Fatigue cracking is a common problem in ships and can potentially lead to a catastrophic brittle fracture. This thesis presents a methodology to quantify the risk of leaving these cracks unrepaired while a ship remains in service. The most important variable affecting the probability of failure is the material fracture toughness. Test results carried out on samples taken from a number of ship steel plates are fitted using a modified 'master curve' approach. The approach links fracture mechanics toughness (characterised by Kjc in MPa√m) to the Charpy 27 joule temperature. A major innovation in the work is the use of time-dependent reliability to account for the variation in toughness as the crack extends by fatigue. The loading applied to the crack tip comprises three components: still water bending; wave induced bending; and residual stress. A number of methods used to calculate the probability of failure are compared, with the convolution integral identified as the most suitable. The methodology is successfully benchmarked against the trend in actual failure statistics from the Liberty ships to date. Target probabilities of failure are used to estimate crack lengths for repair. Taking a figure of 5 x 10⁻⁴ events per year as broadly typical of the observed frequency of brittle fracture in merchant ships the methodology suggests that this is equivalent to a crack length of 300 mm in grade A steel at 0°C. Given that cracks are unlikely to be found until they reach 200mm this provides a small margin of safety. For grade D steel the repair length is increased to over 1000mm potentially allowing the repair to be delayed to coincide with scheduled maintenance periods.
Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:436845 |
Date | January 2007 |
Creators | Kent, John |
Publisher | University of Strathclyde |
Source Sets | Ethos UK |
Detected Language | English |
Type | Electronic Thesis or Dissertation |
Source | http://oleg.lib.strath.ac.uk:80/R/?func=dbin-jump-full&object_id=21659 |
Page generated in 0.0018 seconds