A pseudo non-time-splitting scheme in air quality modeling

Abstract

Air quality standards require better analysis of the impact of emissions from industrial plants and automobiles. The model used for determining the concentration of pollutant consists of a coupled system of diffusion equations which include advection due to the wind velocity, nonlinear chemical reactions of the concentrations, and emission sources. A commonly used method to solve this system numerically in the splitting method: We split each equation into three parts: a diffusion equation, an advection equation, and a system of ordinary differential equations which involves the chemistry. Then we solve in alternate intervals the diffusion system, the advection system and the chemistry system. On February 17, 1994 Pu Sun from North Carolina Supercomputing Center (Research Triangle Park) presented a new computational approach which he calls `a pseudo non-time-splitting scheme.’ He expects that the errors due to the time-splitting scheme may be reduced and that the physical process may be more accurately modeled by this method, while the computational time remains almost the same; alternately, keeping the tolerance level of error fixed, this method should reduce the CPU time. He illustrated his approach with an example of cosine hill rotation with chemical reaction.