ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 50, no.

4, april 2003

441

Simulation of Lamb Wave Propagation for the Characterization of Complex Structures

Valentina Agostini, Pier Paolo Delsanto, Ivan Genesio, and Dimitri Olivero
AbstractReliable numerical simulation techniques represent a very valuable tool for analysis. For this purpose we investigated the applicability of the local interaction simulation approach (LISA) to the study of the propagation of Lamb waves in complex structures. The LISA allows very fast and exible simulations, especially in conjunction with parallel processing, and it is particularly useful for complex (heterogeneous, anisotropic, attenuative, and/or nonlinear) media. We present simulations performed on a glass ber reinforced plate, initially undamaged and then with a hole passing through its thickness (passing-by hole). In order to give a validation of the method, the results are compared with experimental data. Then we analyze the interaction of Lamb waves with notches, delaminations, and complex structures. In the rst case the discontinuity due to a notch generates mode conversion, which may be used to predict the defect shape and size. In the case of a single delamination, the most striking signature is a time-shift delay, which may be observed in the temporal evolution of the signal recorded by a receiver. We also present some results obtained on a geometrically complex structure. Due to the inherent discontinuities, a wealth of propagation mechanisms are observed, which can be exploited for the purpose of quantitative nondestructive evaluation (NDE).

I. Introduction late-like structures, laminates, and composites are extremely common in civil engineering, aerospace, ground transportation, etc. They often are overdesigned to compensate for a lack of ecient techniques for testing the initial integrity of the mechanical structures and/or a reliable monitoring of damage and aging. Consequently, the race for ever improving NDE techniques (in particular ultrasonic) is still open. Also, considerable attention recently has been devoted to the problem of damage assessment by means of a combination of ultrasonic excitation and detection with embedded array sensors and defect signature recognition. From this point of view, a particularly promising technique is the time-reversed acoustic (TRA) and time-reversed ultrasonics (TRUS), which recently was applied in several elds, such as medicine, material characterization, and oceanography [1][3]. Established ultrasonic techniques for NDE take advantage of a wide variety of sources and propagation modes [4].

Manuscript received November 28, 2001; accepted November 12, 2002. This work was supported by the European Community through a Brite EuRam Grant (project No. BE-97.4213 DAMASCOS), and by the INFM Parallel Computing Initiative. The authors are with INFMDip. Di Fisica, Politecnico di Torino, Torino I-10129, Italy (e-mail: [email protected]).

Surface waves are extremely ecient in probing surfaces and thin plates because they can propagate a long distance without appreciable attenuation. Therefore, a large specimen region can be interrogated with a single transducer. In particular, Rayleigh waves (RW) are used to detect subsurface aws in thick plates; and Lamb waves (LW) [5][10] can explore the entire thickness of a thin plate. The wealth of informations provided, however, engenders a nontrivial interpretation of the results, due to the coexistence of at least two modes at any given frequency and the strongly dispersive nature of these modes at high frequencies, resulting in a deformation of the initial wave package along the propagation path. Additional complications arise from the heterogeneity and anisotropy of the specimen, its geometry, and the presence of defects. It has been shown by [11][13] that a single and pure Lamb mode may generate a variety of other modes either by interacting with a surface or subsurface aw or by crossing to another region of dierent impedance or geometry (e.g., with thickness variations). Thus, the output signal may carry more informations, but often it may be dicult to interpret. For a quantitative assessment, a thorough physical understanding of the propagation mechanism becomes a useful tool in the analysis of the experimental data. The local interaction simulation approach (LISA) [14][17] has been designed to provide ecient and reliable simulations for comparison with the experimental data [18], [19] in the eld of ultrasonic wave propagation and other applications. The eciency of the approach (particularly in the case of complex propagation media and pulses), is due to the exploitation of parallel processing, and the possibility of including in the treatment all kinds of local interaction mechanisms. In fact, by putting the processors into a oneto-one correspondence with the cells of the discretized specimen, it is possible to transfer the local interactions between them directly to processors, bypassing the partial dierential equations. Iteration equations then may be obtained directly from heuristic considerations. In short, LISA allows one to perform virtual experiments, in which the propagation of the wave through the whole specimen is followed for as long as desired, even in the case of strongly anisotropic or heterogeneous media [20], [21]. The goal of the present article is to demonstrate the applicability of LISA to the study of LW propagation in complex structures, in particular for the purpose of detection and quantitative evaluation of damages in the structure. In Section II, we analyze several types of aws, starting with the simple case of a passing-by hole, which is considered mostly for the purpose of validating the method (by

08853010/$10.00 c 2003 IEEE

442

ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 50, no. 4, april 2003

comparing the results of the simulations with experimental data). In the following subsections we consider the more interesting cases of a notch and subsurface aw (Section IIB), of a cone-shaped stack of delaminations (Section IIC) and we study the propagation of LW in a more complex structure, such as Y-junction stringers (Section IID). The latter case is particularly relevant in the context of a European collaboration [Brite Euram Project DAMASCOS (damage assessment in smart composite structures)], which has given us the opportunity to get hold of the experimental data from the INSA Laboratories (Lyon, France) [22] and the University of Valenciennes (Valenciennes, France) [23].

Fig. 1. Representation of the GFRP specimen awed by a hole. The transmitter T and receiver R and hole locations H are indicated.

II. Lamb Wave Damage Detection in Plate-Like Structures According to their particle displacement patterns, LWs may be classied into symmetric (sn ) and antisymmetric (an ) modes. Below a certain threshold Q of the product frequency f times plate thickness d, only the zero-th order Lamb modes a0 (fundamental exural) and s0 (fundamental extensional) may propagate. In the following four cases, involving various specimens with dierent kinds of defects, we carry out simulations under the condition f d < Q, in order to avoid the additional complication of higher modes. A. Validation of the LISA Method in the Case of a Passing-by Hole Let us rst consider the propagation of LWs through a 450 250 3 mm glass ber reinforced plate (GFRP), which initially is intact and then damaged by a passingby hole of a diameter ranging from 1 to 10 mm. The hole has been modeled as a void inclusion in the specimen. This simple, well dened geometrical defect was chosen in order to validate our simulation method and obtain a quantitative comparison with experimental results in a controlled and reproducible case. It should be noted that, in spite of its simplicity, this problem requires a full three-dimensional treatment with a very ne space step (0.1 mm). The other parameters were: a0 mode velocity = 3090 m/s, time step 10 ns, density = 1.853 g/cm3 , stiness tensor: 20.3 6.0 6.5 0 0 0 6.0 20.3 6.5 0 0 0 6.5 6.5 16.2 0 0 0 [GPa]. C = (1) 0 0 0 4.3 0 0 0 0 0 0 4.3 0 0 0 0 0 0 7.1 The ultrasonic source/sensor conguration is shown in Fig. 1. In order to visualize the wave propagation inside the specimen, in Fig. 2 we present a snapshot of the LW out-ofplane displacement component in the case of a one with a 10-mm-hole at a time in which the antisymmetric a0 mode is reaching the defect (observe the resulting interference

Fig. 2. Map of the out-of-plane component of the displacement at the time t = 126 s for the GFRP plate with a 10 mm hole.

pattern), while the faster symmetric s0 mode is already being reected back from the edge of the plate. A displacement map, such as Fig. 2, may be quite useful to visualize and understand the propagation mechanisms taking place inside the material, but it cannot allow a quantitative comparison between numerical and experimental results. For this reason we present in Fig. 3 the s0 signal collected at the receiver, normalized with respect to the incident wave amplitude, both for the perfect plate (upper plot) and for the plate with a 10 mm-hole (lower plot). The latter shows, as expected, a reduced wave amplitude due to partial scattering from the hole. In the plots the continuous line represents the results of our simulations; the dots refer to experimental data obtained at the INSA Laboratoires (Lyon, France) [22]. We observe a good quantitative agreement between synthetic and experimental results, although some dierences appear at the end of the signals. These are due to the fact that we have not included in our simulations the transducer damping (the amplitude decrease in the transducer itself) and the material attenuation in the experimental results. Two major eects originate from the interference of a LW with a hole: a wave amplitude reduction due to geometrical dispersion, and a delay in the time signal due to the fact that the wave travels a longer distance because it must circumnavigate the defect. Both eects can be better appreciated in Fig. 4, in which the s0 amplitude and the time delay are reported as a function of the hole diameter . The theoretical results (continuous lines) agree reasonably well with the results of experiments performed both at INSA [22] and at the University of Valenciennes, France [23]. It is interesting to note that the s0 amplitude for = 1 mm is slightly higher than when no hole is present ( = 0). This is due to the fact that we are not considering the total displacement, but only its out-

agostino et al.: characterization of complex structures and lamb waves

443

Fig. 3. The s0 signal vs. the time of arrival at the receiver: comparison between our simulations (continuous line) and INSA experimental data (dots) for the intact plate (upper plot) and for the plate with a 10-mm hole (lower plot) [22].

of-plane (shear) component because that is essentially the quantity measured at the receiver. B. Notch and Subsurface Flaw In this section, we analyze the LWs behavior in a thin Al plate with an embedded aw. The defect is assumed to have an innite extension in the direction orthogonal to direction of the wave propagation, in order to allow a twp-dimensional analysis. Let us consider a thin plate of thickness d of aluminium with a rectangular subsurface defect (void) of length h and width a, located at a depth l from the plate upper surface, as shown in Fig. 5. For l = 0 we have a surface notch. The reected and transmitted signals are recorded by two receivers, R1 and R2 , located on the plate surface. They may be used, after a careful analysis of the signature of the defect, for reconstructing the aw position, size, and shape.

Let us consider the propagation of a zero-th order antisymmetric Lamb mode a0 injected from the left side of a specimen containing a surface notch with l = 0, a = 0.1 d, h = 0.46 d. In Fig. 6 snapshots at four subsequent times of the in-plane (longitudinal) component are shown. The uppermost plot displays the pure a0 mode of the injected (hanning modulated) wave packet well before reaching the notch (located at x = 8000 a.u.). The second plot shows the scattering eect due to the breaking of the wave against the notch. The presence of this discontinuity generates a partial mode conversion from a0 to s0 . This can be clearly observed in the lower two plots, which represent the transmitted wave: in the third plot, the a0 and the mode-converted s0 are still partially overlapping and in the bottom plot they are well separated. The presence of a mode-converted s0 mode supplies additional information, which may be used as shown later, to predict the defect shape and size. The location of the notch may be deduced easily from the separation (spatial at a given time or temporal at a given location) between a0 and s0 (either in the reected or in the transmitted component) and from the well-known phase velocity dierence between the two modes. Again, linear plots are needed for quantitative evaluations. In Fig. 7 the in-plane components of the signals received by receivers R1 and R2 (upper and lower plots, respectively) are reported as a function of time for a rectangular defect with a = 0.1 d and h = 0.46 d (as in Fig. 6). The eects of three dierent depths of the notch are compared. For both the reected and transmitted pulses, we observe rst, from the left, the faster mode-converted s0 pulse and later, trailing behind, the original a0 pulse. All the pulses, with the notable exception of the transmitted a0 signal, exhibit a remarkable dependence on l, which can be used for NDE purposes. In Fig. 8 the reection and transmission coecients of both the a0 and s0 modes are plotted vs. h/d for a surface aw (l = 0). Similar plots for l = 0 or vs. l/d for a given value of h also have been obtained (omitted for brevity). Measuring the amplitudes of the reected and/or transmitted LW modes, it is possible, from these plots to deduce h and l (unless the amplitudes are degenerate functions of h and l) and, with a more detailed analysis, to map approximately the shape of the defect (if the layer thickness may be assumed to be constant). C. Impact-Induced Delaminations A very common kind of failure in composite plates may be represented as a cone-shaped stack of delaminations. We use this representation to model impact-induced damage and proceed to analyze its eects on wave propagation by means of numerical simulations. In Fig. 9 we show carbon bers reinforced plate (CFRP), which we have studied for this purpose. In our model the delaminations (treated as in [17]) allow the transmission of compression pulses (packing together the two contiguous surfaces) but do not transmit either stretching or shear tensions.

444

ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 50, no. 4, april 2003

Fig. 4. The s0 normalized amplitude (left plot) and time delay vs. hole diameter (right plot) in a GFRP. Continuous line, simulations; dotted line, INSA experiments [22]; dashed and dotted lines, Valenciennes experiments [23].

Fig. 5. Representation of the specimen. The darker region represents the defect (a rectangular notch). R1 and R2 are the receivers, the transducer is located to the left of R1 .

Fig. 7. Signal amplitude, normalized to the incident eld, vs. time of a LW propagating in an Al plate with a rectangular subsurface defect located at dierent depths: l = 0.08 d (solid line), l = 0.18 d (dashed line), and l = 0.23 d (dots). Longitudinal components of the reected (upper plot) and transmitted (lower plot) signals.

The density of the CFRP is assumed to be = 1.575 g/cm3 , whereas the stiness tensor is: 85.9 3.72 6.5 0 0 0 3.72 85.9 6.5 0 0 0 3.34 3.34 16.34 0 0 0 [GPa]. C = 0 0 6.59 0 0 0 (2) 0 0 0 0 6.59 0 0 0 0 0 0 7.48 In order to prepare the ground for the more complex case under investigation, we consider rst a plate with carbon bers lying in the plane (x,y) and awed by a single delamination, parallel to the x axis and extending innitely in the y direction (i.e., orthogonal to the propagation direction, allowing a two-dimensional treatment). In a virtual experiment, we assume to inject a hanning-windowed

Fig. 6. Snapshots in an orthogonal slice through the plate of the in-plane displacement at various times for a low frequency Lamb mode propagating in an Al plate with a surface notch (located at a position corresponding to x = 8000 a.u., and, therefore, visible only in the second snapshot). Note the dierent x-ranges and scale in the four plots.

agostino et al.: characterization of complex structures and lamb waves

445

Fig. 8. The LW normalized amplitude at a time after the scattering vs. h/d for l = 0 (surface aws). Modes: reected a0 (circles), transmitted a0 (triangles), reected s0 (diamonds), and transmitted s0 (crosses).

Fig. 10. Snapshots of the displacement eld at time t = 73 s for a 2-mm-thick CFRP sample awed by a single delamination of length 6 mm positioned at 0.3 mm from the bottom surface. The rst plot represents the in-plane (u) component of the displacement. The second plot shows a section at z = 1.8 of the upper plot (continuous line) and the respective unawed case (dotted line). The last two plots represent the out-of plane (v) component of the displacement.

Fig. 9. Specimen used to study a cone-shaped stack of delaminations due to an induced impact (in P). The positions of the transmitter (T ) and of the receiver (R) are indicated.

pulse from the left edge and follow the propagation of the a0 and s0 modes thus generated. In Fig. 10 both the inplane (u) and out-of-plane (v ) components of the pulse are reported for a time at which the faster s0 mode has already reached well beyond the delamination. We recall that, in the u component, the s0 mode is symmetric and the the a0 mode is antisymmetric; the opposite occurs for the v component. In the rst two plots of Fig. 10 (u component) we notice, proceeding from left to right, rst a weak a0 mode. In the delamination region, both the s0 mode and the mode-converted a0 (which in the second plot causes the wiggles on top of the smooth and much larger s0 oscillations), and the pure, much faster and stronger s0 mode. Note that in the u component the s0 mode prevails, but it is almost negligible in the v component. In the third

and fourth plots we nd rst a very strong incoming a0 mode and then, in the delamination region, a very conspicuous mode-converted a0 mode (with a dierent wavelength), which is entirely due to the delamination. In fact it is completely absent in the unblemished plate (dotted line). As already mentioned, our simulation technique yields not only the spatial propagation (as shown in Fig. 10), but also the temporal evolution, which is more useful for comparison with experimental data. In Fig. 11 we plot the received signal recorded as a function of time by the receiver, in the case of a 6-mm thick CFRP plate with a single delamination. Its eect is mostly a time shift (delay) due to the longer path caused by the splitting of the pulse (above and below the delamination) and subsequent recombination. The dependence on position of the delamination is analyzed in Fig. 12, which shows both the time delay and dierence in peak amplitude (between a awed and an unblemished plate) as a function of the delamination location (z ) from the bottom to the top of the plate. Concerning the time shift, the eect is largest at midplate because, at that location, the pulse is completely broken into two parts and the circumnavigation around the delamination takes the greatest toll. On the contrary, the amplitude dierence is larger close to the surface, where the a0 mode displacement is maximal (thus leading to a greater discontinuity), but it vanishes at midplate. The eect of a cone-shaped stack of delaminations is illustrated in Figs. 13 and 14. Fig. 13 displays both the u and v components of the displacement in a similar fashion

446

ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 50, no. 4, april 2003

Fig. 11. Out-of-plane displacement component of a signal hanningwindowed pulse collected at the receiver in a 6-mm-thick CFRP plate (normalized amplitude): the dotted line corresponds to an undamaged plate, and the solid line corresponds to a plate with a single delamination positioned at 0.9 mm from the bottom surface.

Fig. 13. Snapshots of the displacement eld at time t = 76 s for a 6-mm-thick CFRP sample awed by a cone-shaped stack of delaminations. The rst plot represents the in-plane (u) component of the displacements. The second plot shows a section of the upper plot (continuous line) at z = 6 mm, and the respective unawed case (dotted line). The last two plots refer to the out-of plane (v) component of the displacement.

Fig. 12. Dependence on the z -position of the single delamination. Time delay (upper plot) and dierence in amplitude (lower plot) between the reference and the damaged case. (We have considered the out-of-plane v component of the displacement for the a0 mode).

Fig. 14. Out-of-plane displacement component for a signal collected at the receiver in a 6-mm-thick CFRP plate with a cone-shaped stack of delaminations (the dotted line is the reference signal in the undamaged plate).

agostino et al.: characterization of complex structures and lamb waves

447

modulated sine pulse, we can see the a0 mode partially mode converted at the rst discontinuity. As in the previous gures, there is a wealth of propagation mechanisms due to the discontinuous structure of the specimens, which can be exploited for the purpose of quantitative NDE.

III. Conclusions We have presented applications of the LISA simulation approach to the propagation of LWs in complex specimens. The complexity arises from several sources: the heterogeneity and anisotropy of the materials (composites, such as glass bers and carbon bers reinforced plates), the presence of various kind of aws (e.g., holes, notches, subsurface defects, delaminations) and the geometry, which can be very intricate (such as in structures with T- or Yjunction stringers). For the purpose of validating the method, we have presented examples of a direct comparison between numerical and experimental results, showing a quantitatively good agreement. The performed simulations may be useful to provide a deeper insight into the LWs propagation mechanism and, as a consequence, help to optimize ultrasonic NDE techniques. The possibility of applying the LISA approach even to complex structures, such as shown in Fig. 15, clearly demonstrates the applicability and usefulness of the approach. Also a detailed study of the variety of wave propagation eects, which take place on these complex structures, may be very useful for the development and improvement of ultrasonic NDE techniques.

Fig. 15. The Y-junction stringer. Snapshots of the in-plane and outof-plane components of the displacement at dierent time steps.

to Fig. 10. The eect of the stack of delaminations can be observed, particularly in the rst plot. A quantitative evaluation of the eect of the damage can be achieved from Fig. 14, which displays the signal at the receiver. Here, in addition to the phase shift, we notice a much stronger decrease in the amplitude due to the large number of delaminations in the stack. D. Geometrically Complex Structures Many of the fundamental aerospace structures include panels, sometimes of varying thickness, and reinforcing elements. An analysis of the junctions between them is important to evaluate the applicability of ultrasonic techniques to practical structures. Simulations may provide an insight on the interaction between LW s and these basic structures. A detailed analysis of the results of our simulations in these kind of structures is beyond the scope of this paper. We limit ourselves to present in Fig. 15 a twodimensional example of LW propagation in a CFRP Yjunction stringers (bers lying parallel to the length of each element). We show two snapshots of the longitudinal and tranverse displacement, u and v , respectively, at dierent times. After the injection of a hanning-window-

Acknowledgments The authors wish to thank Drs. T. Monnier (INSA Laboratoires, Lyon, France) and S. Grondel (University of Valenciennes, Valenciennes, France) for allowing us to include their experimental results in this paper and also for fruitful discussions.

References
[1] A. Derode, P. Roux, and M. Fink, Robust acoustic time reversal with high order multiple scattering, Phys. Rev. Lett., vol. 75, pp. 42064209, 1995. [2] M. Fink, Time reversed acoustic, Phys. Today, vol. 50, pp. 3440, 1997. [3] P. P. Delsanto, P. A. Johnson, M. Scalerandi, and J. Tencate, LISA simulations of time reversed acoustic and ultrasonic experiments, J. Phys. D: Applied Phys., to be published. [4] J. L. Rose, Ultrasonic Waves in Solid Media. Cambridge: Cambridge Univ. Press, 1999. [5] I. A. Victorov, Rayleigh and Lamb Waves: Physical Theory and Applications. New York: Plenum, 1967. [6] O. Lenoir, J. Duclos, J. M. Conoir, and J. L. Izbicki, Study of Lamb waves based upon the frequency and angular derivatives of the phase of the reection coecient, J. Acoust. Soc. Amer., vol. 94, pp. 330343, 1993. [7] M. J. S. Lowe and P. Cawley, Comparison of the modal properties of a sti layer embedded in a solid medium with the minima

448

ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 50, no. 4, april 2003 of the plane-wave reection coecient, J. Acoust. Soc. Amer., vol. 97, pp. 16251637, 1995. P. D. Wilcox, Lamb wave inspection of large structures using permanently attached transducers, Ph.D. dissertation, Dept. of Mechanical Engineering, Imperial College, London, UK, 1998. O. Poncelet and M. Deschamps, Lamb waves generated by complex harmonic inhomogeneous plane wave, J. Acoust. Soc. Amer., vol. 102, pp. 292300, 1997. C. W. Chan and P. Cawley, Lamb waves in highly attenuative plastic plates, J. Acoust. Soc. Amer., vol. 104, pp. 874881, 1998. N. Guo and P. Cawley, The interaction of Lamb waves with delaminations in composite laminates, J. Acoust. Soc. Amer., vol. 94, no. 4, pp. 22402246, 1993. M. Castaings and P. Cawley, The generation, propagation, and detection of Lamb waves in plates using air-coupled ultrasonic transducers, J. Acoust. Soc. Amer., vol. 100, no. 5, pp. 3070 3076, 1996. Y. Cho, Estimation of ultrasonic guided wave mode conversion in a plate with thickness variation, IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 47, no. 3, pp. 591602, 2000. P. P. Delsanto, R. S. Schechter, H. H. Chaskelis, R. B. Mignogna, and R. Kline, Connection machine simulation of ultrasonic wave propagation in materials. II: The 2-D case, Wave Motion, vol. 20, pp. 295314, 1994. P. P. Delsanto, R. S. Schechter, H. H. Chaskelis, R. B. Mignogna, and R. Kline, CM simulation of the ultrasonic wave propagation in materials. III: The 3-D case, Wave Motion, vol. 26, pp. 329 339, 1994. P. P. Delsanto, R. B. Mignogna, M. Scalerandi, and R. S. Schechter, in New Perspectives on Problems in Classical and Quantum Physics. vol. 2, P. P. Delsanto and A. W. Saenz, Eds. New Delhi: Gordon & Breach, 1998, pp. 5174. P. P. Delsanto and M. Scalerandi, A spring model for the simulation of the propagation of ultrasonic pulses through imperfect contact interface, J. Acoust. Soc. Amer., vol. 104, pp. 2584 2591, 1998. R. S. Schechter, H. H. Chaskelis, R. B. Mignogna, and P. P. Delsanto, Real time parallel computation and visualization of ultrasonic pulses in solids, Science, vol. 265, pp. 11881192, 1994. R. D. Huber, R. B. Mignogna, R. S. Schechter, and P. P. Delsanto, Dynamic full-eld visualization of ultrasound interacting with material defects: Experiment and simulation, Ultrasonics, vol. 35, pp. 716, 1997. P. P. Delsanto, G. Perego, M. Scalerandi, and D. Zoccolan, Efciency of dierent ultrasonic surface waves for subsurface aws detection, in Review of Progress in Quantitative Non Destructive Evaluation. vol. 18A, New York: Plenum Press, 1999, pp. 127133. V. Agostini, J.-C. Baboux, P. P. Delsanto, T. Monnier, and D. Olivero, Application of Lamb waves for the characterization of composite plates, in Non Destructive Characterization of Materials IX, AIP Proc.. R. E. Green, Ed. New York: American Institute for Physics, 1999, pp. 455460. T. Monnier, private communication, April 2000. E. Moulin, private communication, April 2000. M. Deschamps and O. Poncelet, Transient Lamb waves: Comparison between theory and experiment, J. Acoust. Soc. Amer., vol. 107, pp. 31203129, 2000. V. Agostini, P. P. Delsanto, and D. Zoccolan, Flaw detection in composite plates by means of Lamb waves, in Proc. 15th World Conf. Non-destructive Testing, October 2000, pp. 134137. Alamos, NM. Her main elds of investigation are: propagation of acoustic and ultrasonic waves, inverse scattering problem, and modelling of biophysical systems. She has eight publications in international journals and conference proceedings.

[8] [9] [10] [11] [12]

[13] [14]

[15]

Pier Paolo Delsanto was born in Torino, Italy, on March 14, 1941. He is a former professor at the University of Puerto Rico, Mayaguez, Puerto Rico, and presently is a full professor at the Politecnico of Torino, Torino, Italy. For 8 years he has been director of the local Research Unit in Condensed Matter Physics. Previously he spent 3 years at the University of Frankfurt (Germany) with a postdoctoral fellowship, three years as a research physicist at the Naval Research Laboratory (Washington, D.C.), and shorter periods as a visiting professor at Duke University (Durham, NC), University of Montreal (Canada), and University of Melbourne (Australia). He has more than 200 publications in scientic journals and two books in various research elds, ranging from nuclear physics to acoustics, nondestructive evaluation, and biophysics. He has been coordinator of several European and American research programs and invited speaker, session chairman, and/or organizer at many international conferences and workshops.

[16]

[17]

[18]

[19]

Ivan Genesio was born in Bra (CN), Italy, on August 25, 1975. He took his Doctors Degree in physics (Summa cum Laude) at the University of Torino, Torino, Italy, in 1999 with an experimental thesis in collaboration with the National Institute of Nuclear Physics (INFN), Torino, Italy. After some experience in software development for industries, he had a fellowship position at the Politecnico of Torino. The focus of his studies was LISA Simulations of the propagation of Lamb waves in linear and non linear media with special attention on parallel computing techniques.

[20]

[21]

[22] [23] [24] [25]

Dimitri Olivero was born in Carmagnola (To), Italy, on September 29, 1973. He took his Doctors Degree in material science engineering (Summa cum Laude) at the Politecnico di Torino, Torino, Italy, in 1999 with a thesis about the nondestructive characterization of welded joints. In April 2002 he obtained his Ph.D. at the Physics Department of the Politecnico di Torino. The focus of his studies are the numeric methods for the propagation of electromagnetic and ultrasonic waves.

Valentina Agostini was born in Roma, Italy, on November 2, 1970. She completed her Ph.D. degree in physics in 2001 at the Politecnico of Torino, Torino, Italy. She was invited by the International Rectier Corporation as a consultant to their Los Angeles, CA, facilities. During her fellowship at the Politecnico she had teaching experience and took part in a Brite Euram Project with seven European institutional partners. She spent some periods at the Los Alamos Research Laboratory, Los