Train Air Brake Description and History

There is a lot more to train air brake systems than meets the eye. The following writeup was posted to rec.railroad by Bob MacDowell and is as good a description as we have seen. Thanks Bob.

If you still have unanswered questions after reading this, please send a request to brakes@sdrm.org and we will try to answer it.


Early Train Brakes

The first train brakes were very simple. To stop the train, you blew a certain pattern with the engine whistle and brakemen would jump from car to car setting handbrakes. Brakemen, needless to say, had trouble getting life insurance... :-)

The next generation of brakes, invented by George Westinghouse in 1869, added a compressor to the locomotive, and a brake pipe running the length of the train, connected between cars with gladhands, which were symmetrical "non-gendered" connectors that were latched together by hand and would separate by themselves if pulled on. The brake pipe was connected to an air cylinder on each car, which pulled on the handbrake chain when the brake pipe was pressurized.

In other words, charge the brake pipe with air, and the brakes went on. This worked MUCH better than brakemen, but it still took a long time to pump all that air back to the cars. And, all it took was a parted hose or other failure anywhere in the brake system to cause the system to fail entirely.

Westinghouse's Invention

George Westinghouse didn't like the direct air brake either, so in 1872 he invented a thing called a triple valve. This valve, and a reservoir on the car, inverted the behavior of the direct air brake: charging air into the pipe charged the system and released the brakes, and draining air from the brake pipe applied the brakes. He now had a much more responsive system that was fail-safe. It worked, and became the basis of the modern air brake.

The triple valve attached directly to the brake pipe, then had a connection to the reservoir, and to the brake cylinder. It was called "triple valve" not of the three connections, but of its three modes:

(In modern brakes there are two kinds of release actions: Direct Release, in which *any* increase in pipe pressure kicks the brakes off fully. Freight brakes work this way. Graduated Release, in which a partial increase provides a partial, proportional release. This is used in passenger trains, though the valve becomes much more complicated.)

Westinghouse's triple valve improved response times, because it didn't need to move all the air needed to apply the brakes. It only had to move enough air to carry a signal to the triple valve, telling it to apply or release. But still, the signal took awhile to work its way down the brake pipe. This would be improved later...

Improvements: Emergency

An Emergency feature was an early addition to Westinghouse's technology. This added a second reservoir, and made the control valve more complicated, but it added the means to make a harder application of the brakes. And, with a propagation feature called "Quick Action", it made them apply very quickly too.

"Emergency" adds a fourth mode to the brake system. A rapid decrease in brake pressure signals the valve to immediately throw the "works" into stopping the train. Including the full contents of a second, larger reservoir, called the "Emergency" reservoir. (The original reservoir is now called the "Auxiliary" reservoir.

Most freight cars use a "duplex" reservoir, which are two cast halves separated by a steel plate. The steel plate is shaped like a dome inside, which makes the emergency half of the reservoir larger. A tab sticks out of this steel plate, one side labeled "aux" and the other "emerg" so the sides can be identified.

In normal operation, the emergency-equipped control valve operates just like the original triple valve, except, of course, that it also charges the emergency reservoir. But part of the valve is designed to detect a rapid drop in pressure, which trips the emergency mode. The valve will then dump the entire contents of both reservoirs into the cylinder, and when pressure equalizes, there will be nearly full system pressure in the cylinder, 63 pounds or so on a 70-pound brake pipe pressure. This is as hard as the brakes will go, and will often lock up the axles at low speeds, skidding flats in the wheels. The force of an emergency application can also damage lading or even derail the train!

An emergency stop is now the default action almost anytime there's a brake failure. Any rupture in the brake pipe will cause an emergency application, as will a defective brake valve pejoratively called a "kicker" or "dynamiter" (which puts the whole train in emergency.)

The reason a defective valve could disable a whole train, is that part of the "Emergency" feature is another feature called "Quick Action". Since an emergency application requires a rapid drop in brake pipe pressure, there needs to be away to make sure the drop remains rapid, even far back in the train. The locomotive alone can't do that; by the time the pressure drop got 100 cars back, it would not be so rapid.

So each valve repeats and propagates the effect of the emergency brake. That's what Quick Action does. When the valve goes into emergency, Quick Action vents the brake pipe itself, thus _causing_ a rapid drop, amplifying the emergency action and making sure the next valve goes into emergency too.

That also means that if a valve is defective, and produces what is called an "Undesirable Emergency Application" (UDE), the entire train will follow suit.

When conductors rode in cabooses, they knew of four reasons for their train unexpectedly going into emergency:

  1. An air hose has parted.
  2. The train has broken in two.
  3. One of the cars is a "kicker".
  4. The head end is about to hit something.
Usually, they didn't think too much about the cause before grabbing a handhold; the slack action from an emergency brake could tear the stove out of the wall. But as soon as they stopped, they'd radio to the head end "All stopped!" Which gives the engineer some idea of how many pieces the train is in.

The AB Freight Control Valve

The Westinghouse "AB" control valve of 1933 is, essentially, the modern brake. It has all the brake features discussed so far. While there have been technological upgrades ("ABD" to "ABDW" to "ABDX") the basic packaging is standardized, and worth talking about.

The "AB" control valve consists of a pipe bracket, to which all piping connections are made, and two control valve portions (the "Service" and "Emergency" portions) which bolt to the pipe bracket with three bolts. Each of the three pieces weighs about 65 pounds, which is (conveniently?) just light enough to be shipped by UPS.

The beauty of the system is its ease of maintenance. The two portions (which are quite complex inside) simply bolt off; and you don't rebuild them, you just ship them off to someone who does it for you for about $120. Add ten dollars worth of gaskets and filters, and a field diagnostic, and you've done 16-year brake service on a railroad car. The hard part is cutting out the stencil which says "C.O.T.S. 1/4/94" (Clean, Oil, Test and Stencil)

The pipe bracket does not just unbolt; there are pipes attached to it. The five pipes are Brake Pipe, Cylinder, Auxiliary Reservoir, Emergency Reservoir, and Retainer. The last one deserves some explanation.

The retainer is a way of "keeping" some of the brake application even after the brakes are released. When an AB brake releases cylinder pressure, it vents it out the "retainer" port. On most cars, this leads to a retaining valve located on the side of the car. The retaining valve retains pressure in the cylinder when the control valve tries to release it. It can be set for "direct", which lets the air out directly, or "retain 10 pounds" which keeps the last 10 pounds of pressure in the cylinder.

This is used to descend long grades: with the retainers turned up, the cars will hold ten pounds of brakes even while the brakes are fully released and recharging. More advanced retainers added two more settings: retain 20 pounds, and slow release, which would release fully but took about 90 seconds to do it.

On cars that didn't use the feature, a screen was put over the retaining valve port to keep wasps from building nests in the control valve.

More Improvements: the ABD valve

The ABD valve of 1964 added two features which made brake response faster, but they worked mostly like AB valves, and were, of course, compatible. However, they were completely different internally.

The old AB valve used technology of 100 years ago inside the valve: small pistons moved brass slide valves, aligning or blocking ports to make the valve function. They had to be lubricated with graphite, and there were always problems with scoring and leakage. With the ABD valve, rubber diaphragms (like in a car's fuel pump) replaced the pistons, and sliding shafts with rubber gaskets replaced the brass slide valves. Although they did basically the same thing.

Two functional features were added to AB and newer valves: Quick Service and Quick Release. Both worked like Quick Action, manipulating brake pipe pressure to propagate the brake commands more quickly.

Quick Service propagates the effect of a service application. When it sees an pressure drop of 1-1/2 pounds or so, it vents some more brake pipe pressure to atmosphere, assuming a 5-pound application. This means that the next valve sets brake more quickly, and so on.

Quick Release does the same thing, for releases. When it sees an increase in pipe pressure, it adds some air from the emergency reservoir into the brake pipe. This increases the pressure some more, and all in all makes the train release much more quickly.

But... remember the trouble with "Kickers" and the Quick Action feature, where one car could put an entire train into emergency? The same thing can happen here, but now one car can release the brakes on an entire train. This caused major problems until it was learned that certain ways of handling the brake were causing it.

Picture, if you will, a train crew on a hill, trying to uncouple the locomotives from a train. The engineer makes a heavy brake reduction, which causes air in the pipe to move toward the locomotive. Then, a brakeman closes the angle cock (the valve between cars that closes off the brake pipe) on the first car, so they can uncouple. Normal enough, but watch what happens in the brake pipe...

Air is moving through the brake pipe, and suddenly the air passage is shut. But air has momentum, and it's still moving toward the locomotive. So it piles up at the closed angle cock, a little pneumatic traffic jam. And as it does, the pressure at that spot in the brake pipe, increases. Guess what the control valve in that car does. It releases.

This wasn't a problem with the AB valve; at most one or two cars at the front of the train would release. But the Quick Release feature changes all that, because it propagates the release back through the train. That knocks the brakes off the second car, which knocks the brakes off the third, and so on ...

By the time the crew realizes it, their entire train has released, and it's headed down the mountain.

Education has pretty much fixed the problem; wait until air has stopped moving before closing the angle cock; or leave it open a bit so air continues to drain.

Braking down Hills

Diesel locomotives have two brake systems: an air system, which is independent of the train brake; and dynamic brakes, which use the main drive motors as generators, and discard the generated energy out large resistor banks.

Ideally, braking down long grades is done with dynamic brakes. They never overheat or wear out. Realistically, some help from air brakes on the train is often needed, and that's done either by normal use of the brakes, or by stopping and turning up retainers.

I won't talk about how air brakes perform over long periods of use, but I believe they are sufficiently leak-resistant that they can make it down the hill without needing to stop and recharge. However, if the engineer repeatedly applies and releases the air brakes, he will eventually drain the auxiliary reservoirs on the cars (but the emergency reservoirs will be intact...) If an engineer foolishly does this, he may need to use emergency to stop the train. Then he will need to turn up enough retainers to hold the hill, and release the brakes, and pump air into the system until it is recharged. While he waits, the trainmaster will show up and cuss him out, of course...

There have been several rounds of evolution on locomotive brake stands (which together with the control-valve-like Distributing Valve, control the locomotive's and the train's brakes) The very early varieties had three positions: Running (which charged the brake pipe up to a set pressure), Apply (which discharged the brake pipe), and Lap (which did nothing to the pipe). An engineer would move the handle to "Apply" until he got the pressure he wanted, then he would move the handle to "Lap". Unfortunately, leakage would tend to reduce the pipe pressure further, which would cause the cars to set the brakes harder and harder. This caused problems on hills, as described above.

The most common contemporary model is the "26", which is a pressure maintaining brake. Meaning, if you set the handle to 65 pounds, it will automatically hold it at 65, even against leakage. This solves the hill problem.

Before the invention of the pressure maintaining brake, engineers would improvise one by adjusting the "feed valve", which is the pressure regulator that feeds the brake system, and thus determines maximum brake pipe pressure. That's supposed to be set at whatever brake pipe pressure the railroad runs as a practice (70, 80 or 90 pounds) and left there. To maintain pressure in their brake pipes, engineers would dial the feed valve to the application they wanted (say, 60 pounds) then let leakage bring it down to that. The feed valve, thinking 60 pounds was the running pressure, would dutifully hold the brake pipe there, against leakage. It was illegal but it mostly worked.

Runaways

Seems like there's a runaway in almost every train movie. But there's only about three ways a runaway can happen, given modern brakes...

  1. The crew forgot to charge the system in the first place. If you couple onto a string of cars that's been stored for a week, there's no air in it. If you then haul it away without hooking up air or doing an air test, you could lose your train simply because the locomotive is the only brake on the train.

  2. The "Quick Release" bug. Described above. If there's anyone on board capable of pulling an emergency cord, it'll stop the train.

  3. Going so fast down a hill that you gain kinetic energy faster than your brakes can get rid of it. Like running down a steep hill.

  4. The brake pipe gets "bottled up" behind the engine. Either an angle cock vibrated shut (very unlikely) or the brake pipe fell on the tracks and got bent and pinched shut in a way that "bottled the air" in most of the train, so there aren't enough responding brakes to stop it (also very unlikely, but it HAS happened...) Again, if there's anyone back there to pull the emergency cord, they'll stop the train. Or, if there's enough brake pipe leakage in the bottled portion, one of the valves will go into Quick Service, and will cascade at least a minimum application...
These are all pretty bizarre, and they just don't happen much. And when they do, it's in the paper :-)

Return to the FAQ page

Last Update: 11/04/02