Two-channel Impact-Echo

Transcription

1 NDTCE 9, Non-Destructive Testing in Civil Engineering Nantes, France, June 3th July 3rd, 9 Two-channel Ipact-Echo Parisa SHOKOUHI Assistant professor of Civil Engineering, The University of Texas at El Paso (UTEP), El Paso, Texas, U.S.A. Alexander von Huboldt (AvH) Fellow hosted by BAM - Bundesanstalt für Materialforschung und prüfung, Berlin, Gerany pshokouhi@utep.edu or parisa.shokouhi@ba.de Abstract A two-channel cobined Ipact echo (IE) - Spectral Analysis of Surface Waves (SASW) approach used for field evaluation of several concrete slabs is presented here. In a coupled IE- SASW test, the sae ipact source is utilized for both tests, however instead of one single IE receiver, ultiple (at least two) receivers are required. The ain focus of this paper is the data reduction techniques associated with this approach. The relevant traditional (i.e. IE spectru and SASW), less traditional (i.e. attenuation analysis), and non-traditional (i.e. wavelet analysis) schees used for data interpretation are discussed. An alternative procedure to generate the dispersion curve is also presented. Traditional IE data reduction includes calculation of the spectru to identify the thickness resonance frequency of the concrete slab and/or other significant resonance frequencies resulting fro the nearby geoetrical boundaries, internal flaws or heterogeneities within concrete. When the interpretation of the IE spectru is not straightforward, copleentary inforation can be deduced fro the tie-frequency analysis. Attenuation characteristics of IE signals can also provide useful hints to better interpret the results. SASW yields the surface wave velocity profile through the thickness of the concrete slab (i.e. dispersion curve). This study concludes that the full dispersion curve characteristics rather than the average wave velocity provide ore reliable inforation regarding the state of daage in concrete. Résué L utilisation conjointe ipact-echo (IE) Analyse spectrale des ondes (SASW) est présentée ici pour évaluation sur plusieurs dalles de béton. Dans des essais IE-SASW couplés, la êe source d ipact est utilisée pour les deux essais, cependant au lieu d un récepteur IE unique, plusieurs (au oins deux) récepteurs sont nécessaires. L objectif principal de ce docuent porte sur les techniques de réduction des données associées à cette approche. Les schéas traditionnels pertinents (ex. Spectre IE et SASW), oins traditionnels (ex. analyse d atténuation), et non traditionnels (ex. analyse de l ondelette) utilisés pour l interprétation des données sont discutés. Une procédure alternative pour générer la courbe de dispersion est égaleent présentée. La réduction des données IE traditionnelle inclut le calcul du spectre pour identifier la fréquence de résonance de l épaisseur de la dalle de béton et/ou d autres fréquences de résonance significatives, les signatures des surfaces de contour géoétriques proches, les flux internes ou les hétérogénéités du béton. Lorsque l interprétation du spectre IE n est pas directe, des inforations copléentaires peuvent être déduites de l analyse du teps/fréquence. Les particularités de l atténuation des signaux IE peuvent égaleent fournir des inforations utiles pour ieux interpréter les résultats. Le SASW produit le profil de vitesse de l'onde de surface à travers l épaisseur de la dalle en béton (ex. courbe de dispersion). Cette étude conclut que les caractéristiques de la courbe de dispersion fournissent des inforations plus fiables sur l état des doages dans le béton que la vitesse oyenne de l onde.

2 NDTCE 9, Non-Destructive Testing in Civil Engineering Nantes, France, June 3th July 3rd, 9 Keywords Tie-frequency analysis, wavelet transfors, signal attenuation, SASW, dispersion curve Background and Motivation The traditional IE test setup as shown in Fig. consists of a echanical ipact source and one receiver (receiver in Fig. ) located near the ipact (usually at a distance of.5 to. ). The surface of the concrete coponent under test is tapped lightly by the ipact. The resulting vertical vibrations are recorded by the receiver and analyzed to estiate the thickness of the coponent and/or the local condition of concrete in ters of defects and heterogeneities. h VP,VR VP,VR Figure. Two-channel IE-SASW setup. Traditional setup includes only the near receiver. The traditional data analysis includes erely FFT transforation of the wavefor, identifying the significant frequency peaks in the spectru, and relating those peaks to various possible resonance odes of the concrete eber. In case of a concrete plate (e.g. slabs or bridge decks) of P-wave velocity of V P (/s), the resonance frequency f (Hz) and thickness d () are related by: f = β V P d () where the diension-less correction factor β can be accurately obtained through Rayleigh- Lab dispersion equations governing the wave propagation in a free plate []. Soe coercial IE test equipent include two receivers instead of one. A two-channel IE device akes it possible to estiate the wave velocity at every test point. In practical applications, IE tests are conducted in a point-by-point anner. While autoated data collection has revolutionized the state of research in IE, the state of practice is yet to follow. This paper presents a collection of data analysis techniques applicable to two-channel IE testing. A cobination of these techniques provides ore reliable eans of data interpretation than relying only on IE spectru or average wave velocity easureents. Tie-Frequency Analysis using Continuous Wavelet Transfors Frequency-doain analysis of IE is not always straightforward, for exaple the spectru ay include frequency peaks which can not be easily explained or the recognized resonance frequency peaks ay be shifted or distorted. A frequency-doain approach is ideal for odal analysis where the wavelengths (λ ) are greater than the typical diensions of test objects. Tie-doain analysis is on the other hand optial when λ is considerably shorter than the diensions (e.g. ultrasonic pulse-echo testing). Since IE testing lies in the iddle (λ in the range of d), analysis in both tie and frequency doains provides copleentary inforation []. Therefore, a dual tie-frequency analysis ay provide a better eans of evaluation of the IE signals than tie or frequency doain analyses. A two-channel IE field study was conducted to evaluate the state of debonding within concrete slabs in a site located southwest of Houston. The noinal design thickness of slabs

3 NDTCE 9, Non-Destructive Testing in Civil Engineering Nantes, France, June 3th July 3rd, 9 was.33. Scalogras (tie-frequency representations obtained using continuous wavelet transfors) at three core locations are shown in Figs., 3, and 4. The details of the wavelet analysis are not given here but can be found in other references [3]. The corresponding wavefors, spectra, and core pictures are also presented. Nor. Spectral Apl e p litud A a tr l S pec. N or e tud p l i A d e z a li N or z ) H y (K u e n c F re q Figure. Tie history, spectru, and scalogras (gaus4) corresponding to the location of Core IV (Intact Condition- thickness of.34) Nor. Spectral Apl d e u t p l i A d a lize N or H z) K n c y ( u e F re q Figure 3. Tie history, spectru, and scalogras (gaus4) corresponding to the location of Core II (debonded and cracked thickness of.33)

4 NDTCE 9, Non-Destructive Testing in Civil Engineering Nantes, France, June 3th July 3rd, 9 Nor. Spectral Apl..8 d e u t p l i A d e a liz N or H z) K n c y ( u e F re q Figure 4. Tie history, spectru, and scalogras (gaus4) corresponding to the location of Core III (fully debonded thickness of.38) Core IV (Fig. ) was intact, while concrete slabs at test location corresponding to cores II (Fig. 3) and III (Fig. 4) were debonded. Although frequency spectru of Fig. is sufficient to conclude the slab condition, the scalogras of Figs. 3 and 4 provide valuable copleentary inforation. For exaple five different wave coponents are differentiable in the scalogra of Fig. 3. Although the thickness resonance frequency is still the doinant peak in the spectru, the very low frequency slow-attenuating waves give an alaring sign that the slab is debonded. On the other hand, the spectru of Fig. 3 is doinated by the low frequency peak, an indication of a fully-debonded condition. No further inforation regarding the thickness of the slab or the location of debonding can be deduced fro the spectru. Fro the scalogra however, one can still read a resonance thickness frequency of 8 KHz and recognize a high frequency coponent at 5 KHz, corresponding to the thickness resonance frequency of the debonded portion of slab. 3 Attenuation Analysis Attenuation characteristics of IE signals can also provide copleentary inforation. Attenuation (with tie) has been previously used to easure the state of distributed daage within concrete bridge decks. Siilar procedure has been taken here to copare the wavefors previously shown in Figs. to 4. As expected, the response of a sound contact slab is faster attenuated than that of debonded slabs. However, the exponential approxiation of attenuation (as shown in Fig. 5) is not applicable to debonded slabs, unless the attenuation is calculated for different wave coponents within the signal (e.g. each of the five distinguished coponents shown in scalogra of Fig. 3) but this analysis falls beyond the scope of this paper. Attenuation can be also calculated in frequency doain, for exaple by calcualting and coparing the so-called Q-factor fro the resonance frequency and the two neighbouring

5 NDTCE 9, Non-Destructive Testing in Civil Engineering Nantes, France, June 3th July 3rd, 9 half-power frequencies. Such procedure is advantages over exponential estiation for the case shown in Fig. 5, however, it can not be still used for the signal shown in Fig A =.4e R = t.3.. A =.e R =.5 836t.3.. A =.5e R =.9 355t Figure 5. Attenuation of IE signal at Core IV, Core II, and Core III. 4 Two-Channel IE In traditional IE, P-wave velocity V P is usually estiated through a soewhat calibration procedure which involves taking IE easureents on a part of the structure where the thickness is known. The ain advantage of this procedure is that the average throughthickness velocity is directly easured. However, if the concrete under test is of variable quality such that velocity differs significantly fro one location to another, using this procedure leads to inaccurate thickness easureents. Soe later IE coercial test equipent included a second receiver enabling direct velocity estiation (through travel-tie easureents) at every test point. Accurate evaluation of the arrival tie of P-wave is usually difficult. Therefore R-wave velocity is easured and assuing a Poisson s ratio, V P is then calculated. Apart fro the need to assue Poisson s ratio, it is also questionable whether this velocity well represents the velocity through the thickness of the slab. The R-wave velocities (obtained using the arrival ties of the first axia) at the three core locations are calculated and presented together with the wavefors in Fig. 6. Near Far /s ~85 /s ~ /s -..5 x -3 Figure 6. Signals recorded at near and far receivers at locations of Core IV, Core II, and Core III. It is also interesting to note the relative aplitude and attenuation characteristics of the pairs of signals shown in Fig 6. The far signal of Fig. 6 is of uch lower aplitude than the near signal and is very fast attenuated, while the aplitudes of the later parts of the far signals of Fig. 6 and are hardly attenuated. The aplitude of far signal of Fig. 6 occasionally even exceeds that of near signal, a clear indication of excitation of flexural odes.

6 NDTCE 9, Non-Destructive Testing in Civil Engineering Nantes, France, June 3th July 3rd, 9 4. Spectral Analysis of Surface Waves (SASW) SASW gives the R-wave velocity through the thickness of the slab. The average wave velocity (as obtained by SASW) has been previously used to estiate V P and indicate the quality of the concrete and its state of interior daage. It is argued that since the larger defects (e.g. debonding) detectable by IE are usually preceded by the generation of icro cracks, the defected zones of the slab exhibit lower wave velocities. The average frequency spectra along with the corresponding dispersion curves for the three core locations are shown in Fig. 7. In generation of dispersion curves, no extra soothing or filtering was used, except that the data points with coherence values lower than.9 were discarded..5 Near Far.5.5 Phase Velocity (/s) Frequency (khz) Frequency (KHz) Frequeny (khz) Figure 7. Spectra and dispersion curves at Core IV, Core II, and Core III. The characteristics of the dispersion curves are ore representative of the actual conditions observed at these three locations. For exaple, the average R-wave velocity for the fullydebonded slab of Fig. 7 is not uch lower that that for the intact slab of Fig. 7, but the dispersion curve shows a discontinuity at low frequencies. On the other hand, the R-wave velocities for the cracked Core II are significantly lower than those for the two other cores. This is a sign of distributed daage and cracking through the thickness of the slab. 4.3 Alternative Method to Calculate Phase Velocities An alternative procedure to calculate the dispersion curve using wavelet-based filtering of the recorded signals at near and far receivers is presented here. A chart illustrating the steps required to calculate phase and group velocity (C R and C P ) over a certain frequency bandwidth (of center frequency f i ) is shown in Fig. 8. The first step of this procedure involves bandpass-filtering of the signals recorded at near and far receivers (Signals and, respectively) to obtain signal coponents of certain frequency. Having calculated the CWT scalogras for the signals, the filtering operation is reduced to suing up the portion of the CWT around the desired center frequency f i. The second step includes obtaining the envelope and phase inforation of each coponent using Hilbert transfor. The group velocity C P at each frequency f i is estiated by dividing the distance between the two receivers by the difference in the arrival ties of the envelopes of the two signals (t g and t g ). t g is taken as t ph or the reference tie for calculation of phase velocity. In the phase diagra of the near receiver, the phase corresponding to t ph is deterined. Once projected, this phase value corresponds to several tie instances in the phase diagra of the far receiver. The nearest of these values to t g is taken as to t ph. The phase velocity C R at each frequency f i is estiated by dividing the distance between the two receivers by the tie difference t ph -t ph. A good agreeent between the results and the dispersion curve optioned fro SASW is observed. This ethod is siple and does not require phase unwrapping or soothing operations. Since the phase velocity at each frequency is deterined independently and locally (using the inforation around the arrival tie of the corresponding wave groups), the

7 NDTCE 9, Non-Destructive Testing in Civil Engineering Nantes, France, June 3th July 3rd, 9 outcoe is less affected by near-field effects and therefore can provide ore reliable results, especially in case of irregular profiles [5]. Signal Signal CWT Scalogra frequency f frequency f i Hilbert Transfor ax tg dist. C P = t t ax tg g g Group Velocity tg = φ tph = φ tph is closest to tg dist. C = f i t ph t g Aplitude tg tg Phase (rad.) tph tph tg ) s ( / 4 y c it V elo Frequency (KHz) * phase velocity (SASW) + group velocity o phase velocity Figure 8. Procedure diagra, illustration of the necessary steps, and coparison of dispersion curves at the location of Core III. 5 Conclusions Different analysis techniques applicable to data reduction of a two-channel IE-SASW test were discussed. The advantages of using copleentary analysis techniques over traditional data reduction schees were pointed out and illustrated through field easureents. References. Gibson, A., Popovics, J.S. (5) Lab wave basis for ipact-echo ethod analysis, ASCE Journal of Engineering Mechanics, Vol. 3, Nr 4, pp Schubert, F., Köhler, B. (8) "Ten Lectures on Ipact Echo", Journal of Nondestructive Evaluation, Vol. 7, pp Shokouhi, P., Gucunski, N., Maher, A. (6) "Tie-frequency techniques for ipact echo data analysis and interpretation", Proc. of ECNDT, Berlin, Gerany. 4. Ki, D.-S., Park, H.-C. () "Deterination of dispersive phase velocities for SASW ethod using Haronic wavelet transfor ", Soil Dyn. & Earthquake Eng., Vol., pp

8 NDTCE 9, Non-Destructive Testing in Civil Engineering Nantes, France, June 3th July 3rd, 9