Effect of Reference Reflector Orientation for Ultrasonic Inspection of Piping Welds

In this paper, the sensitivity responses from both axial and circumferential notches on piping calibration blocks are compared and their uses discussed when performing ultrasonic inspection of piping welds.


Definitions
For the purposes of this paper, the definitions of axial and circumferential targets are presented as shown in Figure 1. For sensitivity calibration in the axial direction, the reference reflector is machined circumferentially, transverse to the beam. And conversely the target used when the beam is oriented circumferentially is machined in the axial direction. Rather than describe the targets by machining direction, the industry nomenclature AOD for "axial, outside diameter" and COD for "circumferential, outside diameter", referring to beam direction and examination surface, will be used throughout.

Introduction
Piping calibration blocks in accordance with ASME BPVC Sec. V Art. 4 are illustrated showing reference reflectors in both the axial and circumferential directions. Illustration excerpts from the code are shown in Figure 2. The code states "separate calibrations shall be established for both the axial and circumferential scans". While an axial (AOD) inspection is commonly conducted for inspection of girth welds, the circumferential (COD) scan is usually attributed to inspection of long seam welds. These are considerably less common than AOD for ASME piping weld inspections but may occur on headers and reheat lines.
The ASME code also requires a scan for detection of transverse flaws. These would normally be performed using manual skewing techniques and a conventional transducer using a flat wedge, as in most cases the weld cap is left in the as-welded condition. Those inspections typically categorized broadly as AUT, which may utilize fixed-skew angle beam pairs in a pitch-catch configuration, each with a compound axial and circumferential contoured wedge, are not considered.
The decision as to what reference reflector to use to establish the sensitivity level for a transverse scan is a point of contention. Some practitioners use the AOD reflector for both the primary and transverse inspections. And while a scan for transverse flaws uses a probe skewed through a range of angles, somewhere between AOD or COD, some practitioners prefer to use the COD reflector, presuming that because it is oriented in the same direction as a transverse flaw, it must be more appropriate.
This paper will illustrate the challenges in performing a repeatable sensitivity calibration using the COD reflector and a flat wedge for a transverse inspection. The beam physics will be demonstrated using photo-elastic visualization and experimental results.

Coupling Between Flat and Curved Surfaces
To produce a coherent and correctly refracted beam, there must be stable and complete coupling between the wedge and the examination surface. However, when using a flat wedge on a curved surface, the wetted or coupled area is neither complete nor stable.
In theory, the coupling point is simply a line marking the tangent point of contact between the flat wedge bottom and the curved examination surface. In practice, the couplant viscosity and meniscus creates a coupled area of some varying efficiency towards the outer edges of the wetted mark. This is illustrated for a flat wedge on curved pipe surface in the AOD and COD directions in Figure 3. As seen in the COD photo, the coupled area for inspection in the COD direction may literally "slice" the top and bottom of the beam off, resulting in a much smaller and less efficient sound transfer. The wedge may also be easily tilted on the curved surface, moving the exit point and distorting the effective angle of refraction.

Beam Distortion Due to Curvature Effects
Analysis has been performed previously on the sensitivity variations inherent in "rocking" a flat probe side-to-side while inspecting in the axial (AOD) direction (5). While there is a significant sensitivity drop as the probe is tilted side-to-side, this effect is dwarfed by the sensitivity and refracted angle variations when attempting to use a flat probe in the circumferential (COD) direction.
Photoelastic visualization is presented in Figure 5 to Figure 7 using a ½" 5MHz transducer and a 45° wedge on an 8.625-inch OD surface. The probe is tilted slightly forwards and backwards and the effect on the refracted angle imaged.
Pg 4 of 12 The photoelastic images are all taken at the same time position with the only difference being the minor tilt of the wedge. The changes in effective refracted angle are a combination of effects from both a shift in angle and a change in wedge path. The total amount of wedge tilt is only about 5° and creates larger differences in refracted angle than may be initially expected. The effective angle of the refracted wave fluctuates between 37° and 61°, far exceeding the tolerance of ±2° seen in ASME and other codes. The low angle ( Figure 5) is observed near the ID of the simulated pipe surface, whereas the high angle ( Figure 7) has not yet reached half thickness. This illustrates the extreme difficulties in obtaining accuracy and consistency for sensitivity calibration and inspection with a flat wedge in the COD direction.
It may seem surprising at first that pressing lightly on the back of the wedge decreases rather than increases the effective refracted angle. In normal flat surface configurations, wearing down the back of the wedge will increase the incident and refracted angles simply through the principles of geometry and Snell's Law. However, in the case of COD configurations with a flat wedge, pressing on the back of the wedge actually shifts the exit point backwards. This shift reduces rather than increases the angle of incidence. This results in an effect opposite to what would normally be expected. Similarly, when tilting the wedge slightly forward, the exit point shifts forward and the refracted beam angle increases.
The amount of rocking is diminished by using a smaller footprint transducer, but even the beam from a ¼" 5 MHz transducer on a 45° wedge can be similarly manipulated. The response from the OD notch on an 8-inch STD block can be identified at two distinct positions, corresponding to effective angles of 40° and 49° depending on whether a small amount of pressure was applied to the front or back of the wedge. These effective angles are both far beyond the ASME wedge angle tolerance of +/-2°.

Effects on Sensitivity Calibration
During the sensitivity or time-corrected-gain (TCG) calibration, the user is tasked with maximizing the signals from the reference reflectors. With a flat wedge on a curved surface, the responses can vary drastically as the user tilts the search unit to peak the signal. These variations are considerably more pronounced in the COD orientation due to highly variable angle responses shown in the previous section.
Notch reflectors are asymmetric reflectors, producing drastically different responses at different angles. The ability to tilt the wedge allows the operator to easily change the angle and alter the response, especially due to mode conversion effects (6). The beam angle variations will produce erratic echo dynamic traces which make it difficult to generate accurate and repeatable sensitivity calibrations. Side-drilled holes will not incur the same mode conversion challenges as notches, but will still incur effects due to angular differences and changing sound paths. ASME Sec. V Art. 4, ISO 17640 and other codes stipulate a consistency of ±2dB in sensitivity calibration (1,3). Erratic responses when attempting to perform a sensitivity calibration in the COD orientation with a flat wedge run counter to point of establishing a repeatable reference baseline.

Experimental Setup
Responses from AOD and COD notches in 4-inch STD, 8-inch STD and 14-inch Sch. 40 piping calibration blocks were compared using a Sonatest Wave flaw detector and the search units listed in Table 1, Table 2, and Figure 9:   Based on the measured dimensions, the uncontoured wedges above are compared to the ISO 17640 and ASME Sec V Art. 4 limits for the piping blocks in the following table. Wedges marked with a checkmark are sufficiently small in footprint (width or length) for the orientation of calibration noted. For reference, AOD calibrations were first established on each piping block using the AOD notches to generate a 2-point TCG. Using the same search unit, the maximum responses in the COD direction were then examined using the COD notches (example, Figure 10).

Experimental Results
TCG calibrations were performed in the AOD orientation and used as a baseline for comparison to COD results. The numbers shown in the following table below are the differences using the notches in each direction. Numbers greater than zero mean the COD response was higher than AOD, thus producing a lower sensitivity TCG. Conversely, numbers less than zero would result in a higher sensitivity TCG. The average difference was less than 2 dB overall. Focusing only the 45° probes, the difference was less than 1 dB. The 60° beam was too high in many cases to produce a repeatable and coherent echo response. This was due to the high degree of refraction angle variability on the curved surface resulting in a combination of 1 st and 2 nd leg reflections. Figure 11 and Figure 12 show the differences between 45° and 60° responses.

Computer Modeling of 60° Beam
Civa modeling was performed on a simple 60° beam to demonstrate beam interactions on the ID and OD surfaces. Beyond variable wedge tilt, a significant contributor to the messy amplitude variations with a 60° beam is also the angle that the Tx and Rx paths make to the notch. The ideal 60° refraction condition is maintained in Civa but not in manual UT. Even then, the angle returned to the probe from the notch can be well off ideal when it returns. In this image it is off 7.5° in the wedge as it strikes the element. A 3dB beam spread is plotted, and when the front of the beam is traced it never reflects off the ID, while the back of beam makes 3 half-skips.

Conclusions and Recommendations
Using a flat wedge and the COD reflector, it is extremely difficult if not impossible to produce a repeatable and predictable sensitivity calibration. A simple understanding of ultrasonic physics reveals far too many variables, from index point shift, beam distortion due to coupling effects, and beam angle variations. For inspection of girth welds on piping, where the cap is left in the as-welded condition, it is recommended to use the AOD reflector for the primary sensitivity calibration and for the manual transverse scan.

Angle Limitations in COD Direction
For many pipe sizes, it may simply be impossible to establish a consistent TCG in the COD direction using anything but a low angle such as 45°. Angles higher than this use longer wedges with greater incurred tilt, producing more erratic beam behaviours and possible multiple target reflections. And as the pipe curvature greatly limits the useful range of angles, manual phased array inspection is not particularly helpful for transverse scans.

Code Stipulations
The ASME code does not require both AOD and COD reflectors on piping calibration blocks. In response to the question "Does an alternative calibration block in T-464.1.3 require the inclusion of the axially-oriented side-drilled holes if only circumferential welds are to be inspected?", code interpretation BPV V-18-37 [2] states "No, any combination of calibration blocks may be used to meet the requirements of T-464.1.2 and T-464.1.3." These requirements state "separate calibrations shall be established for both the axial and circumferential scans", and thankfully the code does not define a manual transverse skew scan as circumferential.
Based on the number of problems shown in establishing a stable and repeatable TCG on the COD reflector with a flat wedge, a "circumferential scan" should be interpreted strictly for those situations where the beam is directed purely in the circumferential direction. For the most part, this will encompass long seam welds or inspection for transverse flaws on girth welds where the weld cap is machined flush to the pipe body.
It is worth noting that ASME does not actually define the scanning direction as a "circumferential scan" in Article 1 definitions. However, it does state "circumferential direction: direction of sound beam perpendicular to (cylindrical) component's major axis". Clearly, a skewed transverse scan where the beam has to be skewed towards the weld axis by some degree is not by definition in the "circumferential direction".
The approach to defend against the need for the COD target could be part of the rationale that is required in Article 1 Para T-150. As shown, there are minimal differences between AOD and COD sensitivities. Therefore, one could use this demonstration in the procedure to show that the calibration for the transverse scan is adequately addressed with the AOD reflector. However, given the fact that a transverse scan should not be considered "circumferential" by definition, and the improbability of stability in COD sensitivity calibrations, a demonstration of equivalence should be unnecessary.
Furthermore, for those cases where a purely COD inspection is possible, it is recommended that contoured wedges be used that match the OD curvature.

Irregular Beam Coverage on Transverse Scans
Further research is pending by the authors to illustrate the fact that reliable detection based on amplitude responses from transverse flaws is very difficult due to simple beam geometry. Due to curvature, a full skip may be produced only by low angle wedges. Yet on many pipes of standard thicknesses, once a 45° probe is rotated towards the weld axis, the full skip position will no longer interrogate the weld at the cap, and in fact may not even reach the HAZ. Detection of transverse flaws can be shown to occur only due to beam periphery effects.

Minimal Sensitivity Differences Between AOD and COD Calibrations
Because even small flat wedges on curved surfaces can be tilted during calibration, the resultant sensitivity calibrations performed on COD targets are too easily manipulated and variable to be considered repeatable between technicians. At best, calibrating with a flat wedge on a COD target is pointless, as it is shown to produce very close if not identical results to the AOD target. At worst, using the COD target can produce wildly variable results between inspectors, completely missing the point of calibration which is to produce repeatable and predictable inspection levels for flaw detection.