Pollutant Dispersion Simulation in Street Canyons Using the Finite Element Method and Shared Memory Parallelization

Abstract

Air pollution generated by vehicles with internal combustion engines is one of the biggest problems the large cities are facing today. Therefore, the development of numerical techniques for evaluating and controlling the levels of pollutants is essential to maintain the urban environment balance. In the central regions of large cities, the so-called street canyons represent the basic geometric unit, where significant changes can be observed in the wind flow and pollutant dispersion as function of thermal conditions and geometrical configuration of the canyons. In this way, a numerical model based on CFD techniques is proposed in this work to simulate incompressible flows considering heat and mass transfer phenomena. In the present model, an explicit two-step Taylor-Galerkin scheme is adopted where the spatial discretization is performed using the finite element method (FEM) with eight-node hexahedral elements, one-point quadrature and hourglass control techniques. The pressure field is explicitly obtained by using the pseudo-compressibility hypothesis and the velocity and temperature fields are coupled by buoyancy forces according to the Boussinesq approximation. Large eddy simulation (LES) is utilized to analyze turbulent flows, where the sub-grid scales are modeled using both, the classical Smagorinsky`s model and the dynamic model. Programming techniques for shared memory parallelization are also utilized in order to improve the performance of the present numerical code. Pollutant dispersion problems are simulated considering different geometrical and thermal configurations in order to adequately characterize the conditions encountered in urban areas.