A simplified model of aerosol removal by natural processes in reactor containments

Simplified formulae are developed for estimating the aerosol decontamination that can be achieved by natural processes in the containments of pressurized water reactors and in the drywells of boiling water reactors under severe accident conditions. These simplified formulae were derived by correlation of results of Monte Carlo uncertainty analyses of detailed models of aerosol behavior under accident conditions. Monte Carlo uncertainty analyses of decontamination by natural aerosol processes are reported for 1,000, 2,000, 3,000, and 4,000 MW(th) pressurized water reactors and for 1,500, 2,500, and 3,500 MW(th) boiling water reactors. Uncertainty distributions for the decontamination factors and decontamination coefficients as functions of time were developed in the Monte Carlo analyses by considering uncertainties in aerosol processes, material properties, reactor geometry and severe accident progression. Phenomenological uncertainties examined in this work included uncertainties in aerosol coagulation by gravitational collision, Brownian diffusion, turbulent diffusion and turbulent inertia. Uncertainties in aerosol deposition by gravitational settling, thermophoresis, diffusiophoresis, and turbulent diffusion were examined. Electrostatic charging of aerosol particles in severe accidents is discussed. Such charging could affect both the coagulation and deposition of aerosol particles. Electrostatic effects are not considered in most available models of aerosol behavior during severe accidents and cause uncertainties in predicted natural decontamination processes that could not be taken in to account in this work. Median (50%), 90 and 10% values of the uncertainty distributions for effective decontamination coefficients were correlated with time and reactor thermal power. These correlations constitute a simplified model that can be used to estimate the decontamination by natural aerosol processes at 3 levels of conservatism. Applications of the model are described