Ultrasonic methods for assessing fatigue degradation in metals

Abstract

Steel components in nuclear power generation plants suffer from fatigue, due to the cyclic loading to which they are subjected. It is also often the case that the amplitude and frequency of this loading changes in an irregular pattern, as it is directly related to the required power output at any given time. Detecting such damage is made more challenging since the damaged site is often on inaccessible internal surfaces of vessels or pipes. This poses the need to develop a method able to assess the level of such fatigue damage in these components to ensure that they are operating safely.

In this thesis, ultrasound was used as a potential solution to this issue. Previous work with through-thickness longitudinal wave measurements has shown that ultrasound is indeed sensitive to the presence of fatigue, as it will travel more slowly compared with propagating through a non-fatigued volume of a material. However, as fatigue is usually a near surface phenomenon, the bulk longitudinal wave spends limited time inside the fatigue zone, and hence any observed changes are small and difficult to correlate to a specific fatigue state with confidence. Therefore, apart from deriving a fatigue assessment technique, it was also desirable to amplify those changes to minimise the effects of any uncertainties.

The work presented in this thesis begins by verifying the sensitivity of longitudinal waves to the presence of fatigue. The verification was achieved by constructing longitudinal wave speed C-scans for five flat plates containing fatigue damage, each at a different fatigue level, which verified both that the fatigue spot is visible in those maps, and that also, when changes in speed were considered, the speed reduced monotonically as fatigue progressed.

A natural solution to the issue of amplifying the reduction in a wave's propagation speed due to the presence of a fatigue zone is the use of Rayleigh (surface) waves, as they are restricted to propagate in the surface of a material, where fatigue is usually concentrated. Indeed, when Rayleigh wave B-scans were completed for each of the five plates, it was found that the percentage reduction in speed was amplified by a factor of approximately ten. Additionally, using a stiffness reduction approach, a method able to accurately encapsulate the effects which fatigue has on the time-of-flight of longitudinal and Rayleigh waves in a finite element model was also developed. The results from the finite element models were found to agree well with the experimental measurements.

In an attempt to extend the surface waves method to pipes, an extensive numerical study on the feasibility of using creeping waves for fatigue state characterisation was completed, as those are waves which can be excited on the inner surface of a pipe, without having access to it. It was found that for the geometries which are of interest to the nuclear power generation industry, the use of such an approach did not yield satisfactory results, due to multiple unwanted reflections and secondary modes interfering with the monitoring of the creeping wave. Therefore, for pipe geometries, a through-thickness approach was considered, this time using shear waves, deployed and analysed by raster scanning from the outer surface. The challenge of obtaining shear wave C-scans was overcome with the use of EMATs and a stepper motor frame. The resulting C-scans showed an increase in the sensitivity to fatigue of more than double compared with longitudinal waves. Also, the speed-reduction behaviour was explained and verified by measuring the dislocation density of samples corresponding to different fatigue states and comparing the results to existing theoretical models.

Finally, the use of surface waves in the work completed in this thesis motivated a secondary study on the attenuation of Rayleigh waves when propagating over a rough surface. The well-established scattering theory in two-dimensional (2D) analysis was verified using finite element modelling in all three scattering regimes: Rayleigh, stochastic and geometric, and was also extended to roughness parameters outside the region of validity of the theory. The verification also provides useful insight in terms of verifying the three-dimensional (3D) theory as well.