No one had intended to make railroad history on May 5, 1998. It’s just that there was a shortage of locomotives in Phippsburg, CO. Instead of the usual five locomotives, only four were available to pull a 108-car coal train up Union Pacific Railroad’s steep Toponas grade on the western slope of the Rocky Mountains. What followed is, among locomotive builders, legendary.
The locomotives were brand new General Electric behemoths with a twist: their traction motors operated on alternating current rather than direct current. Climbing the Toponas grade that day, the trains slowed to a barely perceptible six meters per minute. No self-respecting engineer would have tried such a foolhardy trick with conventional direct-current motors: wheels would have slipped, the train would have stalled, and the motors themselves would have been fried like an egg. But none of those things happened. Indeed, later investigation showed that the locomotives had been producing more pulling power than was thought possible at that speed. This feat of strength initiated a radical transformation of railroading-a revolution that stems directly from advances in information technology.
Technologically speaking, it is difficult to find anything in railroading that has not changed in the last decade. Dozens of microprocessors in today’s diesel locomotives run almost all of their systems, from fuel feed to cab air conditioning. Pole lines that once flashed past the windows of speeding passenger trains are disappearing in favor of microwave or fiber-optic communications. Experimental new dispatching and control systems may soon tell engineers if they are using the most fuel-efficient throttle settings.
Say “trains” to most people and they think about the passenger variety. But in the United States, the railroads with the greatest economic impact are those that transport cargo. Railroads haul 25 percent of U.S. freight. They are easily the most efficient way to move coal, grain and bulk chemicals. But the railroad companies have long had a sort of love-hate relationship with cutting-edge technology. They only abandoned coal-fired steam locomotives, for example, when General Motors developed the diesel-electric engine and gave demonstrations to railroads around the country in the 1940s. And even then, many railroads stuck with steam for years.
However, over the last decade, railroads have been engaged in their own version of an information revolution. The combination of computers and wireless systems gives railroads greater customer service capacity and better dispatching and cost controls-as well as dispensing with armies of clerks. Charles Dettmann, executive vice president for operations, research and technology at the Washington, DC-based Association of American Railroads, argues that railroads’ competitiveness-perhaps even their existence-depends on their use of information technologies.
“Railroad companies are a very hard sell. I am usually in the position of pushing them farther than they want to go,” says Carl D. Martland, senior research associate at MIT’s Center for Transportation Studies and a consultant to the railroad industry. “They insist on knowing there’s going to be a productivity benefit and will only go as far as that benefit takes them. They have done a very good job of saying, Does this technology do me any good?’”
Just two decades ago, eastern Wyoming’s Powder River basin was a barren, treeless outback with few people and no industry. But the region had something that as of the early 1990s the United States suddenly needed: lots of low-sulfur, relatively clean-burning coal. In fact, a thick seam of coal stretches under the eastern third of Wyoming. And the only practical way to move so much coal out of the remote Powder River basin is by rail.
The two railroads that serve the area-Union Pacific and Burlington Northern Santa Fe-have spent more than $5 billion to build the biggest, most modern industrial rail system in the country. Driven by tightening air pollution regulations, demand for low-sulfur coal is now booming beyond anyone’s wildest dreams; one section of the line has become the first stretch of rail in history to support more than one billion kilograms a day.
And because the railroad and the mines are new, the Powder River operation provides a clean slate for the creation of the most efficient operation possible-without the burden of older infrastructure and the outdated technology that railroads have often kept running. Nowhere to be seen are the workers who once laboriously copied down all car numbers and faxed them to headquarters. As each empty train enters the mine, and as each loaded train leaves, scanners read automatic-identification tags, recording each car number and reporting the data to Union Pacific’s Harriman Dispatching Center in Omaha, NE.
The Harriman Center is the heart of an ambitious effort to direct an entire railroad system from a central location-to dispatch trains using a computer program that chooses the points at which they meet or pass. The Harriman system controls traffic on more than 27,000 kilometers of Union Pacific track in 23 states-though human dispatchers can intervene at any time if they disagree with the computer’s choices-and it allows for the coordination, days ahead of time, of movements over the entire railroad rather than on a single line or division.
Empty trains enter the Powder River basin’s coal-containing silo under the railroad equivalent of cruise control. Trains creep in at around 1.5 kilometers per hour, speeds that only the most skillful engineer could match by hand. Computerized loading chutes fill each car with the planned weight of coal-100,000 kilograms, accurate to within about .2 percent. A train can be filled with coal in 45 minutes, or about twice as fast as previous automatic loaders could manage.
As coal trains pull out of the Powder River fields, the locomotives constantly “talk” to Union Pacific headquarters in Omaha. The stream of data gives a running narrative of the train’s condition, as reported by an array of sensors that monitor, for instance, the oil pressure, operating temperature, horsepower output and the rate of fuel usage. In the old days (say, the early 1990s), engineers knew that something was amiss with a locomotive only when it was already in serious trouble. That’s when alarm bells would ring, or the engine would suddenly shut down or start smoking. Union Pacific is outfitting its entire fleet with onboard computers that constantly track the locomotives’ location and health, then report this information to a maintenance desk at headquarters.
Once the fleet is equipped, a given locomotive will signal the Omaha center that it has a problem long before it tells the engineer. The sensors should usually catch problems hundreds or thousands of kilometers before they become severe enough for the engineer to care. Information that an engine is using 15 percent more fuel than normal, for instance, is of little concern to the engineer but of great interest to the maintenance technicians monitoring the locomotive.
Installing computers on locomotives is not exactly like putting them in the controlled environment of an office. Dirt, vibration and extremes of hot and cold are part of everyday railroad operation. Union Pacific experimented for months with various types of shock mountings and vibration-controlling material. According to chief technology officer Lyden Tennison, lessons were drawn from another enterprise that knows a thing or two about adapting high-tech equipment for inhospitable conditions. “We learned a lot from the military,” he says. Locomotive technicians were at first amused, for instance, to learn that the military kept processors plugged into their sockets under constant vibration by tying them down with dental floss. Amused, but impressed: Union Pacific adopted this solution.
Throughout the diesel age, locomotives worked according to a simple principle: a diesel engine turned a generator that produced alternating electrical current, which was then converted to direct current to run the traction motors that drove the axles. The leap forward that made possible that pull up the Toponas grade depended on a fundamental shift in technology during the 1990s from DC motors to AC motors. This change has been enabled by the availability of fast, inexpensive microprocessors.
Power for both a DC locomotive and an AC locomotive starts its path to the wheels in the same way. In both types, a diesel engine turns a generator that produces AC power, which is then converted to DC. (The starting AC power, at a constant 60 cycles per second, could run the locomotive at only one speed.) Here, though, the technologies diverge. In a DC locomotive, the DC power goes directly to motors that turn the wheels. In an AC motor, the direct current passes through a series of computer-controlled components called inverters, which “chop” the DC power into AC power. This AC is in turn fed to the motors.
Computer chips make AC motors practical by regulating the flow of power with a precision impossible by any other means. The chips monitor and control the DC entering the inverters and make sure that they deliver the proper amount of AC to the traction motors. This is no small feat: each inverter may require as many as 500 on-off commands per second to regulate the AC flow. And while 500 commands per second may seem unimpressive in a day of gigahertz chips, the proper comparison is not with other computers but with human beings. Imagine a train engineer trying to make 500 changes in throttle position every second.
AC motors are more robust than their DC cousins. They’ve been put through brutal tests that demanded maximum possible power production, sometimes for days on end. Those tests went far beyond anything the worst railroad environment could produce, and the motors never came close to overheating, according to Michael E. Iden, Union Pacific’s general director of car and locomotive engineering. As long as the equipment is operating properly, AC motors “really should never burn out,” Iden says. Many railroads are even using AC locomotive power-instead of air brakes-to hold trains stationary on heavy grades, Iden says. This technique, which avoids the time-consuming process of pumping off air brakes, would fry a DC motor in minutes.
Beyond their ability to pull heavier loads, AC motors improve overall efficiency. Each locomotive wheel makes contact with an area of rail no larger than a nickel. The percentage of weight on that wheel that is converted into pulling power is called “adhesion.” While the best DC motors can muster an adhesion of about 30 percent, AC locomotives take advantage of precise computer control of the traction motors to achieve adhesion averaging 34 to 38 percent; each percentage point gain in adhesion provides the pulling power for five additional fully loaded coal cars.
Trains must run on tracks, of course. And once laid, the rail and ties must be maintained and inspected. Information technology is playing a transforming role in this traditionally labor-intensive affair. The last two or three years, for instance, have seen the advent of rail alignment systems that use lasers to gauge distance and direction. Computers then figure a track’s correct curvature and angle of elevation and feed the information to machines that put the rail and ties into place. “The important thing is the ability to measure track geometry rapidly, without depending on human sight,” says Louis Cerny, an independent railroad consultant in Gaithersburg, MD.
One particularly time-consuming rail maintenance job-spreading rock ballast between tracks-is also getting a shot of adrenaline. In June, Herzog Contracting-a railroad construction company based in St. Joseph, MO-delivered a new ballast train to Union Pacific. Unloading 60 cars of ballast normally takes at least two days; Herzog’s train does the job in 30 minutes. As the train chugs along, computers guided by global positioning system satellites decide which car doors to open and how much ballast to pump out (even interrupting the flow at road crossings).
Similar advances are aiding track inspection. This job was once the domain of a lone trackwalker, carrying a few heavy tools, who walked along the track to see if it was shifting, or if spikes were pulling out or rail joints flexing too much. The ultimate in automated track inspection is a system delivered in 1999 to the Federal Railroad Administration by Plasser American, a maker of inspection cars, and Ensco, a manufacturer of railroad inspection hardware and software. This self-propelled mass of sensors and computers, rolling along at up to 145 kilometers per hour, generates readouts of track condition and dispatches crews to the locations of any problems. Most of the major freight railroads in the United States are either using such cars now or have ordered them.
Ensco has also developed remote monitoring systems that can be fitted to any railcar or locomotive. The systems, now in service for Amtrak and several commuter railroads, continually assess track anomalies, ride quality and a locomotive’s mechanical health. When a problem appears, the monitors send an alarm via satellite or terrestrial wireless link. Detailed information on the problem and its exact location can then be accessed through the Internet. Other new inspection equipment uses computerized vision to look for defects in air brake hoses between cars. Pulsing lasers, fanning out in a pie-slice shape, can accurately produce an image of the wheel as it rolls-registering surface defects better than an experienced inspector can when the wheel is standing still. All of these detectors are designed to report trouble spots to the train crew or the dispatcher before a small problem grows and causes a wreck.
Down the Line
With the cost of technology constantly falling, railroads may be poised for another round of automation. The first candidate is an idea railroads have so far shunned called positive train control. The computers that control a locomotive’s throttle and brake would be equipped with global positioning system receivers that tell them precisely where they are and how fast they’re going. The modification was originally proposed as a safety enhancement, to prevent collisions: if an engineer sped past a stop signal, the system would signal the computer to slow or stop the train. That application failed to win over the railroads, though. “It would have cost a lot of money for a minimal safety improvement and so wasn’t cost effective,” explains MIT’s Martland.
But many railroad officials are beginning to understand the business case for positive train control: the same technology provides continual updates on the location of every locomotive on the railroad. Advanced tracking and control technology is already in place on high-speed passenger trains such as those on the Boston to Washington line. The technology is also under development at a number of companies, most prominently Pittsburgh-based Union Switch and Signal.
Combining satellites with computers to govern a train’s speed is only one step toward completely computer-automated operation. Subways routinely operate this way; the driver goes along for the ride. But a freight train is not as simple as a subway. A long train may be climbing one grade and descending another at the same time, for example. And every freight train has its own braking characteristics, which an engineer must quickly master; mishandling a train can cause serious damage, like torn couplers and perhaps even derailments. Some railroads, however, are experimenting with computers that can learn a train’s characteristics as fast as an engineer can. For example, computers have taken control of heavy ore trains in Minnesota, operating efficiently and stopping smoothly at red signals.
The next logical step is fully automatic operation, with an engineer on board only as a monitor. While the technology to implement this largely exists, other factors stand as barriers. The hefty up-front costs, for example, discourage railroads from installing new systems that don’t provide an obvious bottom-line benefit. Safety is another concern; automated control systems must be proved extremely reliable before they can be trusted to replace human operators, and it is not until such a substitution is possible that the technology has much of an economic payback.
Computerization has already made it possible for railroads to operate with fewer people. The newest developments represent an assault on the jobs of the two most important people who run a train: the engineer and the conductor. And deployment requires renegotiating contracts with labor unions representing workers whom new systems might displace.
It is looking as if the long climb up the Toponas grade is just the beginning of an accelerating journey into a computer-automated future. Says Union Pacific’s Iden, “We’re just starting to tap into the benefits of technology.”
Five poems about the mind
Work reinvented: Tech will drive the office evolution
As organizations navigate a new world of hybrid work, tech innovation will be crucial for employee connection and collaboration.
I taught myself to lucid dream. You can too.
We still don’t know much about the experience of being aware that you’re dreaming—but a few researchers think it could help us find out more about how the brain works.
Is everything in the world a little bit conscious?
The idea that consciousness is widespread is attractive to many for intellectual and, perhaps, also emotional
reasons. But can it be tested? Surprisingly, perhaps it can.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.