CFD-based sectional modeling of aerosol dynamics: particle formation in an aerosol reactor

In many current applications computer modeling of aerosol dynamics needs often to be associated with multidimensional CFD calculations. The sectional approach can offer high accuracy and general applicability because it may describe complex and arbitrary forms of the particle size distribution. The basic disadvantage of the sectional approach is its heavy computational cost, especially when implemented in a Eulerian computational framework. Accordingly, alternative methods, like modal methods, are often used in multidimensional CFD-based aerosol flows calculations.

The numerical solution of aerosol dynamics is challenging in cases where the evolution of the particle size distribution is dominated by a strong competition between homogeneous nucleation and condensational growth. The vapor phase change in this case is characterized by extremely shorter time scales in comparison with the other processes like transport. The implementation of the sectional method in a Eulerian CFD framework is problematic when standard CFD techniques are used (Pyykonen & Jokiniemi 2000).

The numerical solution of aerosol dynamics is challenging in cases where the evolution of the particle size distribution is dominated by a strong competition between homogeneous nucleation and condensational growth. The vapor phase change in this case is characterized by extremely shorter time scales in comparison with the other processes like transport. The implementation of the sectional method in a Eulerian CFD framework is problematic when standard CFD techniques are used (Pyykonen & Jokiniemi 2000).

The methodology was assessed by simulating laminar flow aerosol reactor experimental data (Nguyen et al. 1987). The calculated bulk average particle size distribution at the reactor outlet for various particle size grid resolutions is shown in figure 1. The nucleation rate from the classical nucleation theory was used, multiplied by a correction factor C , as commonly made in this type of simulations. As can be seen numerical diffusion is efficiently combated even with 10 size bins/decade. Satisfactory accuracy is achieved with 20 size bins/decade. The required CPU times were 70 min and 160 min respectively, with a common PC. We conclude that the sectional method can be efficiently used in multidimensional CFD-based aerosol simulations, by employing the proposed methodology.