Subsea pipelines are increasingly being used both around offshore drilling facilities and for long distance oil and gas transport. Accidents can have devastating environmental and economic impact, amplifying the need for accurate, reliable detection and characterisation of pipeline defects. Inspection of these pipelines for corrosion and other defects is crucial for safe operation. Radiography holds a significant advantage over many other inspection methods in that it does not require surface preparation or insulation removal.

Subsea pipeline radiography is a relatively new technique, and underwater conditions are not covered by radiographic standards. Water can have a significant impact on a radiographic image and access is very difficult, meaning standardised above-water methods may not be applicable. This is particularly the case for defect characterisation; standard methods often call for calibration objects to be included in the setup, which can be a very complex operation in subsea conditions. There is also a lack of experimental data for research, due to the difficulty and high costs associated with subsea radiography. Simulation is one of the key ways of assessing inspection problems, however radiographic simulation models have not been validated for subsea inspections.

This thesis addresses the two problems of accurate subsea simulation and alternative defect characterisation methods. Firstly the accuracy of a radiographic simulation model applied to subsea pipeline inspections is investigated. Experimental measurements of a sample in a water tank are used to adjust the simulation, with the aim of matching image quality parameters - such as signal-to-noise ratio and contrast. The simulation has been partially matched to experiment, with some differences found in contrast-to-noise ratio. Possible causes of the differences are analysed, with the most likely cause found to be detector backscatter and additional scatter from out-of-setup objects within the experimental exposure bay.

The simulation model is then used to provide data for development and testing of a defect characterisation method. The method relies on knowledge of the setup geometry and use of multiple images, and does not require calibration objects to be included in the setup. It is specifically aimed at use in situations where access is difficult such as in subsea pipeline inspections. The method is tested on simulated and experimental flat bottomed hole defects and simulated corrosion patch defects. Results demonstrate a good, consistent ability to calculate lateral and axial defect dimensions. Defect thickness calculations are more difficult and as such errors are more significant. However, errors in thickness are due to overestimation, meaning the calculation could be used to place a maximum limit on potential defect size rather than as an actual estimate of the thickness.