Multimode Virtual Source Aperture Imaging for Non-Destructive Testing

Introduction

There exist several forms in which acoustic energy may travel through a component. The two main modes considered in application for NDT are longitudinal (particle oscillation in the direction of propagation) and transverse (particle oscillation perpendicular to the direction of propagation). Sound may convert propagation modes and refract when hitting a reflective surface. These mode-converted signals are present in collected data, but are often ignored in favour of the stronger single mode signals.

However, in some cases it may be beneficial to analyse the mode-converted signals. Consider Figure 1 which depicts sound paths for a multimode (Figure 1a) and a single mode (Figure 1b) inspection. The multimode sound path is more likely to return to the transducer to be detected, whereas the single mode is less likely, due to the difference in nature of the refraction angles for single mode and multimode flight paths.

It is necessary for the VSA algorithm to be able to calculate the correct path that sound has taken between any given two points. This is often difficult when the sound has traversed a refractive boundary, which is the case when sound has mode-converted (multimode sound paths). Single mode sound paths are often easier to calculate, due to the symmetry between incident and reflected angles.

When calculating the path of sound that has experienced refraction, Fermat’s principle of least time is often used to find the correct path (the correct path is the path that takes the least amount of time for sound to travel between the source and point of interest). This usually involves a computationally time consuming process, whereby the refractive boundary is traversed. The time of flight is calculated over a range of points, before settling on the least time. This method can be acceptable for a single refractive boundary e.g. between coupling material and the inspected component, but is usually too slow when accounting for multiple refractive boundaries, as occurs with multimode paths.

Conclusion

Compared with other iterative approaches, the new approach to solving Fermat’s principle of least time has reduced the path calculation time by a factor of two. This is significant for increasing the frame rate for inspections where the surface geometry relative to the transducer is constantly changing (e.g. in immersion inspection).

The developed VSA multimode algorithm was a success, as it could be used to analyse more of the information contained within the ultrasonic data. The effect of this can be seen in Figure 2. Here, real VSA data was collected using a sample containing a surface breaking flaw. The data was reprocessed using a single mode algorithm (figure 2a) and a multimode inspection (figure 2b). It can be seen that the base of the flaw can only be distinguished using the multimode algorithm. This is significant as the additional information provided by the multimode algorithm could be used for improving the characterisation of the flaw and sizing accuracy.

Please visit TWI’s software website for more information on TWI Crystal in relation to FMC and VSA inspection.

This project was funded by TWI’s Core Research Programme.