The test conducted on the rough cracks has shown both the advantages and limitations of the procedure presented in Figure 5.2 . Firstly, it has shown that rough cracks can be detected through the enhancement features on the B-scan images. The machine learning program explained Section 5.1 can be used to recognise the features and classify them into normal/angled defect. Secondly, two cracks positioned closely together can be distinguished in the B-scan and through the Rayleigh wave amplitude measurements, as shown in Figure 5.8 and Figure 5.9. If the cracks are within the near-field of each other, the inclination of the first crack encountered in the scan can be characterised
using the FER value, with the reference value calculated in the near-field of the crack.
For the second crack, the reference can be estimated from the dip after the enhancement of the first crack.
The procedure shows a good estimation of the crack depth if the crack is approxi- mately uniform, as shown by Crack 1, but if the crack depth has a varying depth, such as is the case for Crack 2, the estimation will give an average depth over the area covered by the EMAT. For a very deep crack, a threshold depth may need to be introduced, so that the procedure gives a “deeper than” output for cracks outside its detection range. This range can be evaluated from the frequency bandwidth of the signal used and from the calibration shown in Figure 4.6.
Chapter 6
Effects of crack width on Rayleigh
wave transmission
The study of Rayleigh wave interaction with surface defects has been an active area of research [23,53,54,56,57,59,62,65,69,72]. While the previous research investigated these interactions, the majority of the research were based on the assumption that the crack width is infinitely long compared to the incoming Rayleigh wave. This certainly is not true for many more realistic defects, especially when the crack has a finite width and is comparable with the size of the Rayleigh wave beam. For example, rolling contact fatigue (RCF) cracks founds in rails have typical crack width of 5 mm to 20 mm [3]. The term narrow-width crack is used in this thesis to refer to this type of crack. This issue is investigated in this chapter.
This chapter proceeds as follows. Experiments were initially performed using an EMAT transmitter with a laser interferometer as receiver, to map the diffraction around 90° and 25° cracks of different widths, which were chosen to represent normal and inclined cracks. A second set of experiments were then performed using an EMAT transmitter and an out-of-plane EMAT receiver with the measurements compared with those using the laser receiver. The spatial averaging effect of the EMATs receiver is discussed here. The viability of a pair of EMATs to be used for real crack detection, where the width may be less than the EMAT width, is explored. Results from EMAT measurements of cracks of different widths are discussed in the light of characterising realistic surface cracks and suggestions are made on how to deal with situations where the EMATs are expected not to be able to size the crack depth accurately using Rayleigh wave transmission. Based on the viability study, EMAT techniques show promising
possibilities for use real applications, in particular for detecting RCF cracks in rails.
6.1
Rayleigh wave diffraction around narrow-width
cracks
When a propagating plane wave encounters a small obstacle, some of the wave bend around the obstacle to get past it. The is what a diffraction is. Real surface cracks typically have a surface length or width of finite size. Thus, when a Rayleigh wave is used to characterise them, it will suffer from diffraction effect around the crack edges. The wave transmission will be affected, due to the delay of Rayleigh wave component that has been diffracted at the crack edges. As the wave transmission is crucial to crack depth estimation using the method explained in Chapter 4, the diffraction at the crack edges will most likely to affect the depth estimation and may cause some errors. It is important therefore to know how much the diffraction affects the transmission. Another important point is whether, with the additional diffraction factor, the strong alternating black and white enhancement pattern can still be observed and used for inclined cracks.
The diffraction was studied experimentally using an EMAT as transmitter and a laser interferometer [107] as receiver on aluminium bars, as shown in Figure 3.12. The full experimental details are explained in Section 3.4. The laser interferometer was chosen as the method for detection because it detects surface displacements over the small area illuminated by the laser. This allows for detailed mapping of the ultrasonic wave around the crack during a scan, which cannot be done easily using the EMAT receiver due to the finite size of the coil. The EMAT transmitter was fixed at 130 mm from the crack, while the laser receiver was scanned from 20 mm away from one side of the crack to 30 mm on the other side of the crack. The width of the aluminium bar, as shown in Figure 3.12 is divided into two,with only one half scanned to record the diffraction, and the results are mirrored to give a full mapping across the width of the bar. This one-half section was divided into 27 sections, each 1 mm apart, and a scan was made along each section. Surface displacements measured along each scan line were used to reconstruct the interaction and thus give surface displacement snapshots at a given time.
Two sets of cracks, each at an angle of either 90° or 25° to the sample surface, have been used as, explained in Chapter 3. For each type of crack, two crack widths have
been considered; 20 mm and 10 mm. The first one was chosen to investigate the case when the size of the EMAT is comparable with the crack width, and the second one is for the case when the crack width is significantly smaller than the size of the EMAT.