Parallel Numerical Simulation of Reacting Flows in Air Quality Models

Abstract

Air quality models (AQMs) play a fundamental role in forecasting air pollution phenomena and are an important tool in the development of control strategies for reducing the damaging effects of heavy concentrations of pollutants. The space and time scales for an effective simulation of air pollution problems on urban, regional, or global domains, and the wide range of chemical–physical phenomena included in AQMs require very powerful computational resources. Therefore, the development of efficient parallel software for AQMs is an active research field. The choice of the numerical solution method for the problem is driven by several considerations. The most widely used computational approach for the solution to the problem is based on splitting techniques. Two basic splitting approaches can be identified—that is, physical and dimensional splitting. The vertical turbulent diffusion and chemical kinetics are usually coupled because they have similar time scales, especially for highly turbulent wind fields. In a parallel distributed-memory environment, the dimensional splitting does not seem to be very effective, in terms of efficiency and scalability.