When considering the possible hazard or nuisance associated with a release of toxic or malodorous gas into the atmosphere, large concentrations are especially important, but relatively little work has been done on measuring or modelling the probability distribution of large concentrations of a contaminant dispersing in a turbulent flow. We have previously applied statistical extreme-value theory to field experiment measurements, analyzing large concentrations at a handful of positions. We have also been involved in using a related moment-based method to give more comprehensive spatial coverage for a steady line source in a wind tunnel. In the latter case, however, we used a method which would be expected to have shortcomings, and we did not calculate confidence intervals for the estimates. In the present paper we use an improved method, with confidence intervals derived from bootstrapping. Again we analyze the measurements from steady line-source wind-tunnel experiments, with particular emphasis on the spatial variation of the estimated maximum possible concentration. We show that the moment-based method agrees well with maximum-likelihood fitting to exceedances of a high threshold. We find that the centreline maximum concentration, normalized by the centreline mean concentration, increases downwind from a value just greater than 1 near the source, to a peak value of 5–6, before decreasing with distance from the source. Across the plume the maximum concentration only varies slowly for the downwind distances in these experiments. These observations are explained in terms of the physical processes of turbulent advection and molecular diffusion.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)