Rejection sampling from a spherical harmonics angular flux functional expansion for a discrete ordinates to Monte Carlo splice

Splice methods offer the ability to accurately solve larger and more complicated problems than traditional solution techniques. In advanced splice methods, two (or more) distinct numerical methods are applied. Such methods allow users to increase local solution accuracy and reduce computational expense by judiciously selecting the method best suited to a particular region within the problem space. The primary challenge with developing effective splice methods of this type is the development of algorithms for coupling different solution techniques at the splice interfaces. This paper develops a method of representing angular flux data from a primary discrete ordinates solution as a function using spherical harmonics. Rejection sampling from the function is then used as a source in a secondary Monte Carlo solution. The three dimensional discrete ordinates radiation transport code PARTISN and the Monte Carlo radiation transport code MCNP were used to splice across a simple neutron transport problem exhibiting a forward-peaked angular flux distribution at the selected splice location. Functionality was added to MCNP to permit rejection sampling from spherical harmonics. Standalone global solutions from both codes are presented and compared to the spliced solution. (authors)