Study On Radionuclide Variable Wind Direction Atmospheric Diffusion And Source Term Inversion Model Based On Mixed Layer Modification

Despite extensive protective measures implemented in nuclear facilities,the risks of nuclear accidents and radionuclide releases still exist.Rapid and accurate assessment of the radionuclide diffusion consequences and accident source term is of significant importance for understanding accident progression and formulating emergency measures.In view of the complex meteorological conditions characterized by temperature inversions and variable wind directions,the challenges in conducting realtime voxel model dose assessments,and the inaccuracies in source term inversion based on limited monitoring data,this paper carries out research on radionuclide variable wind direction atmospheric diffusion and source term inversion model based on mixed layer modification.Furthermore,verification studies are conducted through the application of typical scenarios.(1)In response to the complex meteorological conditions of "temperature inversions" and variable wind directions in urban areas,we develop a radionuclide variable wind direction atmospheric diffusion model based on mixed layer modification.This model modifies the inhibitory effect of temperature inversions on the radionuclide vertical diffusion,the impact of variable wind direction on the radionuclide horizontal diffusion,and the attenuation effect during the radionuclide diffusion.Compared to non-modification models,the statistical indicators of the modified model are closer to the observed data,meeting the recommended standards.The performance of the modified model is superior to AERMOD model,demonstrating excellent overall predictive capability.(2)To improve the accuracy and real-time for public dose from radionuclide diffusion,we conduct research on high-precision real-time dose calculation method for human voxel model.We develop a voxel model dose assessment method based on GPU acceleration,which enables real-time dose assessment for the high-precision human voxel model Rad-HUMAN composed of millions of voxels.This method provides realtime interactive dose calculation support for multiple high-precision voxel models,addressing the challenge of rapid and accurate assessment of public dose from radionuclide diffusion in urban areas.(3)To address the problems of inaccurate source term inversion based on limited monitoring data,.we develop a source strength inversion method based on single monitoring point data and a radiation field reconstruction method based on inverse distance weighting.We establish a multidimensional source term inversion model for nuclear accidents based on monitoring data,matching the gradient correlation as the optimal similarity measure for source term inversion models,and develop an optimization algorithm based on adaptive search space evolution strategy.The above methods realize rapid source strength inversion and radiation field reconstruction,improving the optimization performance of the multidimensional source term inversion model.Based on the modified radionuclide atmospheric diffusion model and source term inversion model,we conduct validation studies on typical scenarios.To verify the universality of the models in nuclear energy systems and nuclear technology application facilities,we select traditional nuclear reactor types and fourth-generation advanced nuclear reactor types,conducting studies on the radionuclide diffusion of the Fukushima nuclear power plant accident and the hypothetical accident of China leadbased reactor(CLEAR-I).We also conduct studies on source term inversion models for accidents in a fusion deuterium-tritium neutron source(HINEG-I).The simulation results demonstrate that the positions and times corresponding to typical airborne radionuclide concentrations are consistent with the observed data.Inversion temperature play a significant role in suppressing the atmospheric dispersion of radionuclide.By the wind direction modification,the dose in the model is lower than that in the non-modified model,aligning with the model.The error of the inversion results is within 1.1%to 5.0%,satisfying the requirements for radiation protection and emergency decision-making support.