STUDY OF NUMERICAL DIFFUSION IN A DISCRETE-SECTIONAL MODEL AND ITS APPLICATION TO AEROSOL DYNAMICS SIMULATION

The discrete-sectional model is an important tool used to study aeroso l dynamics, although it suffers inherent numerical diffusion errors th at are dependent on the model parameters including section spacing, nu merically conserved aerosol property, and number of discrete sizes. An analytical evaluation of the errors due to the deviations of the deri ved aerosol properties in the sectional formulation and the interchang e between discrete-size and sectional aerosols is carried out. The err ors are larger if the difference in the aerosol property function inde x (xi) between the derived and conserved properties is larger. Results of parametric studies are reported for condensation and coagulation s ystems. The results for condensation systems show that the v-model is better than the n-model and the upsilon(2)-model in predicting the int egral properties of the size distribution. However, the deviation in t he results between the n-model, upsilon-model, and upsilon(2)-model re duces as the section spacing is decreased. A finite value of the geome tric standard deviation (sigma(g)) is obtained for a specific section spacing and is independent of the chosen conserved aerosol property. T he results for coagulation systems also show that the upsilon-model is better in predicting N, V, and upsilon(g), although the upsilon(2)-mo del is better in predicting the second volume moment, V2, and sigma(g) . The inclusion of a larger number of discrete sizes gives more precis e description of aerosol dynamics for the molecular clusters, although the effect is rather insignificant for the number greater than 20. It is also found that the discrete formulation possesses numerical diffu sion in simulating a condensation process. (C) 1998 American Associati on for Aerosol Research.