Because inhalation toxicity of an airborne contaminant usually depends nonlinearly on the agent density, time-dependent, naturally-occurring density fluctuations in open-air release scenarios have a disproportionately large effect on the time symptoms will onset. A wide range of integrated toxic loads is found for different realizations of the same release and thus the onset of health-effect symptoms occurs over a wide range of exposure durations. This naturally occurring variability makes integrating a model of a representative or average agent density time history of relatively little value to determine the toxicity one can expect. Rather, the statistical distribution of possible toxic load integrals must be taken into account and this is heavily dependent on low-probability realizations that are difficult to obtain in field trials, wind tunnel experiments, and simulations. What real-time users of a predictive model need to know is whether the symptoms of whatever adverse health-effect being considered can occur with unacceptably high probability and what is the earliest possible time this symptom onset can occur. This paper reports a major simplification in computing these toxic agent health effects that accounts for the presence of large fluctuations and the corresponding low-probability scenarios that result from these fluctuations. This new approach requires minimal agent time history data and so identifies the most dangerous, plausible worst-case health-effects for urban scenarios. We will present the distributions of possible toxic load integrals in terms of two time independent quantities, the ensemble maximum density and earliest agent arrival time at each location and we will show how this information gives a simple, fast method to compute the earliest (i.e. most conservative) symptom onset time for real-time contaminant transport software systems such as NRLs CT-Analyst.
展开▼