Smart Sensing of the RPV Water Level in NPP Severe Accidents Using a GMDH Algorithm

Abstract

The reactor pressure vessel (RPV) water level is critical information for confirming the condition of core cooling in severe accident situations. However, the measured RPV water level signal cannot be trusted during severe accidents due to the unknown integrity of the sensor. In this study, the RPV water level was predicted under severe accident conditions using a group method of data handling (GMDH) algorithm. The prediction model was developed using data obtained from numerical simulations of the optimized power reactor 1000 (OPR1000) using MAAP4 code and validated using independent test data. The developed GMDH model performed very well. In addition, to investigate the effect of uncertainties in the input variables, the model was tested using input data with an artificially added random error. It was accurate enough to predict the RPV water level in severe accident situations when the RPV water level sensor cannot be trusted. Therefore, the developed GMDH model will be helpful for providing effective information for operators in severe accident situations.