Analysis of RA-3 experiments for the validation of a neutronic model

Abstract

A 3D neutronic model for the RA-3 reactor was developed on the basis of previous experience and validated with selected experimental data. Control rod calibrations were reproduced in N94 and N136 cores. The calculated values are shown to be dependent on relative position of the rods and the procedure that gives the best estimation of the rod value is the one performed following the experimental method compensating small rod insertions with small extractions. Rod worth calculations differ from the measured values in less than 2%. Rod-drop experiments were used to evaluate rod effectivities. The experimental results showed discrepancies between estimators derived from the point reactor model, and from spatial modal kinetics. Discrepancies are also observed when using different detectors. Even when using the spatial modal kinetics approach, the estimators obtained from different detectors disagree when one of them is located near to the rod, but differences are considerably reduced with respect to point reactor model because in this case only the delayed evolution is considered. We can say that all estimators give fairly similar results when the detector field of view is not influenced by the local perturbation introduced by the falling rod. This indicates the existence of spatial effects which are not completely accounted for in the spatial modal kinetics approach. Also, the importance of verifying the form function behaviour during the delayed evolution. The rod drop experiments were simulated using the improved quasi-static model and static evaluations. The rate of overestimation static/dynamic is constant in both core configurations and varies between 18% and 23% for the analyzed rods. The dynamic model allows comparing also thermal flux ratios at the detector positions. The neutronic model is considered reliable for design and fuel management analysis. The estimations of criticality, control rod calibrations and excess reactivity are satisfactory and simple models representing the core components without any kind of correction factors have been used to achieve these results.