The objective of the results analysis and interpretation activity is to derive an understanding of those aspects of plant design and operation that have an impact on the risk.
In addition, an important part of this task is to identify the key sources of uncertainty in the model and assess their impact on the results.
Uncertainties can be thought of as being of three main types:
•
Parameter uncertainty: these are uncertainties in the values of the initiating event frequencies, component failure probabilities, human error probabilities, etc. These uncertainties can be propagated through the analysis to generate an assessment of the uncertainty on the overall quantitative results using standard methods. Parameter uncertainties are addressed in Section 11.
•
Model uncertainty: There are questions with how to model certain failures (e.g. RCP seal LOCAs), or how to represent the impact of plant conditions on system success criteria, for example, for which there is no universally accepted approach. Typically, in PSAs, these model uncertainties are dealt with by making assumptions and adopting a specific model. In relatively rare cases, alternate models may be incorporated into the PSA, weighting each model by a probability representing the degree of belief in that model as being the most appropriate.
•
Completeness uncertainty: This is the most difficult to deal with as it represents those contributors to risk that are not included in the model. If the PSA model only includes internal initiating events at power, the contributors to risk not modelled include external events, and alternate modes of operation. At a more subtle level, typically PSAs do not include contributions from errors of commission.
The significance to risk of individual contributors (initiating events, accident sequences, functional failures, system failures, component failures, human failures, etc.) are explored to derive an understanding of the risk profile of the plant, i.e., what is the impact of various aspects of plant design and operation on risk. The impact of uncertainties and assumptions on the PSA results are addressed in order to determine the robustness of the conclusions concerning the risk profile. The analytical tools used for the analysis of results are importance analyses and sensitivity analyses.
12.2. Results analysis and interpretation tasks and their attributes
Table 12.2 lists the main tasks for the PSA element ‘Results Analysis and
Interpretation’. Tables 12.2-A through 12.2-C present the description of general and special
attributes for these tasks.
Table 12.2 Results Analysis and Interpretation Tasks Task ID Task Content
RI-A Identification of Significant Contributors RI-B Assessment of Assumptions
RI-C Documentation
Table 12.2-A Attributes f or Results Analys is and Interpreta tion: Task RI-A ‘Ide ntif ication of Sign ificant Contributors’
Task / GADescription of Task/General Attributes Identifier and Description of Special Attributes (in Italics) Rationale/Comments/Examples for: General Attributes and Special Attributes (in Italics) RI-A The task includes identification of significant contributors to the risk profile of the plant. RI-A01Significant contributors to CDF are identified. The contributors are, in increasing level of resolution: -Initiating events -Accident sequences -Key safety function failures -System failures -Basic events The basic events include: -Equipment failures or unavailabilities -Common cause failures -Human failure events RI-A02 Significant contributors to accident sequences, key safety functions, and systems are identified. RATIONALE: Reviews of the solutions to system models, functional models and accident sequences are an essential part of the validation of the structure of the plant logic model, and furthermore, provide additional insights on the risk profile of the plant. RI-A03 When assessing the significance of basic events using importance measures that involve setting failure probabilities to unity, such as the risk achievement worth (RAW), the assessment is performed by resolving the PSA model rather than re-quantifying a pre- solved cutset list.RATIONALE: A cutset list generated with too high truncation value will exclude some components and therefore their RAW values are identically unity. An alternative to resolving the model is to use a cutset equation solved at a lower truncation value.
Table 12.2-B Attributes f or Results An alysis and Interpreta tion: Task RI-B ‘Assessment of Assum p tions’
Task / GADescription of Task/General Attributes Identifier and Description of Special Attributes (in Italics) Rationale/Comments/Examples for: General Attributes and Special Attributes (in Italics) RI-BAssessment of the significance of key assumptions and model uncertainties is performed. RI-B01 Significant sources of model uncertainty are identified (see the comment). COMMENT: A model uncertainty is one associated with the modelling of an issue or phenomenon, for which there is no consensus approach. Such model uncertainties are typically addressed by adopting one of a number of models, or making an assumption about the impact of a phenomenon on the operability of a system or function. A significant source of uncertainty is one where the adoption of a different model or a different assumption can alter the significance of a contributor. Since different applications make use of different results, the significant sources of uncertainty will differ between applications. RATIONALE: When the results of the PSA are to be used for decision making, the decision-maker needs to be aware of the impact of the uncertainties on the PSA results used in the decision (see RI-B02). EXAMPLES of potential sources of model uncertainty include: -Success criteria -RCP seal LOCA model -Assumptions about the necessity for room cooling -Choice of quantification model for human error probabilities. RI-B02 The effect of a significant assumption on the results is assessed by performing sensitivity studies, using different plausible assumptions. RATIONALE: Understanding of the impact of modelling uncertainty on the results of the PSA is crucial to a decision-maker. An exception can be when a particular model has been approved as the standard model. COMMENT: Some sensitivity studies can be performed by changes to parameter values, or turning off parts of the model to represent groups of failure. Others may involve adding new portions to the model. These are done in a manner consistent with the relevant attributes. RI-B03 For model uncertainties or assumptions affecting the same parts of the PSA model, sensitivity studies are performed simultaneously to determine whether there are synergistic effects. RI-B04 The choice of the specific assumptions or models adopted for the base case model is justified. RATIONALE: Understanding the reasons for choosing a specific assumption or model is important for the decision-maker.Table 12.2-C Attributes f or Results Analysis and In terpreta tion: Task RI -C ‘Doc um enta tion’
Task / GADescription of Task/General Attributes Identifier and Description of Special Attributes (in Italics) Rationale/Comments/Examples for: General Attributes and Special Attributes (in Italics) RI-C Documentation of results is performed in a way that facilitates understanding of the technical basis for the significance of contributors. RI-C01 The results of the analysis of the significance of contributors are presented in a variety of ways to characterize the risk profile of the plant, i.e., what are the aspects of design and operational practices that have an impact on risk, how they impact risk, and why. RI-C02 The results of the sensitivity analyses are documented so that the impact of each significant assumption is characterized appropriately, and the choice of the assumption or model for the base case justified.
Dans le document
Determining the Quality of Probabilistic Safety Assessment (PSA) for Applications in Nuclear Power Plants | IAEA
(Page 120-125)