University: University of Bristol  Supervisor: Dr Jie Zhang, Dr Nicolas Larossa and Dr Alexander (Sasha) Velichko  Start date: October 2021

How much of the actual ‘reserve life’ of UK high integrity assets can we use by accounting for their local topology, e.g. flaw tip acuity, in structural integrity (SI) assessments and thus relaxing the conditions for repair or replacement? How much of the actual over-conservatism can we remove in design standards by incorporating damage tolerance arguments for non-sharp defects? In most cases, the answer to these questions is currently NOTHING. The application of methods considering flat tip acuity (or defect bluntness) is limited by the lack of credible NDE methods to distinguish between sharp and non-sharp defects and by the unavailability of testing procedures for notch-fracture toughness measurement. 

This project aims to address these issues through the development of blunt defect characterisation approaches. This will build on current defect characterisation expertise at Bristol which is generally built on the application of techniques to describe defects scattering responses. These will be developed in this case for the problem of blunt defects. The project will link closely with the structural integrity team to explore how varying quality of characterization affects the results life extension possible, therefore closing the link between NDT and structural integrity. 

As well as exploring existing characterisation approaches, machine learning, will be used to analyse both ultrasound images from the defects of interest and their associated scattering responses (S-matrices), and set up appropriate networks to measure the acuity of the defects.  A small number of FMC data sets will be measured experimentally from a few representative defects, meanwhile, a large number of FMC data sets from a wide region of defects will be simulated from an efficient hybrid wave scattering model. These data sets will be used for determining characterization performance of the scattering based approaches, training machine learning networks and finding the optimized parameters. The performance of the trained networks will be tested using the FMC data measured experimentally from real defects. A key part of the project will be the comparison of the performance from each of these different methodologies, which will feed into the structural integrity assessments proposed. 

The work will be primarily supervised by Dr Jie Zhang in the Bristol Ultrasonics and NDT group. In addition Dr Nicolas Larossa will provide support in the development and application of the associated structural integrity methods. Dr velichko will provide expertise and support in the scattering based modelling and characterisation approaches. Together this supervision team and the environment of the research group at Bristol will provide an excellent environment for the student on the project. 

Funding: 

This studentship covers fees at the home/EU rate, a stipend of £16877 per annum and the full technical and professional training programme as part of the FIND CDT.