WSJ account of GPS used on Rail Roads:
1998
Railroads Test Satellite Positioning
In Effort to Improve Safety, Efficiency
By WILLIAM M. CARLEY
Staff Reporter of THE WALL STREET JOURNAL
As a Union Pacific freight -- three locomotives pulling 83 cars -- rumbled through a foggy night near Seattle, the engineer was startled to see a Burlington Northern freight coming head-on. The BN train had run a red signal.
The Union Pacific engineer hit the brakes, but it was too late. In the head-on collision, locomotives and cars smashed together and burst into flames, killing the UP engineer, his conductor and three Burlington Northern crewmen.
Rules Change for Nation's Railroads, Paving Way for High-Speed Trains
Fast-forward from that Nov. 11, 1993, accident to this June 3 near Seattle, where another train, belonging to what is now Burlington Northern Santa Fe, is operating under the guidance of global-positioning satellites. Engineer Dax Riddle will try to pass Milepost 110, which for test purposes represents a red signal.
Automatic Brakes
"We're trying to go right through," he says, gunning three locomotives and a string of cars to 45 miles an hour. But the global-positioning satellites, linked to computers in the locomotive, sense that the train is nearing a forbidden zone. A warning siren screams in the cab, and when Mr. Riddle doesn't slow down, the brakes go on automatically. The train stops just short of Milepost 110.
"If there had been another train beyond that "red signal," the system would have prevented a collision," says Larry Milhon, a BNSF official riding in the locomotive.
The satellite/computer system for guiding trains, being tested jointly by Burlington Northern Santa Fe Inc. and Union Pacific Corp. for three years in Washington and Oregon, could do more than prevent collisions. An enhanced system pinpointing train locations more precisely than is possible on most of today's rail lines could improve operations and scheduling and thus cut the bottlenecks and delays now plaguing U.S. railroads and their customers. In addition, the railroads could get economic benefits, including lower crew and fuel bills and better use of expensive locomotives, cars and track.
"We could squeeze more through the same pipeline," Mr. Milhon says.
Systems Complex and Costly
That, at least, is the goal. But such systems are complex and costly. Even if further tests go well, these systems may take years -- on some railroads, nearly a decade -- to deploy broadly.
Nevertheless, some railroad executives say satellite-based systems are coming. "Ten years ago, this was a pipe dream, but with the vast improvement in technology, these systems are now within reach," says Matthew Rose, BNSF's senior vice president and chief operating officer.
Other rail officials say BNSF is close to deciding on implementing a satellite-guidance system on a major route where on-time performance is critical. If the carrier goes ahead, perhaps as early as at a July 16 board meeting, it will be inaugurating the industry's first commercial use of such a system.
The system tested by BNSF and Union Pacific was developed by a joint venture of General Electric Co.'s locomotive division and Harris Corp., a Melbourne, Fla., company that makes Air Force electronic-warfare systems. The GE Harris Railway Electronics team, led by GE Chairman John F. Welch, briefed CSX Corp. executives in January. CSX Chairman John Snow was impressed. "These systems have tremendous potential," Mr. Snow says, to enable the railroad to operate more efficiently and improve service.
CSX Studies Begun
Mr. Snow says CSX has begun studies feeding data from two typical CSX routes into GE Harris computers to analyze potential improvements in operations. If results, due in the fourth quarter, are favorable, Mr. Snow says CSX will move ahead. But he cautions that the cost of implementing such systems on a major railroad would run hundreds of millions of dollars. "If they don't work," he notes, "we'd be shooting ourselves in the foot."
Some railroad men, worried about costs and the possibility that computer simulations may be exaggerating economic benefits, are holding back. Norfolk Southern Corp. and the Union Pacific are each installing computerized dispatching as a precursor but will await further studies before implementing a satellite-guidance system. Right now, analysts say, UP is too busy clearing up its widely publicized delays to embark on a costly new system.
A few areas wouldn't benefit from such systems. In corridors heavily used by passenger trains, such as the Boston-New York-Washington line, tracks are already wired and monitored electronically, locating trains accurately and stopping wayward ones automatically. But on some passenger lines and on vast stretches of track mainly used by freights in the South, Midwest and West, trains are governed only by track-side signals, which have no automatic stopping system and give dispatchers only a rough idea-sometimes within only five or 10 miles -- where a train is.
Global-positioning satellites have been tested on a railroad before, from 1987 to 1992, when then-Burlington Northern used a system designed by Rockwell International Corp. to prevent collisions among trains hauling iron ore from Minnesota mines to a Lake Superior port. But amid debate over the system's benefits, the railroad scrapped it.
Another project, sponsored by the Association of American Railroads, the Federal Railroad Administration and the state of Illinois, is just getting under way. It will test a satellite-based guidance system on 123 miles of UP track between Chicago and Springfield, Ill. The job of building the system will be up for bids. In addition to GE Harris, others working on such systems include Rockwell, which is building a system for CSX to test on track from Spartanburg, S.C., to Augusta, Ga., and Harmon Industries Inc., which is testing one for an Amtrak line in Michigan.
To safety officials in Washington, D.C., who have been pushing the satellite systems for years, the sooner railroads implement them, the better. "Half the major accidents we investigate could be prevented by these systems," says Robert Lauby, director of railroad safety at the National Transportation Safety Board. He ticks off collisions he thinks could have been avoided:
On June 22 last year, two Union Pacific freights collided head-on in Texas, killing four. Weeks later, another UP freight on a siding in Kansas moved onto the main line, hitting a passing UP train and killing an engineer.
In 1996, on a CSX line used by both passenger and freight trains, a commuter train operated by a CSX crew rammed an oncoming Amtrak express in Silver Spring, Md., killing 11 passengers and crew.
In 1994, two Southern Pacific trains hit head-on near Marathon, Texas, killing two, and three Burlington Northern trains collided near Thedford, Neb., killing two.
Human Error at Fault
All these accidents, as well as the 1993 UP-BN collision near Seattle, were caused by human mistakes, such as engineers misinterpreting signals or falling asleep or dispatcher errors, Mr. Lauby says. "We need a safety net to catch these human failures, and we've been frustrated" because railroads haven't quickly installed systems to stop trains automatically.
Railroad men contend that safety alone doesn't justify the expense, adding that collisions have declined in recent years and the number of fatalities is relatively small. (Last year, nine people were killed in train collisions; 448 died in grade-crossing accidents.) And Jerry Davis, UP's president, says demands for immediate implementation are unrealistic anyway. "The NTSB thinks we can put this in overnight, like throwing a switch," he says. But just installing the necessary computers and other gear on each of UP's 6,500 locomotives is a major project that would take three years, he adds.
BNSF and CSX may move ahead anyway because of the economic benefits, especially the chance to improve the woefully poor use of equipment. Because of bottlenecks in rail corridors and waits for trains in yards, locomotives sit idling as much as 40% of the time, GE's locomotive-building unit says.
A major bottleneck involves "meets and passes" -- two freights meeting on a single track with one directed off to a siding so the other can pass. One train often waits an hour or more for the other to arrive and go by it. If train locations could be pinpointed more precisely, computers in dispatchers' offices could indicate just how fast each train should go to meet at the siding at about the same time, eliminating a lot of waiting.
Even Union Pacific sees potential benefits. The railroad, says Jeffrey Young, general director of operating systems, has 2,300 "meets and passes" a day. "If we could save just five minutes waiting at each, it would mean huge savings on crews and equipment," he says. Cutting waiting increases a railroad's average speed. Mr. Young figures that for every mile-per-hour rise in average speed, UP would need 200 fewer locomotives, which cost nearly $1.6 million each. (Because of the current extraordinary bottlenecks, UP's average speed in May was only 14 mph.)
Savings in Fuel
Another advantage: GE Harris President Gregory Lucier says that by eliminating "a lot of the hurry-up-and-wait you see now and by moving at more constant speeds, trains could cut fuel use." Mr. Rose, the BNSF operations chief, agrees. He estimates fuel use could drop 3% to 5%, saving $24 million to $40 million on BNSF's $800 million annual fuel bill.
But although the idea of satellite guidance sounds beguilingly simple -- sailors have used satellites to determine their positions for years -- developing a system is more difficult on a railroad. Satellite signals might indicate a position that is off by 100 yards, and railroad-guidance systems must be able to determine on which of two parallel tracks a train is located and thus must be accurate within a few feet.
In the tests on 850 miles of UP and BNSF track in Washington and Oregon, GE Harris equipped locomotives to pick up signals from Coast Guard radio towers, as well as satellites, to improve accuracy. Yet both signals can be blocked when trains are behind mountains or in tunnels. So GE Harris incorporated a third system, an inertial-navigation device similar to those on airliners, which measures every forward, backward or turning motion of a locomotive. Those movements are continuously plotted on a precise map of the rail system stored in the locomotive's computer. GE Harris also included a fourth system, based on an odometer in the locomotive and the computerized map.
In recent tests by BNSF and Union Pacific, this hybrid system has proved "very successful," narrowing a locomotive's position to within the required few feet, says Nicholas Marsh, a BNSF assistant vice president.
But some bugs remain. During tests, one UP locomotive became "lost"; GE Harris engineers found a conflict between instrument outputs. Calculating how long it will take to stop a train before it enters a forbidden zone also is proving difficult. Algorithms must take into account speeds, grades, curves, the weight of shipments, the weather (rain degrades braking capability), and the cars' age (some are 30 years old and don't brake as well as newer ones). A UP official says calculations of safe braking distances can be off as much as 8%. William Matheson, GE Harris technical director, says the calculations are being refined.
The UP-BN Accident
Despite problems, some executives are still enticed by the economic benefits. And there's also the hope of reducing collisions, such as the UP-BN accident near Seattle, which the NTSB says illustrates why satellite systems are needed.
In the fog that night in 1993, visibility was bad. And as the Burlington Northern train headed south, the crew was distracted. The conductor in the locomotive cab radioed Longview Junction Yard to alert the clerk for a friendly wave.
"Hi there, Longview... . Hey, Longview... . Hey, Randy, good night," the BN conductor radioed.
"Hello, did somebody call Longview Junction? Over," the clerk replied.
"Well, hi there... . Don't sneeze, you'll miss me," the conductor said.
Moments later, the Burlington Northern rolled past the yard, with the conductor and the engineer probably looking right to wave to the clerk, according to the NTSB.
Seconds later, the train, traveling 40 mph, passed a yellow signal on the left that the distracted conductor and engineer apparently never saw. The signal, which in the fog may have been visible only six seconds, indicated that the train should slow down to 35 mph and be prepared to stop. But the engineer increased power. Two miles later, he applied the brakes, but by then the train was going too fast to stop for the red signal -- and for a switch that would have diverted the northbound UP train to another track.
Rounding a sharp curve, the UP engineer saw the BN train and braked -- too late to avoid a crash that, miles away, sounded like rolling thunder. "Had there been positive train separation," the NTSB's Mr. Lauby says, "the system would have slowed the BN train at the yellow signal and stopped it at the red, and there never would have been a collision."
Return to top of page | Format for printing
Copyright c 1998 Dow Jones & Company, Inc. All Rights Reserved.
Personal Path
|
