Speaker
Description
Verification and validation are essential components of the nuclear data development process, since only those data that have been demonstrated to accurately simulate real-life applications can be relied upon for those applications. While most nuclear data work relies on so-called 'differential' measurements, the validation of this work requires data from full systems also called Integral Benchmarks. Thousands of integral benchmarks are available and their use for Nuclear Data validation requires an automated process to minimize work time and improve Quality Assurance. For this purpose, an automated tool is developed at LANL.
The automated Nuclear Data Validation Tool is built as a tool kit. It is designed to be open source and user friendly. The automated tool is composed of five modules. When using all the modules successively, it aims at processing one or several ENDF format data file to be used by the MCNP transport code to run a selection of integral benchmarks and provide simulation results into graphs or tables. Each module can be run separately as well.
The first module encapsulates the NJOY data processing code to create a full or a partial nuclear data file library in the ACE format and its associated xsdir file. Parameters available to users are temperatures and the NJOY version code. The second module constitutes a database of MCNP input files associated to a search functionality. About 1,200 of input files from the International Criticality Safety Benchmark Evaluation Project (ICSBEP) are available. Benchmarks and benchmark input files can be selected through metadata and isotope occurrence. User-defined libraries or the regular MCNP libraries are available. The library definition operates at the isotope level. The third module is an MCNP run launcher. For a list of cases, it submits runs either sequentially or simultaneously. Workstation as well as HPC platforms are supported; 6.1 and newer versions of MCNP are accessible. Parallelism through MPI, OMP or mixed MPI/OMP is user-defined as well as the architecture of the computing platform. The postprocessing of the MCNP output files is performed in the fourth module. Results are stored in json files. This module can also perform comparisons to experimental values.
The last module is dedicated to documentation and the presentation of results through graphs and/or tables. For each plot, all parameters are user-defined. Plot parameters can be saved and stored into a list of pre-defined graphs for further use.