the concrete surface [8].The velocities of L and S waves from two concrete specimens were tabulated in Table 2,where the velocity of L waves at P4was not listed as the corresponding wave packet was overlapped with the inherent E /M noi becau of short propagation distance.It is also noticed that the genuine arri-val time for L waves are barely visible at P3and P6becau of sig-nificant wave attenuation whereas the combined packet of S and R waves is still strong.By using the averaged velocity values cap-tured at P2and the measured density of the concrete as per the approximation given in Table 1,the dynamic Young’s modulus and Poisson’s ratio of the rebar-reinforced concrete beam could be inverly deduced from Eqs.(1)and (2)as 19.48GPa and 0.223,respectively.The results are almost identical with the val-ues for the specimen samples in Table 1;the difference could ari from the error of velocity measurement and the contribution of the embedded rebar that enhanced the stiffness of the concrete beam.A four-point bending test was then applied on the two con-crete beams to generate surface cracking which was clo to nsor P2,and other internal cracking,shown in Fig.3a,with histories of displacement and force in terms of time in Fig.3b.Fig.4compares the wave signals captured by nsor P2before and after the testing for two specimens.It is evident that the wave magnitude de-cread significantly due to the existence of cracking,which pre-vented effective wave
propagation in the concrete.Another reason for this phenomenon is possible debonding degradation of PZT disks during the testing.More importantly,a delay in the arri-val time of L waves,with an averaged velocity of 3084.4m/s,was obrved after the occurrence of cracking.With the assumption that there was no discernible change in density and Poisson’s ratio,
the dynamic Young’s modulus of the damaged concrete beam was calculated as 17.72GPa using Eqs.(1)and (2)again,the difference being attributed to the adver effect of cracking.In consideration of the proportional relationship between wave velocity and Young’s modulus,the result was reasonable compared with the va-lue of the intact condition.
It can therefore be concluded that as long as the excitation fre-quency is carefully lected,slim PZT disks are capable of success-fully activating and capturing ultrasonic elastic waves with a reasonable distance in reinforced concrete structures.They can also determine the effective elastic properties of concrete struc-tures and monitor changes for the existence of damage,thus pro-viding the possibility of quantifying the verity of damage.3.Identification of rebar damage
3.1.Wave propagation along rebar in concrete
To investigate the capability of piezoelectric elements for iden-tifying damage at the interface betwee
n embedded rebar and con-crete,six concrete beams were prepared with the same dimensions and configuration of reinforcement rebar (two pieces of rebar in each concrete beam)as described above,except that the diameter of the rebar was incread to 12mm.Before casting 12pieces of re-bar in concrete beams,simulated damage in the form of material removal of 1/3in depth and 75mm in length,at a position 600mm away from one end,was introduced in two pieces of rebar.Damage with the same depth and position but 150mm in length was introduced in another two pieces of rebar.The four pieces of damaged rebar were cast parately in four concrete beams,each paired with a piece of intact rebar.The other four pieces of in-tact rebar were cast in another two concrete beams.As a result,a total of 12pieces of rebar (four damaged and eight intact)in six concrete beams were characterized individually in this study,di-vided into three groups as Benchmark (rebar 1–8),D075(rebar 9–10with 75mm damage)and D150(rebar 11–12with 150mm damage).Rectangular PZT patches with the dimensions of 20mm (L)Â5mm (W)Â1mm (T)were attached at the expod ends of each rebar,acting as the actuator and nsor respectively,shown in Fig.1b.The same experimental ttings as described in Section 2were
applied.
Schematic diagram of concrete beam and PZT positions (unit:mm)(a)concrete surface-attached PZT disks and (b)rebar surface-attached Table 1
Mechanical properties of concrete.Young’s modulus E (GPa)Poisson’s ratio t
Density q (kg/m 3)
18.99
0.2虚拟化服务器
2238
The propagation properties of guided waves along the rebar were evaluatedfirst,at central frequencies from20to100kHz.A Hilbert transform[22]was adopted to obtain the energy envelope for monitoring changes in the amplitude of wave packets.It was obrved that although the wave magnitudes in terms of excitation frequency were not exactly identical among the rebar specimens, the magnitude of thefirst wave packet generally reached its max-imum around30kHz,shown in Fig.5a,with typical waveforms shown in Fig.5b.Subquent investigations were therefore con-ducted at the excitation frequency of30kHz,where the wave mode demonstrated maximum wave energy and the wavelength was much greater than the rebar diameter,minimizing the inter-ference of rebar ribs[11,13].It was also noticed that the propaga-tion velocity of thefirst wave packet along the rebar embedded in concrete at30kHz was measured around2441m/s,which was substantially lower than the theoretical velocity of5139.65m/s for L waves in the bare rebar.This effect is mainly attributed to t
he interaction between concrete and rebar.Becau of the similar
Table2
Velocities of L and S waves at different positions for two specimens.
Sensors L waves(m/s)S waves(m/s)
P23199.3/3119.21848.8/1923.7
P3N/A1849.2/1815
P4N/A1823.1/1868.6
P53099.6/3111.31820.4/1845.2
P6N/A1822.9/1825.5
Y.Lu et al./Construction and Building Materials47(2013)370–378373
impedances of rebar and concrete,a considerable portion of wave energy will be transmitted across the boundary between rebar and concrete.Since wave propagation velocity in the concrete is slower than in the steel,the superposition of the two different speed waves caus the lower wave velocit
y on the rebar surface,to-gether with wave dispersion [5,11,14,15]3.2.Principal component analysis
In general,direct comparison between the signals from differ-ent specimens can lead to incorrect conclusions,as many parame-ters can influence the captured wave signals.For example,Fig.6compares the signals from a benchmark rebar and a damaged rebar with simulated damage 150mm in length.It is evident that the magnitude of transmitted wave signals in the damaged rebar was greater than that in the benchmark,which obviously contra-dicts the intuitive understanding that the transmitted waves dissi-pated by the damage would have a lower wave magnitude than that of benchmark becau of wave reflection.This finding is mainly attributed to complex wave propagation mechanisms interacting with damage as well as differences in the casting of individual concrete beams and bonding of PZT elements on the re-bar.However,if the amplitude is simply normalized by the respec-tive maximum value,information regarding wave dissipation during wave transmission becau of the existence of damage would be lost,which is unhelpful for effective damage asssment.
On the other hand,it is appreciated that the reciprocal signals for each actuator–nsor pair located at the two ends of a rebar are identical if no damage exists in the rebar,on the basis of direct and conver effects of piezoelectric transducers.In contrast,the interaction between guided waves and t
he damage (partial re-moval of rebar material in this ca)would cau wave reflection,transmission and possible mode conversion,schematically shown in Fig.7.The captured wave signals therefore compris not only the direct transmitted signal propagating along the remnant of the rebar,but also the wave reflection from the boundary of the damage (surfaces 2and 3)and the ends of the rebar (surfaces 1and 4).As a result,if damage exists in the rebar,the signal received at one end of the rebar (signal superposition of A,B,C and D)would be different from the signal captured at the other end (signal superposition of E,F,G and H)becau of the different distances between damage and nsors,except for the unique situation when the damage is exactly in the middle of the rebar.However,eking information about hidden damage via calculation of the di-rect algebraic difference between the two kinds of signals can give ri to significant error becau of the potential pha shift of signals acquired in different testing scenarios.In this respect,the differences between certain statistical parameters extracted from wave signals captured at both ends of the rebar could act as damage indices to describe the existence and verity of damage,whereas differences in damage indices are expected to be marginal in an intact rebar.
500
1000
1500
2000
25003000
35
40
45
50
Displacement Force
Time [s]
05
10
方案书模板
15
20
25
Crack
PZT disk无奈的爱情
(a)
(b)
3.Four-point bending test for concrete beams (a)surface cracking and displacement and force
histories.
374Y.Lu et al./Construction and Building Materials 47(2013)370–378
