Practical application of 'creeping' waves

TWI Bulletin, September 1986

by Peter Smith

Peter Smith is Laboratory Supervisor in the NDT Research Department.

Ultrasonic 'creeping' waves exhibit the main properties of longitudinal (compression) waves but follow the surface at which they are generated. This combination of characteristics makes them extremely useful for certain non-destructive testing (NDT) applications where other wave modes have limitations. Examples are given in this article of their use in detection of shallow surface breaking defects and detection of defects under weld caps.

The most commonly used NDT methods for detecting surface breaking defects are dye penetrant and magnetic particle testing, but these are sometimes inappropriate. For example, dye penetrant testing may not function properly when a surface has been prepared by grinding, and magnetic particle inspection indications may be difficult to interpret at weld toes. Moreover, these methods do not give any indication of defect depth. In such circumstances ultrasonic techniques may be more suitable for detection of surface defects.

Surface breaking defects can be detected ultrasonically by using Rayleigh (surface) or Lamb (plate) waves. Rayleigh waves travel along the surface of the test specimen to a depth of one wavelength (approximately 0.75mm at 4MHz). Particle displacement is elliptical, and the major axis perpendicular to the test surface.

Surface waves depend upon a stress free surface for good propagation; weld spatter and surplus couplant in front of the probe may greatly reduce signal amplitude. A heavily pitted surface damps the surface waves, and indeed, could cause no signal propagation at all. Similarly, a weld toe causes an interruption to surface wave propagation and results in a reflected signal being detected even in the absence of a defect.

Lamb waves only propagate successfully in plates where thickness is a few wavelengths of the ultrasound. There are typically two modes of propagation: the symmetrical or dilational wave and the asymmetrical or bending wave. The generation of the two modes depends upon elastic characteristics of the material being tested, angle of incidence and frequency. At high frequencies, many harmonics are produced and the presence of this multitude of wave modes makes interpretation of defect signals difficult.

Longitudinal angle waves

Longitudinal angle wave probes were originally developed to overcome problems with inspection of austenitically welded components and surfaces clad with stainless steel. The principal difficulties were signal scattering and attenuation when shear wave probes were used because of the grain structure of the material. It was found that these problems could be reduced by use of longitudinal wave angle probes because longitudinal waves were scattered less by the material structure than shear waves. Longer wavelength longitudinal waves also reduced the amount of scatter compared with shear waves of the same frequency. Longitudinal angle probes of 70° were designed for ultrasonic examination of clad surfaces, and it was during use of these probes that the creeping wave was observed. ^[1]   These are waves which follow the surface ( i.e. creep along it), but otherwise exhibit compression wave properties. The signal indications were very weak. However, further developments in probe design enabled a 90° beam of compression waves, described by the developers as creeping waves, to be produced with adequate sensitivity for flaw detection. These probes are now marketed commercially. ^[2]  

Probe construction and operation

Creeping wave probes are designed with separate transmitter and receiver crystals. The transmitter crystal generates compression waves at an angle between 70-90°; therefore the waves propagate virtually parallel to the test surface and travel just below it. The proximity of the beam angle to the critical angle causes a head wave to be produced by total internal reflection at the testpiece surface. This is the creeping wave. Shear waves are also generated by mode conversion from this head wave, with an angle of incidence between 33-38°. Figure 1 illustrates the mechanism of creeping wave generation. The creeping wave follows the surface but is not affected by weld spatter, surplus couplant or a change of geometry, such as at a weld toe, because it is derived from the compression wave travelling just below the surface.

The tolerance of creeping waves to surface conditions can be demonstrated by the fact that signal amplitude is only slightly decreased when the surface in front of the probe is damped, for example by placing a finger on the surface,unlike surface waves which would be heavily attenuated. However, because of the continuous generation of shear waves ( Fig.1), creeping waves have a steep energy decay and can only exist for a short distance along the surface. To maximise efficiency, probes are constructed with separate transmit/receive crystals, i.e. they are twin crystal probes. The crystals are also angled towards each other to provide a focusing effect, focal length being about 20-30mm depending on type of probe. The useful range of the creeping wave is small,about 30-40mm maximum, which provides a signal/noise (S/N) ratio of ≥ 6dB in austenitic stainless steel. The ratio may be better than this depending upon the grain structure of the metal. The longitudinal wave component can be detected at a range of 150mm at the surface, with a refracted beam angle of about 76-80°.

In the range over which the creeping wave exists there is no significant divergence between the surface, creeping, component and main compression wave beam. Therefore the probe is only sensitive in a surface zone a few millimetres deep. In this region shear waves do not interfere with inspection because they travel at approximately half the velocity of the creeping wave and compression wave, so that any signal arising from them is much further along the flaw detector timebase.

Beyond the range of the creeping wave, the probe is no longer highly sensitive at the testing surface and responses may be observed from reflectors deeper in the test piece. Responses may be obtained by either the main compression wave beam, a combination of shear and compression waves, or shear waves alone ( Fig.2). Compression wave response (A) is distinct from that of shear wave components because compression waves travel faster. Nevertheless in other instances (see B-E in Fig.2) the complexity of beam paths and wave modes causes serious problems in interpretation and the use of these probes for detection of deeply embedded defects is not recommended.

Practical application

Calibration

Because a wide head wave is produced by these probes, it is impossible to calibrate the timebase of a flaw detector in the normal way by using a standard calibration block and calibrating the screen to the required range; indeed it is difficult to determine the probe index point. However, because creeping waves run directly along a surface, the index point position is not important as it can be compensated by a suitable delay setting. Therefore an arbitrary index is scribed on the probe.

To calibrate the flaw detector, a normal compression wave probe of the same frequency as the creeping wave probe is used to set the range of the timebase, because the wave velocities are identical. Once the timebase has been adjusted, the stand off distance at which the creeping wave probe will be used can be set. To do this the creeping probe is placed on the test specimen itself at a known distance from a free edge and the range delayed to the required distance, so that the echo appears at the correct range ( Fig.3).

A Röntgen type 80-640 SEK probe was used, operating at 2MHz. It has twin 6 x 28mm crystals, squint angle (toe in) 27.6°, focal length 20mm, and was connected to a Krautkramer USIP11 ultrasonic flaw detector.

Sensitivity to small surface defects

A 10mm thickness plate of HR 15 ferritic steel containing notches of 2.0, 0.5 and 0.1 mm depth and with machined surfaces was used to test the probe's ability to detect near surface defects. The probe was placed on the notched plate facing a free edge. The stand off distance was set on the timebase as described above. To obtain a reference sensitivity, the signal amplitude from this edge was set to full screen height. The 2mm deep notch was scanned from the notched surface at 20mm stand off ( Fig.4). Signal amplitude from the creeping wave at 2omm range was recorded and found to be the same as the reference level ( Fig.5), indicating that virtually all the creeping wave energy is concentrated within 2mm of the surface.

To determine the limiting depth of defect which can be detected using creeping waves a number of other notches were examined using the same probe at the above sensitivity. All notches were the same length, which was greater thanprobe width. Consequently, notch depth was the only varying parameter. Responses from the other notches of 1.0, 0.5, 0.25 and 0.1mm depth can be seen in Fig.6-9, respectively. It is apparent that the limit of detection is around 0.25mm: the 0.1 mm notch was not detected. Also, the signal amplitude decreased as notch depth decreased, but not quite in the manner expected. The Table shows results obtained compared with expected response based on relative reflecting areas of the notches. The smaller notches give greater signals than expected, suggesting that creeping wave energy available forreflection is concentrated at the surface. This aids detection of shallow defects.

Comparison between defect depth and response amplitude

Weld inspection

A major benefit of use of this type of ultrasonic probe is the ability of creeping waves to travel underneath the reinforcement of a T butt or fillet weld. This was demonstrated on 12mm thickness plate of BS 4360 Grade 50D steel with a T fillet weld which was fatigued and a crack grown in the toe of the weld to a maximum depth of 6mm. The sensitivity of the probe was set by placing it facing a free edge, see Fig.10. The resulting signal was set to full screen height at a stand off of 20mm (48dB of gain).

The probe was then placed touching the toe of the weld and the whole weld length scanned along the cracked side ( Fig.11). The crack was detected at the focus distance and its length accurately measured by the 6dB drop technique. A signal amplitude of full screen height was recorded with the longitudinal indication of slightly greater strength, at 29mm range. The probe was then placed on the opposite side of the T joint ( Fig.12). From this position the crack was detected at 46mm range. The ultrasound in this instance travelled under the T butt weld to reach the crack; length was again successfully measured using the 6dB drop technique.  

These examples show that creeping wave probes are extremely useful for detection of shallow surface breaking defects, even at weld toes or where access is only possible from the far side of the weld. However, use of creeping waveprobes for general manual testing is difficult and is not recommended. The probe should be scanned at a fixed distance from the weld and the signals are generally complex and therefore difficult to interpret. Further, it must beremembered that the effective inspection zone of the creeping wave is small and that reflectors from features in the body of the weld may rise from the main compression beam or the shear waves also present. Consequently, evaluation ofweld body defects is not generally possible.

References