Developments in Ultrasonic Phased Array Inspection III

Ultrasonic Phased Array Inspection of Welded Pipes Using Wave Mode-Converted at the Inner Surface of the Pipe

R. Long, P. Cawley, Imperial College, UK; J. Russell, Rolls-Royce, UK

ABSTRACT

The ultrasonic inspection of welded pipework found in the nuclear industry often proves challenging. The inability to access both sides of the weld might warrant the use of waves that are mode converted at the inner surface of the pipe to propagate a wave at the desired angle which would maximize a specular reflection from a defect with a postulated inclination. Even where access to both sides of a weld is available, inspections using mode converted waves might be preferred rather than using waves that propagate directly through the weld since the structural complexity of austenitic stainless steel welds can disturb the ultrasonic wave propagation. As most phased array controllers and stand alone software do not readily provide for focused mode conversion inspections a solution was developed to use Full Matrix Capture (FMC) and process the data ourselves. Inspections were modeled using the CIVA software and compared to experimental results conducted on a welded test piece.

INTRODUCTION

Rolls-Royce and Imperial College, London are working together to develop a phased array inspection of a welded pipe with a completely undressed weld cap [1,2]. The target application is a section of large bore stainless steel pipe-work with a wall thickness of greater than 50mm and contains an austenitic weld. Defects can occur anywhere within the weld and Heat Affected Zone (HAZ) of the pipe-work. Any inspection should be capable of finding flaws with a through-wall dimension of 10% of the pipe wall thickness. For defects occurring on the lower or upper fusion face, ideally we would like to use waves that are reflected and mode converted at the inner surface of the pipe. However, most commercial phased array controller software does not currently provide for this. In addition, we would like to superimpose the B-scan images over the region of the component that is being inspected to aid defect location. For focused on-the-skip and mode conversion inspections, no known currently available phased array controller software allows this. In order to conduct such inspections and display results in a manner that we would like, our solution was to use Full Matrix Capture (FMC) [3] and process the data ourselves. FMC is a data acquisition technique which involves the collection of the complete set of time-domain data (A-scans) for all combinations of transmit and receive elements.

Comformable phased array Membrane device

The application of conventional phased array inspection is usually performed using a solid wedge placed between the transducer and component under test to refract the ultrasonic beam into the test piece. The beam is steered and focused to a chosen location by applying delay laws to the individual transducer elements. For inspections above an undressed weld cap there will be a mismatch between the profile of the surface under test and the plane base of the wedge. This will produce an irregular coupling layer leading to beam distortions and a loss of sensitivity which contribute to reduce the inspection performance. A cost effective solution that we have chosen is to couple a standard phased array to the surface under test via a water path which is made more convenient by encapsulating the fluid with a conformable synthetic rubber membrane [4, 5]. A single device is used for surface profile measurement (required for updating of delay laws) and inspection, allowing rapid scanning of components with irregular surfaces without the need for multiple angled probes and time consuming mechanical scanning. Figure 1 shows a photograph of the 3rd generation membrane device that incorporates a standard linear 2 MHz, 128 element, 0.75 mm pitch phased array probe from Imasonic France [6]. For the through weld inspections, the device housing is designed such that the phased array

is angled at 18 and the fluid stand-off of the first element to the surface under test is set at 20mm. For mode conversion inspections the phased array was angled at 7 and the fluid stand-off set at 7mm. These parameters have been chosen to minimise the possibility of an internal reflection appearing in the Bscan images [7]. The design allows a constant pressure configuration using a header tank of water to provide adequate pressure for the membrane to conform over the irregular surfaces of interest, such as weld caps. The membrane material used is a low loss castable polyurethane rubber, with an acoustic impedance similar to water, which has been developed with the help of Rolls-Royce, Derby, UK. The device housing has been designed to allow the membrane to be readily changed should this prove necessary. An integrated irrigation system pumps water from an isolated reservoir through eight narrow bore tubes that direct water, which acts as an acoustic couplant, onto the surface under test. When testing above an irregular surface profile the use of delay laws computed for a plane surface may lead to beam splitting and loss of the original focal point [4]. The inspection performance can be recovered with the application of updated delay laws requiring knowledge of the surface profile under test. A convenient method is to utilise the phased array incorporated in the membrane device to scan and to measure the surface profile at each test location [1].

Irrigation feed Couplant distribution

Low profile Imasonic2MHz 128 element phased array

Header tank for constant pressure encapsulated water column

Figure 1 - Photograph of 3rd generation conformable phased array membrane device mounted on welded test piece seen in constant pressure configuration and featuring couplant

distribution.

Upper Fusion face

Defect 2

Appro

x 60

mm

Defect 1 =135mmDefect 2 =185mm

7mm

Immersion Phased Array80 el - 2MHz- angle 8o

Lower Fusion face

Defect 1

Figure 2 - Set up for obtaining simulated and experimental data showing phased array arrangement and stainless steel block where each test piece comprises one of the defects

shown.

Modelling of test scenario using CIVA software

The CIVA software [8] was used to obtain simulations for phased array inspection of defects on the weld fusion face. A single FMC data set was obtained for inspection of each defect, which when processed using software developed by Imperial College, provided inspection possibilities using direct, on-the-skip or mode converted waves for various incident angles and aperture sizes. This is one of the advantages of obtaining FMC data over conventional data in that a multitude of inspections scenarios can be obtained from one data set [9]. The intention of the exercise was to model proposed primary inspection techniques for each defect, which are normally based on a specular reflection that provides the greatest amplitude. The FMC data also allowed processing to find possible secondary inspection techniques to aid confidence that a given defect has been identified and may additionally provide for defect sizing without the need for additional probes or scanning.

Figure 2 shows a schematic of the pipe wall with the two defects that will be considered for this paper. Defect 1 and Defect 2 represent an 8mm through wall lack of fusion defect on the lower 25 degree and upper 10 degree weld fusion face respectively. For these simulations the phased array was modelled as an 80 element immersion array with a 1.25mm pitch angled at 7 and with a 7mm stand-off to the surface under test. The phased array is positioned such that the inspection of a given defect does not require that waves first propagate through the weld material.

The simulated results for inspection of Defect 1 are shown in Figure 3 with amplitude displayed on a logarithmic scale. A single simulated FMC data set was obtained and processed with beam forming optimised for inspections that use focused direct longitudinal (L), direct shear (T) waves, mode converted shear into longitudinal waves (TL) and on-the-skip shear waves (TT). The B-scan shown in Figure 3a displays the crack tip diffraction resulting from the L wave along with later arrivals attributed to TL and T wave interactions, whose images also fall on the lower weld fusion face. Inspecting with a direct longitudinal wave could result in false calls for additional defects. Even so, if well understood, techniques such as this can provide defect confirmation or crack sizing if the signal is sufficiently high amplitude, and has high enough signal to noise to provide confidence that any signals are related to the defect. Similarly the T wave inspection (Figure 3b) displays additional features in the B-scan. The primary inspection technique for this defect uses TL wave inspection. Investigations suggest that a technique using TT wave approach would be a possible secondary technique for crack sizing and defect confirmation.

The simulated results for inspection of Defect 2 are shown in Figure 4. Beam forming was optimised for inspections that use focused L, T, mode converted shear into longitudinal waves, with direct longitudinal wave reflected off the defect (TLL) and TT wave inspections. The technique using TLL wave inspection (Figure 4c) is the preferred primary inspection technique for this defect. Applying similar reasoning as described earlier, the technique that uses TT wave inspection would be a possible secondary technique for crack sizing and defect confirmation.

Noise was not included in the CIVA simulations, so whether the techniques investigated could provide adequate defect responses in practice had to be found by experimental results.

Direct T45 deg 30el

Mod con TL29 deg 25el

On-skip TT40 deg 30el

(a)

(b)

(c)

(d)

Direct L45 deg 30el L

T

TLL

TT

TLL

TLT

TL

TL

Figure 3 - Simulated results for inspection of Defect 1 with delay laws optimised for (a) direct longitudinal waves, (b) direct shear waves, (c) mode converted shear into longitudinal waves and (d) on-the-skip shear waves. Results normalised on a logarithmic scale with background set to 20dB below the maximum amplitude in Bscan image.

Direct L73 deg 14el

Mod con TL 29 deg 24el

On-skip TT 45 deg 24el

Mod con TLL29 deg 24el

(a)

(b)

(c)

(d)

L

TL

TLL

TT

TLL

Figure 4 - Simulated results for inspection of Defect 2 with delay laws optimised for (a) direct longitudinal waves, (b) mode converted shear into longitudinal waves, (c) mode converted shear into longitudinal wave with direct longitudinal wave reflected off defect and (d) on-the-skip shear waves. Results normalised on a logarithmic scale with background set to 20dB below the maximum amplitude in Bscan image.

EXPERIMENTAL DATA

All experimental testing has been completed on flat plate welded testpieces with the weld cap left undressed to replicate the target application. Each test piece was manufactured by Sonaspection UK [10] from two flat stainless steel plates of greater than 50mm thickness, with lower and upper weld fusion faces of 25 and 10 degrees respectively. An Electric Boat (EB) insert was used in the base of the weld giving a more uniform, but not flat, root. The first 6mm of the weld was laid using Manual Metal Arc (MMA) welding and the rest by Tungsten Inert Gas (TIG). Surface breaking defects were machined using Electro-Discharge Machining (EDM) with a gape of 0.3mm. Lack of Side Wall Fusion defects were embedded using an in-house Sonaspection technique. The defects reported in this paper are Defect 1 and Defect 2 (both 8mm through the wall dimension) shown in Figure 2, which lie on the lower and upper fusion faces respectively. The 3rd generation membrane device was used for the inspections that incorporated a standard linear 128 element, 2 MHz, 0.75mm pitch phased array probe from Imasonic, France [5]. The phased array is angled in the device housing at 7 and the height of the first element above a plane surface is 7mm. A Peak NDT phased array controller [11] was used to obtain experimental FMC data. The application of fluid coupling via the irrigation holes in the membrane device base plate was utilised for these results. A single FMC data set was obtained for each inspection, which was processed and displayed using the Imperial College software. The sampling frequency used was 25MHz. All Bscans were normalised to the maximum amplitude in the image and are shown on a logarithmic scale.

The experimental results for inspection of Defect 1 are shown in Figure 5. The weld fusion face is drawn in for reference only and should not be taken to be an accurate representation. The background was set at 15dB down from the maximum amplitude seen in the Bscan image. When using the primary TL inspection technique, shown in Figure 5a, a strong response is observable in the image suggesting the presence of a defect located on the lower fusion face. Figure 5b shows the confirmation of defect presence by the secondary TT technique, which also provides for defect sizing using the separation distance of the signals diffracted from the crack tips. The background was set at 9dB down from the maximum amplitude seen in the Bscan image. The amplitude of the defect response for the secondary technique is 6dB down from the primary technique, resulting in a lower signal to noise ratio.

The experimental results for Defect 2 are shown in Figure 6. The backgrounds for the primary and secondary inspection techniques were set at 15dB and 9dB respectively. The presence of a defect lying on the upper fusion face is identified using the primary TLL technique and is confirmed by the secondary TT technique. Figure 6b shows the confirmation of defect presence by the secondary TT technique, which also provides for defect sizing using the separation distance of the signals diffracted from the crack tips. The amplitude of the defect response for the secondary technique is 8dB down from primary technique.

Defect Identification

Fluid

Steel

(a)

Defect confirmation

Fluid

Steel Weld fusion face profile

Phased Array

0dB 9dB 0dB

(b)

15dB

Weld fusion face profile

Phased Array

Figure 5 - Experimental results for inspection of Defect 1. (a) Primary TL inspection using 29 degree incident shear wave and 32 element aperture (b) secondary TT inspection using 44 degree incident shear wave and 50 element aperture. Results normalised on a logarithmic scale to maximum the amplitude in Bscan image.

Defect Identification

Fluid

Steel

(a)

Defect confirmation

Fluid

Steel Weld fusion face profile

Phased Array

0dB 9dB 0dB

(b)

15dB

Weld fusion face profile

Phased Array

Figure 6 - Experimental results for inspection of Defect 2. (a) Primary TLL inspection using 29 degree incident shear wave and 30 element aperture (b) secondary TT inspection using 40 degree incident shear wave and 42 element aperture. Results normalised on a logarithmic scale to maximum the amplitude in Bscan image.

Specular Reflection off Defect 10

Phased Array

Fluid

Steel

Dominant featuredue to irregular surface

(a)

Weld fusion face profile

Specular Reflection off Defect 10

Fluid

Steel

Weld fusion face profile

Phased Array

12dB 0dB 12dB 0dB

(b)

Figure 7 - Experimental results using 3rd generation membrane device on welded test piece for the negative direction inspection of Defect 1 using 65 degree direct longitudinal wave primary inspection technique and 40 element aperture. Bscan images shown for a) plane surface delay laws and b) delay laws updated for approximate weld cap profile. Results normalised on a logarithmic scale to maximum the amplitude in Bscan image.

Some experimental results are shown in Figure 7 for the inspection of Defect 1 using waves that are required to propagate through the weld material on their way to the defect. The 3rd generation membrane device was used for obtaining the results incorporating a 128 element, 2MHz, 0.75mm pitch phased array from Imasonic France [6]. The phased array was angled at 18 and the fluid stand-off of the first element to the surface under test is set at 20mm. A Peak NDT phased array controller [11] was used to obtain experimental FMC data. The FMC data was processed for beam forming using a 65 direct longitudinal wave which provides a specular reflection off a defect lying on the lower fusion face.

Figure 7a shows experimental results for the through weld inspection where plane surface delay laws are used for beam forming. The background was set at 12dB down from the maximum amplitude seen in the Bscan image. The Bscan displays a feature, seen on the right hand side, whose amplitude dominates the reflection off the defect. This feature is due to some of the waves propagating through the weld cap profile for apertures on the far right of the phased array. Figure 7b shows the experimental results where the delay laws have been updated for the surface profile under test measured using the same phased array incorporated in the membrane device. The effect is to remove the dominant feature from the Bscan image such that now the reflection off the defect can be readily identified. The SNR for the reflection relative to the background noise is 7dB which is half that achieved in the results when using the mode conversion technique shown in Figure 5a. The reason is that the wave propagation through the structural complexity of austenitic stainless steel welds is disturbed and produces back scattering. In addition the location of the defect from the position of the reflection becomes uncertain. Rolls-Royce and Imperial College, London are working together to develop a solution to this problem [12].

Experimental results confirmed that the use of primary and secondary inspection techniques can be useful in identifying and confirming the presence of a defect. The more differing techniques that can be used on any one inspection the greater the confidence when calling the presence of a defect. An advantage of obtaining FMC data over conventional data is that a multitude of inspection scenarios can be obtained from one data set. Results show that focused waves that are reflected and mode converted at the inner surface of the pipe have often proved most appropriate for these inspections, though most commercial phased array controllers do not currently provide for this.

CONCLUSION

Focused waves that are reflected and mode converted at the inner surface of the pipe often prove most appropriate for inspection of defects that might lie on the weld fusion face of welded pipe work. The results for inspections using focused mode converted waves have shown significant improvement of

SNR ratio over waves that are required to directly propagate though the weld material. It is hoped that in time commercial phased array controller beam forming and results presentation software will readily provide for mode conversion inspections. For our research UT phased array inspection data was acquired as FMC which allowed us control over data processing and display of results. A single FMC data set provided a multitude of inspection possibilities from which inspections were chosen which displayed the best defect response. When inspecting critical components, an advantage of FMC data over conventional methods is that the complete data set can be archived for future processing when advanced algorithms might allow improved identification of defects.

REFERENCES

1) R. Long and P. Cawley, Phased array inspection of irregular surfaces, in Review of Progress in Quantitative Nondestructive Evaluation 25, edited D. O. Thompson and D. E. Chimenti, AIP Conference Proceedings vol. 615, American Institute of Physics, Melville, NY, 2006, pp. 814-821.

2) Long, R., Russell, J., Cawley, P., Habgood, N., Non-Destructive Inspection of Components with Irregular Surfaces using a Conformable Ultrasonic Phased Array, Proc. 6th Intl. Conf, on NDE in relation to Structural Integrity for Nuclear and Pressurised Components, Budapest, Oct 2007.

3) C. Holmes, B. Drinkwater and P. Wilcox, Post-processing of the full matrix of ultrasonic transmitreceive array data for non-destructive evaluation, NDT & E International, 38, 2005, pp. 701-711.

4) R. Long and P. Cawley, Further Development of a Conformable Phased Array Device for Inspection Over Irregular Surfaces, in Review of Progress in QNDE, Vol. 24, op. cit. (2007), pp. 754-761.

5) J. Russell, R. Long, P. Cawley and N. Habgood, Inspection of Components with Irregular Surfaces using a Conformable Ultrasonic Phased Array, 2009, in Review of Progress in QNDE, in press

6) Imasonic, 15 rue Alain Savary, 25000, Besanon, France. www.imasonic.com 7) Russell, J., Long, R., Cawley, P., Habgood, N, Inspection of Components with Irregular

Surfaces using a Conformable Ultrasonic Phased Array, in Review of Progress in QNDE, 28, op. cit. (2009), In Press

8) P. Calmon, E. Iakovleva, A. Fidahoussen, G. Ribay and S. Chatillon, Model based reconstruction of UT array data", Review of Progress in QNDE, ed. by D. O. Thompson and D. E. Chimenti 27, (AIP Conference Proceedings 975, Melleville, 2008), pp. 699-706.

9) R. Long and P. Cawley, Ultrasonic Phased Array Inspection of Flaws on Weld Fusion Faces using Full Matrix Capture, 2009, in Review of Progress in QNDE, in press.

10) Sonaspection, Lancaster, UK. www.sonaspection.co.uk 11) Peak NDT, Derby, UK. www.peakndt.com 12) G. Connolly, M. J. S. Lowe, and A. Temple, Simulation and Modelling of Ultrasonic Wave

Propagation in Austenitic Steel Welds, 2009, in Review of Progress in QNDE, in press.