Abstract:
Quantifying the variability in the prediction of atmospheric dispersion code from the Influences of variations in meteorological data is investigated in this study. Historical meteorological data from 2016 to 2020 by the National Centers for Environmental Prediction (NCEP) and the overlap hypothetical accident Loss Of Offsite Power (LOOP) and Large-Break Loss Of Coolant Accident (LBLOCA) are used as initial and boundary conditions. The Fangchenggang nuclear power plant in China, close to Thailand, is considered a study location. The Nuclear Accident Consequence Assessment Code (NACAC) is used as a simulation tool for the investigation process. The NACAC prediction performance is verified by comparing the predicted result with the Java-based Realtime Online DecisiOn Support system (JRODOS). It found that different computational schemes cause variations in dispersion distances of about 200 km and activity concentration of about one order of magnitude. A sensitivity test with various meteorological input data is performed in NACAC to demonstrate the Influences of meteorological characteristic changes on the predicted results. Variations in rain, wind, and atmospheric stability class data affected radionuclides' depletion, dispersion range, and dispersion boundary. The scenario with low rain intensity, low wind speed, and stable atmospheric stability class (F class) causes the highest average radionuclide concentration. The influences of variations in meteorological data on NACAC predicted results are investigated. The high variants of each meteorological data are found in the middle of the year. This variability causes differences in dispersion characteristics and activity concentration for each year. Utilizing five years of meteorological data for simulation yields more comprehensive predicted results than a single year. The high disparity in both predicted results is found at the 50th percentile. The average correlation coefficient of the total effective dose equivalent value over a year in the short, medium, and long dispersion distances of predicted results at the 50th percentile are found at 0.79, 0.81, and 0.66, respectively.
