Feinstein: Railroads must install collision avoidance system

[unhappycamper]

http://www.mcclatchydc.com/2013/12/05/210728/feinstein-railroads-must-install.html



An Amtrak Capitol Corridor train from Sacramento, Calif., arrives at Diridon Station in San Jose on Aug. 10, 2012. Amtrak's California corridors have the highest ridership in the country outside the Northeast

Feinstein: Railroads must install collision avoidance system
By Curtis Tate
McClatchy Interactive
December 5, 2013

Noting similarities between a fatal weekend commuter train derailment in New York and an accident in California that killed 25 people five years ago, Sen. Dianne Feinstein, D-Calif., said Thursday that the nation's rail operators must install a collision avoidance system by the end of 2015.

"Sunday's crash was preventable," Feinstein wrote Sen. Jay Rockefeller, D-W.Va., the chairman of the Senate Commerce Committee.

Feinstein wrote the bill that requires railroads to install the system, called Positive Train Control, and opposes giving them more time to install it. It would automatically stop a train if the engineer fails to obey a signal or exceeds the posted speed.

Four rail systems, Metrolink, Amtrak, Alaska Railroad and BNSF, will meet the 2015 deadline. Other railroads have lobbied Congress for a five-year delay.

[ladjf]

1. An Iphone app with gps would have been able to throttle back from 82 mph to 20 mph based upon the

gps location of train as it approached the 30 mph zone. Trains only have two direct controls for speed, (1) the throttle and (2) the brakes. Or the train company could control all of their moving trains from a central location, handling both the speed issues but anti-collision as well.

Why such capabilities aren't used on all trains is beyond comprehension.

[Not Sure]

4. It's only beyond comprehension if you don't comprehend what it takes to operate a train

Sure, it's possible to have a sophisticated computer system automatically operate trains on the main line (we actually have that now and it works well under ideal circumstances), but any deviation from what the computer identifies as "normal" and the computer throws control of the train to the engineer. That can put you in a bad spot as the engineer, when the train is trying to slow for a speed restriction and determines it cannot be done so all of a sudden the train is speeding and it's under your control. There are bugs in the programming, but it's getting there.

Positive train control isn't the same thing as this automated control system, but it will work in concert with it. Positive train control isn't about running the train as much as it is about stopping the train before it leaves the limits it's authorized to operate within. It will also stop a speeding train if there is no response to warnings to slow down. Currently the alerter system in place on freight locomotives will stop the train if there is no response to a visual and audible warning, and it goes off every few minutes to a couple times a minute depending on the speed of the train (faster train, more frequent alerts). I'm not a passenger locomotive engineer, so I can't comment on whether passenger locomotives have alerters, but I can't imagine they aren't so equipped.

So back to the automated control system, like I said it works pretty well, but only for running under ideal conditions. For any unusual conditions, such as mechanical failure, inclement weather (air brakes work much slower in very cold weather), or failure of the railroad's physical plant (track or signal defects), there is no substitute for a qualified engineer at the controls. And it doesn't hurt to have a second set of eyes in the cab, too, which is something I wish was required for passenger railroads (it is required by my employer and most other freight railroads).

A computer cannot feel how the train is behaving (a remote operator definitely cannot do this). It cannot see the track ahead and predict when pedestrians or vehicles may or may not attempt to cross ahead of the train. It cannot respond to communication failure. It cannot move at restricted speed (looking out for and stopping short of objects or broken rail in the train's path). It cannot take a mandatory directive. It cannot make a judgement of an emergency situation unless it can be sensed by measurable data.

In other words, a computer can run a train. But a computer cannot do the work of an engineer. Maybe just not yet. But for now, an engineer is the only one who can do the job all the time.

Sometimes humans make mistakes and it is terrible when they do. All of us feel it when one of us makes a mistake like this. It makes me redouble my efforts to keep my crew safe, keep my train and its cargo secure, keep my community and those I travel through safe from harm and remain alert and vigilant when I'm at the controls. My heart goes out to the families of those who were lost, those who were injured, and to the crew of the train.

[ladjf]

5. My post only addressed the issue of improper speed of certain geographic points along the route.

I did not suggest replacing the train operator.

The throttle and brake controls that were available to the engineer at the front of the train were wired to the physical controls back at the engine.

1. Any GPS will determine the exact location of the instrument within a few feet.

2. The programmed trip director would have been set to negotiate the river turn at 30 mph.

3. At the appropriate location prior to the turn, the computer would have signaled the throttle to be reduced so as to achieve a rate of 32 mph. If needed, appropriate brakes could be applied. These controls would be "talking" the the engine and brakes using the same circuitry that was available to the engineer.

4. After the train successfully negotiates the turn, the trip director would resume whatever speed is desired to continue.

One doesn't need to know anything about trains other than the throttle and brakes use electronic devices to regulate both and GPS positioning, working in conjunction with the required trip plan will determine when and how much to use throttle and brakes.
Thousands of airplanes and ships are now operating all over the world using far more complex computerized control systems that the one I suggested for the train. Passenger and freight airplanes routinely may totally automated landings requiring no input from the pilots.

[Not Sure]

6. My reply was directed toward the points you made about controlling trains from a central location

[ladjf]

7. I threw that in as a possibility. But, I favor some type of on-board monitoring that would kick in

if the engineer were to become incapacitated.

[phantom power]

2. that sounds great, but what if we just put two drivers in the cab?

[madokie]

3. That would be the first thing I'd suggest

[ladjf]

8. Yes. Two people on the job is the must obvious first step. n

[happyslug]

9. Old Accounting Rule. "To put two people in charge, is to put no one in charge"

No, you have to have one engineer. You can have an assistant, but it must be clear he (or she) is the assistant to the engineer. That assistant Engineer may even be able to operate the train (with the Engineer's approval) but the engine is always the responsibility of the Engineer.

Pilots on planes follow the same rule, the Pilot is in Charge, the Co-Pilot can assist the pilot and even fly the plane, but the Pilot always stays in charge, the Pilot is always the "Driver" of the plane.

Thus a mandatory assistant Engineer would be good, but two "Drivers" would be bad.