Hazardous Reading Material

A co-worker of years gone by sent me a link to the ProPublica article “The True Danger of Long Trains.” ProPublica advertises itself as “an independent, nonprofit newsroom that produces investigative journalism with moral force. We dig deep into important issues, shining a light on abuses of power and betrayals of public trust—and we stick with those issues as long as it takes to hold power to account” with a mission “to expose abuses of power and betrayals of the public trust by government, business, and other institutions, using the moral force of investigative journalism to spur reform through the sustained spotlighting of wrongdoing.”

For what it’s worth, I looked forward to reading the article, hoping that, finally, somebody, somewhere had tracked down enough data to make a data-driven case that extra-long main line trains presented increased risks to operating safety, and thus were both a bad operating practice and a bad financial practice.

The article starts out with a dramatic third-person, first-hand account of the horrific August 2017 CSX derailment near Hyndman, Pa., and proceeds to make several dramatic assertions:

1. “Driven by the efficiency goals of precision scheduled railroading, companies are forgoing long-held safety precautions, such as assembling trains to distribute weight and risk or taking the proper time to inspect them, ProPublica found. Instead, their rushed workers are stringing together trains that stretch for 2 or even 3 miles, sometimes without regard for the delicate physics of keeping heavy, often combustible tanker cars from jumping off the tracks.”
2. “Before that morning in Hyndman in August 2017, regulators had already investigated seven long-train accidents in which the length was a culprit, and the nation’s largest rail worker union had sounded alarms about a pattern of problems.”
3. “None of this caused the Federal Railroad Administration, the agency in charge of train safety, to intercede—even as more long trains crashed in the years after the Hyndman derailment, sending cars spilling into other communities.”
4. “To make sense of this gap in information, ProPublica reviewed court and regulatory records of thousands of incidents involving trains of all lengths, as well as technical and investigative notes in federal files from nearly two decades of long-train incidents. We conducted more than 200 interviews, including candid conversations with rail personnel who described how companies have sidestepped best practices when building and running long trains.”

I don’t mind bold assertions. If you got it, flaunt it. Strut, preen, crow, brag, rag—provided you deliver. It’s the lack of delivery in this article that frosts me.

Notice how the ProPublica article locates the “foregoing long-held safety precautions” with the Class I railroads adoption/adaptation of PSR, which was embraced by all the Class I’s except BNSF by 2017. When conducting its investigation, reviewing records of thousands of incidents, the time frame expands to “nearly two decades,” much of which was prior to PSR gaining status as an operating paradigm, before average train lengths exceeded 7,000 feet. Indeed, the trend to longer, heavier trains is persistent in railroad operations, not because of PSR, but because of the greater efficiencies of railroads hauling high-tonnage loads.

The U.S. GAO (General Accountability Office) did a study in 2019 of the increase in train lengths. The GAO reported: “According to officials from all seven Class I freight railroads and representatives from AAR, FRA, STB and other stakeholders we interviewed, freight train length has increased in recent years; however, the data are limited. According to data that two Class I railroads provided to us, their average train length increased over the 10-year period of 2008 through 2017 by about 1,500 feet for one railroad (from about 6,000 to 7,500 feet, or up to about 1.4 miles) and about 1,200 feet for the other railroad (from about 4,900 to 6,100 feet, or up to about 1.2 miles). These data represent an increase in the average length of a train of about 25% for both railroads. Two additional Class I railroads reported average train lengths between about 5,800 and 6,600 feet for the year 2017.”

You know what else was reported between 2008 and 2017? The 14% percent decline in train accidents per million train-miles.

“Regulators had already investigated 7 accidents in which train length was identified as the cause.” What accidents? What investigators? That is not made precisely clear in the rest of the article. The reporters do provide us with more than a dozen links to FRA accident reports.

Let’s take a look at the reports and identify what FRA determined to be the primary cause of each incident:

1. HQ 2005-75, BNSF train collision with Union Pacific train in Missouri: “The accident cause is H221, failure of the BNSF crew to comply with an automatic block signal or interlocking signal displaying a stop indication. It is also determined to be appropriate to indicate the contributing cause of H605, failure to comply with restricted speed in connection with the restrictive indication of a block or interlocking signal.” The “H” in the cause code stands for “human factors.” Indeed, the human crew failed to comply with a restricted speed requirement which led to the violation of the signal displaying “stop.” Length and weight of both trains were not a factor.

2. HQ 2006-67, UP trains collide in Utah: “The probable cause of the accident was the failure of the engineer on Train #1 to stop short of a controlled signal displaying a stop indication. A contributing cause is operation of a locomotive by uncertified/unqualified person.” Train length and weight were not a factor.

3. HQ 2008-25, BNSF tank car rupture in Minnesota: “The probable cause of the incident was the complete failure of the tank car. The failure occurred due to the presence of two pre-existing fatigue cracks that were located at the inboard terminations of the welds joining the extended stub sill cradle pad to the tank. These cracks propagated in a brittle manner completely around the tank, causing it to fracture into two separate halves. This was a result of the dynamic forces imposed by the trailing cars as they ran toward the head end cars, which had been slowed by the air hose separation at the 24th and 25th cars in the train.” Train weight, length and make-up did not cause the accident.

4. HQ 2006-10, Norfolk Southern derailment in Alabama: “The primary cause determined for the derailment of NS Train 23GV509 was the improper use of dynamic brake during running release of automatic brake application, which caused slack run-in of 288,000 pounds, forcing a lightly loaded 89-foot car to derail to the high side of a 5.7-degree curve. A contributing factor was the train make-up. The train make-up recommended in the NS System Timetable Equipment Restrictions, under EQ-9, states that heavier loaded articulated 5-well double-stack equipment should be handled in the head 25% of the consist. The much heavier double-stack equipment in the train was located in the last half, creating buff forces that contributed to the derailment.” Train make-up cited as a contributing factor, not the cause.

5. HQ 2015-1094, BNSF derailment in Wisconsin: “The Federal Railroad Administration’s investigation determined the probable cause was the rapid adjustment of the dynamic brakes, with an additional contributing factor identified as buffing action caused by the train make-up.” Train make-up cited as contributing factor, not the cause. BNSF has not endorsed PSR as an operating policy.

6. HQ 2017-1204, UP derailment, Iowa: “The FRA determined the probable cause of the derailment is cause code H504, buffing or slack action excessive, train make-up. There was no contributing cause code for this accident.” Train make-up cited as the cause. In response to this accident, UP changed its instructions for building trains.

7. HQ 2019-1344, UP derailment, Nevada: “The FRA investigation determined the probable cause of the incident was H504, buffing or slack action excessive, train make-up.” UP made additional changes to train building instructions.

8. HQ 2019-1350, UP derailment, Nevada: “The FRA investigation determined the probable cause of the accident was H504, buffing or slack action excessive, train make-up, the descending grade and track curvature.” Train make-up cited as the primary cause.

9. HQ 2019-1359. UP derailment, Texas: “The FRA determined the probable cause to be H504, buffing or slack action excessive, train make-up.” Train make-up cited as the primary cause.

10. HQ 2019-1363, UP derailment, Illinois: “FRA determined the locomotive engineer’s performance caused the accident.” This accident boggles the mind, or at least my mind. The locomotive engineer cited was essentially moving 183 cars to make up a train without air braking available in the cars. Shoving the cars with the “slack” (the “play” between cars) “bunched,” the locomotive engineer made a full application of the locomotive brake to slow the cars. As the slack “ran out” due to the braking of the locomotive, a coupler broke and one section of the train began to roll free. DS2F (disaster sure to follow).

11. HQ 2019-1368, NS derailment, Georgia: “The FRA determined the probable cause of the derailment to be E7AL, on-board (locomotive) computer failure to respond.” Train make-up cited as a contributing factor.

12. HQ 2019-1388, UP derailment, Idaho: “FRA determined the probable cause to be H504, buffing or slack action excessive, train make-up.” Train make-up cited as the primary cause.

13. HQ 2020-1389, BNSF derailment, Arizona: “FRA has concluded the probable cause of the accident was cause code H519, dynamic brake, too-rapid adjustment. Contributing to the cause of the derailment was a block of 5 empty cars ahead of 6,900 trailing tons, cause code H504, buffing or slack action excessive, train make-up. FRA fatigue analysis found fatigue was present in both the engineer and conductor, which may have had a contributing effect.”

Saying it’s so doesn’t make it so. The ProPublica article asserts what’s supposed to be proven. FRA didn’t intercede? Well, obviously FRA did, convincing UP to alter train make-up instructions. FRA didn’t regulate? It must be established that there is a need for regulation.

So, let’s examine the 13 cases (besides Hyndman) investigated by FRA. First, we get rid of cases 1 and 2, with a suggestion to our ProPublica reporters: Learn the difference between a human factor stop signal violation and in-train dynamic forces. Anyone who tries to pass off a signal violation as product of train length simply has no business “investigating” the business.

Next, we get rid of case 3: Propagation and expansion of previously existing cracks at welds is a defect in the equipment, not in the train make-up.

Next, case 11 goes away, as the cause was determined to be a computer failure. Then there’s case 12: Try as we might, we haven’t been able to eliminate gross stupidity and/or insanity among our operating employees. We try, but somehow, somewhere, sometime, a lunatic will decide he or she can control the movement of 100 or 200 cars with the independent brake.

Train make-up plays a role in 8 cases (plus Hyndman), split 4 and 4 between primary cause and contributing factor. This does not mean there aren’t other cases where train make-up, length and/or weight has directly caused a derailment. It does mean that stretching back to 2005, the ProPublica reporters have provided us with only these 9 examples.

The CSX derailment at Hyndman was caused by attempting to operate the train with 33 handbrakes applied, a procedure authorized by CSX supervision that only confirms insanity is not restricted to the rank and file.

After reviewing “thousands of records and court documents,” the reporters came up with 6 cases. Look, Class I railroads initiate approximately 2,000 main line trains each day (my estimate, and I may be off). If I’m close in that estimation, then railroads have been either extremely lucky, extremely good or both. If I’m off in my estimate—well, either way our intrepid reporters must do a better job.

Truthfully, I’m a bit disappointed. If I had examined thousands of court records and documents and reports, I’m pretty sure I would have unleashed the power of Excel and sorted the data by year, starting with the simple stuff, like this:

1. For each year, the total number of main track derailments, total ton-miles, and total train-miles. Determine the rate of derailments per year, per million ton-miles. Determine, in thousands, tons per train. Compare years. Maybe a trend emerges.
2. For each year, find the mean length and weight of trains. For each year, the median length and weight of trains. Compare years. Maybe a trend emerges. Develop correlation between this trend (2) and the trends in (1) above. Correlation isn’t causation but it’s a start.
3. For each year, find all derailments where buff and draft forces, train make-up, train-handling is listed as the primary cause. Eliminate from the calculation all instances of gross stupidity and/or obvious madness. Compare years. Maybe a trend emerges.
4. Compare trends in (3) above with those determined in (1) and (2) above. Highlight areas where the rate of change in the trend in (3) differs markedly from the rate of change in the trends in (1) and (2) above.
5. For all years, repeat with derailments where buff and draft forces, train make-up and train handling are listed as a contributing factor. Repeat calculations and comparisons.
6. Get granular: Repeat calculations for train weights of 2,000-4,000 tons; 4,000-6,000 tons; 6,000-8,000 tons, 10,000-12,000 tons; and above 12,000 tons.

Now, this isn’t a perfect method, but we’re not after perfection. We’re after a calculus, a probability, a factor of risk. And we’re looking for a rational basis for minimizing, regulating that risk.

Unlike rational analysis, all good stories have a villain, and the ProPublica reporters know how to tell a good story and finger the villain. Here comes old flattop, everybody’s favorite CVO (Chief Villainy Officer): E. Hunter Harrison, who almost single-handedly convinced an entire industry to follow him down the rabbit hole into transportation hell—allegedly.

It’s always convenient to have a villain, but Hunter Harrison? Hell, I’d like to blame Harrison too, because I hate “oversized” trains. I hate them because monster trains paralyze a yard in their arrival, classification, make-up, and departure. I bet if we could get our hands on the data, the classification, assembly, and dispatchment of monster trains correlates with a decline in yard cars dispatched per crew hour, despite the reduction in yard forces.

I hate monster trains because they occupy too many slots for other trains when in transit. Take a single track main of 200 miles with an average speed of 25 mph and 6 passing sidings, each with a capacity of 130 cars. Try dispatching that railroad when 8 trains are moving, 4 in each direction, when only 2 trains moving in each direction can fit into the sidings. Let me know how long it takes you to seek alternate employment.

But Hunter as super-villain? I don’t think so. Like all capitalist enterprises, railroads are driven to reduce costs, doubly-so when business declines. Decline it has, and across more than just the rail industry. If you take a look at the U.S. BEA (Bureau of Economic Analysis)Table, Gross Output by Industry, Chained Indices, and modify the table for annual figures 2014-2022, you’ll find a steep dropoff and failure to recover in “mining: except gas and oil.” Or in railroad business talk, “coal is gone.”

Manufacturing declined as a whole, plateaued during this period before collapsing in 2020, and barely recovered to the 2012 level by 2022. The collapse is particularly severe and sustained through 2022 in the non-durable goods sector.

If you modify the chart again, requesting the data for 2006-2014, you’ll find the “mining except gas and oil,” activity peaked in 2006. Manufacturing as a whole peaked in 2007. Auto and auto parts have been strong, but many other categories are at 90% of the 2012 base.

The “growth industries”? Information, Professional Services, Health Services, Real Estate and Finance—and when they use rail, it’s regional/commuter, not freight. Add to that the fact that imports from Asia have declined for past several months and you, if you’re a railroad, are between the rockiest rocks and the hardest places.

Say what you want about Hunter Harrison, and I don’t recall ever saying anything good. I think Hunter recognized the revenue loss in the decline of coal, tried to pre-empt the effects of that loss, and intensely sought an end-to-end partner for a transcontinental merger. This doesn’t mean he was right or good or a hero or a villain. Nor does it mean we should not regulate the length and weight of trains. It does mean we have to measure what we want to regulate and tailor the regulations to the measure.

I’m not here to sell a story, like ProPublica.

David Schanoes is Principal of Ten90 Solutions LLC, a consulting firm he established upon retiring from MTA Metro-North Railroad in 2008. David began his railroad career in 1972 with the Chicago & North Western, as a brakeman in Chicago. He came to New York in 1977, working for Conrail’s New Jersey Division. David joined Metro-North in 1985. He has spent his entire career in operations, working his way up from brakeman to conductor, block operator, dispatcher, supervisor of train operations, trainmaster, superintendent, and deputy chief of field operations. “Better railroading is 10% planning plus 90% execution,” he says. “It’s simple math. Yet, we also know, or should know, that technology is no substitute for supervision, and supervision that doesn’t utilize technology isn’t going to do the job. That’s not so simple.”

Editor’s Comment: The ProPublica story is yet another example of the mountains of misinformation being fed to the public through sloppy research and reporting conducted by people with little or no knowledge or understanding of railroads. “An independent, nonprofit newsroom that produces investigative journalism with moral force. We dig deep into important issues, shining a light on abuses of power and betrayals of public trust—and we stick with those issues as long as it takes to hold power to account … to expose abuses of power and betrayals of the public trust by government, business, and other institutions, using the moral force of investigative journalism to spur reform through the sustained spotlighting of wrongdoing.”? Yeah, OK. Give me a break. This story derailed, crashed and burned in its first paragraph. And the public believes this malarkey. What are we doing as an industry to counteract it? Thank you, David Schanoes, for taking the time to pick apart ProPublica’s “investigative journalism.” – William C. Vantuono