The verification and validation procedures are used for checking that the code fulfills its purpose according to given specifications. Each software release must be tested continuously during its development and, in particular, before and during the verification and validation process.
The verification process used for the MORET code includes respect of requirements and specifications, code review, testing (according to non-regression of operability installation, and portability).
In the development phase, the verification procedure involves performing specific tests to model the new capability introduced in the code and then proceed to an analysis of the results. When the development is verified and finalized, another testing phase has to be done. Many capabilities and functions result from the interaction of code components. A new feature could change the overall operation of the software. A large set of verification tests is launched to ensure that the "new code" still meets previous specifications (this is also known as non-regression tests).
These checks ensure that the coding is performed without program errors and, where appropriate, meets programming rules. However, they never completely guarantee that the model is correctly represented.
The validation process of a calculation route, which aims at predicting the accuracy of a calculation code associated to a given configuration, is a prerequisite in support of safety reports. It is based on comparisons between calculations and experiments. The observed discrepancies are interpreted and eventually transposed to real configurations.
The experimental validation does not mean the validation of a code, but validation of a calculation route made of a nuclear data library and codes used with some parameters, options and methods.
+ Validation database
The experimental validation database covers a broad variety of configurations in terms of fissile medium, moderator, reflector and neutron spectrum (to validate different materials and cover a wide moderation ratio range). It is made up of 2300 critical experiments (included in 244 series of experiments) divided into 8 categories. They are mainly taken from the Handbook of the International Criticality Safety Benchmark Evaluated Experiments Project and from other experimental programs, such as the MIRTE program performed in French facilities.
Category distribution of the database
Some efforts are made to continue and expand the validation of the code for reactor applications. This work is still in progress.
+ Methodology for validation
Two calculation routes are validated. The first one is the multigroup calculation route of the CRISTAL package, where the deterministic cell code used for producing macroscopic cross sections is APOLLO2. The second route is a continuous energy route. In this case, 1200 experimental cases contained in the whole database are performed for the moment.
The validation is done through the comparison of the calculated keff with the benchmark keff. If the discrepancy between these two keff is higher than the combined standard deviation of the benchmark uncertainty and the Monte Carlo standard deviation, a bias can be identified.
Then, various libraries are tested to assess the impact of nuclear data evaluations.
Finally, thanks to the two calculation routes, the keff results of multi-group codes can be compared to the keff of the continuous energy route using the same libraries. These comparisons allow enhancing biases due to the multigroup treatment and physical models implemented in multigroup codes.
+ Some results using the JEFF-3.1 library
The average C-E discrepancies (given in pcm) for the two calculation routes are reported per 4 categories (more than 75% of database experiments). For LATTICES and METAL categories, two averages are calculated depending on the reflector type.
- Continuous energy route: consistency between calculated and benchmark keff (overestimation for experiments with lead due to the inelastic scattering reaction).
(C-E) for continuous energy route [pcm]
- Multi-group route: good agreement C/E except for experiments with metallic reflectors or absorbing canisters (spotted in different colours on the graphics). The discrepancies depend on the reflector thickness and the distance to fissile. The multi-group treatment (self-shielding of material) may be improved.
(C-E) for multi-group route [pcm]