Back to Basics

S-scan Coverage with Phased Arrays

by Ed Ginzel* and Michael Moles^†

Here is an excellent article that effectively makes the case for phased array ultrasound for weld testing. It should become required reading in courses that cover phased array ultrasonic testing. We note again the need to include the code authorities in the loop. Roderick K. Stanley Associate Technical Editor

 

This paper describes ray tracing to show coverage using phased array S-scans for encoded linear scans. A variety of weld configurations were modeled, from simple butt welds to double V and complex T-welds, including a variety of thicknesses. For simple butt welds below 10 mm (0.4 in.), a single S-scan provided coverage; however, thicker walls and more complex structures require multiple S-scans. No definitive rules are possible, as welds vary so much in configuration, though some guidelines are given.

INTRODUCTION

Construction welds in pressure vessels and other components typically require testing to "guarantee" structural integrity. Normally, construction welds are tested to code; however, codes are, by necessity, well behind technology. While radiography has been the historical testing method, ultrasonic testing has become more common in recent decades. In the last few years, phased arrays have become much more prevalent.

Ultimately, modeling shows that a simple guideline is possible for
whether one or more S-scans are required.

Ultrasonic phased arrays are a recent technology that can perform S-scans, raster-type electronic scans, time of flight diffraction, and tandem and manual tests. S-scans are a new technology, with their own learning curve. ASME defines an S-scan as follows (ASME, 2006):

An S-scan (also called a Sector, Sectorial, or Azimuthal scan) may refer to either the beam movement or the data display.

1. When used to refer to the beam movement, it refers to the set of focal laws that provides a fan-like series of beams through a defined range of angles using the same set of elements.
2. As a data display, it is a two-
   dimensional view of all A-scans from a specific set of elements corrected for delay and refracted angle. Volume-corrected S-scan images typically show a pie-shaped display with defects located at their geometrically correct and measurable positions.
   

Phased arrays can be used in two basic modes: manual and encoded. This paper refers to encoded scanning only, specifically encoded linear scanning for welds. Linear scanning, called one-line scanning in some industries, is a way of testing the weld in a single pass, with the scan running parallel to the weld at a fixed distance from the weld axis. Here, the position of the arrays relative to the weld and the beam coverage are critical.

One issue that is frequently discussed concerns the coverage possible with S-scans, and whether one S-scan with linear encoded scanning is adequate for thin-walled components. This paper determines the limitations of S-scans for covering welds of different profiles and thicknesses using modeling. Coverage has already been addressed in Section V of the ASME Boiler and Pressure Vessel Code (2007), which requires a scan plan to show coverage. Actual coverage will depend on array position, angular range, thickness and so on. Ultimately, modeling shows that a simple guideline is possible regardless of whether one or more S-scans are required.

PHASED ARRAYS

Ultrasonic phased arrays are similar to sonar, radar and other wave physics array technologies. Briefly, a series of active elements are pulsed, time-delayed, received, converted from analog to digital format and summed independently. The waves interact in constructive and destructive interference to produce wavefronts, which act very similarly to conventional ultrasonics. The arrays can be made in many different shapes, sizes and frequencies, and the elements are all ultrasonically insulated from each other. Phased arrays are described in depth elsewhere (R/D Tech, 2004).

Besides being able to generate unusual scan patterns, phased arrays have significant advantages over conventional monocrystal ultrasonics:

• Speed - for welds, corrosion and other components, linear scanning can increase scanning speed significantly (and hence, reduce costs).
• Imaging - S-scans, E-scans and other 2D and 3D imaging can give much better and more interpretable discontinuity assessments.
• Flexibility - phased arrays can perform a wide variety of scans for different types of discontinuities on a number of different components.
• Data storage - full or even partial data storage and display allows better discontinuity interpretation and can be used for archival purposes.
• Reproducibility - though not demonstrated yet by third party trials, using the same set-up and procedure with phased arrays gives much more reproducible results than with manual ultrasonics.

In addition to advanced generation/reception, true depth or volume-corrected S-scans are popular due to their apparent coverage and imaging. Figure 1 shows an example of an S-scan with easy discontinuity location analysis. Discontinuities located on the B0 (half skip) horizontal line are on the inner diameter, and those on the T1 line (full skip) are on the outer diameter. Signals in between B0 and T1 are midwall. While S-scans, as Figure 1 shows, give apparent coverage, it is not fully clear that all sections of the weld are actually tested, or how many S-scans are required for full coverage. This is the purpose of this paper.

Figure 1 — S-scan of a weld with inner and outer diameter discontinuities located.

S-SCAN COVERAGES AND BEVEL INCIDENCE ANGLES

Coverage for manual testing using S-scans is not an issue, as procedures normally require the operator to raster scan over the full weld (this is the same as conventional manual testing). However, for encoded linear scans using S-scans, coverage is important. In order to get the benefits of faster scanning with smaller instruments, a limited number of S-scans is desired. Clearly, S-scans may cover the whole weld, if the wall is thin. For thicker walls, multiple scans are obviously needed: either a single pass with a suitably long array, or multiple passes.

The situation is complicated by the requirements to be close to the bevel incidence angle (ASME, 2007), or to use "appropriate" angles. How close is close enough? Theoretical work (Charlesworth and Temple, 1989) showed that an incidence within 5° of the bevel could be considered close enough, while at 10°, off-incidence angle signal amplitudes start to drop off significantly. Effectively, there is no consensus at the practical level.

In addition, monocrystal probes normally only come at fixed angles such as 45, 60 and 70°. If the designer (legitimately) mandated a weld bevel angle of 37.5°, it would never be possible to test within ±5° of bevel incidence angle. Overall, it appears that there is no uniformity of practice here.

Modeling of the bevel incidence angle was performed with a basic graphic design program that provides angular information on a simple butt weld. The results are shown in Figure 2. Depending on the bevel incidence angles required, either two S-scans (for ±10°) or three (for ±5°) will provide "appropriate" angles (ASME, 2007). This work did not define "coverage," particularly for thin-walled components.

Figure 2 — Model of bevel incidence angle: (a) ±5° bevel incidence angle coverage (requires three S-scans for coverage); (b) ±10° bevel incidence angle coverage (requires two S-scans for coverage); (c) ±10° coverage on thicker wall (still requires only two S-scans for coverage).

CODES

Inspection codes generally accept new technologies like phased arrays. This applies specifically to the three main North American codes: ASME (2007), API (1996) and AWS (2006). However, the techniques and technologies need approval. All these codes have methodologies for adopting new technologies and techniques.

The general conditions to be met for any weld examination in ASME include "an appropriate angle" and "full volume coverage" (ASME, 2007).

MODELING PROCEDURES

This work used simple computer modeling to determine coverage of various weld profiles and thicknesses to show whether full coverage is obtained or not. The software used does not calculate bevel incident angles commercially yet, but readily shows coverage using S-scans. The program is simple to use, and fulfills the ASME scan plan requirements. It is typical of the type of simple ray tracing programs that are, or should be, available on standard phased array instruments to evaluate ASME and other code scan plans.

Modeling was performed on three basic welds: butt welds, T-welds and K-welds. The assumption is that the T-welds and K-welds are reasonably thick, while butt welds can be any thickness. The array was moved to optimize the test, on the basis that all operators will generate a scan plan and follow it during testing.

The software uses standard commercial arrays and wedges, which can be selected as appropriate. These probes are then placed on the modeled component, including the weld profile, to view the beam patterns. Using S-scan parameters (for example, angular range), S-scan coverage can be immediately visualized. Moving the probe around or changing the setup parameters allows the operator to optimize the test, and print the results for a scan plan and auditable report. Multiple S-scans and multiple arrays can be modeled, as well as more complex components. The dotted lines beside the ray trace fan are the 6 dB drop-off points - that is, what is normally considered adequate coverage by code.

A prototype software version has bevel incidence angle calculated, as shown in Figure 3. Here, selected angles on a 25 mm (1 in.) pipe wall with a single S-scan are shown; interestingly, essentially full coverage is obtained, while bevel incident angles are only just above 10° at worst.

Figure 3 — Ray tracing using advanced software with sample bevel incidence angles (arrowed) calculated on 25 mm (1 in.) wall double-V weld.

MODELING RESULTS

Single V Weld

This weld configuration is typical of shielded metal arc welding prep: it has a 2 mm (0.08 in.) land, 60° included angle and a 2 mm (0.08 in.) gap. The probe/wedge combination for thin wall examples has been selected to minimize the dimension to the nose of the wedge (Figure 4a). This is necessary to ensure optimum approach for the root region. The modeling was performed on walls from 5 to 25 mm (0.2 to 1 in.). The 5 MHz array modeled used a 16 element, 10 mm (0.4 in.) aperture. Figure 4b shows the dimensions of the 5 MHz 64 element array used for thicker walls.

Figure 4 — Dimensions of: (a) probe/wedge used for thin wall examples; (b) larger 64 element array and wedge.

Thin wall sections will be problematic in that the root regions may not be accessible, even with a 70° refracted angle on the half skip. Coverage using a half skip may be possible with wedge modification. Figures 5a to 5d show a series of welds from 5 to 9 mm (0.2 to 0.35 in.) wall, all covered by a single S-scan. Figure 5a shows the coverage on 5 mm (0.2 in.) with a single skip. Depending on the geometry and parameters, it may be possible to obtain volume coverage using a 1.5 skip technique, as shown in Figure 5b.

Figure 5 — Modeling results with single S-scan on thin walled butt welds: (a) 5 mm (0.2 in.) wall using standard 45 to 70° refracted shear wave S-scan; (b) 5 mm (0.2 in.) wall using 1.5 skip 50 to 65° refracted shear wave S-scan; (c) 7 mm (0.28 in.) wall using standard 45 to 70° refracted shear wave S-scan; (d) 9 mm (0.35 in.) wall using standard 45 to 70° refracted shear wave S-scan; (e) 9 mm (0.35 in.) wall using standard 45 to 70° refracted shear wave S-scan with increased standoff.

As thickness increases the traditional approach using a 45 to 70° S-scan sweep is feasible, even with a relatively wide cap. In both cases (5 and 7 mm [0.2 and 0.28 in.]) the angles of approach for the root rely on corner reflections and the skip up to the upper weld bevel provides reasonable angle approach, albeit slightly more than 10° off the bevel angle at the upper extreme (Figure 5c). However, at the upper extreme (cap region) the corner effect will again ensure high probability of detection especially at the 45 to 50° refracted range.

Of course the gap is critical at this juncture. If the gap gets too wide, the 1.5 skip option would need to be used to ensure suitable root coverage. By the time the wall thickness reaches 9 mm (0.35 in.), the coverage of the root-region by a refracted 65 to 70° angle can be reasonably assured for most conditions of cap geometry (Figure 5d). However, at this point the cap region could suffer a reduced coverage in the heat affected zone of the probe side of the weld. If the gap is not excessive, this can be adapted to by a slight increase in standoff (Figure 5e).

The minimum angle could be extended to 40 from 45° and still achieve the volume coverage (if calibratable at this range). However, the assumption that the root is properly addressed assumes that the gap is not varying and the center of the weld cap can be used as a guide to adequately assure the probe position for alignment. Both of these assumptions cannot be guaranteed to be true.

In addition, this is a 30° weld bevel; other weld bevels and component geometries are possible, so the best solution is modeling on the specific weld bevel to produce a scan plan. These modeling results can only be regarded as a guideline.

At thicknesses over 9 mm (0.35 in.), single V bevels should be tested using a minimum of two S-scans with two different standoffs from each side of the weld. This may be achieved by two passes or, if the probe/wedge design is suitable, it may be possible to leave the probe in one position and generate two sets of S-scans per side. Results from 10 mm (0.4 in.) and up are shown in Figures 6a and 6b. The amount of standoff difference will depend on the configuration, and is typically 0.5 to 1.5 aperture dimension. The illustrations in Figure 6 use a 10 mm (0.4 in.) aperture (16 elements with 0.6 mm [0.02 in.] pitch).

Figure 6 — Modeling results for walls 10 mm (0.4 in.) and up: (a) 10 mm (0.4 in.) wall using standard 45 to 70° refracted shear wave S-scans with two standoffs (b) 25 mm (1 in.) wall using standard 45 to 70° refracted shear wave S-scan with two standoffs.

For thicker sections, the ratios remain the same; however, the probability exists that the volume of concern extends to greater dimensions due to the heat affected zone. Two standoff positions not only ensure required volume coverage, but also provide increased probability of detection by providing a second angle of approach on the bevel face.

These results were essentially established in earlier modeling and experimental work (Moles and Zhang, 2005). Here, the concern was more about the bevel incidence angle than coverage; however, the conclusions are similar. For thin walls, enough ultrasound was "showered" back to the probe for detection with a single pass. For thick walls, multiple S-scans were required for "appropriate" angles. Note that two smaller probes were used here, though a single larger array should be technically feasible.

One last point about the modeling images: while the software may not give direct measurements of the bevel incident angle yet, the images show clearly that in some cases the angles are not "appropriate." This does not apply so much on the near bevel, but Figures 5 and 6 show that angles on the far side weld bevel are clearly inappropriate. The codes generally require testing from both sides, which is supported here. Though angles may be inappropriate, coverage may be adequate on the far bevels.

Double V

Double V weld preparations are typically used where wall thickness increases and access is available from the two sides. One purpose of double V weld preparations is to reduce the weld volume that could result from the wide openings that could occur if the single V was continued. The angles used will be greatly dependent on the welding process and access. Since no single preparation geometry exists, the one used for illustration here will consider a relatively wide opening with symmetrical geometry left and right and inside and out.

The probe/wedge combination for heavier wall examples need not be as concerned with the approach restrictions of cap dimensions, and can be larger than those required for thin walls. A typical off-the-shelf 5 MHz 64 element probe can be used for most applications. Dimensions of such a probe are shown in Figure 4b.

The likely scenario for increased thickness would be to use larger apertures, so this has been increased to 20 elements, or approximately 12 mm (0.47 in.). Figures 7a to 7c show the modeled results. Because of the shorter height from the root to the near and far surfaces, the openings at the surfaces are smaller than was the case for the single V. This means that a single 45 to 70° S-scan could be used to cover the volume of interest. This is seen in the blue S-scan in Figure 7a. However, in order to better address the possibility of incomplete root penetration (midwall on this illustration) a second S-scan is presented that covers the root region with a direct soundpath.

Figure 7 — Modeling results for symmetrical double V welds: (a) 25 mm (1 in.) wall using standard 45 to 70° refracted shear wave S-scans with second standoff for root; (b) 50 mm (2 in.) wall using standard 45 to 70° refracted shear wave S-scans with second standoff for root, showing lack of coverage; (c) 50 mm (2 in.) wall using standard 45 to 70° refracted shear wave S-scans with three standoffs.

Although the "volume" can in some cases be fully addressed by a single focal law, the added coverage of a second S-scan (from both sides of the weld) should be a requirement, which it is in ASME (2007). In Figure 7a, the apertures are 20 elements, with the first start element at #7 and the second aperture starting at #40. Hence, almost two aperture dimensions separate the focal laws. This distance will vary depending on the weld dimensions.

The ability to address both the weld outer edges at the heat affected zone and still provide direct approach to the root area cannot extend continuously and will be limited by the probe dimensions. As an example, using the same design of bevel and probe/wedge combination on a 50 mm (2 in.) wall, the coverage cannot be maintained (Figure 7b). This would require a third S-scan and a repositioning of the probe/wedge, as shown in Figure 7c.

Sometimes asymmetry of the weld bevel is used to overcome welding approach difficulties. Bevel angles may then be reduced in some cases, but the relative root position and weld openings must be considered. For thicker sections this will in most cases require more than two S-scan standoffs. Figures 8a to 8c illustrate some examples of asymmetric welds and possible testing solutions using S-scans. In Figure 8a, the weld is well covered using three S-scans. In Figure 8b, the (interior) root is poorly addressed as a result of only two S-scans. Note that Figure 8b also shows a less than ideal angle approach to the weld prep at about 1/3 to 1/2T depth. This can be improved by using a hybrid triple S-scan, as shown in Figure 8c. (Figure 8c still shows a less than ideal angle approach to the weld prep at about 1/2T depth with this setup. The region between the green and the red shows that a small portion is not covered with a direct first half-skip approach.)

Figure 8 — Modeling results for asymmetrical double V welds: 50 mm (2 in.) wall using standard 45 to 70° refracted shear wave S-scans with three standoffs for asymmetric bevel; (b) 50 mm (2 in.) wall using standard 45 to 70° refracted shear wave S-scans with two standoffs for asymmetric bevel, but now inverted; (c) possible hybrid triple S-scan approach for testing double V welds with asymmetric profiles.

Other Weld Configurations — T-welds

Examination of welds is not always as simple as that presented for butt welds. Curved surfaces or angled intersections can provide other considerations. Also, access is always a consideration since some surfaces are often unavailable for testing.

The general rule for ultrasonic testing is to inspect the weld from the surface on which the bevel is made. For T-welds, this will be dependent on the preparation type. As with butt welds, the preparation can be single or double-sided. Similar considerations must be given for access (probe size) and volume coverage, as was done for butt welds. Access to the vertical face (unbeveled) may be limited or the wall thickness may be such that no practical skip effect can be used to approach the weld from those surfaces. Fusion faces provide a major concern for weld quality and the vertical land at the root is again given extra coverage where possible. This will usually involve two standoff positions when using S-scans (T-connections are not likely encountered in thinner wall plate, such as less than 10 mm [0.4 in.]). Figures 9a and 9b show two sample T-weld profiles and testing strategies.

Figure 9 — T-weld profiles and testing strategies: (a) 20 mm (0.8 in.) beveled T-weld using standard 45 to 70° refracted shear wave S-scans with two standoffs; (b) 20 mm (0.8 in.) beveled plate on 20 mm (0.8 in.) plate using standard 45 to 70° refracted shear wave S-scans.

Although a single fusion-face is well-approached from the plate via a skip, and a single standoff may suffice, little useful information is gained and weld surface geometry may present limitations for access and cause mode conversions.

When access is possible from the opposite wall, the ideal access is made to view the longest fusion face and also provide access to angle beams to assess the beveled faces. Where concern exists for toe cracking on the unbeveled plate, a second set of S-scans (or E-scans) can be made (Figures 10a and 10b). In Figure 10a, a normal incidence is all that is required for the flat fusion line of the member the probe is on. The beveled fusion lines would normally be covered by the angular approach from the lower member.

Figure 10 — Scanning from the opposite wall: (a) 20 mm (0.8 in.) plates using 0° E-scan, and 0 to 35° S-scans at two standoffs with another two standoffs for 0 to –35° S-scans using refracted compression mode; (b) 20 mm (0.8 in.) plates using two compression mode S-scans 25 to 45° and –25 to –45° for toe cracking.

DISCUSSION

One question that has been asked is whether we really need to perform two S-scans on thin plates to get coverage. The modeling shows clearly that on butt welds of less than 10 mm (0.4 in.), coverage can be achieved in a single S-scan. However, not all bevel incidence angles will be optimum. Will we miss discontinuities? The answer, based on previous modeling (Moles and Zhang, 2005), is probably no for thin plates, as the probe aperture is relatively large in comparison with the wall, and should collect reflected ultrasound well.

However, this is not the case with all welds. Above 10 mm (0.4 in.), two S-scans are definitely required for coverage. In addition, two S-scans are required for appropriate angles. As the wall gets thicker, more than two S-scans may be required. For complex weld configurations, no predictions are possible. However, it is generally straightforward to model the component, and then determine what the optimum S-scan positions are, and whether coverage is possible. A future version of the software will have bevel incidence angles, so both coverage and appropriate angles can be determined with a simple program.

Codes typically require some knowledge of "coverage," so this trend of using modeling should continue, primarily for encoded linear scans. Note that coverage and appropriate angles are not the same; one can have coverage but inappropriate angles. In such a case, little may be detected.

CONCLUSION

Simple ray tracing programs can show coverage of welds using S-scans for linear encoded scan plans. To answer the question of whether more than one S-scan is required for thin walls, the modeling shows that a single S-scan can obtain coverage for a butt weld up to around 10 mm (0.4 in.). For codes such as ASME Section V (2006) that require scan plans for linear encoded scanning, simple modeling appears to be the best solution to show coverage and appropriate angles. Realistically, it is not possible to give rules from this modeling due to the wide configuration variations in welds (and probes); these results should be treated as guidelines only.

REFERENCES

AchAPI, Recommended Practice for Ultrasonic and Magnetic Particle Examination of Offshore Structural Fabrication and Qualification of Technicians, third edition, Washington DC, American Petroleum Institute, 1996.

ASME, ASME Boiler and Pressure Vessel Code, Code Case 2557: Use of Manual Phased Array S-scan Ultrasonic Examination, per Article 4, Section V, New York, American Society of Mechanical Engineers, 2006.

ASME, ASME Boiler and Pressure Vessel Code, New York, American Society of Mechanical Engineers, 2007.

AWS, D1:1 2006, Structural Welding Code - Steel, Miami, Florida, American Welding Society, 2006.

Charlesworth, J.P. and J.A.G. Temple, Ultrasonic Time of Flight Diffraction, Hertfordshire, England, Research Studies Press, 1989.

Moles, M. and J. Zhang, "Construction Weld Testing Procedures Using Ultrasonic Phased Arrays," Materials Evaluation, Vol. 63, 2005, pp. 27-33.

R/D Tech, Introduction to Phased Array Ultrasonic Technology Applications, Quebec, R/D Tech, 2004.

 

* Materials Research Institute, 432 Country Squire Rd., Waterloo, ON N2J 4G8, Canada; (519) 886-5071; e-mail eginzel@mri.on.ca.

^† Olympus NDT, 73 Superior Ave., Toronto, ON M8V 2M7, Canada

Copyright © 2008 by the American Society for Nondestructive Testing, Inc. All rights reserved.