To invent the train is to invent the rail accident of derailment.”  —Paul Virilio

When most of us think of train derailments, we think of horrendous events, like the Lac-Mégantic rail disaster in 2013 that destroyed the town in Quebec and killed 47 people.

Fortunately, the vast majority of train derailments associated with the chemical process industries bear absolutely no resemblance to the Lac-Mégantic rail disaster or other rail disasters. Fortunately, most train derailments are not disasters, but inconveniences that occur inside the plant at very low speeds. Fortunately, the consequence of most train derailments is that a crew must set the car back on the tracks.

The Vast Majority

The Federal Railroad Administration (FRA) compiles data on rail accidents. Railway companies are required to report this data and the FRA investigates accidents resulting in, among other things, fatalities, derailment of 15 cars or more, extensive property damage, or accidents “likely to generate considerable public interest.” In 2017, there were 1232 derailments, of which about 30% involved hazardous materials. Of those, 12 derailments involved the release of hazardous materials.

These statistics are compilations of the data that railway companies are required to report. Cars derailed on industry tracks by non-railroad employees are explicitly excluded from any reporting requirements. [49 CFR 225.15(e)(1)] The FRA does not collect data on in-plant derailments.

So how many in-plant derailments are there?  We can only guess. In our years of experience as PHA facilitators, every plant we have ever worked with that receives or ships material by rail has acknowledged in-plant derailments as an occurrence. There are typically one of two estimates that plants give: “Not many—about once a month” or “A lot—about once a month.” A study published in 2014 looked at a single plant that reported 348 derailment incidents over a period of 42 months, or about 2 per week.

There are about 13,500 chemical plants in the U.S. If half of them have in-plant rail, and on average, the U.S. chemical plants in that half each experience one in-plant derailment per month, that suggests that there are about 80,000 in-plant derailments per year in chemical plants, a number that dwarfs the 400 or so hazardous material train derailments reported to the FRA each year.

What Causes Derailments?

There are three basic causes for derailments: track failures, running gear failures, and operating errors.

Track failures include broken rails, gauge spread, buckling, and worn rail. Some of these effects can be compounded by weather—extreme heat, extreme cold, and extended dryness—but each is the natural result of aging. Rails require regular inspection and maintenance. For in-plant track, it is important for the plant maintenance policies and procedures to address track upkeep and to assure that track is properly tended.

Running gear consists of wheel sets, trucks, and suspension. The failure of any of these components can lead to derailment. The care of running gear is well beyond the scope of any plant maintenance program. The best that a plant can hope to do is run their rails as though they know the running gear is defective. That mean slower speeds and more gradual stops and starts.

Operating errors that lead to derailment include overspeed, sudden stops or starts, collisions, and switching errors. It should be no surprise that overspeed on curves can lead to derailment. In addition to the obvious concern of flange-climbing derailment during overspeed, the lateral forces of overspeed on curves can also gradually lead to gauge spread, so that a derailment happens under perfectly normal speeds. As for sudden stops and starts, the front part of the train experiences the forces first, so that cars at the back of the train either overrun the train and potentially derail, or are jerked into motion upon starting, which can lead to derailment.

Switching errors that lead to derailment include putting or leaving a switch in the wrong position and incomplete switching. In the first case, the switch misdirects the train, which then introduces the potential of a collision and derailment. In the second case, the switch is not fully in position, so that when a wheel rides over the switch it “splits the switch” and ends up derailing.

Foreign objects on the tracks can also lead to derailments. While there is no instance of a penny on the track ever causing a derailment, a brick on the track has.

Should We Be Alarmed?

In-plant rail movements happen at slow speeds. In-plant derailments generally don’t result in equipment damage, much less release of hazardous materials. Even on main lines, where the data must be reported to the FRA, over half of derailments of trains carrying hazardous material do not result in damaged trains. Of those that do result in damaged trains, about 10% result in a release of hazardous material (about 5% of all derailments). The portion resulting in release at low speed is much, much less.


Photo credit:  Derailment of train EC 107, by Ludek

Even a miniscule portion of a huge number may still be very large.  If one in a thousand in-plant derailments results in a release, that would amount to 80 releases per year. Yet the literature suggests that a rate of one in a million in-plant derailments resulting in a release is a more accurate estimate. That works out to about once every 12.5 years for the entire industry.

A search of the literature revealed two instances of in-plant derailments that made the news.  One was a chlorine release at a plant in South Carolina in 2005. Nine people died and 72 people were hospitalized. Another incident that made the news was a derailment of sulfuric acid cars in Martinez, California in 2016, where there was no release and no injuries. Otherwise, derailments involving releases of hazardous materials occurred on main lines and are included in the FRA statistics.

Risk is a product of consequence severity and likelihood. While the likelihood of an in-plant derailment is significant, the likelihood of an in-plant derailment resulting in a release is extremely low. Although the consequences of an in-plant derailment with release can be quite severe, the likelihood of an in-plant derailment with release is negligible, meaning that the risk of an in-plant derailment with release is negligible.

Some organizations, however, are driven, not by risk, but by consequence severity. In that case, they will feel compelled to address the problem of in-plant derailments.

Near Misses

Fortunately, almost all in-plant derailments result in nothing more than the need to rerail the car. Every derailment, however, is a near miss, a warning that something is not right. Take the time to investigate the derailment. Was it because of track failure, running gear failure, or error? Is there something that needs to be done? In the portfolio of hazards to worry about, in-plant derailments should be near the bottom of the list. Although, as one plant manager put it, “If your maintenance team is good at putting derailed cars back on the tracks, they are doing it too often.”