NDTnet - January 1997, Vol.2 No.01
Evaluation of ultrasonic indications by using PC-based synthetic aperture focussing technique (PCSAFT) By J. Pitkänen, P. Kauppinen, H. Jeskanen & V. Schmitz* Technical Research Centre of Finland, Finland *Fraunhofer Institute for Nondestructive testing (IzfP), Germany
- Selection of probe step distance
- Coupling effect
- Signal amplification
- Use of SAFT in Finnish VVER440 reactors
- Ultrasonic inspection of thin walled piping
- Future developments
One of the classical tasks of ultrasonic NDT is the detection, sizing and characterization of material damages like cracks in welds, lack of fusion, foreign material or delaminations. To increase the reliability and with respect to quality assurance, automatic scanning systems are used together with a u-processor controlled high frequency data acquisition system for the complete high-frequency signal. The applied imaging scheme is based on the backpropagation of the elastic waves which have been received by the ultrasonic probe back into the component. The application of the implemented method: Synthetic Aperture Focussing Technique (SAFT) covers a wide range field of testing ferritic and austenitic material and is used for testing pipes, turbines, plates, vessels or pump housings.
The original ultrasonic application of SAFT followed from the radar experience in early 70'ies and the first digital implementation of one-dimensional SAFT was demonstrated in 1976.
During data acquisition an aperture is scanned along a line. For simplicity we assume that the same probe transmits and receives the ultrasonic pulses. At all probe positions x the complete time-dependent amplitudes (A-scans) will be stored. Calling the reconstruction module, at each probe position, the complete A-scans are loaded into the PC memory. All amplitudes of the digitized A-scan
are backprojected into the material. Amplitudes which are received due to scattering on real defects are summed up in phase and amplitudes which are due to stochastic noise will be averaged.
To be able to perform such a process, the coordinates of the probe positions have to be known. Therefore the hardware consists of
an ultrasonic equipment (USD10, Krautkraemer)
- Personal computer (PC486)
- Fast FIFO-DMA-lnterface (by IzflP)
- Manipulator and control unit with coordinate interface.
As a result of the imaging process a side-view image is obtained with the following features:
correction of the depth dependent beam width
- constant horizontal resolution (approximate 1/3 probe diameter)
- good axial resolution (according to bandwidth of the probe)
- high detectability due to data acquisition including noise level
- good coupling because preferably small sized probes should be used
- signal-to-noise ratio (SNR) improvement due to the automatic time shift corrected summation.
The effect of the synthetic aperture is shown in the figure 1.
|In automatic ultrasonic testing, the probe is moved by a manipulator along a distance x and the amplitudes as a function of the time-of-flight (TOF) are stored. Point like defects form hyperbolic shaped curves, the position of the segment of the hyperbolic function depends upon the selection of the insonification angle, the length of the segment upon the beam opening angle. The following example shows horizontally the movement of the probe and vertically time-of-flight in its, figure 2a. The length of these TOF curves correspond to the beam width plus the size of the defect. SAFT removes the beam divergence and allows to image the defect close the true size. The drastical reduction of the dynamic curve - Fig. 2a - to the extension of defects Fig. 2b - is clearly seen.
The axial resolution of SAFT is equal to the pulse length and not related to the aperture. The pulse length is a function of the broadband characteristic of the system. Probes with 80% bandwidth allow to generate image spots in beam propagation direction with a length of nearly 3/2 of a wavelength.
The lateral resolution of SAFT for point like probes and in pulse echo mode is equal to 1/2 of the wavelength. If the probe is larger than 1/2 of the used wavelength, then the lateral resolution is equal to k x d with d equal to the effective probe diameter and k a factor which lies between 0 and 1. A practical value for k is 1/3.
4 Selection of probe step distance
This parameter influences directly the SNR and herewith how "clear" the image of a defect is displayed above the noise. The parameter does not influence directly the resolution of the image.
Synthetic Aperture means, that if the sound field transmitted by a probe passes a defect, this defect has to be insonified as often as possible. If the defect is only one time within the sound field the effect of SAFT is zero. A probe step size of some 1/1000 of a wavelength does not make a sense, because the information between two probe positions is - from the ultrasonic point of view - not independent. On the other hand small probe step distances increase the amount of data which have to be stored. Therefore one has to choose a compromise. To start with, it is recommended to select a probe step distance of 0.3 mm if the frequency range lies between I MHz and 5 MHz (shear or long. waves). In general, a probe step size should lye around 1/5 of the effective crystal diameter of the probe.
Sizing of the crack is independent upon the selected probe step distances. The SNR for all three selected probe step distances is always high enough; the maximal reconstructed values vary between 7200 and 2330. This is due to the fact, that the flaw is 70 mm below the surface and the beam opening angle allows to get enough sample points at the surface. Another advantage of using Synthetic Aperture Focussing Technique is the fact, that the accuracy of an image does not rely on exact positioning of the probe. This means that it is possible to vary probe step sizes without the need that one of the probe positions gives exactly the maximum amplitude. In fig. 3 a crack has been imaged which lies 70 mm below the surface and has a depth extension of 40 mm. A contact technique probe of 2 MHz with 45° insonification angle has been used. From left to right, probe step sizes have been selected to be 0.3 mm, 0.6 mm and 1.2 mm. Obviously the SAFT-image does not change drastically and the depth evaluation in all three images gives a value of 40 mm. With increasing probe step size fewer A-scans contribute to the image and the dynamic range of the image will be reduced. This is indicated by the figures in the lower right corner of each picture (figures 3 a,b,c) demonstrating that by increasing the probe step size by a factor of 4 results in a reduction of the highest reconstructed value by a factor 4.
Figure 3: Saft -image from a crack
5 Coupling effect
In conventional techniques using the DGS-diagram, amplitude variations influence immediately the equivalent flaw size: e.g. a drop of 6 dB would reduce an equivalent flaw size of 6 mm down to 4 mm. Therefore possible variations in coupling directly affect the reliability of inspection. SAFT relies less on amplitude information but more on time-of-flight information. The image spots are formed by summing up information from different probe positions. As long as there are many probe positions the image intensity does not change much if there is loss of coupling during some probe positions. Loss of amplitude influences only the SNR in the image.
6 Signal amplification
The data acquisition in SAFT is based upon the complete high-frequency ultrasonic signal, without time gate or amplitude threshold. Due to the time-corrected signal-to noise averaging capability of SAFT, signals which are buried within the noise can rise to amplitudes above the noise. Therefore it is important to digitize the ultrasonic signals down into the noise. On the other hand the highest amplitudes should not exceed the maximum amplitude which is allowed by the transient recorder. Today 8 bit transient recorders are used which correspond to 40 dB dynamic range. This is not sufficient for detecting and sizing very small defects together with specular reflecting (corner echo, back wall, root of a weld) indications. Therefore it may be necessary to repeat the data acquisition at a 20 dB higher or lower amplification. In the future, this could be avoided by using a transient recorder with a higher dynamic range than 40 dB.
7 Use of SAFT in Finnish VVER440 reactors
The Technical Research Centre of Finland (VTT) has used PC-SAFT in the inservice inspection of primary components in Finnish PWR type VVER440 (Loviisa 1 & 2). The indications detected in manual ultrasonic inspections exceeding predefined amplitude-level, were analysed with the SAFT-method. This was simply realized by connecting the cable from the ultrasonic equipment to the PC-SAFT -equipment. By this way analysis of the defect could be performed directly after detection. From every indication several reconstructions at different probe positions along the circumference were computed.
One example of application is the inspection of welds in main gate valves. The scanning manipulator was placed on the nozzle between the main cooling pipe with a diameter 500 mm and valve. The geometry didn't allow direct insonification in inspection, but full skip (reflection from the backwall) had to
be used. Therefore the sound path was more than 300 mm and because of strong attenuation of the ultrasonic pulse in the austenitic forging, the SNR was not sufficient to make a satisfactory evaluation of the indication. This was verified in laboratory tests using a full-scale mock up (calibration block of the valve), figure 4.
|Figure 4: Valve geometry and ultrasonic sound path|
As an example a reconstruction of a defect, 20 mm high in through-wall direction is shown in figure 5a. It is difficult to detect this defect located behind the weld because the defect can't be seen in the normal A-scan due to the low SNR. With help of PC-SAFT algorithm SNR was improved remarkably which allowed to locate the defect from the noise. The original reconstructions are shown in false colours and in the black and white presentation in figure 5a the defect is marked with a white rectangular.
Figure 5b is showing real defects detected in the valve. Based on the normal procedure used to analyse indications of mechanized inspection the depth of defect was estimated to be more than 40% of wall thickness. The PC-SAFT reconstruction shows however that the indication is formed from two separate single indications, figure 5b. According to ASME XI Code (Fig. IWA-3300-1) each single defect can be evaluated separately.
|Figure 5a: Reconstruction from a |
reference reflector behind the weld
|Figure 5b: Reconstruction of real|
defects in the weld
8 Ultrasonic inspection of thin walled piping
In thin walled pipes (thickness between 10 and 30 mm) the inspection of areas near to the scanning surface are problematic. With help of a special composite probe, made by IzfP sizing with SAFT is possible. As a reference results from an austenitic calibration block are shown (material AISI 304). In the block there are 1.5 mm side drilled holes. The thickness of the calibration block was 25 mm. Measurement with a 4 MHz composite probe gave good results for sizing throughout the thickness, see figure 6. Additionally the applicability of SAFT-technique to size cracks was tested using a reference block containing artificial IGSCCs. The cracks were analysed with SAFT using the same kind of composite probe and we can see clearly corner reflection in the HAZ and very small indications, which can be caused by tip-diffracted signals. In the probe echoes are received not only from the material under inspection but from the wedge of the probe itself which was the reason that some areas couldn't be evaluated, see figure 7.
Demonstration of sizing capability|
with side-drilled holes
|Figure 7: SAFT -reconstruction|
from a IGSCC
9 Future developments
Based upon the concept of inverse scattering theory it is shown that SAFT only performs a perfect reconstruction, if a couple of assumptions are fulfilled. Besides the scalarisation, linearization, bandwidth and aperture restrictions, the real world situation is three-dimensional. The applied algorithms - in most of
the cases - are two-dimensional and allow to image slices like B-scans or C-scans. Research work is currently under progress to expand the capability to reconstruct three-dimensional volumes and to extract from these volumes those slices which carry the most important information.
The application of PC-SAFT-technique in the inservice inspections of Finnish nuclear power plants is mainly focussed on evaluation of indications detected in ASME-hased manual inspections. Originally the technique could only be used to components having relatively high wall thicknesses but with special transducers SAFT is also applicable to thinner pipes.
International Conference "Computer Methods and Inverse Problems in
Nondestructive Testing and Diagnostics", 21-24 November 1995, Minsk, Belarus
Contact for the proceedings: DGZfP
Deutsche Gesellschaft für Zerstörungsfreie Prüfung e.V.
Motardstraße 54 , 13629 Berlin, Germany
Telephone: + 49 (030) 386 29 911, Fax:.. /29 918,
© Copyright 1. Jan 1997 Rolf Diederichs, firstname.lastname@example.org
/DB:Article /SO:DGZfP /AU:Pitkaenen_J /AU:Kauppinen_P_ /AU:Jeskanen_H /AU:Schmitz_V /CN:DE /CT:UT /CT:SAFT /CT:instrument /ED:1997-01