Illustration by Bloomberg View
Illustration by Bloomberg View

The death of four passengers in a train derailment this week in New York City has greatly lessened chances that Congress might postpone a 2015 deadline for railroad companies to install Positive Train Control systems. The derailment appears to have been caused by operator error, which is just what the automated mechanisms are designed to address.

The engineer of the Metro-North commuter train told investigators he was zoned out when the train entered a curve at 82 miles per hour (132 kilometers per hour), almost three times the speed limit. With Positive Train Control, a computer could have sensed the danger and slowed or stopped the train.

Such enhancements will be costly -- the industry estimates a price tag of more than $5 billion, and has asked for more time to install them. But they probably will not improve safety significantly. Operator error plays a role in only 5 percent of train-accident fatalities -- which totaled 696 in 2012. To significantly lower the toll, officials will have to address a much bigger cause: collisions between trains and road vehicles.

Such crashes account for 31 percent of railway fatalities, killing someone in the U.S. every two in three days.

These collisions have decreased over the decades, as the number of railway-highway intersections has diminished, and enhanced safety precautions have been put in place. Since 2009, however, progress has stalled. Preventing more deaths will require better management of train crossings.

Warning drivers isn’t enough; intersections of roads and railways are always marked by an X-shaped crossbuck sign, but many drivers ignore stop signs and even descending gates when they think a train may not really be approaching, studies have shown. Some traffic engineers argue against even placing stop signs at low-use crossings, because they don’t work.

At so-called passive rail crossings, with no flashing lights or descending gates, drivers are expected to listen and look for oncoming trains. At many such crossings, however, drivers don’t have a clear view. In a study of collisions at train crossings from 2001 through 2005, sight obstruction was found to be a condition at the scene of 689 crashes that resulted in 87 deaths and 242 injuries.

States have been slow to adopt laws requiring regular inspections of railroad crossings and removal of obstructions. According to the last survey by the Federal Railroad Administration, 31 states had no regulations at the end of 2009.

Active crossings equipped with flashers and gates could be made safer, too. The biggest factor in crashes at these sites is too-long warning times -- longer than 30 seconds -- which give drivers time to become impatient and drive around the gates. Here, the problem is often outdated technology. With most existing systems, approaching trains activate warnings when they reach a certain distance from the crossing, calculated so that the fastest train on the track gives a 20-second warning. That means the slowest trains may trigger the warnings long before they are visible to waiting drivers.

State-of-the-art systems take speed and distance into account to provide a consistent warning time. However, only about 30 percent of the approximately 65,000 active crossings on public property are equipped with them. These systems are only marginally more expensive than their predecessors, and states can use federal funds to pay for them. The railroad administration should mandate their use, with relevant exceptions -- for instance, in places where trains are accelerating as they approach a crossing.

Without such measures, even if Positive Train Control eliminates many of the spectacular accidents, hundreds of people will continue dying yearly in preventable train collisions.

To contact the Bloomberg View editorial board: