Evaluating the solution of nucleation and condensational growth by a sectional moving grid - remapping method

In this study, a sectional moving grid-remapping method for the numerical solution of condensational growth is evaluated. The main novelty of the method is linked with the remapping step, which is based on splitting the particles over all underlying fixed bins, assuming a linear shape for the bin distribution. Extensive comparison with a high resolution ND (Qamar et al. 2006) and a first order discretization scheme is presented. The method is tested against analytical solutions for pure condensation/evaporation. Further testing is carried out for condensation of sulfuric acid involving a multi-modal distribution. The method is then evaluated for a case of an ideal one-dimensional aerosol reactor, characterized by intense competition of nucleation and growth. The method converges stably and quickly, as the number of bins increases and mitigates numerical diffusion as efficiently as, or even slightly better, than the TVD scheme. The computational cost is comparable to that of the TVD scheme when a Runge-Kutta ODE solver is employed. The number of formed particles is predicted with less than 8 bins/decades. To reproduce all the details of the particle size distribution a higher resolution is needed, mainly due to insufficient density of nodes near the distribution peak. The moving grid-remapping method is independent from convergence conditions. This allows for using a simple explicit time discretization, that reduces the computational cost more than ten times, in the aerosol reactor case. The first order discretization scheme showed extensive numerical diffusion and required more than ten times finer particle size resolution. [GRAPHICS] .