IRSN FRO400099

APPLICATION OF PHASED ARRAYTECHNIQUES TO COARSE

GRAIN COMPONENTS INSPECTION

G6rard CATTIAUX,Steve MAHAUT, Jean-Louis GODEFROIT, Olivier ROY

Rapport IRSN/D6partement d6valuation de &W No 624

Octobre 2003

I NSTIT UT DE RA DI P ROTECTION ET DE SO R ETt N UCLtA I RED t P A R T E M E N T D t V A L U A T I N D E S R E T t

RAPPORT DES/624

APPLICATION OF PHASED ARRAYTECHNIQUES TO COARSE

GRAIN COMPONENTS INSPECTION

G6rard CATTIAUX *,Steve MAHAUT, Jean-Louis GODEFROIT, Olivier Roy

ProceedingsUltrasonic International 2003

Juillet 2003

IRSN/DES/SAMSCEA/LIST/SISC SACLAY

Octobre 2003

APPLICATION OF PHASED ARRAY TECHNIQUES TO COARSE RAINCOMPONENTS INSPECTION

Steve NtAHAUT.-, Jean-Louis GODEFROIT, Olivier ROY a, G6rard CA'171AUX b.Commissariat a PEnergie Atornique, LIST/SISC, CEA Saclay, Bdt. 611, 91191 Gif-sur-Yvette cedex, Franceb nstitut de Radioprotection et de SCiret6 Nucl6aire, DES/SAMS, BP 17, 92262 Fontenay-aux-Roses cedex,France

Abstract

Ultrasonic inspection of cast stainless steel components from primary and auxiliary cooling circuits ofFrench Nuclear Power Plant has to face with major difficulties due to the coarse grained structure ofthese materials. Attenuation losses and structural noise are encountered, which limits the performances ofdefect detection ability, mostly in terms of degraded signal-to-noise ratio and poor sensitivity. Toovercome such problems, theoretical and experimental studies have been achieved at the French AtomicEnergy Commission, with support from the French Institute for Radiolo ical Protection and Nuclear

Safety.Experimental studies have been performed over stainless steel specimen of known coarse structure(equiaxial grains and/or elongated grains), containing artificial reflectors (cylindrical holes and electro-eroded surface breaking notches). Those mock-ups have been inspected using contact probes of differentarray designs (linear or matrix splitting), and using pulse echo or dual-element techniques. Such arraysallow to control the ultrasonic beam so as to investigate different inspection angles and focusing depths.Experiments were carried out using oblique longitudinal waves, using delay laws computed by a specificmodel, taking account of acoustical and geometrical properties of the probes and the inspected

component.

In addition, specific reconstruction techniques have been investigated to enhance the signal-to-noise ratioas well as spatial resolution. These techniques are based on beam-forming summation and multi-angleinspections. Experimental results show that such techniques allow to reduce the speckle noise and tooptimise the beam resolution. Those increased performances allow to detect and to size small planardefects located at the inner wall of a thick specimen, using comer and tip diffraction echoes.Keywords phased arrays, stainless steel, beam forming image and signal processing.�, Corresponding author. Tel: 331 69 08 47 14; Fax: 331 69 08 75 97; e-mail:steve.mahautgcea.

1. Introduction

Cast stainless steel is widely used in pmary and secondary cooling components of pressurized waterreactors (PWR), because of its good corrosion resistance, high strength and weldability. However, theperformances of ultrasonic inspections of such materials are degraded because of its coarse grainstructure, which leads to attenuation losses and scattering noise. Such phenomena have been specificallystudied in the nuclear industry, and one may find relevant synthesis about these effects and somerecommendation for probes optimization in ternis of frequency and operating mode in [1]. Basically,longitudinal waves generated by low frequency probes (about I MHz) are usually suggested to reducethe backscattered noise and therefore to enhance the signal-to-noise ratio 2 Large aperture focusing

probes are therefore required to preserve a thin lateral resolution (hence a good defects sizing accuracy)in spite of this low frequency, which is difficult to be carried out for contact inspections, because ofpossible internal reflections in the wedge, as well as unperfect contact between the specimen and theprobe. Studies have then be perforined at the French Atomic Energy Commission (CEA), supported bythe Institute for Radiological protection and Nuclear Safety (IRSN), to investigate the application ofphased arrays techniques for cast stainless steel inspection.

2. Simulation of phased array techniques

2.1 Considerations for the phased array probes design

As already mentionned, large aperture focusing probes operating at low frequencies are conventionnalyused to reduce the attenuation losses and backscattered noise These characteristics cannot be readilyimplemented for contact probes, because large wedges may also give rise to internal echoes inside thewedge which lead to a "blind" detection zone close to the probe (near surface echoes may not bedistinguished from the wedge echoes). For such cases, dual-element probes, made of separatetransmitting and receiving probes over half-wedges acoustically isolated, are very efficient to remove thisdead zone, and the possibility to use large aperture makes them suitable for thick inspection of coarsegrain components 3,4]. At last, both transmitting and receiving part of the probe have been splitted intomatrix arrrays in order to focus and steer the beam both in the incidence and perpendicular plane. Theconception of this dual-element array probe has been made using the modeling tools gathered in the Civaexpertise software developed at the CEA [5]. Those tools allow to predict ultrasonic fields radiated byvarious transducers into complex (geometry and structure) specimen 6], as well as complete inspectionresult (accounting for beam-defect interaction for various paths between the probe and the defect - director comer echoes - and interaction with the boundaries of the specimen).

2.2 Prediction of ultrasonic field in the specimen

Figure I shows the ultrasonic fields radiated by the dual-element matrix array, designed to inspect planaror cylindrical specimens (using a shaped wedge to match a pipe of 450 mm external radius) of 70 mmthickness with 45' longitudinal waves. Dimensions and array splitting design of this probe, operating at800 kHz frequency, are also reported. Fields radiated in the incidence and perpendicular plane aredisplayed, corresponding to longitudinal (L) waves steered at 45' and focused at 70 mm (at backwall).One can observe that the delay laws applied to the array probe allow to obtain the desired beamcharacteristics into both planar and cylindrical specimen (focused 45' L-waves, very low array lobeslevel, almost no shear waves).

2.3 Prediction of the inspection performances

Figure 2 displays the simulated inspection of the planar and cylindrical specimen containing verticalplanar defects emerging at backwall, from 3 to 15 nun height. As the inspection is performed using 45'L-waves, the echoes received by the dual-element array probe are related to the comer echo path(reflection at backwall and at the defect), and the tip diffraction echoes (with or without reflection atbackwall), which allow to size the defect. For both configurations, one can observe that the tip diffractionecho is clearly distinguished from the comer echo for the IO mm height defect. The overall responses of

2

defects from 3 to 10 mm height do not exceed large variations (less than 6 dB), which tends to indicatethat such defects would be detected within a reasonable signal-to-noise ratio.

3. Application of the phased array inspections over realistic mock-ups

3.1 Description of mock-ups and array probes

Cast stainless steel specimen of equiaxial and uniaxial (elongated grains) structures have been used toevaluate the performances of different contact array probes. Artificial defects (side drilled holes andelectro-eroded planar notches at backwall) have been implanted in those specimen to evaluate defectdetection and sizing abilities. Figure 3 shows typical macrographs of both structures as well as theimplantation of the artificial defects (corresponding to the defects dimensions already used for theprevious simulated results). The designed dual-element matrix probe has been used to inspect thesecomponents, as well as an additional existing contact array probe in pulse echo mode, which was notdesigned for this application, of 2 MHz central frequency, 128x5Omm' 64 elements).

3.2 Comparison of inspections of a planar stainless steel mock-up with both arrayprobes

Figure 4 shows a comparison of performances modification (compared to a erritic steel specimencontaining calibration side drilled holes) observed on a cast stainless steel component inspected withboth probes previously described. Attenuation losses and signal-to-noise ratio are drastically degradedcompared to the ferritic steel inspection using the 2 MHz probe (an amplification of 20 dB is set from theferritic to the stainless steel inspection, and the SNR in the stainless steel component is about 4 dB),whereas the 800 kHz dual-element array probe allows to obtain a SNR about 14 dB, while the attenuationlosses are about 6 dB. It also has to be mentionned that these performances are obtained thanks to thefocusing pattern applied to the matrix probe.

3.3 Inspection of a cylindrical mock-up

The dual-element matrix array probe has been used to inspect a cylindrical forged stainless steelcomponent. For this configuration, both matrix arrays used at Transmission/Reception were mountedonto an adapted cylindrical wedge. Figure shows the experimental Cscan (213 raster scanning patternover the cylindrical specimen, as illustrated on the simulation result on figure 2 and Bscan views (crosssection images of the inspection) related to the different planar defects from 3 nun to 15 nun height. Allthe defects are detected, and defect sizing using the tip diffraction echo is performed from the mmheight defect. These results confirm the ability of this array design to optimally focus and steer the beamto provide a high resolution and sensitivity (although experiments over a cylindrical cast stainless steelhave not been perfon-ned, comparison with the planar stainless steel mock-up shows that for similarmaterials, it should still be allowed to detect all of these defects with similar SNR as those obtained onthe planar mock-up).

3.4 Application of multi-angle reconstruction for the planar cast stainless steel mock-upinspection

3

In addition to the inspections carried out with the phased arrays in "conventional" mode (application ofone set of delays at transmission and reception for the whole scanning displacement), another technique,making use of multi-angle inspections and summation has been investigated. This technique may bedenoted as tomoSAFT", as it relies on summation of echoes received at different scanning positions, buteach of these echoes is received following a given arbitrary inspection angle applied both at transmissionand reception thanks to a multi-angle delay law settings. This procedure therefore acts as follows as thearray probe is continuously scanning the specimen, consecutive shots are made to transmit and receivelongitudinal waves focused at the backwall within an arbitrary range of angles. The echo reconstructionis obtained using the echoes from each angle inspection, temporally and spatially shifted according to thedefect location at backwall. Figure 6 displays the comparison of the inspection carried out over the planarmock-up using 450 L-waves, and the presented technique with three different angles inspections 40'/45'/50'. The experimental Bscan images from the 45' L-waves inspection and from the"tomoSAFT" technique are displayed, which shows that backscattered echoes have been decreasedcompared to the defect echoes. Defect echoes are also slightly more focused, as the technique leads to alarger synthetic aperture, both at transmission and reception. Echodynan-iic curves also displayed on thisfigure also allow to estimate the improvement of the SNR, which is about dB using this technique. Itcan also be pointed out that some backscattered noise echoes still remain as the technique is only appliedfor a given focusing depth (at backwall).

4. Conclusion

Experimental and theoretical investigations of phased array applications have been carried out toimprove the detection and sizing accuracy of defects located in thick cast stainless steel specimen. Aspecific matrix array probe used in dual-element mode has been designed and experimentally validatedover mock-ups with representative material structures, containing electro-eroded planar notches andcylindrical holes for calibration purposes. Experiments have been made with longitudinal waves focusedat back-wall, and show that the lateral and temporal resolution allows to detect all defects (from 3 to 5mm height) and to size them (from 10 mm height) in a 70-mm-thick specimen) with a fairly good sgnal-to-noise ratio (higher than 10 dB). Additional tests have also been performed over a cylindricalspecimen. At last, experimental studies have been made to investigate multi-angle reconstruction, inorder to increase the signal-to-noise ratio and the lateral resolution. Preliminary results obtained withsuch a technique, which can only be applied with a phased array thanks to its ability to electronicallysteer and focus the beam at transmission and reception while continuously scanning the specimen, showthat one can significantly improved the signal-to-noise ratio.

5. References

[I] - G. Maes, B. Hansoul, P. Dombret, Proc. of the 12 th inter. Conf. on NDE in the nuclearand pressure vessel ind., 1994, 197.

[2 - M. Serre, P. Benoist, D. Villard, N. Demathan, op.cit. , 191.

[3 - Y. Kurozwni, in Insight, Vol. 44, N' 7 2002, 437.

[4 - M. Delaide, G. Maes, D. Verspeelt 2 nd Int. Conf. On NDE in relation to Structural

4

Integrity for Nuc. And Press. Components, 2000, 347.

[5 - S. Chatillon, A. Lh6mery, F. Cartier, P. Calmon, in D. 0. Thompson, D. E. Chimenti

(Eds.), Rev. of Prog. in QNDE American Institute of Physics, 2001, 20A, 71 .

[61 - S. Mahaut, 0. Roy, C. Beroni, B. Rotter, in Utrasonics, Vol. 40, N' 1-8, 2002, 165.

5

U12 elem4BxB mm

130 m mm

--.. db.-

Figure 1

6

scanning

Scanning BO mmScanning 250 mm

Ra

00

Ep0 U

CX

Figure 2

70

Figure 3

7

Ferritic steel Cast stainless steelSide drilled holes -;-- 55 mm Side drilled holes 10 -- 70 mm

Scanning (mm) 330 Scanning (mm) 232

N CM SNR 4 dBGain 20 dB

X

E EP

can

Ea B

25

Figure 4

Scanning (80 mpi)-,

,'oft%

'Tr

r.0E0;wUa

... . ......... .. ......... .- -. ............... ....... ......... ...

Figure 5

Figure 6

9

Figure 1: Simulation of UT fields radiated by a twin-crystal matrix array probe through a planar (left)and a cylindrical (right) specimen

Figure 2 Simulation of inspections perforined by the twin-crystal matrix array probe through a planarand cylindrical mock-ups containing planar notches at backwall

Figure 3 Macrographs and artificial defects implantation in the experimental mock-ups

Figure 4 Experimental inspection of ferritic steel (left) and cast stainless steel nght) planar calibrationmock-ups with array probes at 2 MHz (top) and 00 kHz (bottom)

Figure 5 : Experimental inspection of a cylindrical forged stainless steel mock-up with the twin-crystalarray probe

Figure 6 Comparison of 45' longitudinal waves inspection (left) and multi-angles reconstructiontechnique (right) carried out with the twin-crystal array probe.

10