A lot of factors influence the magnitude of fluctuations. These include the degree of solar heating or ground cooling, the terrain, presence of buildings and other structures, the height at which the wind is measured, and weather.

Now let us see what happens if we release 1.50 kg/sec of carbon dioxide at ground level for two minutes into the atmosphere. A sensor measuring the carbon dioxide concentration downwind of the release point might yield a time plot as shown above. For these tests, about 90 carbon dioxide sensors were employed at various locations; this is an example. The average concentration as seen by the sensor was about 9000 ppm carbon dioxide (ignoring the initial buildup and tail-off at the end), but because of wind fluctuations and extent of mixing into the air, the instantaneous concentration was as high as 16000 ppm.

In addition to small-scale fluctuations in concentration, the toxic cloud can meander in and out of the centerline location. In the example shown above, the cloud centerline meandered away from the sensor resulting in lower concentrations towards the end of the two-minute release period. Also, the duration of the cloud as it passed over the sensor was somewhat longer than two minutes. The cloud tends to spread out laterally, horizontally, and vertically as it travels downwind.

The gas dispersion models in the public domain including the models in the PEAC tool predict average concentrations and do not predict spikes or peak concentrations. Peak concentrations are of most concern when dealing with highly toxic chemicals where one or two breaths may be fatal or incapacitating. Meander can be corrected by assuming that the highest concentration passes over the receptor, e.g., the worst case. The PEAC tool assumes that the toxic cloud has “meandered” such that the highest concentration is at the protective action distance downwind of the release. However the models in the PEAC tool (and also ALOHA, SLAB, and other models) do not consider fluctuations, where “slugs of toxic chemical” might pass over the receptor. Some models ask the user to specify a term called the “concentration averaging time”; if the user specifies a short averaging time (e.g. one minute), the model will predict higher concentrations at the cloud centerline. A long time (e.g. 1 hour) specified would predict a lower concentration because of normal meander. However concentration spikes occurring because of localized wind shifts and eddies formation are not predicted.

What Happens if the Winds Die Down Completely?

A test at the DOE Haz Mat Spill Center (part of the Kit Fox series of tests performed during August and September 1995) fit this weather condition. Carbon dioxide was released at ground level at a constant rate for a period of 6 minutes under almost nighttime conditions and clear skies. The winds, which were about 1 meter per second at the start of the test, essentially quit at the end of the test. The carbon dioxide pooled over the ground and remained there during the night. But there were still considerable concentration fluctuations as the air slouched around the measuring sensors.

Chlorine was released through windows from a building at Springfield MA one June evening about 13 years ago as a result of an accident. Weather conditions were overcast, with no wind at all. Residences in the area within a few miles of the release site reported chlorine odors during the night, and evacuations took place. There was no correlation between elevation or direction from the source. The chlorine gas apparently skipped around with no apparent pattern. Chlorine odors were present at some locations far from the site and absent at closer locations.

The Urban Environment

Most models including the models in the PEAC tool allow the user to either input a surface roughness length or select between choices of (1) flat, open terrain, (2) cropland or light residential, or (3) urban or forest conditions. The effect of buildings and other structures is to help break up and disperse the toxic chemical cloud as it travels downwind. Therefore the toxic cloud is wider, higher, but less concentrated at the centerline than if it passed over a flat surface.

Plywood structures simulating buildings were placed in the path of the plume cloud for the carbon dioxide release tests as part of Kit Fox (August-September 1995 tests in Nevada). The tests verified that the plume cloud was taller and wider compared with flat terrain because the structures tended to break up the cloud. But the tests also demonstrated many anomalies. These included (1) parts of the carbon dioxide cloud traveling upwind, (2) a taller carbon dioxide cloud than predicted by models, (3) sometimes higher concentrations measured at several meters above the ground than at the ground surface, and (4) a long time to scour out the residual carbon dioxide from the structures after the cloud passed. Some of these anomalies could also be demonstrated in a wind tunnel, but the wind tunnel could never simulate the large-scale tests outdoors under stable, nighttime conditions (the so-called “F” stability).

Gas dispersion models predict average conditions. They do not predict peak concentrations because of wind fluctuations or anomalies because of wind patterns around buildings. Some anomalies are as follows: