An Experimental Field Dataset with Buoyant, Neutral, and Dense Gas Atmospheric Releases and Model Comparisons in Low–Wind Speed (Diffusion) Conditions

Abstract

Abstract A unique field dataset from a series of low–wind speed experiments, modeling efforts using three commonly used models to replicate these releases, and statistical analysis of how well these models were able to predict the plume concentrations is presented. The experiment was designed to generate a dataset to describe the behavior of gaseous plumes under low-wind conditions and the ability of current, commonly used models to predict these movements. The dataset documents the release and transport of three gases: ammonia (buoyant), ethylene (neutral), and propylene (dense) in low–wind speed (diffusion) conditions. Release rates ranged from 1 to 20 kg h−1. Ammonia and ethylene had five 5-min releases each to represent puff releases and five 20-min releases each to represent plume releases. Propylene had five 5-min puffs, six 20-min plumes, and a single 30-min plume. Thirty-two separate releases ranging from 6 to 47 min were conducted, of which only 30 releases generated useful data. The data collected included release rates, atmospheric concentrations to 100 m from the release point, and local meteorological conditions. The diagnostics included nine meteorological stations on 100-m centers and 36 photoionization detectors in a radial pattern. Three current state-of-the-practice models, Aerial Locations of Hazardous Atmospheres (ALOHA), Emergency Prediction Information code (EPIcode), and Second-Order Closure Integrated Puff (SCIPUFF), were used to try to duplicate the measured field results. Low wind speeds are difficult to model, and all of the models had difficulty replicating the field measurements. However, the work does show that these models, if used correctly, are conservative (overpredict concentrations) and can be used for safety and emergency planning.