·Home ·Table of Contents ·Methods and Instrumentation

Comparative Analysis of the Ways to Increase Signal to Noise Ratio at NDT Inspection of Austenitic Welds V.Grebennikov, V.Badalyan, D.Grebennikov, A.Vopilkine Scientific and production centrer "ECHO+", 1, Kurchatov sq., Moscow, 123182, Russia, Tel: +7-095-196-91-91; Fax: +7-095-935-73-90; E-mail : echo.ndt@ g23.relcom.ru Contact

The studies of welded joints of 40 and 55 mm thickness were executed by means of ultrasonic holographic system. The advantage of two-frequency and two-mode methods in combination with of acoustical holography for austenitic welds testing is shown.
KEY WORDS: austenitic welds, signal to noise ratio, two-frequency method, two-mode method.

INTRODUCTION

The method of ultrasonic testing of austenitic welds by longitudinal waves is widely implemented in Russia and other countries [1, 2]. Moreover, for the realisation of the testing method TR-probes focusing ultrasonic ray on determined zone of thickness are used. The testing of the welds of significant thickness (more than 30 mm) is performed step by step using different probes that complicates the testing procedure. The interpretation of testing results is complicated due to presence of shear wave signals (probes can transmit and receive both longitudinal and shear waves). One should note that detection of the defects in the whole volume of tested weld by the longitudinal wave inspection requires the removing of weld reinforcement. Moreover, the surface and undersurface plate-type defects (cracks, lack of fusion) are bad detected when testing by longitudinal waves.
To increase signal/structure noise ratio by austenitic weld inspection the multifrequency testing method was developed and implemented in Russian atomic energy industry [2, 3]. Its essence is that the weld is tested by the pulses of different frequencies, received echo-signals are compared by time of reception and amplitude; the signals with equal time of reception and amplitude exceeding determined level of cut-off are selected. In case of optimal choice of frequency band and angle of ray incidence the multifrequency (two- frequency) method can increase the signal/structure noise ratio by 6 - 12 dB. The multifrequency method can be used by both longitudinal and shear wave inspection, by automatic and manual testing. One of possible variant of two frequency method is two- mode method, using at the same time two wave modes - longitudinal and shear one.
High resolution (exceeding in several times the resolution when testing by conventional ultrasonic inspection methods) allows to hope on successful use of the holographic system for testing the austenitic welds of different thickness.

EXPERIMENTAL

Testing Device.


Fig 1:The defect models in the defect free austenitic weld (the sample1)

The ultrasonic testing was performed using the ultrasonic system Augur 4.2 equipped by linear scanner [5].

Test Specimens.
The studies were performed on austenitic weld samples made of 12H18N10T stainless steel, 40 and 55 mm thickness. The welding was executed automatically under flux using the wire SV-04H19N11M3. All welds had V- and X-shape. Two of them were apriory defect free with following fabrication in them models of defects (Fig.1). The defect models were shaped as notches and side drilled holes. The other samples had specially embedded real defects (Fig.2, 3). The types of the real defects were lack of fusion, lack of fusion located at weld boundary, slag inclusions.

Fig 2: The sample 2, three lacks of fusion.

Fig 3: The sample 3, two slag inclusions and three lacks of fusion on boundary of the weld

Testing Methods.
The ultrasonic holography testing was executed through automatic transverse scanning (axis of priority scanning was perpendicular to the weld axis). The aperture of transverse scanning (across the weld) was 70-100 mm, step - 0,3 mm. The step of longitudinal movement (along the weld) was 5 mm. The testing was performed: by shear wave - direct and reflected beam and by longitudinal wave - direct beam only. The scanning wasn't executed on the surface of the weld reinforcement.

Probes
The angle beam longitudinal and shear probes with frequencies 1.65 ; 2.5 MHz and beam incidence angles 45^o, 50^o, 55^o, 60^o, 65^o were used. The used probes are marked by the following way: l - longitudinal wave; s - shear wave; 2 - frequency 1.65 MHz; 3 - frequency 2.5 MHz; h-one half directivity diagram angle -12^o-20^o; (d)- one half of directional diagram angle -10^o-15^o; two next numeral are incidence angle of probe.

Methods of Signal/Noise Increase.
To increase signal/structure noise ratio by holographic testing, two-frequency and two-mode methods were used in the work.
Two-frequency and two-mode methods were realized through combination of images that corresponds to the two frequencies (1.65 and 2.5 MHz) or the two wave modes (longitudinal and shear, both 2.5 MHz) using the following algorithm:

RESULTS

The signal/structure noises ratio (A/A_n) for volumetric-type defect models (thickness of austenitic weld 40 mm (the side drilled holes of diameter 3 mm and length 15 mm) are shown at Fig. 4-5 (the path in the base metal wasn't taken into account). Each group of points placed vertically corresponds to hole (two holes) detected from the left or the right from the weld axis. The mark of drills at Fig. 4-5 corresponds to the Fig.1. The holes marked by asterisk correspond to the maximal path through the fused metal (testing through the weld). One can conclude from the presented graphs that when testing by conventional method using probes by incidence angle 60^o and frequency 1.65 MHz and 2.5 MHz the signal/noise ratio for side drilled holes under the maximal way of passing through the weld metal is always less than the necessary criterion 6dB. So the weld can't be tested according to Russian testing rules [2]. The use of the holographic method increased the above mentioned ratio: by 4 - 5 dB in case of shear wave testing and by 6 - 10 dB in case of longitudinal wave testing. In addition the two-frequency method increased the signal/structure noises ratio by approximately 6 dB.

Fig 4: Dependence of signal/noise ratio on traversed path in austenitic weld metal. Longitudinal waves. Straight beam.

Fig 5: Dependence of signal/noise ratio on traversed path in austenitic weld metal

Fig 6: Dependence of signal/noise ratio on notch height in austenitic weld. Longitudinal waves (l2(3)-60). Straight beam.

Fig 7: Dependence of signal/noise ratio on notch height in austenitic weld. Shear waves (s2(3)-60). Straight beam.


The graphs of dependencies of defect signal/mean level of structure noises (A/A_n) on the notch height (h) are presented at Fig.6-7. One can see that this ratio is no more than 2 dB by testing by conventional method (one-frequency) using probes mentioned above. The ratio of the useful signals from notches to the mean level of structure noises is increased more than 6 dB when using the same shear wave probes and holographic method. The possibility of notch detection using shear wave is much better than in case of longitudinal one. Using the latter wave type one can't detect practically notches of 1mm height. The best results were obtained by using holographic method combined with two-frequency method: the useful signal exceeded noise level by 14 - 16 dB. Similar results were received by investigation the 55mm austenitic weld (Fig.8). In general the signal/noises ratios A/A_n were less by 4-6 dB than the same ratios for the 40mm austenitic weld.

Fig 8: Dependence of signal/noise ratio on notch height in austenitic weld of 55mm thickness.Straight beam.

Fig 9: Dependence of signal/noise ratio on notch height in austenitic weld of 40mm thickness. Shear waves (s2(3)-60). Reflected beam.


Ultrasonic testing of austenitic welds with non-removed reinforcement demands applying reflected beams. The investigation showed it is impossible to select useful signals from structure noises when conventional testing by shear waves reflected from the bottom surface: the mean level of structure signals exceeds the useful ones (A/A_n< 1). The ultrasonic testing using reflected from bottom surface longitudinal waves is impossible due to low sensitivity. The use of the two-frequency method in combination with holographic one allowed to detect successfully all notches (and notch of 1 mm height too) (Fig.9). One should note that conventional criterion of defect detection on noise background: +6 dB can be lowered to +3 dB in case of testing by holographic method. In this case we deal with defect image (the image can be easily selected from the chaotic noises).
The ratio A/A_n for side drills holes is practically independent on incidence angle within band 55-65^o for longitudinal wave. Maximum signal/structure noises ratio for notches is achieved by shear wave incidence angle 60^o. So 60^o is the optimal incidence angle for investigation austenitic welds by longitudinal and shear waves.

Fig 10: The non-holographic image (A - scan visualization) of defect models (see Fig. 1). Longitudinal waves, f=2,5MHz, a=60º. There are the crosscuts of the weld (B - to the left, D -to the right).	Fig 11: Holographic image of notches in austenitic weld (Fig. 1). There are the crosscut (B - to the left) and the plan of the weld (C -to the right). Longitudinal waves, f =2,5MHz a=60º. Satellite images occur due to the wave transformation.
Fig 12: Holography image of notches in austenitic weld. The crosscut (B - to the left) and the plan (C - to the right) of the weld. Shear waves, S3-60, f=2,5MHz a=60º.	Fig 13: Holography image of notches in austenitic weld (Fig.1). Two-frequency method (1.65MHz+2.5MHz), shear waves,a=60º. The images of the side drilled hole of Æ3 mm (z=20mm) are to the right side. Æ3mm.


The analysis of acoustical images of artificial defects in austenitic welds (side drilled holes and notches fabricated in a priory defect free sample, Fig.1) approved conclusions about most effective methods of signal/noise ratio increase based on signal/noise analysis (Fig.4-9). Using visualization images (A-scan visualization without holography) one can't distinguish notch images from color spots corresponding to structure noises (Fig.10). The holographic data processing applied to shear and longitudinal wave data allows make a good difference between images of defects and structure heterogeneities (rather worse in shear waves), (Fig. 11, 12). The additional advantage can be obtained using combination of images measured at two frequencies (two-frequency method, Fig.13, 14) or at two wave types (two-mode method, Fig.15).
    The efficiency of real defect detection by the two-mode method was studied on the austenitic weld samples with preliminary embedded defects (Fig. 2, 3). For example, one can see at Fig.16, 17 the images of lacks of fusion of various lengths.
    The studies executed at weld samples of 55 mm thickness gave the analogous results: reliable detection of side drilled holes of 3 mm diameter, notches of 3 mm height and more. The notches of 1 mm height are detected by shear waves - reflected beam at noise level. The application of the two-frequency or two-mode methods allows it detection at signal/noise ratio exceeding 6 dB. At the weld samples of such thickness with embedded defects one can reliable detect inclusions of 2 - 3 mm diameter and more than 5 mm length, lack of fusion of 2 - 3 mm height and length more than 10 mm, weld side lack of fusion of 2 mm height and more than 5 mm length.

Fig 14: Holographic image of notches and side drilled hole (z=20mm)in austenitic weld (see Fig. 1). Two-frequency method (1.65MHz+2.5MHz), longitudinal wave,a=60º.	Fig 15: Holographic image of notches and side drilled hole (z=20mm) in austenitic weld (see Fig. 1). Two-mode method, longitudinal +shear waves,a=60º.
Fig 16: Holographic image of two lack of fusion in the 40mm thickness austenitic weld ( 50mm and 15mm length , Fig.2). Two-mode method, a=60º.	Fig 17: Holography image of two lack of fusion on the boundary of the austenitic weld (50mm and 10mm length, Fig.3). Two-mode method, a=60º.


One should note that resolution of holographic method when testing by longitudinal waves is about 2 times worse than one by shear waves (2 - 2,5 mm comparing 1 - 1,5 mm). To confirm it, one should compare images of notches measured by different wave type. At the notch images measured by shear waves, in differ from longitudinal waves, the upper lighting point corresponding to the upper edge of the notch doesn't merge with down one corresponding to the down notch edge (see Fig.11and Fig.12). Hereupon, there is the possibility of exact measurements of defect depth. Moreover, the diffracted signals are presented in reflected beam data. This important information is used in data interpretation.
    The comparison of results of measurements of signal/noise ratio for both shear and longitudinal waves confirms that the plate type defects are much better detected using shear waves than longitudinal ones. On the contrary, the volume type defects are better detected by longitudinal waves than shear ones. In this connection it is reasonable to test using simultaneously two types of waves: longitudinal and shear (two-mode testing). To realize the two-mode testing the combination of images corresponding to the different wave type is executed according to the above described algorithm (1). The two-mode testing must increase the signal/noise ratio due to both difference in spatial frequencies of two types of waves and the difference in nature of dispersion of elastic waves of different type.

CONCLUSIONS

It is possible to measure the real sizes of defects in three coordinate planes by mean of Augur 4.2 system. Experimentally determined measurement inaccuracy of defect depth does not exceed 2-3 mm at longitudinal waves, and 1,5-2 mm at transverse waves.
    Basing on the analysis of detection possibility of artificial and real defects one can make the conclusion that the NDT system Augur can detect in austenitic welds of 40-60 mm thickness real defects if its length is more than 5 mm and size in depth is more than 2 mm. The optimal angle of beam incidence was 60^o and work frequency - 2,5 MHz.
Thereby, the holographic method in combination with two-frequency or with two-mode methods allows to conduct noise independent inspection of the austenitic welded joints with sensitivity and resolution in several times exceeding values, reached when use conventional (one-frequency and non-coherent) methods.

ACKNOWLEDGEMENTS

The authors wish to acknowledge Dr. V. Krylov for providing austenitic specimens.

References

1. Handbook on ultrasonic examination of austenitic welds. Doc. IIS-IIW-836-85.
2. Testing of austenitic steel welded joints, PNAE G-7-032-91, Moscow, CNIIatominform, 1992 (in Russian).
3. V.V. Grebennikov, A. K. Gurvich, M. V. Grigoriev, Multifrequency method of ultrasonic testing of austenitic welds, Defectoskopiya, 1974, No1, p. 81-89 (in Russian).
4. V.V. Grebennikov, A. K. Gurvich, State of art in ultrasonic testing of austenitic welds, Defectoskopiya, 1985,No 9, p. 3-12 (in Russian).
5. V. G. Badalyan, A. K. Vopilkin and others, Computer based system for ultrasonic testing with coherent data treatment, Defectoskopiya, 1993, No 1, p.3-15 (in Russian).
6. V. G. Badalyan, E.G. Basulin, Digital reconstruction of images of reflectors using projection in spectral space technique. Soviet Acoustical Journal, vol. XXXIV, 2, 1988 (in Russian).

© AIPnD , created by NDT.net

|Home|    |Top|