Abstract
In this thesis novel concepts were applied to common problem areas in fracture mechanicsincluding: welding residual stress, microstructural embrittlement, crack-tip bluntness, (out-of-
plane) plastic constraint and fracture mode mixity.
The relative contributions to weld fracture toughness variability from microstructural
embrittlement and longitudinal welding residual stress were quantified. Data quantifying
either variable were gathered and then assessed using a combination of Weibull methods
and bivariate data analysis. This investigation suggests that variability in microstructural
embrittlement is the major contributor to variability in weld fracture toughness.
Using multivariate methods, existing data measuring longitudinal welding residual stress
variability of similar welds were analysed in combination with several other factors. The
results suggest that cumulative heat input from the weld tool is the most significant driver of
variability.
Multivariate analysis of mixed variable data was used to assess if pre-cracking via fatigue or
electrical discharge machining has a significant effect on fracture toughness of test
specimens. Analysis accounting for pre-crack technique and crack growth orientation,
suggests that neither factor can be used to reliably predict fracture toughness.
The effect of out-of-plane constraint loss on fracture resistance was investigated. Failure data
over a range of specimen geometries and constraint levels were gathered experimentally.
These experiments were then simulated, modelling failure using local approach criteria
based on both maximum stress and dilatational strain energy density. These models saw
various degrees of success based on failure criterion. Whilst relatively undeveloped,
dilatational strain energy density is an adequate failure criterion, though further development
is warranted.
Lastly, the statistical methods used in this thesis were assessed for validity using hypothesis
testing and error analysis. Several shortcomings were found in the Weibull parameter
estimation techniques used throughout this thesis. However, given prior validation with
physical evidence, the validity of these methods is not entirely undermined
| Date of Award | 1 Oct 2024 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Mahmoud Mostafavi (Supervisor) & Martyn J Pavier (Supervisor) |