A common location for defects to appear is at the surface of a component. While at the near surface, many non-destructive evaluation (NDE) techniques are available to inspect for these, at the far surface this is much more challenging. Ultrasonic array imaging is proposed to enable far surface defect detection, location and characterisation. In this thesis, array imaging is explored for the practical application of subsurface imaging and characterisation of the near backwall defects, especially surface-breaking cracks. In this application, the component has two parallel surfaces, the crack is initiated from the far side and the phased array is attached on the near side.

One specific challenge here is the presence of a strong reflection from the backwall, which can often mask the relatively small response from a defect. Ideally, a pure scattered field from a defect is needed for the correct estimation of the scatterer through the imaging algorithm, thus, the presence of the backwall will result in strong effects in the resulting image. Several post-processing methods, which have to be used together with specific imaging algorithms, are developed in this thesis to eliminate the strong effects introduced by the presence of the backwall. The effectiveness of these proposed methods are validated in both simulations and experiments.

In this thesis, the Factorization Method (FM) is explored for the application of subsurface imaging and characterisation of the surface-breaking cracks. The performance of the FM algorithm and relevant backwall effects eliminating methods were tested with the simulated and experimental data. The experimental results showed a good consistency with the simulated results. It is shown that the FM algorithm can generate high quality images to provide a good detection of the crack and an accurate sizing of the crack length.

The quality of the array imaging highly depends on the imaging algorithm used. One way to make array imaging more applicable is to extend the library of the imaging algorithms. Based on this aim we study the application of two innovative imaging algorithms, i.e. the Direst Sampling Method and the Fast Topological Imaging, in ultrasonic array imaging. Firstly, these imaging algorithms are tested with numerical simulations to image different type of defect. After that, several experiments are performed to assess their performance in practice.

The development of the backwall effect eliminating methods is based on the assumption that the backwall is perfectly smooth. However, for real components experiencing harsh environments, the surfaces are rarely perfectly smooth. Therefore, the developed backwall eliminating methods are also validated on components with a rough backwall to check the tolerance to the roughness.