"Right On Track," Feature Article, June 2007

FEATURE FOCUS: TRANSPORTATION

right on track

A federal drive that looks at issues ranging
from sleep time to hairline cracks is improving the safety record of U.S. rail traffic.

by Harry Hutchinson, Executive Editor

When the Federal Railroad Administration announced earlier this year that rail accidents had declined for the second year in a row, the agency attributed the gains not to luck, but to steps that it has taken to improve the reliability of equipment and of the people who operate it.

Under a program that it calls the National Rail Safety Action Plan, the agency is addressing numerous factors that can contribute to accidents. Some of its solutions are procedural, addressing human behavior and the effects of fatigue. Others involve evolving technology meant to assure that the road itself is safe.

A new track-inspection vehicle operated by the U.S. Federal Railroad Administration, called the T-19, carries laser equipment that monitors rails.

Two of the largest causes of all rail accidents are not completely under the control of railroads. They are collisions at grade crossings and with trespassers on tracks. In the first case, the railroad can control the quality of its signals that warn traffic that a train is coming, but the most common cause of an accident at a crossing is that someone thought he could beat the train. And the railroad has no control over a driver's poor judgment.

In the case of trespass, incidents may arise from a commuter's decision to take a shortcut to the neighborhood station or a feeling of adventure among kids on a midnight stroll. Or just about any other situation in which someone ignores the repeated warnings about how dangerous it is to walk along the railroad tracks. Barring the invention of an impenetrable shield to seal off the railway, the only control the railroads have over trespass is to keep repeating those warnings.

Even so, the overall rate of accidents has been declining over the past few years. There were nearly 19 accidents for every million train miles in 2004. Last year, that rate had fallen to about 16 percent—to 15.98, to be exact.

The statistics include accidents that cover a range of severity. To an outsider, the phrase "train accident" implies destroyed vehicles, and a high probability of crippling injury or death. But spokesmen for the FRA point out that it is not always so.

Accidents are reportable at certain thresholds. For instance, a train accident is any in which there is more than $8,200 damage to track or equipment. On the other hand, any impact with highway equipment, even if damage is minor, must be reported, and so must any kind of accident that results in an injury requiring more than first aid. According to an FRA spokesman, if a single wheel were to derail under a train carrying hazardous materials, the incident must be reported.

When the FRA talks specifically about "train accidents," the term describes incidents that involve only railroad equipment and personnel, in the yards and elsewhere. According to the FRA, the most frequent causes of train accidents are human factors and track problems. Together they account for more than two-thirds of all train accidents. Everything else—including signal and all other kinds of equipment problems—account for less than a third of train accidents. Railroads and regulators have a great measure of control over these events.

The rate for train accidents—those involving only railroad equipment and employees—was 4.38 per million train miles in 2004, the highest in the past 10 years. The rate fell to 4.10 in 2005 and to 3.54 in 2006.

High-Tech Eyes on the Road

The FRA has various technologies to monitor the tracks. In the old days it was the job of the track walker. It was probably great work, at least in mild weather, to stroll through the countryside and take note of signs of wear or weakness in the rails and the supporting roadbed.

Today, the plan is to get the job done better and faster by using accelerometers, lasers, high-resolution cameras, or computer-controlled machinery. And when any of these technologies identifies a questionable piece of track, the global positioning system makes it possible to record the exact location of potential trouble.

There are different classes of track, rated for different speeds. As the speed limit rises, the gauge tolerance gets tighter. The FRA operates several railcars that travel the nation's tracks to monitor gauge and strength of rails in an initiative formally known as the Automated Track Inspection Program.

Two older models use an instrumented drive axle to check the gauge of track and also to apply a lateral load simulating a heavy train. A newer model, known as the T-18, has a deployable fifth axle that allows it to test track at a faster rate.

The cars also carry sensors that monitor ride quality to identify suspect stretches of track.

Jo Strang, associate administrator for safety at the FRA, said the technology was developed by the agency in cooperation with Ensco Inc. Two more track-testing cars from Ensco are being added to the FRA's fleet. Strang said three are assigned to the agency's office of safety and two to the office of R&D.

Kevin Kesler, vice president of Ensco's rail division, said the T-18's deployable fifth axle represents the second generation of the company's Gauge Restraint Measuring System. It can test track at speeds up to 50 miles per hour while it applies a 14,000-pound lateral load. The older technology, using one of the load axles for testing, has a top speed of about 30 mph, he said.

When the gauge measurement system identifies a possible defect, the site is marked by GPS and that information is sent to a database by wireless transmission.

The deployable fifth axle on the T-18 track-inspection car can test the gauge and strength of rails at a rate of 50 mph as it applies a 14,000-pound load. The system was developed by Ensco Inc.

The two newer track testing vehicles, T-19 and T-20, differ from the T-18. They carry a laser system that measures gauge and detects the outline of a cross-section of rail. By comparing the image to an ideal, the system can detect defects that may develop from wear and other surface damage.

Another track inspection technology uses a high-speed line scan camera and is compact enough to be mounted on a hi-rail maintenance vehicle. The camera specifically checks joint bars, the plates that link lengths of rail.

According to Strang, a 5 mm crack can degrade a joint bar as much as 80 percent. A few years ago, in January 2001, there was a serious rail accident in North Dakota that was caused by a faulty joint bar that caused a track to fail, she said.

The National Transportation Safety Board and Congress directed the railroad industry to improve inspection of joints after several serious derailments.

The hi-rail maintenance vehicle is a light truck that has a set of steel wheels in addition to its tires, so that it can run on the highway or on the train tracks. While the vehicle travels at speeds up to 55 mph, the camera takes a picture every half millimeter along each joint bar. It is looking for hairline cracks that usually start at the top of the joint bar.

According to Kesler, the camera's resolution permits the discovery of cracks less than a tenth of an inch long. He added that Ensco is doing research to modify the system so that it can one day confirm the presence and tightness of bolts and inspect other features of bars.

Other track-monitoring technology does not need maintenance vehicles, and uses locomotives instead. Arrays of accelerometers mounted on locomotives detect unusual accelerations from track and report anomalies, which are pinpointed by GPS satellites.

So far, the system has been placed on about 100 locomotives, Kesler said, and many railroad companies have expressed interest. He's sufficiently optimistic about the system to predict that it "will probably be on every locomotive in the next five to ten years."

The track technology promises to let the railroads perform predictive rather than reactive maintenance on their roads. That can save money, and improve both safety and reliability.

"Where we're going," Kesler said, "is that all of your track will talk to you all the time."

Getting Humans to Err Less

Human factors account for almost 40 percent of accidents involving only rail equipment and personnel. The driver of a train passes a stop signal. Someone fails to return a switch to its proper position. Someone forgets to set a hand brake. These are all examples of universal human frailty, and they all have the potential to result in human disaster.

What's more, the FRA has identified a strong likelihood of fatigue playing a role in one out of every four of those incidents.

Researchers looked at the work histories of crews for the 30 days leading up to each of about 1,400 train accidents. The researchers found that employees' estimated level of alertness correlated strongly with their chances of being involved in an accident caused by human factors.

The FRA says its National Rail Safety Action Plan, begun in 2005, is working. After a three-year plateau, the total number of accidents declined slilghtly, about 2.4 percent, in 2005 and dropped by 8.5 percent in 2006.

A bill before Congress, H.R. 1516, the Federal Railroad Safety Accountability and Improvement Act, which is intended to reauthorize the federal rail safety program through 2011, contains provisions that may make it possible for the FRA to set hours-of-service regulations for railroad personnel. Current hours of service, although they are enforced by the FRA, were mandated by federal law.

In a paper released last November under the considerable title of "The Railroad Fatigue Risk Management Program at the Federal Railroad Administration: Past, Present and Future," the FRA discussed the hours-of-service rules and other research into fatigue.

The current rules, mandated by law in 1907 and last updated 30 years ago, say that a train service employee can work a single stretch of no more than 12 hours. After that, the employee must have at least 10 hours off duty. An employee working less than a 12-hour shift must get at least 8 consecutive hours off duty every 24 hours. That means an individual can put in 11 hours and 59 minutes, be off duty for 8 hours, and be called back to work.

Pushed to the limit, the rules don't assure a lot of time for sleep. According to the FRA, a locomotive engineer could operate a train for as many as 432 hours in a single month, or the equivalent of more than 14 hours a day, and still be operating within current federal rules.

The FRA argues that the rules are not based on scientific observation, and in an attempt to correct that, the agency is developing statistical models based on work histories that will predict when fatigue may be nearing a critical point. The FRA has said this work is at least partly inspired by similar research by the Department of Defense to predict the fatigue and alertness levels of troops.

Models are turning up new points to consider. According to the report, it seems that not only the length of the day, but also the time of day counts. The FRA said research has shown that the circadian rhythm, the internal clock that sets the rhythm of the day for most animals, including people, may play a part in human factors accidents. According to the agency, the risk of human factors accidents rises as much 20 percent among crews working between midnight and 3 a.m.

Close, But No Report

Accident statistics for U.S. rail operations don't include the close calls—accidents that almost happened, but were avoided at the last minute by the quick reactions of an individual or, sometimes, by the intervention of blind luck. They leave no record, so there's no telling just how serious a situation was, or how it may be avoided in the future.

To get a fuller idea of the operational safety of American railroads, the FRA wants to hear about the accidents that almost happened, but didn't. Earlier this year, the agency started a program that is expected to bring new data on close calls, in order to give a broader picture of safety conditions that need to be addressed.

Under the FRA's program, called the Confidential Close Call Reporting Pilot Project, Union Pacific Railroad, the Brotherhood of Locomotive Engineers and Trainmen, and the United Transportation Union have agreed to let railroad employees report possibly dangerous situations and near accidents to the Bureau of Transportation Statistics.

Strang pointed out that the program makes it possible for reports to be placed anonymously.

An optical inspection system using a high-speed line scan camera looks for tiny cracks in the bars that join rails. The equipment is compact enough to be mounted on a hi-rail maintenance vehicle.

According to the FRA, incidents can range from careless lifting on the job that puts an employee at risk of minor injury to more serious behavior, such as a train operating in non-signaled territory proceeding beyond its track authority, or a crew member's failure to properly test an air brake.

The pilot project started at Union Pacific's railroad yard in North Platte, Neb., in February and is being extended to other yards, including BNSF Railway in Lincoln, Neb., and Canadian Pacific in Portage, Wis.

Right now, the plan calls for reports to be collected for the next five years to build enough information for thorough analysis. The FRA said a review team will evaluate reports as they come in, so that officials can make safety recommendations for situations that require immediate attention. The FRA said it is talking with some commuter railroads about the possibility of adding another project location.

It seems the National Rail Safety Action Plan is yielding significant results. According to the FRA, the number of derailments in 2006 declined 8.3 percent and collisions between trains decreased by 27.1 percent from 2005.

According to a statement by the U.S. Department of Transportation Secretary, Mary E. Peters, human error was still the leading cause of all train accidents, but human factors accidents declined 20.2 percent in 2006. Train accidents caused by track issues decreased 5.8 percent, and those caused by equipment failure fell by 8.2 percent. Accidents linked to signal problems declined 27 percent.

Of course, a safety action plan or any other program won't be able to eliminate all accidents, not even those whose causes are under the control of railroad operators—not as long as trains remain heavy, can't stop on a dime, and run under the supervision of mere mortals. Application of improved practices and technologies can, however, bring us ever closer to the impossible goal of zero accidents.

The numbers say that, for the past two years, the safety action plan has done exactly that.