AAR undesired emergency brake study complete - Association of American Railroads Vehicle Track Systems Newsletter

Railway Age, Dec, 1990

The AAR Undesired Emergency (UDE) study began in 1984 at the request of the Vehicle Track Systems TTD implementation Officers. The intent of the study was to discover the cause of UDEs and to make recommendations to the industry aimed at reducing or eliminating UDEs. After a five-year period encompassing 10 revenue train tests, laboratory analysis, and a full scale train test at the Transportation Test Center in Pueblo, the final report has been completed.

UDEs typically occur when a minimum service brake application is made. UDEs are normally sporadic and unpredictable, and occur more often on moving trains than on standing trains. Finding the control valve which initiated the UDE is an almost impossible task. At the beginning of the study, the major challenge was locating one or more control valves which positively initiated a UDE. If and when this was accomplished, it was felt that inspection of the control valves would lead to identifiable defects which could then be corrected. As it turned out, finding the kickers" (control valves that initiate UDEs) was relatively easy and finding the causes proved to be very difficult.

The initial phase of the UDE study began with train tests on Canadian Pacific Locotrol-equipped unit coal trains, Santa Fe TOFC trains and Chicago & North Western freight trains. The CP train tests yielded ten kickers," all of which were located in parts of the train where the quick service activity was highest. The control valves were promptly inspected by the AAR and by the air brake manufacturers and no defects were found. However, recordings of brake pipe pressure during service brake applications showed that pressure pulses consisting of small, rapid reductions of short duration occurred on some of the brake applications. Exhibits I and 2 show minimum brake applications made on the same car of the same train, yet one of the reductions shows a 23.5 psi/sec reduction lasting for 12 milliseconds. The sharpest reduction rates were typically about 20 to 30 psi per second and lasted for about 10 to 15 milliseconds.

Tests of conventional TOFC trains on Santa Fe produced seven UDEs initiated by four control valves, and once again no defects were found. However, the slack action eastbound was much more severe than it was westbound, which led to the theory that slack action contributed to the increased number of UDEs on the eastbound trip. The brake pipe pressure recordings showed pulses very similar to those seen on the earlier CP tests.

Tests on Chicago & North Western confirmed that slack action did have some effect on brake pipe pressure. Brake pipe pressure was recorded on the last car of the train during slack run-ins and run-outs without any brake applications. Exhibit 3 shows the brake pipe pressure during a heavy run-in on a 53-car mixed freight-TOFC train.

A theory was developed to explain the effect seen on these tests. Air has mass, and therefore has inertia. When a slack run-in occurs, each car decelerates suddenly. As the car and its brake pipe decelerate, the air within the brake pipe continues at its original velocity. Me air in effect "sloshes forward," which causes a pressure rise in the front end of the pipe, and a pressure drop in the back end of the pipe.

To check this theory, the AAR contracted with the University of New Hampshire to develop a brake system computer model which would simulate the effects of slack action on the air in the brake pipe. The computer model confirmed the slack theory, and also predicted that as each car ran-in, it would produce a negative pressure pulse which would propagate back in the brake pipe at nearly the speed of sound. If the slack run-in speed was close to the propagation speed, the pulses would add to each other, resulting in a rapid, short duration I to IV2 psi reduction (Exhibit 4). Furthermore, the computer model predicted that the pulse would double on itself as it bounced off the rear angle cock, producing up to a 2V2 psi reduction (Exhibit 5). UDEs were created on the computer model, using severe slack run-ins combined with normal quick service activity.

Laboratory testing performed at the AAR Chicago Technical Center further confirmed that the reduction rates predicted by the computer model could indeed cause a "good" control valve to go into emergency. In addition, a possible fix" was developed in the form of a 0.43-inch-diameter choke between the pipe bracket and the emergency portion.

All of this activity led to the following theory: Slack action could, under the right conditions, cause sharp, short-duration pressure fluctuations which when combined with the quick service activity of a normal brake application cause "good" control valves to go into emergency. Due to the unpredictable and unrepeatable nature of slack action, this theory seemed to fit the facts.

To check the slack theory, and to evaluate the effects of the 0.43-inch pipe bracket choke, a 36-car test train was operated at the Transportation Test Center. The test consisted of instrumenting five cars spaced throughout the train for brake pipe pressure, buff force, and coupler movement. The train was handled such that the pressure pulses created by the slack run-ins coincided with minimum service brake applications. The TTC tests proved conclusively that slack action alone, without any brake applications, can produce sharp brake pipe pressure reductions of up to 2 psi. While no UDEs occurred during the tests, the reduction rates created were very nearly equal to those required to produce UDEs in earlier laboratory testing. Exhibit 6 shows the brake pipe pressure plots and the draft gear compression for cars 1, 12, 23, 30 and 36 during a hard slack run-in. Note that the brake pipe pressure of car 36 dropped 2 psi due to slack action alone. ne maximum pressure reduction rate was 27 psi/sec for about 20 milliseconds. This rate is very close to the rates required to create a UDE under laboratory conditions.