Control Rod Calculation in Axially Heterogeneous Fast Reactors—II: Impact of Three-Dimensional Homogenization on Core Parameters

Advanced sodium-cooled fast reactors with improved safety features such as the French Advanced Sodium Technological Reactor for Industrial Demonstration (ASTRID) CFV (French acronym of Coeur à Faible effet de Vide sodium, meaning low sodium void effect core) core concept are characterized by an axial heterogeneous core that will present a challenge for the homogenization procedures used today, taking into account all the different axial material transitions. Reliable modeling of the control rod and accurate prediction of the control rod worth are essential to determining the shutdown margins and to ensuring safe operation.

In this work (part II of two companion papers), two different homogenization schemes are compared. One is based on the traditional reactivity-equivalence procedure in two dimensions, and the other is a newly implemented three-dimensional (3-D) version of the reactivity-equivalence procedure, with approximations based on the results in the companion paper. The deterministic results are compared with a Monte Carlo reference.

Both cross-section sets from the two homogenization schemes yielded results within the requested ±5% error margin in reactivity. The largest discrepancy was found for the classical procedure for the case with a slightly inserted control rod (normal operating conditions). Both cross-section sets yielded similar power profiles in the fuel subassembly neighboring the control rod within the 2σ Monte Carlo standard deviation. Neither of the cross-section sets was able to predict the large gradients in capture rates close to the internal control rod interfaces.

The study showed that the traditional two-dimensional (2-D) reactivity-equivalence procedure produces homogenized cross sections that yield reliable results in a CFV-type core. One exception from this was found for slightly inserted control rods, where the effect of the follower-absorber interface could not be fully captured by the 2-D scheme, and for such cases, 3-D modeling is recommended.