Benchmarking a CAD-Based Monte Carlo Code Using Fusion-Specific Experiments

Abstract

Researchers at the University of Wisconsin-Madison Fusion Technology Institute and Argonne National Laboratories have recently developed a computer-aided-design-based Monte Carlo code (DAG-MCNP5) to perform nuclear analysis of complex three-dimensional systems such as ITER. In this work, DAG-MCNP5–calculated results will be compared to native MCNP5–calculated results and to measured results for ITER-specific benchmark experiments in order to provide additional quality assurance for DAG-MCNP.

Calculated results are compared for the bulk shield mock-up and the helium-cooled pebble bed (HCPB) breeder blanket mock-up, which utilize the 14-MeV Frascati Neutron Generator facility. Neutron flux was measured at different depths in these experimental mock-ups using activation foils that cover the neutron energy range of 0 to 14 MeV. Additionally, tritium production in Li₂CO₃ pellets was measured in the HCPB experiment.

Results of the foil activation calculations for the bulk shielding experiment and the HCPB breeder experiment show agreement within statistical uncertainty for DAG-MCNP5 and native MCNP5. Calculated results for tritium production in the HCPB mock-up also agree within statistical uncertainty for the DAG-MCNP5 and native MCNP5 calculations. Timing results showed that DAG-MCNP5 is 5.3 times slower than native MCNP5 for the bulk shield mock-up. For the HCPB mock-up, DAG-MCNP5 is 4.8 times slower than native MCNP5.

It is concluded that the close agreement of calculated foil activation and tritium production between DAG-MCNP5 and native MCNP5 in these complex and ITER-relevant geometries provides additional quality assurance for the DAG-MCNP5 code and the mcnp2cad tool used in this work.