Effect of lattice-level adjoint-weighting on the kinetics parameters of CANDU reactors

Abstract

Space–time kinetics calculations for CANDU ® 1 CANDU ® ( CAN ada D euterium U ranium) is a registered trademark of Atomic Energy of Canada Limited (AECL). 1 reactors are routinely performed using the Improved Quasistatic (IQS) method. The IQS method calculates kinetics parameters such as the effective delayed-neutron fraction and generation time using adjoint weighting. In the current implementation of IQS, the direct flux, as well as the adjoint, is calculated using a two-group cell-homogenized reactor model which is inadequate for capturing the effect of the softer energy spectrum of the delayed neutrons. Additionally, there may also be fine spatial effects that are lost because the intra-cell adjoint shape is ignored. The purpose of this work is to compare the kinetics parameters calculated using the two-group cell-homogenized model with those calculated using lattice-level fine-group heterogeneous adjoint weighting and to assess whether the differences are large enough to justify further work on incorporating lattice-level adjoint weighting into the IQS method. A second goal is to evaluate whether the use of a fine-group cell-homogenized lattice-level adjoint, such as is the current practice for Light Water Reactors (LWRs), is sufficient to capture the lattice effects in question. It is found that, for CANDU lattices, the generation time is almost unaffected by the type of adjoint used to calculate it, but that the effective delayed-neutron fraction is affected by the type of adjoint used. The effective delayed-neutron fraction calculated using the two-group cell-homogenized adjoint is 5.2% higher than the “best” effective delayed-neutron fraction value obtained using the detailed lattice-level fine-group heterogeneous adjoint. The effective delayed-neutron fraction calculated using the fine-group cell-homogenized adjoint is only 1.7% higher than the “best” effective delayed-neutron fraction value but is still not equal to it. This situation is different from that encountered in LWRs where weighting by a fine-group cell-homogenized adjoint is sufficient to calculate the correct effective delayed-neutron fraction. It is concluded that lattice-level weighting with a detailed fine-group heterogeneous adjoint is desirable for the correct calculation of the effective delayed-neutron fraction in CANDU lattices.