No abstract

Different time-dependent mechanisms such as creep, environmental surface oxidation or internal material degradation due to aging and irradiation will subject structures to the possibility of premature failures. In this paper a micro-scale finite element mesh consisting of multiple elements encased in ∼50-150µm sized grains with designated grain boundaries is used to replicate shapes and dimensions to simulate an isotropic metallic microstructure. The grains are encased in pseudo-grain boundary element sets which can have different material and damage parameters compared to the grains. In this type of mesh random crack paths for intergranular and transgranular cracking conditions are allowed. It is shown that creep cracking using a uniaxial ductility constraint-based model coupled with a functionally distributed time-dependent environmentally assisted corrosion/oxidation/material degradation damage model acting on surface or internally can be realistically predicted using this model. It is also evident material properties input data have scatter especially at the sub-grain level where the measurement methods are new and not always standardised. This is dealt with in the model by employing a normal distributive probabilistic method to allow for statistically varied random damage and crack growth development. In this way it is possible to take into account the inherent variability in material input properties at the analysis stage without the need to change material properties following each run. The method could negate the need for knowing the exact material properties, which in any case is impossible to derive at the microstructural level, as results of each run can be varied using a statistically distributed critical damage criterion specified for each element.

No abstract

No abstract

The effect of residual stress on component failure has been investigated using the distributions from current failure assessment procedures, and a residual stress profile simple to apply with less conservatism has been proposed for the weld geometries of T-plate and tubular T-joint. The Stress intensity factors (SIFs) in the two weld geometries under various types of loads have been calculated using the Green's function method. The Green's functions were determined not only for the T-plate but also for the tubular T-joint with the built-in ends. The use of a linear (bending) stress profile, derived from an analysis of measured residual stress distributions in T-plate and tubular T-joints, has been examined. The profile was validated with experimentally measured residual stress distributions in two materials, a high strength and medium strength ferritic steel and two geometries, a T-plate joint and a tubular T-joint for crack lengths up to half the plate or pipe thickness. Whereas the recommended residual stress distributions are geometry and material specific, it is shown that a simplified linear bending profile provides a possible guideline, applicable to a range of materials and geometries, where detailed information on weld procedures or residual stress profiles are unavailable.