Nuclear analysis supporting the design and licensing of ITER is traditionally performed using MCNP and the reference model C-Model; however, the complexity of C-Model has resulted in the geometry creation and integration process becoming increasingly time-consuming. Serpent 2 is still a beta code; however, recent enhancements mean that it could, in principle, be applied to ITER neutronics analysis. Investigations have been undertaken into the effectiveness of Serpent for ITER neutronics analysis and whether this might offer an efficient modeling environment.

An automated MCNP-to-Serpent model conversion tool was developed and successfully used to create a Serpent 2 variant of C-Model. A version of the deuterium-tritium plasma neutron source was also created. Standard reference tallies in C-Model for the blanket and vacuum vessel heating were implemented, and comparisons were made between the two transport codes assessing nuclear responses and computer requirements in the ITER model. Excellent agreement was found between the two codes when comparing neutron and photon flux and heating in the ITER blanket modules and vacuum vessel.

Comparing tally figures of merit, computer requirements for Serpent were typically three to five times that of MCNP, and memory requirements were broadly similar. While Serpent was slower than MCNP when applied to fusion neutronics, future developments may improve this, and Serpent offers clear benefits that will reduce analyst time, including support for meshed geometry, robust universe implementation that avoids geometry errors at the boundaries, and mixed geometry types. Additional work is proceeding to compare Serpent against experiment benchmarks relevant for fusion shielding problems. While further developments are needed to improve variance reduction techniques and reduce simulation times, this paper demonstrates the suitability of Serpent to some aspects of ITER analysis.