4.20.
Calculation methods such as computer codes and nuclear data used in the criticality safety analysis to calculate keff should be verified to ensure the accuracy of their derived values and to establish their limits of applicability, code bias and level of uncertainty. Verification is the process of determining whether a calculation method correctly implements the intended conceptual model or mathematical model [2].
4.21.
Verification of the calculation methods should be performed periodically and should test the methods, mathematical or otherwise, used in the model and for computer codes, and should ensure that changes of the operating environment, i.e. operating system, software and hardware, do not adversely affect the execution of the codes.
4.22.
The results of the calculations should be cross-checked by using independent nuclear data or different computer codes when available.
4.23.
After verification of the calculation method is complete and prior to its use in performing a criticality safety analysis, it should be validated. Validation relates to the process of determining whether the overall calculation method adequately reflects the real system being modelled, and enables the quantification of any calculation/code bias and uncertainty, by comparing the predictions of the model with observations of the real system or with experimental data [2]. The calculation method should be validated against selected benchmarks that are representative of the system being evaluated. The relevance of benchmarks for use in performing validation should be determined from comparison of the characteristics of the benchmarks with the characteristics of the system of fissile material being evaluated.
4.24.
In selecting benchmarks, consideration should be given to the following:
Benchmarks should be used that have relatively small uncertainties compared with any arbitrary or administratively imposed safety margin.
Benchmarks should be reviewed to ensure that their neutronic, geometric, physical and chemical characteristics encompass the characteristics of the system of fissile material to be evaluated. Examples of neutronic, geometric, physical or chemical characteristics that should be used for all materials include, but are not limited to, the following:
Molecular compounds, mixtures, alloys and their chemical formulae.
Isotopic proportions.
Material densities.
Relative proportions or concentrations of materials, such as the moderator to fissile nuclide ratio. Effective moderators are typically materials of low atomic mass. Common materials that can be effective moderators include water (i.e. hydrogen, deuterium and oxygen), beryllium, beryllium oxide and graphite (i.e. carbon). In the presence of poorly absorbing materials, such as magnesium oxide, oxygen can be an effective moderator.
Degree of homogeneity or heterogeneity and uniformity or non-uniformity, including gradients, of fissile and non-fissile materials (e.g. spent fuel rods, settling of fissile materials such as waste).
Geometric arrangements and compositions of fissile material relative to non-fissile material such as neutron reflectors and including materials contributing to the absorption of neutrons (e.g. cadmium, hafnium and gadolinium are commonly used, but other materials such as iron also act as slow neutron absorbers).
The sensitivity of the system to any simplification of geometry, for example elimination of pipes or ducts.
Neutron energy spectrum.
Calculation methods should be reviewed periodically to determine whether relevant new benchmark data have become available for further validation.
Calculation methods should also be re-verified following changes to the computer code system and periodically.
4.25.
Once the calculation method has been verified and validated, it should be managed within a documented quality assurance programme as part of the overall management system. The quality assurance programme should ensure that a systematic approach is adopted in designing, coding, testing and documenting the calculation method.