University of Bristol
Supervisors: Dr Sergio Cantero Chinchilla, Prof Paul Wilcox, Prof Mahmoud Mostafavi
Industry Partner: Rolls Royce
This PhD project will examine the effect of intervals of inspection when estimating the life of a safety critical component. Non-destructive evaluation (NDE) is currently used at intervals throughout the life of a component to detect if any cracks over a specified length have been formed. If a crack is found and its length measured, then the remaining useful life of the component can be estimated. The inspection frequency and remaining useful life (RUL) are based on a suitable crack growth model and operational loading. All aspects of the process (residual stress levels at manufacture, crack length measurement, material properties used in the crack growth model, the crack growth model itself, operational loading etc) involve assumptions and uncertainty.
The purpose of this PhD is to investigate:
- how uncertainties propagate through the process,
- how information obtained from NDE can be used to update crack growth model parameters to reduce uncertainty,
- the optimal inspection interval (which may not be constant) to achieve the most accurate outcome, involving a cost-benefit argument and considering whether the added cost of permanent monitoring is worthwhile.
The project will investigate the damage growth through fatigue testing in metallic samples with damage initiators. These samples will be extracted from plates of interest to Rolls-Royce (RR). Ultrasonic NDE measurements will be frequently taken during the fatigue testing in order to estimate the damage extent throughout the experiments. In order to validate the NDE technique, the damage growth will be monitored in parallel through techniques such as full field measurements or X-ray tomography. All fatigue experiments will continue until failure to build a statistical database of failure loads. An adequate methodology will be developed to estimate the optimal inspection intervals that minimises the inspection costs while maximising the RUL of the samples.
This overall process will be applied and developed for different samples: (1) standard sample extracted from parent material; (2) welded ferritic plate with damage initiators; and (3) dissimilar metal weld (DMW). A DMW poses the following scientific challenges that will be addressed: (a) heterogeneous microstructure at the joints causing biased characterisation of stress concentrations and (b) large grain size increasing the uncertainty associated with ultrasonic measurements. If successful, the method will be also applied to a ferritic plate welded to an austenitic plate to generate a DMW to align with the research currently conducted at Bristol-Manchester-Imperial on data-centric engineering.