Integrating Probabilistic Approach into British Defect Assessment Procedure: State of the art review on existing PFM procedures
TWI Industrial Member Report 1205-2024
By Dr Yin Jin Janin
Industrial Need
Engineering structures and equipment (such as onshore/offshore structures, pressure vessels, pipelines) are expected and often designed to not impose ‘unacceptable’ risks to human life. This is not always successful as indicated by some high-profile structural failures. The investigation of those failures, and development of the appropriate codes and standards, have helped reduce failure rates throughout the industrial era.
The available deterministic flaw assessment procedures, eg BS 7910 (BSI, 2019), API 579 (API, 2021), R6 (EDF, 2019), RSE-M (AFCEN, 2020), etc, do not provide a consistent measure of component risk or quantify the component margin. A typical flaw assessment allows the users to determine flaw acceptability by incorporating safety factors, margins and conservatisms to account for model and input uncertainties. Probabilistic methods to quantify component margin are deemed to be the more consistent approach when the target reliability acceptance criteria are well defined.
The probabilistic assessment approach to determine reliability can be complicated and is not always explicitly described in currently available procedures. This work package includes a state of the art review of existing PFM procedures (BS 7910, API 579-1, R6, CSA Z662 and SSM Report 2008:01), focussing on the determination of probability of failure as well as the differences in the treatment of variables.
Key Findings
- In addition to R6 and BS 7910, only a few procedures offer some guidance in performing probabilistic assessments.
- API 579 and RSE-M do not offer direct advice to establish POF. RSE-M only incorporates reliability assessment using partial safety factors.
- There is some similarity in the treatment of variables between R6, BS 7910, CSA Z662 and SSM Report 2008:01. But so far, only R6 offers a user-friendly guidance on performing a probabilistic assessment and determination of POF (especially when considering more than one variables).
- The R6 probabilistic procedure is applicable to non-nuclear applications as it is written based on simple mathematical principles to determine POF.
- There are minor differences in the definition of limit state functions between BS 7910, SSM and CSA Z662 but they largely based on a simple FAL principle and can be executed using a BS 7910 Option 1 procedure (only using yield and tensile strengths).
- Among these procedures, there are differences in the treatment of input data, their relevant distributions and time-dependent properties.
- R6 and BS 7910 appear to be the most user-friendly procedures for probabilistic assessments.
- Industry-specific environmental effects, eg fracture toughness properties of irradiated materials, may be used as a reference to determine an indicative value but they are rarely applicable outside their own remit as it is not possible to read cross from two completely different environments eg from nuclear to sour service.
Impact
This work package (WP) has identified the differences between the available defect assessment procedures. There are some industry-specific treatments of input data which may not be beneficial for other industries. Advice on statistical distributions of material properties in different environments is not readily available. This is currently under investigation in other work packages.
Most procedures recommend direct testing to determine representative material properties. R6 and BS 7910 appear to be the most user-friendly procedures for probabilistic assessments but in areas to which these two procedures do not apply, care should be taken when employing approaches from other documents.