Using Large-Eddy Simulation and Wind-Tunnel Data to Investigate Peak-to-Mean Concentration Ratios in an Urban Environment

Abstract

The estimation of short-term-averaged maximum concentration is of foremost importance, for instance, for the impact assessment of odorant sources, flammable gases and the accidental or intentional release of toxic gases. As dispersion models only give 1-h averaged concentration values, a simple formulation (the power-law function) has been widely used in practical applications to overcome this limitation. The present study investigates the potential of large-eddy simulation (LES) to assess the influence of turbulent eddies on averaged concentration over short time intervals and, thus, on dispersion within a building array, with LES results compared with wind-tunnel data. The results indicate that the LES approach underpredicts the concentration fluctuation intensities governed by the smaller eddy motions and we conclude, not surprisingly, that the particular choice of subgrid-scale model and grid size is important in describing the smallest wavelength concentration motions. However, even though the LES results are not able to predict peak-to-mean values for very short averaging times, the fit of the power-law function can be extrapolated to produce a valid relation for shorter averaging times, implying the LES technique can be used to assess the p value (the exponent) in the commonly-used power-law function. This is found to be smaller (by about one half) for sensors in the central position within the array than for those located in short streets or at intersections, and it also decreases more slowly with distance from the source. No substantial difference is found between sensors located at the canopy height H and at half the canopy height, i.e. within the canopy. In contrast, there is a significant difference for sensors located above the building height at 1.5H.