Comparison of Geant4 and MCNP6 for use in delayed fission radiation simulation

Neutron induced fission fragment distributions and delayed fission radiation are extremely important with reactor applications in fission cross sections and heating. Data on the fragment distributions are sparse so simulations use models or interpolations between known neutron energies. Different simulations perform different treatments of the distributions, and have different capabilities and flexibility in use. MCNP is a typical workhorse for fission simulations and coupled with burn-up codes such as CINDER can provide delayed radiation from fission. Geant4 is an extremely flexible physics based Monte Carlo simulation framework, but is not typically used for fission research. In this work the applicability of Geant4 for delayed fission radiation simulations is examined, with comparison to MCNP6 coupled with the CINDER2008 burn-up code. The Fisher and Engle fission experiment with the Godiva II subcritical assembly as a fission neutron source is used as a test case. Both simulations are adapted from that experiment and simulation results are compared with that experiment. Following Fisher and Engle, photons/fission/sec, MeV/fission/sec, and MeV/photon are examined. For the first two quantities results from both simulation codes are similar and are lower than experimental values, with Geant4 giving a higher value for earlier time bins and MCNP6/CINDER giving a higher value for the later time bins. For the last quantity both simulations are usually within uncertainty of the experimental values, with MCNP6/CINDER values consistently higher than both experimental and Geant4 values.