AUTOMATED VARIANCE REDUCTION OF MONTE-CARLO SHIELDING CALCULATIONS USING THE DISCRETE ORDINATES ADJOINT FUNCTION

Although the Monte Carlo method is considered to be the most accurate method available for solving radiation transport problems, its applica bility is limited by its computational expense. Thus, biasing techniqu es, which require intuition guesswork, and iterations involving manual adjustments, are employed to make reactor shielding calculations feas ible. To overcome this difficulty, we have developed a method for usin g the SN adjoint function for automated variance reduction of Monte Ca rlo calculations through source biasing and consistent transport biasi ng with the weight window technique. We describe the implementation of this method into the standard production Monte Carlo code MCNP and it s application to a realistic calculation, namely, the reactor cavity d osimetry calculation. The computational effectiveness of the method, a s demonstrated through the increase in calculational efficiency, is de monstrated and quantified. Important issues associated with this metho d and its efficient use are addressed and analyzed. Additional benefit s in terms of the reduction in time and effort required of the user ar e difficult to quantify but are possibly as important as the computati onal efficiency. In general, the automated variance reduction method p resented is capable of increases in computational performance on the o rder of thousands, while at the same time significantly reducing the c urrent requirements for user experience, time, and effort. Therefore, this method can substantially increase the applicability and reliabili ty of Monte Carlo for large, real-world shielding applications.