What is Reliability in NDE?
Inspection configurations are extremely diverse, and in a context of more and more demanding requirements (security, cost, performance, and environment) with a lot of emerging and innovative techniques, to apply standards is often not sufficient to guarantee that an NDE inspection will reach the required and expected level of performance.
A qualification process is then necessary, and has to demonstrate that the involved NDE will meet the requirements in the conditions in which it will be applied.
Indeed, inspection relies on measurement and as any measurement, it is subject to variability and uncertainty, as well as deterioration factors. This will impact the performance, and you have to assess this variability to ensure and demonstrate the reliability of the inspection.
This is a costly process. Knowledge, methodologies and relevant tools are then a prerequisite if you want to be cost efficient!
Why register to this session?
You wish to:
- Get some knowledge on:
- Existing process and methodologies for NDE qualification, standards, reference documents and recommended practices,
- How to account for influential parameters / variables,
- The statistical indicators used to evaluate the reliability of an inspection (sensivity analysis, POD, false alarms, etc.)
- Understand and master underlying statistical hypotheses and reasoning,
- Use tools that will help you perform reliability studies in a NDE context,
- Discover how simulation can help you reach your targets and reduce costs!
This training is for you!
The “Plus” of our training
- A global and plurisectoral approach aiming at offering a global perspective on the existing methodologies, in order to select an adapted implementation for you,
- Many practical exercises on various examples and tools are planned (statistical computation and analysis, modelling, quiz sessions, etc.) aiming at a pragmatic and pedagogic approach to these methodologies,
- The use of simulation, which is currently becoming a really powerful and reachable way to support you in this context.
The program
This training session takes place over 3 days:
Day 1:
- Measurements and uncertainties: Definitions and statistical tools
- Density function, cumulative distribution function, expected value, variance, quantile, linear regression modelling, etc.
- Workshop: Experimental approach of the uncertainty, binomial law application examples, building a linear regression model
- To produce a sensivity analysis:
- Influential parameters (definition and classification), interaction and correlation, methodologies, screening methods, Sobol indices, metamodels, etc.
- Workshop: Establishing and classifying a list of influential variables, using simulation and metamodels to produce sensitivity analysis
Day 2:
- Statistical indicators to evaluate reliability in NDE:
- POD (binomial approach, parametric approaches), Probability of False Alarms, Specificity, NPV, PPV, ROC curves, Sizing uncertainties, POR
- Workshop: Computing POD values with binomial model, building and analyzing POD curve with â vs a or Hit-Miss parametric models, PFA
Day 3:
- The standard framework:
- General overview, Deterministic and statistical approaches, Details on the main reference documents and standards (RSEM, ASME-BPVC, DNV OS F101, MH1823A, ENIQ RP41)
- Workshop: Synthesis on the presented reference documents
- Innovative methods:
- Benefits of simulation, MAPOD, using a combination of simulations and experimental trials, "advanced POD" (POD curves with Min (POD)>0 and Max (POD)<1, non-parametric POD, multi parametric POD), pending questions and perspectives, bibliography
- Workshop: MAPOD using simulation and metamodels
Intercompany sessions organized in our premises are proposed, but we can also perform dedicated training in your facilities, on demand.
For more information on prices and scheduled sessions, take a look at our Training Course Catalog.
The Tools
- Microsoft Excel and Microsoft PowerPoint
- CIVA (CEA License)
- MH1823 POD (courtesy of Charles Annis).
- OpenTurns - Python
