Derailments Caused By Emergency Braking?

[zugmann]

But at least with slack, you still had that option available if needed.

I ran the robot engines, and they give you limited control over what you can do (and no DPs for us).  Always had to start the cut of cars on a grade, then there was a little more of a severe hump about 5 car lengths away in front of you.  So if you didn't get them moving just a bit by then, you'd stall or slip (and when a RCO slips, it stops). 

So taking slack was pretty routine for us (even on jobs with old head engineers).  I always tried without first, but most of the time you didn't get too far. 

So for us at that terminal, for those jobs, for how we switched cars, slack was an asset. Eliminate slack, and well, we'd just have to find other ways to switch.  Probably by taking smaller cuts.  Perfectly doable, of course, but I don't see slack as a tool of the devil, either.

[Anonymous]

I found that article by George Hilton called SLACK in Trains magazine February 1976.  It covers a lot of what fits perfectly into this thread.  Mr. Hilton does not mention anything about slack being a Great Satan, but he does refer to slack this way:

“The end result of most of what is wrong with railroading.”    

[BaltACD]

Slack control is what separates Engineers from those that just operate the throttle and brakes.  Virtually any idiot can open and close the throttle, apply and release the brakes - Engineers do that and keep the buff and draft forces of slack under control.

I have had the opportunity to operate a simulator under the direction of one of our Road Foreman of Engines - the machine let one load in the consist of a existing train, including various locomotive consists as well as specific territories on my carrier.  While the simulator did not impart the feel of motion and slack action - there was a display that showed a graphic of the consist and how the slack was positioned in the train and numeric values of that slack and if the forces were either buff (run in) or draft (run out) - with every manipulation of the throttle or brake you could watch the shift in the location and value of the slack action.  At some physical locations it was necessary to apply a minimum brake reduction - not to control speed but to keep the slack under control.

Slack action separates Engineers from poseurs.

[edblysard]

Jeff,

Yup, most stuff is bigger, but yeah, pulled that number out of thin air!

It’s like trying to explain to folks outside the industry how much a locomotive really weighs…you can give them the pounds, or tons, and they still stand there and stare at you with that blank look that tells you they have no clue…usually followed by a silly question like “why don’t locomotives have disc brakes like my car?”

Even if you tell them getting hit by a Dash 9 or a SD70 is like 225 Chevy full sized trucks hitting you all at one time, they still don’t get it….

 

Taking slack for us is normal, but then again, our MK1500Ds weight 250000lbs each with 1500 HP, so it’s a needed skill here…I would imagine with a pair of Dash 9s it would make little difference….now switching is different, slack is absolutely necessary in flat switching.

And when kicking against a cut f cars, the slack in the drawbars and couplers acts like a cushion, stepping down the impact car by car.

[Anonymous]

Mr. Hilton’s article on slack covers not only slack, but also the issue of switching impact and the overlap of the function of devices to control both causes of impact.  Switching impact damages lading, while slack impact creates dangerous train dynamics.

Here is his comment about the functional necessity of slack for starting trains:

“A contemporary lashup of diesels may exert a tractive effort of over 300,000 pounds, triple what a big articulated steam locomotive could put out.  Roller bearings have reduced rolling friction.  All of this has reduced the significance of the original function of slack:  allowing cars to be started one at a time.  In fact, slack generally has been eliminated from passenger trains, and some engineers think the same could be accomplished with freights, even given the existing technology of locomotives and separate, unpowered cars.”

Interestingly, Hilton says the probability of a derailment of a train rises more than proportionately to the length of a train.  The longer the train is, the higher is the probability of its having a dynamiter in the consist, the greater will be the force of cars striking one another in braking, and the higher is the probability of having extremely unstable cars in the consist.  He cites statistics showing that the probability of derailing from equipment-related causes is five times higher for a 200-car train as it is for a 100-car train.

Train separations are 1 out of 6 for trains of 250 cars; and 1 out of 136 for trains under 100 cars.

In an average derailment, trains of 100 cars have 4.6 cars derailing; and trains of 250 cars have 7.1 cars derailing. 

[schlimm]

With the adoption of many groups of cars permanently joined with drawbars, isn't the total slack reduced considerably in trains utilizing that type of equipment?

[aegrotatio]

beaulieu

There was an Emergency Braking induced derailment at Ellicott City, MD yesterday. The Engineer of a CSX coal train initiated an Emergency Braking application when he saw trespassers on the tracks, 21 coal cars derailed, unfortunately the two trespassers were killed. Train crew was unhurt, but shaken up.

This is not determined to be the cause of the derailment. The cause is not yet known.

[oltmannd]

Slack is more than free-play in the couplers. In fact, it's only a small part of it.  It's mostly the springiness of the draft gear.  The coupler is attached to a yoke.  The yoke holds the draft gear and transmits the forces to the center sill.  The draft gear, in days of old, was a big spring.  Now, it's made of rubber pads http://www.minerent.com/products/dg_SL-76.php

[Anonymous]

In the Trains article about slack, author George W. Hilton brings up the point that the spontaneous action of a “dynamiter” or “kicker” can cause slack to derail a train.  However, he makes no mention of the fact that slack can cause a “dynamiter” or “kicker” to act.

I wonder why he would leave that out of an article about slack. 

[Anonymous]

Mr. Hilton, in his Trains piece called SLACK, includes a table of coupling impact forces expressed in foot-pounds.

He says that when a 75 mph, 100-car train goes into emergency, the slack run-in force can reach the level of a 14 mph single car impact during switching.  This impact would amount to 937,860 foot-pounds.

To put that into perspective, that would be the energy of 468 tons of force sustained over a distance of one foot.  468 tons is roughly the dead weight of two modern locomotives or three loaded cars.

Mr. Hilton says this much impact force will aggravate the cars ahead enough to derail the train.   

[greyhounds]

Bucyrus

Mr. Hilton, in his Trains piece called SLACK, includes a table of coupling impact forces expressed in foot-pounds.

When a 75 mph, 100-car train goes into emergency, the slack run-in force can reach the level of a 14 mph single car impact during switching.  This impact would amount to 937,860 foot-pounds.

To put that into perspective, that would be the energy of 468 tons of force sustained over a distance of one foot.  468 tons is roughly the dead weight of two modern locomotives or three loaded cars.

Mr. Hilton says this much impact force will aggravate the cars ahead enough to derail the train.   

And if doesn't wreck the train it will wreck anything remotely fragile, such as TVs, washing machines, etc. in the boxcars.

 

[cefinkjr]

Bucyrus

It does not seem very likely that the derailment could be unrelated to the brake application where the two events just happen to occur about the same time.   

No, but, without witnesses, you've got a chicken and egg situation.  Did the emergency application cause the derailment or did the derailment cause the emergency application?

Kind of reminds me of most derailments I "attended" as an Operating Department trainee many years ago.  The Transportation Dept official says the rolling stock caused the derailment, the Mechanical Dept guy insists that the track was at fault, and the MoW man is certain the train was going too fast.

[Anonymous]

Chuck,

I understand your point as it would apply to a emergency application not made by the engineer at the same time a derailment happens.  That would be the chicken and egg situation, as you say. 

But my statement that you quoted was made on the condtion where a wreck occurred during an emergency application made by the engineer.  In that case, it would seem conclusive that the emergency application caused the wreck.  Otherwise it would have to be an incredible coincidence such as an engineer dumping the air for some reason, and an axle just happening to break at the same moment, for instance.   

[Paul_D_North_Jr]

Bucyrus

In the Trains article about slack, author George W. Hilton brings up the point that the spontaneous action of a “dynamiter” or “kicker” can cause slack to derail a train.  However, he makes no mention of the fact that slack can cause a “dynamiter” or “kicker” to act.

I wonder why he would leave that out of an article about slack. 

  I believe that cause was not generally well-known or understood at the time.  There were many other possible causes, and the research and computer re-enactments during the ConRail era that Don/ oltmand mentioned above had not yet occurred.  Plus, Hilton was/ is an economist, not an engineer or other mechanically-qualified technician, and a magazine article - even in Trains - has its limits on the exploration and exposition of such phenomena.  It would, however, be more interesting to see what air brake expert David G. Blaine has had to say or write about the subject, then and now.

- Paul North.   

[zugmann]

I was watching a training video about train dynamics today. 

Learned there are two types of slack: free slack and spring slack. Free slack is the slack between the knuckles, while spring slack is the slack in the spring-loaded draft gear in cars.

 

I'll admit I chuckled when the video started discussing slack.

[Paul_D_North_Jr]

Just as Don Oltmann said about 6 or 8 posts above ! 

(except for the training video and the chuckle part . . .  ). 

[Anonymous]

In George Hilton’s article called SLACK, he discusses the incredible impact force when slack runs in during an emergency application of a 100-car train.  He cites an impact force of 468 foot-tons occurring wherever the force reaches its maximum potential.  But I am not completely clear on his analysis, or how often this maximum run-in occurs.

Does every emergency application that occurs while pulling with slack stretched lead to a slack run-in?  On one hand, the run-in seems logical because the brakes are applying from front to rear.  So trailing cars will run up against a gathering resistance as the head end cars set up and slow.

However, the run-in process takes time, and this time has to factored against the time it takes for the emergency application to propagate from front to rear.  And the run-in time itself depends on the rate of deceleration of the head end.  And not only does the emergency application take time to activate front to rear, but it also takes time for each brake to fully apply once it is activated.

So, when an emergency application is made, it sets off a wave of brake application from front to rear, and potentially, a wave of slack run-in from front to rear.  I can see how the brake action might stay far enough ahead of the slack to stop its the run-in before it reaches the end of the train.  I suppose the makeup of loads and empties would have a big effect on how it all plays out.

I wonder what that dynamic model would look like with a 100-car train of equally loaded cars, running 60 mph on straight, level track, and dynamiting from the head end.  What would the slack status be once the train stopped?  

[BaltACD]

Bucyrus

In George Hilton’s article called SLACK, he discusses the incredible impact force when slack runs in during an emergency application of a 100-car train.  He cites an impact force of 468 foot-tons occurring wherever the force reaches its maximum potential.  But I am not completely clear on his analysis, or how often this maximum run-in occurs.

Does every emergency application that occurs while pulling with slack stretched lead to a slack run-in?  On one hand, the run-in seems logical because the brakes are applying from front to rear.  So trailing cars will run up against a gathering resistance as the head end cars set up and slow.

However, the run-in process takes time, and this time has to factored against the time it takes for the emergency application to propagate from front to rear.  And the run-in time itself depends on the rate of deceleration of the head end.  And not only does the emergency application take time to activate front to rear, but it also takes time for each brake to fully apply once it is activated.

So, when an emergency application is made, it sets off a wave of brake application from front to rear, and potentially, a wave of slack run-in from front to rear.  I can see how the brake action might stay far enough ahead of the slack to stop its the run-in before it reaches the end of the train.  I suppose the makeup of loads and empties would have a big effect on how it all plays out.

I wonder what that dynamic model would look like with a 100-car train of equally loaded cars, running 60 mph on straight, level track, and dynamiting from the head end.  What would the slack status be once the train stopped?  

 

MOST emergency brake applications ARE NOT initiated by the emergency brake valve on the locomotive - they are initated from somewhere in the train - they will propagate in both directions from the point of occurence.  Most, if not all locomotives have a delay programed into their emergency braking system that have the engines continue pulling under power for a period of time when a train initiated emergency application is detected (the theory for this is to keep the head end of the train moving LONGER & FURTHER than the rear of the train and thus minimize the occurrence of one end of the train colliding with the other end of the train when the train has broken in two sections - the video of the tornado derailment in Illinois a couple of years ago is a example of what happens when the head end of the train stops before the rear end of the train).

The most slack action is not created in 'high speed' occurrences as the sheer inertia of thousands of tons of train and the rate of brake application through the entire train will still have the entire train moving when the brakes are fully applied throughout the train.  Extreme slack action happens in low speed occurrences -  15K ton, 9000 foot train moving 5 MPH - engineer initiates emergency application - head end of the train can be stopped before the application has propagated to the rear of the train - the rear portion of the train is then running into a virtual immovable wall.  Such occurrences is why most trains now are required to have '2-way EOT's' - the emergency application can be initiated both from the engine and the EOT and thus minimize slack movement throughout the train, the EOT's application being initiated at the speed of radio communication.

[Paul_D_North_Jr]

Bucyrus

In George Hilton’s article called SLACK, he discusses the incredible impact force when slack runs in during an emergency application of a 100-car train.  He cites an impact force of 468 foot-tons occurring wherever the force reaches its maximum potential.  But I am not completely clear on his analysis, or how often this maximum run-in occurs.

That doesn't seem near large enough to me.  468 foot-tons (936,000 ft.-lbs.) is the same as dropping about 4 loaded normal freight cars (117 tons = 234,000 lbs. each) about 1 foot.  Or looking at it another way, that's the same as the kinetic energy of 1 such car dropping 4 feet vertically. 

More to the point, in the horizontal direction, physics tells us that the 4 ft. of height energy = velocity squared/ 2 x the Gravity constant (32.2 ft. per second per second).  Doing the math, that's the eqivalent of the one car moving at just over 16 ft. second (or about 10.9 MPH) running into a brick wall (the rest of the train at rest).  Alternatively, if it's just 4 cars doing that, each one then needs only 1 ft.'s worth of kinetic energy, which they will have at 8.0 ft./ sec. (5.5 MPH). So that's why that cited value seems way too low to me for a mainline train coming to a crashing stop - instead, it seems more appropriate for a short cut of cars being switched at 4 or 6 MPH to bump up against another standing cut of cars or the rest of the train, etc.  (Those interested can do their own math for other such plausible scenarios.) 

Bucyrus

Does every emergency application that occurs while pulling with slack stretched lead to a slack run-in?  On one hand, the run-in seems logical because the brakes are applying from front to rear.  So trailing cars will run up against a gathering resistance as the head end cars set up and slow.

However, the run-in process takes time, and this time has to factored against the time it takes for the emergency application to propagate from front to rear.  And the run-in time itself depends on the rate of deceleration of the head end.  And not only does the emergency application take time to activate front to rear, but it also takes time for each brake to fully apply once it is activated.

So, when an emergency application is made, it sets off a wave of brake application from front to rear, and potentially, a wave of slack run-in from front to rear.  I can see how the brake action might stay far enough ahead of the slack to stop its the run-in before it reaches the end of the train.  I suppose the makeup of loads and empties would have a big effect on how it all plays out.

I wonder what that dynamic model would look like with a 100-car train of equally loaded cars, running 60 mph on straight, level track, and dynamiting from the head end.  What would the slack status be once the train stopped? 

Some years ago I saw - in the trade magazine Railway Age, I think - a depiction of a computer simulation-based 3-dimensional plot of the slack action in a train while in motion.  It looked like a series of irregular model railroad mountain ranges on a large table top (mathematicians call this kind of function and study "topology', I believe - kind of like topography is for surveyors and mapmakers.)  The one horizontal axis of the plot - from the foreground to the right background, commonly referred to as the "X" axis - was of course the location of the point of interest (cars) in the train.  The other horizontal axis, sometimes referred to as the "Y" axis - from the foreground to the left background - was Time, from the start in the foreground to the end of the time period under study at the background horizon.  The vertical axis upwards ("Z") was, of course, the amount of compression/ 'buff' when the slack ran in; tension / 'draft' was a negative value, below the 'plane' or 'tabletop' of the plot.  By choosing a particular point (car) in the train and following a line parallel to the time ("Y") axis backwards, you could see what happened to a particular car during the study's time frame.  Alternatively, by choosing a particular time and then following a line parallel to the train location axis ("X"), you could see the state of the slack forces in the entire train at that instant; and, you could choose many such instants.  Viewing all the data provided an overview of the event, from start to finish, for the whole length of the train.  It was of course easy to pick out the highest 'mountaintop' in that depiction, which was of course the peak slack value.

The good news, I know I cut out that article and saved it.  The bad news is, I'm not sure in which file or box.  Further, even if I had it in front of me right now - how could I scan it and post it so that everyone here could view it ?  So hopefully my 'word picture' description above will suffice for now. 

- Paul North.   

[Anonymous]

Paul,

I had not pondered slack action during emergency braking until getting into this thread.  I had always visualized it fundamentally running in with brake service applications unless the train is pulled to keep it stretched during braking.  But when I consider emergency application from the head end, I am not so sure what happens.

I see a lot of reference sources listing the speed of emergency propagation as well as the speed of brake setup on each car.  Some references talk about how these numbers have changed over time and differ with different brake pipe pressures.

Hilton says that slack runs in or out at 200-400 feet per second.  I don’t think that statement is adequately qualified to stand just like that.  He goes on to say, “If it takes 90 seconds for brakes to go on in the whole train, the train will have contracted 100 feet or more in the interim, mainly in the first few seconds.”

Again, I believe there are missing components in that analysis.  Hilton is referring to 100 cars, with one foot of slack in each.   If the train is 5000 feet long and the slack runs in at 300 feet per second; that will take 16.6 seconds.

When he says 90 seconds for the brakes to go on in the whole train, he means completely set up.  But the emergency propagation rate is 900 feet per second, so that means that when the engineer puts the brakes into emergency, it will only take 5.5 seconds until the last car goes into emergency.  But it will take another 84.5 seconds until all of the brakes in the train fully apply maximum braking pressure, and the last car to fully apply will be the last car in the train.

So that means that each car requires 84.5 seconds to fully build up maximum braking effort once the application begins.

Hilton says that the slack will completely run in “mainly in the first few seconds.”  I don’t see that.  If it requires 84.5 seconds to fully apply, how much braking are you going to develop in just “the first few seconds”?

But the larger question is this:  If the brakes on the hind end start to apply only 5.5 seconds later than the brakes on the head end—and if it takes 84.5 seconds to fully apply, once a brake begins to apply—what is the slack really inclined to do?

Is that 5.5-second difference in the initiation of such a long application phase really enough to cause the degree of run-in shock that Hilton describes?  As the run-in begins, all the cars will have initiated some degree of braking.  And as the run-in wave moves back, the braking on the stretched cars will be further increasing.

Hilton says, “The cars will have banged into one another with an impact which at its worst is equal to a switching impact of 14 mph.”  From that figure, he calculates a force of 937,860 foot-pounds.

For this scenario to make sense, enough braking will have had to occur on the bunched part of the head end to decelerate by at least 14 mph while hind end cars are coming on with no deceleration yet.  That is the part I can’t see.  By the time the first cars have dropped 14 mph, the rear cars will have also started to decelerate.  So I am not convinced that there will ever be a 14 mph discrepancy anywhere in the train. 

 

[BaltACD]

Sounds like Hilton has a economists understanding of train dynamics - not a railroaders.

[Paul_D_North_Jr]

Bucyrus

  [snipped and emphasis added - PDN] Hilton says that slack runs in or out at 200-400 feet per second.  I don’t think that statement is adequately qualified to stand just like that.  He goes on to say, “If it takes 90 seconds for brakes to go on in the whole train, the train will have contracted 100 feet or more in the interim, mainly in the first few seconds.”

Again, I believe there are missing components in that analysis.  Hilton is referring to 100 cars, with one foot of slack in each.   If the train is 5000 feet long and the slack runs in at 300 feet per second; that will take 16.6 seconds.

This is a nomenclature confusion or conflation problem.  The cited rate is likely a 'peak rate' of 200 to 400 feet of 'actual slack' run-in (not 'train length') per second.  The theoretical elapsed time for 100 ft. of slack  to 'run in' at 300 ft. per second = 1/3 second; for the range of the other values of 200 to 400 ft. per second is 1/2 to 1/4 second, respectively, of course.  But slack doesn't instantly start to run in at that high rate, it has to build up to it gradually though quickly, as the emergency brake application pressure-decrease wave propagates to the rear of the train.  The slack run-in rate will be very low - near zero - just behind the first car to apply its brakes, and then increase going rearward.  For instance, if the first car is going 1 MPH slower than the following car, it will take 1 ft./ 1.47 ft. per sec. = about 2/3 of a second for the 2nd car to catch up with the first.  By then the first car is going even slower, so the next 1 ft. of slack may be 'caught up' at a rate of 2 MPH = 3 ft. per second, or about 1/3 of a second.  Keep going with this kind of progression or series, and I think you'll wind up with a total in the 5 to 10 second range, as Hilton stated.  A 'quick-and-dirty' Excel spreadsheet with 100 iterations of this calculation might be fun to try, for those who are able and so inclined . . .     

- Paul North. 

[BaltACD]

Let me reiterate - Severe slack action takes place at low speeds, nominally 10 MPH and less.  At higher speeds the inertia of the train and the brake application rate will minimize the severity of slack change as the brakes will be fully set while the entire train is still in motion.  At slow speeds, one end of the train can get stopped before the brake application is fully propagated through the train and the remainder of the train that is still applying the brakes runs into a virtual 'brick wall'.

Personal observation over the years, trains shatter more knuckles when making stops for Stop Signals and other legitimate stopping situations - Engineer is applying the air brakes, normally to complete the stop after having done most of the speed reduction using extended range dynamic braking - in applying the train brake a 'kicker' places the train in emergency at slow speed and severe slack action shatters one or more knuckles.  The kicker may be from the actions of the brake valve on one of the cars, it may be from a 'bunch leak' that occurs as the slack moving toward the locomotives in a controlled manner cause a shift in the positions of the glad hands on the air hose coupling....whatever the cause the low speed UDE causes violent slack change in the train - slack change more than the weakest knuckle can stand.

[Anonymous]

BaltACD,

I understand exactly what you are saying about slack run-in being muted at higher speeds and potentially very severe at slow speeds.  That is exactly my point in questioning Mr. Hilton’s analysis in my previous post.  At higher speeds, I see the brakes getting somewhat of a hold on the entire train before the slack gets a good opportunity to run.

And also, at low speed, there is not that much range to decelerate before reaching dead stop.  So that 5.5 second delay in propagation might very well allow the head end to actually stop while the hind end is still moving.  And once the cars stop, their resistance to movement jumps way up—something about bodies at rest wanting to remain at rest.

But if a train is moving fast, all the slack will have run in while the head end cars are still moving, so they are not able to build up that maximum resistance that occurs with stopped cars.

There might also be some effect arising from the fact that braking friction between the brake shoes and wheels is lowest at the highest speed and highest at the lowest speed.  So if the head end dynamites, the brakes will tend to really grab and pinch down the momentum.  Although, I don’t know how that exactly shakes out because they would grab the trailing cars as well just a few seconds later.       

[NP Red]

It's kind of like that in an automobile.  If you slam on the brake at 3 mph, the passengers will fly worse that if you slam on the brakes at 70 mph.