BENCHMARKING THE MCNP MONTE-CARLO CODE WITH A PHOTON SKYSHINE EXPERIMENT

The MCNP Monte Carlo transport code is used by the Los Alamos National Laboratory Health and Safety Division for a broad spectrum of radiati on shielding calculations. One such application involves the determina tion of skyshine dose for a variety of photon sources. To verify the a ccuracy of the code, it was benchmarked with the Kansas State Universi ty (KSU) photon skyshine experiment of 1977. The KSU experiment for th e unshielded source geometry was simulated in great detail to include the contribution of groundshine, in-silo photon scatter, and the effec t of spectral degradation in the source capsule. The standard deviatio n of the KSU experimental data was stated to be 7%, while the statisti cal uncertainty of the simulation was kept at or under 1%. The results of the simulation agreed closely with the experimental data, generall y to within 6%. At distances of under 100 m from the silo, the modelin g of the in-silo scatter was crucial to achieving close agreement with the experiment. Specifically, scatter off the top layer of the source cask accounted for approximately 12% of the dose at 50 m. At distance s >300 m, using the Co-60 line spectrum led to a dose overresponse as great as 19% at 700 m. It was necessary to use the actual source spect rum, which includes a Compton tail from photon collisions in the sourc e capsule, to achieve close agreement with experimental data. These re sults highlight the importance of using Monte Carlo transport techniqu es to account for the nonideal features of even ''simple experiments.' '