A TEST OF MAIN STREAM PIN POWER RECONSTRUCTION METHODS

Preliminary to implementing a pin power reconstruction scheme in the n odal core calculations of the ELCOS system, the ''main stream'' method s and elements thereof were tested against fine-mesh calculations of a number of benchmark ''small cores'' consisting of uranium, controlled uranium, and mixed-oxide assemblies. Overall, the results do not clea rly favor one of the methods. However, test details conduce us to pref er the 32-term expansion for corner-point fluxes over their determinat ion by the separability assumption, and the 21-term expansion of the i ntranodal flux over the 13-term expansion. There is little difference whether the factorization of the pin power distribution into global an d form factors is imposed on the group fluxes or on the power. Data tr ansfers and matrix inversions connected with the many-term flux expans ions slow down the nodal calculation. This condition may be alleviated in some cases by an approximation leading to fewer matrix inversions.