Chapter 4: Safety Assessment

Approaches for deterministic safety analysis

Deterministic safety analysis is performed with a number of computer codes. It is important that the computer codes used to perform accident analysis are verified and validated for the accident scenarios of interest. A number of codes which are widely accepted as well validated for different accident scenarios have been developed. Furthermore, care should be taken to ensure the quality of the analysis, by using high-quality input data and by limiting the influence of 'user effects' on the results. The latter can be accomplished by guaranteeing that the analyst performing the calculation has a deep knowledge of both the code used in the analysis and the physics involved in the accident sequence which is simulated.

Various options are currently available for performing deterministic safety analyses with different levels of conservatism associated with the computer code used, the assumptions made about the availability of systems, and the initial and boundary conditions applied for the analysis. Usually, four options (summarised in Table 2.4-1) are considered:

Table 2.4-1: Options for performing deterministic safety analysis.

 
Option

 
Computer Code Type

 
Assumptions on system availability

 
Type of initial and boundary conditions 

  1. Conservative

 Conservative

 Conservative

 Convervative

 2. Combined

 Best estimate

 Convervative

 Convervative

 3. Best Estimate Plus Uncertanity (BEPU)

 Best estimate

 Convervative

 Best estimate; partly most unfavourable conditions

 3. Realistic

 Best estimate

 Best estimate

 Best estimate



Conservative

In the conservative approach, both the assumed plant conditions and the physical models are set conservatively. The parameters need to be allocated values that will have an unfavourable effect in relation to specific acceptance criteria. The conservative approach was commonly adopted in the early days of safety analysis to simplify the analysis and to compensate for limitations in modelling and knowledge of physical phenomena with large conservatisms. It was assumed that such an approach would bound many similar transients in a way that the acceptance criteria would be met for all bounded transients.

Experimental research has resulted in a significant increase in knowledge of physical phenomena, and the development of computer codes has improved the ability to achieve calculated results that correspond more accurately to experimental results and recorded event sequences in nuclear power plants. Owing to the improved capabilities of computer codes and the possible drawbacks of the conservative approach (e.g. potential masking of important phenomena, and conservatisms in different parameters potentially cancelling each other out), conservative analysis is rarely used for new nuclear power plants and is not suggested for current safety analysis, except in situations in which scientific knowledge and experimental support is limited. This option remains relevant, however, as it may have been used in legacy analyses.

Combined

This option is a combined approach based on the use of best estimate models and computer codes instead of conservative models and codes. Best estimate codes are used in combination with conservative initial and boundary conditions and with conservative assumptions made about the availability of systems, assuming that all uncertainties associated with the code models are well established and that the plant parameters used are conservative, based on plant operating experience. The complete analysis requires use of sensitivity studies to justify the selection of conservative input data.

Best Estimate Plus Uncertainty (BEPU)

This option allows the use of best estimate computer codes together with more realistic assumptions. A mixture of best estimate and partially unfavourable (i.e. somewhat conservative) initial and boundary conditions may be used, taking into account the very low probability that all parameters would be at their most pessimistic value at the same time. Conservative assumptions are usually made about the availability of systems. In order to ensure the overall conservatism required in analysis of design basis accidents, the uncertainties need to be identified, quantified and statistically combined.

Realistic

This option allows the use of best estimate models and computer codes, and best estimates of system availability and initial and boundary conditions. Hence, this option is appropriate for realistic analysis of anticipated operational occurrences aimed at the assessment of control system capability and in general for best estimate analysis of design extension conditions (see paragraph below), as well as for the purpose of justifying prescribed operator actions in realistic analysis. Deterministic analysis for operating events that may necessitate a short term relaxation of regulatory requirements may also rely on best estimate modelling.

Analysis approach for DECs and severe accidents

Design extension conditions without significant core degradation or with core melt (severe accidents) are usually modelled with either of the two last approaches (BEPU or realistic).

For more information see SSG-2 Rev. 1 Read more →