Electronically Controlled Pneumatic Brake

Technical Briefing
Prepared by Paul Niquette
Revised March 6, 1997
Copyright ©1997 Morrison Knudsen Corporation


Air-operated brakes date back to the nineteenth century in the railroad industry. A brake pipe runs the length of the train. Air pressure in the pipe is controlled by the train operator.

In a conventional pneumatic brake system, brake pipe pressure (BPP) does two things:

    [1] BPP energizes the brake system onboard each car and
    [2] BPP commands the brakes into operation throughout the train.
The same can be said for the hydraulic 'BPP' in an automobile, truck, or airplane -- with a crucial difference: In a train, these two functions conflict with each other. High BPP provides energy to the braking system, but for safety's sake, the brakes are commanded into operation by low BPP. {FootNote 1}

In briefest summary, BPP (70-110 psi) from the locomotive delivers air into a pressure vessel called an 'air reservoir' or 'auxiliary reservoir' onboard each car, accumulating pneumatic energy distributed throughout the train. {FootNote 2}

The train operator signals for the application of brakes by opening a 'control valve' in the locomotive that causes a release of air from the brake pipe and the reduction in BPP (by 5-20 psi for service application).

The 'brake pipe reduction' (BPR)...

    [a] activates a 'brake valve' in each car,
    [b] prevents the escape of air from each reservoir back into the brake pipe, and
    [c] allows air to flow from the reservoir into the 'brake cylinders' that operate the 'brake shoes' against the 'wheel treads.' {FootNote 3}
The consequent force applied by a piston to the brakes is regulated to be proportional to the magnitude of the BPR signalled from the locomotive.

Releasing the brakes throughout the train requires venting the individual brake cylinders to the atmosphere, which takes place as an all-or-none event. The train operator operates a valve that commands air to be pumped into the brake pipe, increasing the BPP, which is sensed onboard each car causing an exhaust valve to open. A spring retracts the brake shoe from the wheel tread. Meanwhile, the increasing BPP recharges the reservoir through the check valve.


...with the conventional pneumatic brake system described above:

Problem #1 -- Non-Simultaneous Application of Brakes

The command for brake application begins with the opening of a valve at the locomotive end of the train. Air has to flow out of the brake pipe to produce BPR, which acts as a 'signal' to the brakes on the train.

  • The BPR takes the better part of a second to propagate from car to car. Thus, brakes are applied in sequence beginning at the first car behind the locomotive. Depending on train length, it can take more than a minute before the brakes are applied throughout the train. That extends stopping distance by a significant factor over that which would be experienced if all the brakes could be applied simultaneously.
  • Moreover, during the braking process, cars farther back continue unbraked, producing a potential for what is known as 'slack run-in.' To overcome the dangers of this condition, the train operator routinely continues to apply traction power -- even increasing it -- during the initial stages of braking, a procedure called 'power braking,' which also extends stopping distance and wastes fuel.
Problem #2 -- Recharging Delay

During the stopping of a train, the brake pipe is fully committed to the function of 'signalling' and cannot be used for 'energizing' the reservoirs. After stopping the train, therefore, the conventional system reverts to the function of replenishing the air in all reservoirs. It can take several minutes to 'recharge' the brake system -- time during which the train is out of productive service.

Problem #3 -- All-or-None Release

As described above, inherent in the design of the conventional pneumatic systems on trains, there are conflicting interactions between the functions of 'energizing' the brake reservoirs and 'signalling' the brakes into operation. A given BPR eventually commands a certain intensity of braking effort throughout the train. As the train decelerates, the operator may determine that more braking is required and increase the BPR -- decrease the BPP -- for more deceleration.

  • The opposite -- 'graduated release' -- is not supported by conventional pneumatic braking systems on trains. If the amount of braking appears to be excessive for the desired stopping 'target,' the operator is faced with a Hobson's choice: either
    (a) to release the brakes completely or
    (b) to allow the train to stop short.
  • Operating rules sharply limit the first option (releasing the brakes while the train is still moving), inasmuch as there may be insufficient air remaining in reservoirs in each car for subsequent re-application of the brakes.
Stopping short on a siding may leave the mainline blocked by the rear of the train. That condition will prevail until air pumped into the brake pipe can restore the pneumatic energy in the reservoirs onboard each car to a safe level. As described above, recharging the brakes can take as long as several minutes, potentially delaying movements of other trains on the mainline.


...separates the two functions of supplying energy to the reservoirs and signalling the application/release of brakes.

    On a train equipped with ECP Brake, the brake pipe is dedicated to the solitary purpose of energy supply -- continuously pumping air from the locomotive into all the reservoirs on the train. The signalling function is performed over an electronic network.
Two alternatives are under consideration for the signalling means that will be adopted as the interoperable standard for the 'physical layer' (a) wireless and (b) trainline.
    The signalling protocol selected by the railroad industry is LonWorks by Echelon, a decision made by the Association of American Railroads (AAR) in consultation with operators and suppliers. {FootNote 4}
Microprocessor based nodes in each railroad car support intelligent distributed control (IDC) functions in a sophisticated, peer-to-peer, multi-domain network architecture. The signalling method is exceptionally robust, and the brake system architecture is made vital by virtue of 'expectant state' timing features.


Inasmuch as commands are delivered at electronic speed, brakes throughout the entire train operate simultaneously. Braking distance is thereby reduced, and 'power braking' is obviated. The system supports 'graduated release,' which enables operators to stop with precision and safety. The reservoirs throughout the train are continuously being energized, which reduces -- often eliminates -- the recharging time following a train stop. Benefits include...

  • Since trains equipped with ECP Brake can stop faster, they can run faster.
  • Shorter test times as well as faster brake readiness allows trains to be put in service more quickly.
  • Fuel efficiency is improved.
  • Through improved brake management, trains are easier and safer to operate.

With the Echelon-based IDC signalling in place, each train becomes equipped for the first time in history with a two-way signalling infrastructure -- an onboard 'information superhighway.' Many VMS-based benefits to railroading become feasible, including...

  • Onboard Hotbox Detectors
  • Derailment and Impact Monitoring
  • Acoustic Signature Analysis
  • Brake Wear / Slack Measurements
  • Automatic Rostering
  • Cargo Status Monitoring
  • Load Cell Measurements

By all indications from industry working sessions and published reports, the railroads are expecting high returns from their investment in ECP Brake. With over a million and a half freight cars in the U.S. fleet, the logistical numbers get huge in a hurry. Here is one preliminary 'bill of materials' for products that will be participating in the ECP Brake opportunity.

Car Control Device (CCD) 
  • Modification Kit for Pneumatic Components 
  • Intra-Car Wiring Kit 
  • Electronic Control Node with Signalling Interface 
  • Battery and Power Management
End of Train (EOT) Unit
Electronic Control Node with Signalling Interface
Onboard Power Generation 
  • Alternator and Power Management 
  • Battery Backup and Installation Kit
Wireless Transceivers or 
Glad-Hand Connector (for wired system) {FootNote 5}
Head End Unit (HEU) 
  • Modification Kit for Locomotive 
  • Pneumatic Controls 
  • Operator Control Unit and Display


As celebrated in official industry documents, many of the highest payoffs are expected to be derived from the vehicle monitoring opportunities, which require only incremental investments for electronic intensive smart sensor devices. Some expected to be addressed by suppliers...

  • Wayside hot box detectors are mandated under state laws to one every 20 miles. There's talk of making that 10 miles. But a bearing can go bad in 1 mile, and a seized bearing can cause a lot of expensive damage and service delays. Then, too,...
  • Acoustic sensing with sophisticated 'signature' tracking will be the way to go. Bearings give warnings before they get hot, but problems cannot be reliably detected from the wayside because of ambient audio noise and the 'air gap' problem.
  • Brake-feedback is also desired by the railroad industry. This will provide real-time confirmation of the application and release of the brakes on each and every car. Also on-line information about slack adjusters will displace maintenance labor and improve safety.

The magnitude of the installation effort asserts a singular logistical challenge. At n hours per car at m installation sites, the total job spans 1,500,000 n/m hours (that's 240 n/m years, working 3-shift, 5-day weeks).

Total interoperability is a long way off. An early emphasis will be given to equipping 'unit trains.' Some railroads have expressed reluctance to keep a 100-car train out of service longer than two weeks. That makes for an upper bound of 2.4 hours per car. If unit trains constitute half the fleet, they might be outfitted with ECP Brake over a period of 50 / m years -- ten sites, starting in 1997 would finish in 2002.


Steel wheels on steel rails enable trains to be the most efficient form of transportation (kwh/kg/km, hp-hr/ton/mile) in the world. What is beneficial for rolling is not so good for stopping. The coefficient of friction is on the order of 0.2 ('adhesion' of 20% in railroad parlance). {FootNote 6}

Brake effort, which is upper bounded by the 'coefficient of static friction' between wheel and rail, must be adjusted -- off-line and open loop -- to prevent wheel slide. Since changes are inconvenient, typically no more than 65% of tare weight is used for setting the upper limit of brake effort. Conventional braking administered by 'brake-pipe reduction' (BPR) commands the same intensity of braking throughout the train. That means, of course, that a freight car contributes its proportionate share of maximum braking only when empty -- precisely the situation when the train needs that car's braking least. Heavily loaded trains, therefore, suffer excessively long stopping distances. Moreover, the train becomes subject to harsh and unpredictable 'slack-action,' in which drawbar forces get routinely challenged by uneven braking, the worst antagonist being slack run-in ('buff'), especially troublesome with loaded cars coupled behind unloaded cars.

With ECP Brake, of course, each car administers its own braking by interpreting a coded command (% braking effort) over the Echelon network. The concept of intelligent distributed control (IDC) applies. Each freight car can be equipped with load cells that provide sensor data permitting the ECP Brake system to increase the local braking effort to match the load.

Transit vehicles derive slip/slide protection based on tachometer feedback from the wheels -- not unlike anti-lock braking systems (ABS) on cars, trucks, and airplanes. Before ECP braking, freight trains have nothing that corresponds to ABS. Still, apart from detecting slides in emergency brake application, ABS is going too far, in my opinion. And also not far enough: In a steel-on-steel environment, the tachometer can only tell the system after-the-fact that the slide has already occurred, which is not appropriate in routine, non-emergency braking.

Traditionally, ABS operates in a 'duty-cycle modulation' mode. The algorithm relies on detecting discrepancies between/among wheel rotating speeds, which works well enough in maximum brake effort situations. In railroading, of course, the left and right wheels on a given axle always turn at the same RPM, so discrepancies have to be detected between/among axles. Unlike rubber tired wheels, however, steel wheels skid rather abruptly down to Zero RPM -- tantamount to a wheel lock-up. Then too, it takes time to re-accelerate the wheel after the brakes are pulsed off due to low adhesion. Finally, brakes on both axles in a truck are mechanically linked for common operation. Actually, on most freight cars, brakes on all four axles operate from a common brake cylinder. Remedial pulsing has to apply to the whole car, even though only one of its four axles has locked up. That puts in doubt the ABS approach as used elsewhere: Discrepant rotating rates would have to be detected at axles between/among more than one car.

Clearly a load-cell node product -- 'kit' -- is a natural VMS application that fits in with the highest priorities associated with in the freight application of ECP Brake. {FootNote 7}

Another approach is to offer support for a 'downloadable' load parameter (wireless or otherwise) tied to the lading and billing procedures.


For all the reasons set forth herein, the Class I Roads want ECP Brake -- sooner the better.

{FootNote 1} "Whoever said 'Let's kill two birds with one stone' was a wishful thinker not a successful hunter." 

-- Sophistication: How to get it...then what!{Return}
{FootNote 2} The reservoir has two chambers, 'auxiliary' and 'emergency.' Their functions and others are not germaine to the present explanation. {Return}

{FootNote 3} The brake valve is also called 'triple valve,' since it performs three functions: charge, apply, and release. {Return}

{FootNote 4} Either signalling method is compatible with LonWorks, which supports all seven layers of the Open System Interconnect (OSI) standard: physical, link, network, transport, session, presentation, and application. MK/ASD weighed in early during the industry-wide deliberations with technical analyses, with the consequence that LonWorks prevailed as an interoperable standard on its objective merits against ardent challenges on behalf of disparate and heterogeneous alternatives from vested proprietary interests. {Return}

{FootNote 5} The brake pipe on a freight car terminates with a rubber hose on each end of the vehicle. The glad-hand connector is a genderless, quick-release device that connects the hoses from car to car. {Return}

{FootNote 6} 'Slip/Slide' limits dominate the performance of a train. Slip sets an upper boundary on useable 'tractive effort' and slide does the same for 'braking effort.' The former causes wheel wear; the latter does too, but in a worse form, flat spots (a flat spot does not heal by the rolling of the wheel; it gets nothing but worse. {Return}

{FootNote 7} As of this writing there are AAR-sponsored interface specifications but no known product announcements. The freight railroad markets are dominated by shipments in bulk, all-or-none carloadings -- coal being the largest (40%). Even a one-bit resolution on load would support better braking and shorter stopping distances for the train than what is done today (almost certainly a load cell subsystem would apply an analog-to-digital encoder of at least a byte). {Return}

Home Page
Back to References
Top of Page