GPT error correction technique for shutdown margin optimization calculations

The evaluation of shutdown margin (SDM) for each core loading pattern sampled during a reload optimization is one of the most important constraints which fuel designers must consider. Past studies have shown that SDM constraints can be most efficiently estimated via perturbation theory. However, the accuracy of that approach has been deemed inadequate under a few isolated cases where heavily skewed (e.g., stuck rod) power distributions are involved. This study presents a robust, yet very efficient, error reduction technique which adaptively 'steps in' to refine the accuracy of SDM perturbation theory estimates should their error exceed a preset specification. The additional computational effort required for error correction remains small enough that the overall technique is still 5 et 6 times faster than a standard forward nodal neutronics eigenvalue calculation. (author)