Prediction of Reactor Core Power Distribution Based on Non-intrusive Reduced-Order Model

In order to perform fast calculation of the reactor core power distribution in support of online monitoring of the core status and/or the development of nuclear reactor digital twin system, a non-intrusive reduced-order modeling method was investigated. This method performs a POD process over reference snapshots of the core power distributions computed by high-order transport model, and obtains the POD expansion coefficients by projecting the snapshot solutions onto the resulting POD orthogonal basis. The prediction of the power distribution for previously unknown core status is computed by interpolating the expansion coefficients over the reference core status points consisting the snapshots. The efficacy of this method was examined with various core status metrics, including core temperature, burn-up, and control rod depth. The results show satisfactory consistency between the fast prediction and transport reference solutions. In conclusion, the method proposed proves to be capable of performing fast calculation of the core power distribution with reasonable accuracy.