|Mr Nader Zentuti|
For assessing a component operating at high temperatures for creep-fatigue damage which can lead to crack initiation, the stress state during nominal operating conditions is required. These stresses are usually obtained from complex thermal and mechanical FE. However, using these models to predict the stress evolution for the entire lifespan of a component would be computationally prohibitive, especially if a probabilistic assessment is required. Conventionally, a fixed stress state is assumed to persist during periods of the components history which can lead to over conservatism in the assessment results. Therefore, an approach for accounting for the variability in the stress state over the entire history of a component is required.
An approach was formalised (see Fig.2) with the aim of incorporating stress uncertainties in creep-fatigue assessments. This was achieved constructing a predictive (surrogate) model which captures the FE model behaviour for a wide range of possible input operating conditions (see Fig.3). Therefore, the stress state can be treated as an uncertainty in the creep-fatigue damage assessment.
A number of benefits can be derived from this work:
- Incorporating stress uncertainties provides better representation of historic loading conditions and reduces conservatism.
- Feed into a methodology for treating stresses in probabilistic creep-fatigue crack initiation assessments e.g. EDF’s R5 Volume 2/3 procedure.
- Promote a shift from over-conservative deterministic assumptions (e.g. fixed stresses) towards a systematic probabilistic framework.
Ultimately the aim of this project is to quantify and incorporate real-life uncertainties in high temperature assessment procedures.