Ultrasonics 44 (2006) e951–e955 www.elsevier.com/locate/ultras

Simulation and data reconstruction for NDT phased array techniques S. Chatillon *, L. de Roumilly, J. Porre, C. Poidevin, P. Calmon Commissariat a` l’E´nergie Atomique, LIST, CEA Saclay Baˆt. 611, 91191 Gif-sur-Yvette cedex, France Available online 5 June 2006

Abstract Phased array techniques are now widely employed for industrial NDT applications in various contexts. Indeed, phased array present a great adaptability to the inspection configuration and the application of suitable delay laws allows to optimize the detection and characterization performances by taking into account the component geometry, the material characteristics, and the aim of the inspection. In addition, the amount of potential information issued from the inspection is in general greatly enhanced. It is the case when the employed method involve sequences of shots (sectorial scanning, multiple depth focusing etc) or when signals received on the different channels are stored. At last, application of electronic commutation make possible higher acquisition rates. Accompanying these advantages, it is clear that an optimal use of such techniques require the application of simulation-based algorithms at the different stages of the inspection process: When designing the probe by optimizing number and characteristics of element; When conceiving the inspection method by selecting suitable sequences of shots, computing optimized delay laws and evaluating the performances of the control in terms of zone coverage or flaw detection capabilities; When analysing the results by applying simulation-helped visualization and data reconstruction algorithms. For many years the CEA (French Atomic Energy Commission) has been being greatly involved in the development of such phased arrays simulation-based tools. In this paper, we will present recent advances of this activity and show different examples of application carried out on complex situations. Ó 2006 Elsevier B.V. All rights reserved. Keywords: Phased array; UT simulation; Reconstruction

1. Introduction Over last years, significant technological advances have been made in phased arrays techniques, mostly in terms of acquisition systems (versatility, miniaturization) and transducers technology (piezocomposite probes with highly reduced acoustical cross-talk). Such progress makes the phased array technology a very powerful tool in terms of adaptability and versatility for a wide range of industrial applications. However, an optimal utilization of such system, in particular for complex applications such as beamsteering, electronic commutation or multiple parameters settings, requires modeling-based conception and exploitation. Modeling is applied to compute suitable delay laws, to simulate the inspection, allowing feasibility study or per-

*

Corresponding author. E-mail address: [email protected] (S. Chatillon).

0041-624X/$ - see front matter Ó 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.ultras.2006.05.060

formance demonstration and reconstruction of acquired data. In a first time, we will briefly present the UT simulations tools developed at the French Atomic Energy Commission (CEA) for several years and their application on a basic case. In a second time, a Model-based ray data reconstruction is proposed an applied on two examples of simulated data. 2. Simulation tools for phased array techniques Modeling tools allow to simulate realistic NDE configurations inspections. These models, gathered in the Civa software [1,2], aim at being able to conceive, optimize and predict the performances of various NDE methods. Such models may also be used for experimental data inversion [3] or complex results interpretation. A very broad range of realistic configurations has to be dealt with, in terms of specimen, probes, scatterers, inspection method, etc. Such configurations also need to be

e952

S. Chatillon et al. / Ultrasonics 44 (2006) e951–e955

modeled with high speed computation codes for parametrical studies. Semi-analytical models have therefore been developed: they include the simulation of the beam propagation [4], as well as defect scattering, [5–7]. Phased array modeling tools allow to deal with various transducer designs: 1D or 2D linear, annular or encircling array probes. The pulse responses of the field radiated by all elements of the array are individually computed and stored. Therefore the beam radiated by the array may be readily synthesized as the sum of these individual pulse responses, time shifted and weighted according to the delay and amplitude laws without any need of new calculations [8]. As an illustration of these simulation tools abilities [9], the following example demonstrates a compensation of aberration effects due to an irregular profile. A linear array is used at immersion over a specimen made of an irregular profile. Simulated transmitted beam profiles have been reported on Fig. 1, for two different materials: Plexiglas and ferritic steel, with respective sound celerities in Longitudinal mode 2720 m/s and 5900 m/s, for each following labeled configuration: ‘‘no delays’’ (all elements are simultaneously shot), ‘‘phase compensation’’ (delays are optimized to recover a planar wave front in the material), and ‘‘phase compensation + focusing’’ (delays are optimized to focus the beam at the profile depth, taking account of the irregular profile). If no delays are applied, the profile variation clearly leads to beam distortions, both in terms of amplitude and arrival time, even in the case of slight aberrations (on the irregular profile, the altitude variations are about ± 1 mm over a 40 mm distance, within slow slope variations (see Fig. 1). These effects are obviously stronger for the water/steel interface than for the water/Plexiglas interface. Adapted delays allow to compensate these aberrations, either if the aim is to transmit planar waves or to focus the beam.

3. Model-based data reconstruction The objective of reconstruction is to display the timedependent ultrasonic echoes in real spatial coordinates, in order to get an optimal positioning of defects. A standard and simple processing of UT data consists in applying simple coordinates transformation, i.e., displaying the data according to spatial coordinates (X, Z) related to the specimen, rather than displaying the echoes according to acquisition parameters (time and scanning length for instance). Such a transformation postulates that one can get an accurate knowledge of the refraction angle and the velocity of the material, so that the only determination of the time of flight of the ultrasonic echo allows to give its position in spatial coordinates, assuming that one can define the propagation as a ray along which ultrasonic echoes are displayed. Such an approximation is classically applied for sectorial scanning procedure, as illustrated below. However, it has been pointed out that such a simple procedure may give rise to inaccurate positioning of echoes – and consequently bad positioning of defects – if the complete propagation path is not correctly taken into account. Fig. 2 illustrates this point for an immersion linear array parallel to a plane water/steel interface: a snapshot of the ultrasonic field radiated by the array is displayed, as well as two possible rays which would correspond to the UT path, the first one emanating from the central axis of the probe projected to the interface, the second one taking into account the deflection of the beam in the coupling medium. As a delay law is applied to steer and to focus the beam out of the axis of the probe, the ‘‘ray’’ which shall model the propagation in the material has to be shifted from the projection of the axis over the interface, as the propagation in the water also needs to be taken into account. Fig. 3 presents a comparison between two reconstructed sectorial images, the first one obtained by considering rays

Fig. 1. Simulation of phase aberration effects due to an irregular geometry and their compensation using a linear array with adapted delay laws.

S. Chatillon et al. / Ultrasonics 44 (2006) e951–e955

Fig. 2. Snapshot of ultrasonic beam radiated by a linear array used in steering and focusing mode, and two possible ‘‘ray’’ propagations assumed in dashed and dotted lines.

emanating from the same origin impact point, corresponding to the center of the probe projected over the interface, whereas the second one assumes rays on distinct impact points (one per applied steering angles), which abscissa

e953

have been calculated using simulation tools previously discussed. The previous reconstruction has been applied to the sectorial scanning inspection of a complex (irregular profile) component containing two series of 2 mm diameter side drilled hole, from 20 to 50 mm depth. The first set lies below a planar profile, while the second one lies below an irregular profile, which exhibits depth variations about ± 0.5 mm. Fig. 4 presents the reconstructed cross-section views associated to the three different cases (from left to right): planar profile with delays computed for the plane interface, irregular profile with delay laws adapted for the plane interface, and irregular profile with delays computed to take account of the actual irregular profile. It can be readily pointed out that the use of delay laws (and, subsequently, of rays reconstructions) nonadapted to the irregular profile below the probe cannot lead to an accurate positioning of the defects. As displayed in the figure, where the actual positions of the defects are also reported, the maximum amplitudes of the echoes arising from the side drilled hole are mostly shifted (about more than 20 mm

Fig. 3. Comparison between classical and model-based ray reconstruction.

Fig. 4. Reconstructed cross-section views for the following configurations (from left to right): regular interface, irregular interface with laws computed for a planar interface, and irregular interface with adapted laws.

e954

S. Chatillon et al. / Ultrasonics 44 (2006) e951–e955

on left side of actual positions), and the echoes are enlarged compared to the planar profile case. The use of adapted delays and reconstruction allows to compensate this effect, so that similar performances may be obtained both for the planar and irregular profile. A second example deals with the assessment of a composite stiffener with sharp bending radius. For such a highly bended component, standard inspection should lead to limited performances due to uncovered scan area and beam

disorientation. To overcome these difficulties, a 64 elements semi-circular linear array with a 10 MHz central frequency has been designed using the Civa simulation tools. Fig. 5 shows how electronic commutation enables a full coverage of the small radius part, keeping 0° L-wave inspection. The electronic commutation technique is one of the most usual application for phased arrays techniques: it consists in using a limited number of active elements (at transmission and reception) multiplexed over a large array.

Fig. 5. Semi-circular linear array + electronic commutation to keep 0° L-wave inspection.

Fig. 6. Simulation of the inspection of the stiffener with planar delaminations embedded along the curved part. a/ aligned b/ tilted and/or misaligned.

S. Chatillon et al. / Ultrasonics 44 (2006) e951–e955

In this case, the first eight elements are used at transmission and reception, then this aperture of eight elements is successively translated with a step of one element until to cover the whole array. This technique is classically apply to reach very high acquisition rates with a simplified mechanical device, since one of the displacement direction can be replaced by an electronic commutation. However, the limitation of this technique is that the inspection resolution depends both on the beam spot and the distance between two adjacent elements. To illustrate the performances of this technique, the inspection of this component with several planar delaminations embedded along the bended part have been simulated. Fig. 6 shows the simulated result represented in the inspection coordinates (electronic steps) and in the component coordinates using the previous data reconstruction method. Two sets of defects are considered, one with defects aligned along the curved part and one with tilted and/or misaligned defects. For the first set, we can see that all the defects are correctly detected and positioned in the component. In particular, the control of the beam orientation (0° L-wave inspection) enables to obtain the same sensitivity on all the defects. For the second one, all the defects are correctly detected and positioned. We can clearly see significant loss of the sensitivity on the two first defects due to their disorientation according to the beam. 4. Conclusions Modeling tools developed at CEA, in terms of wave propagation and echo formation simulations, allow both to optimize and to predict performances of phased array techniques. In a first time, simulation tools have been used to illustrate the ability to predict and to compensate beam distortions or deviations which may occur through a complex shaped specimen. In a second time, a ‘‘model-based

e955

ray data reconstruction’’ technique has been proposed. This development aims at displaying the echoes as accurately as possible in the spatial coordinates of the components, so that one can efficiently locate and to characterize flaws. Its application to sectorial scanning inspection and to electronic commutation leads to satisfying results. References [1] P. Calmon, A. Lhe´mery, I. Lecoeur-Taibi, R. Raillon, Integrated models of ultrasonic examination for NDT expertise, in: D.O. Thompson, D.E. Chimenti (Eds.), Review of Progress in QNDE, vol. 16, Plenum Press, New York, 1997, pp. 1861–1868. [2] A. Lhe´mery, P. Calmon, I. Lecœur-Taı¨bi, R. Raillon, L. Paradis, NDT E Int. 37 (2000) 499–513. [3] G. Haiat, P. Calmon, F. Lasserre, Application of ultrasonic modeling to the positioning of defect in a cladded componentReview of Progress in QNDE, vol. 23A, AIP publishing, 2004, pp. 103–109. [4] N. Gengembre, A. Lhe´mery, Calculation of wideband ultrasonic fields radiated by water-coupled transducers into heterogeneous and anisotropic mediaReview of Progress in QNDE, vol. 19, AIP publishing, 2000, pp. 977–984. [5] R. Raillon, I. Lecœur-Taı¨bi, Transient elastodynamic model for beam defect interaction. Application to nondestructive testing, Ultrasonics 38 (2000) 527–530. [6] M. Darmon, P. Calmon, C. Bele, Modelling of the ultrasonic response of inclusions in steelsReview of Progress in QNDE, vol. 22A, AIP publishing, 2003, pp. 101–108. [7] N. Gengembre, A. Lhe´mery, R. Omote, T. Fouquet, A. Schumm, A semi-analytic-FEM hybrid model for simulating UT interaction involving complicated interaction of waves with defectsReview of Progress in QNDE, vol. 23A, AIP publishing, 2004, pp. 74–80. [8] O. Roy, S. Mahaut, M. Serre, A. Lhe´mery, Application of ultrasonic beam-forming and self-focusing techniques to defect characterizationProceedings of the First International Conference on NDE in Relation to Structural Integrity for Nuclear and Pressurized Component, Woodhead publishing limited, Cambridge, 1999, pp. 480–487. [9] S. Mahaut, S. Chatillon, E. Kerbrat, J. Porre´, P. Calmon, O. Roy, New features for phased array techniques inspections: simulation and experiments, in: Proceedings of the 16th World Conference on NDT.