East Palestine Train Derailment and Modeling
A little before 9:00 PM on February 3, 2023, a Norfolk Southern freight train derailed in East Palestine, Ohio, which on the Ohio/Pennsylvania border about half-way between Cleveland and Pittsburgh. Roughly three dozen of the 141 cars derailed, and of those that derailed 11 had hazardous material on them. While nobody was injured in the derailment the resulting fire burned for several days and nearby residents and businesses were evacuated.
The United States Environmental Protection Agency (EPA) started conducting air monitoring that night, and over the next few days residents as far as 20 miles to the north reported chemical smells.
As the railcars burned, Norfolk Southern became concerned about the possibility of an explosion that could send shrapnel up to a mile away. As a result, on February 6, they punctured small holes in the five railcars containing vinyl chloride, let the contents drain into a ditch, and then lit the ditch on fire to burn off the hazardous chemical. While this controlled release and burn eliminated the concern of a catastrophic explosion, it also produced a mammoth cloud of black smoke.
By February 9 EPA reported that air quality both outdoors and indoors was at normal levels, and the evacuation order was lifted. Since that time there have been reports of sick children as well as ill or dead pets/wild animals. On February 15 there was an emotional town hall meeting in which many local residents were upset about everything from how this had happened to the response of both Norfolk Southern and state and federal officials. Norfolk Southern choose not to attend the meeting, citing a “growing physical threat to employees and members of the community.”
I’m not here to opine on who was right and wrong, or what the environmental impacts to the community may or may not be. Those are very complicated issues and I certainly have not delved into them deeply enough to render an informed opinion.
Rather, what I’d like to talk about is how as soon as I first saw this on the news I immediately thought “air dispersion modeling will play a key role here.” As lawsuits come about both plaintiffs and defendants will be establishing how air quality in the area was impacted from this accident—and that is best done with air dispersion modeling.
When conducting air dispersion modeling the first thing to decide is the appropriate model to use. That decision is based on a number of factors, including the source being modeled. In this case the source to be modeled is basically a series of fires—some at discrete points (e.g., individual railcars), some over long areas (e.g., the controlled release and burn of the vinyl chloride). The workhorse model of the modeling community, AERMOD, isn’t designed for those kinds of sources. For these situations you need an emergency release model such as SLAB, ALOHA, or DEGADIS.
Another important consideration is how to characterize the source. Modeling of this situation is much more complicated than a simple power plant stack; while the power plant stack doesn’t move or change its source characteristics very much, the sources in this case did change location and characteristics over time (i.e., different railcars caught fire at different times, burned more intensely (hotter) than other times, etc.). Also, the calculation of pollutant emission rates is challenging as you have to determine which cars had which chemicals, which cars released its chemical emissions at what times, how the chemicals may have interacted with each other in the atmosphere to possibly produce other chemicals, etc.
Finally, using representative meteorological data is very important. The Beaver County Airport is about 8 miles to the southeast of East Palestine, but it is about 200 ft higher in elevation with several ridgelines/valleys between the site of the accident and the airport. I checked a network of amateur weather observers that I participate in and found a station only about 3 miles from the accident site, to the northwest—but care has to be used when using data from “backyard” weather stations as they may not be reliable.
So as you can see, while air dispersion modeling is exactly the tool that will need to be used in analyzing air quality impacts from this accident, this isn’t going to be a straightforward modeling analysis. Great care will need to be taken in executing this analysis to produce reliable—and defensible—results.