BaltACD Euclid & Rude - sell the FRA on your ideas. Simple! Once they are sold it can be mandated by the end of the year!
Euclid & Rude - sell the FRA on your ideas. Simple! Once they are sold it can be mandated by the end of the year!
Especially the current version of the FRA!
No, wait, the current version of the FRA can't do anything that fast!
-Don (Random stuff, mostly about trains - what else? http://blerfblog.blogspot.com/)
oltmannd ruderunner Here's a thought for a cheap simple and rugged derailment sensor: a set of limit switches located near the kingpin and contacted by an arm mounted to the truck. Set them so that the arm contacts a switch if the truck rotates too far from straight ahead. Example if one truck rotates 10 degrees but the other stays straight ahead that indicates a truck off the rails. 10 degrees is just an example number, the actual number would vary by car length and track curvature. This wouldn't have to effect a brake application but just an alarm to alert the crew. Heck it could even include a "switching" setting to prevent false alarms when working tight trackage at low speeds. It doesn't need to be extremely precise either, if the truck is within normal range of rotation then its likely tracking correctly. But once outside of that range we need to stop the train and find out what happened. I'm not looking to prevent derailments but rather prevent simple wheels off rails from turning into a flaming pileup. No doubt something like this could be worked out. I'd look for rate of change rather than absolute limits. It could react faster. Lots of research needed, however. Not simple or easy to do. Expensive to validate since you have to actually derail a car at speed, on purpose. But, that's not the biggest problem. The biggest problem is failed compontents. Mainline derailments are exceedingly rare. Component failures are not. How you mount them, wire them, etc. Are you allowed to run with failed components? What do you do when a train stops for a failed sensor? The trick is not to create more trouble than you solve.
ruderunner Here's a thought for a cheap simple and rugged derailment sensor: a set of limit switches located near the kingpin and contacted by an arm mounted to the truck. Set them so that the arm contacts a switch if the truck rotates too far from straight ahead. Example if one truck rotates 10 degrees but the other stays straight ahead that indicates a truck off the rails. 10 degrees is just an example number, the actual number would vary by car length and track curvature. This wouldn't have to effect a brake application but just an alarm to alert the crew. Heck it could even include a "switching" setting to prevent false alarms when working tight trackage at low speeds. It doesn't need to be extremely precise either, if the truck is within normal range of rotation then its likely tracking correctly. But once outside of that range we need to stop the train and find out what happened. I'm not looking to prevent derailments but rather prevent simple wheels off rails from turning into a flaming pileup.
No doubt something like this could be worked out. I'd look for rate of change rather than absolute limits. It could react faster.
Lots of research needed, however. Not simple or easy to do. Expensive to validate since you have to actually derail a car at speed, on purpose.
But, that's not the biggest problem. The biggest problem is failed compontents. Mainline derailments are exceedingly rare. Component failures are not. How you mount them, wire them, etc. Are you allowed to run with failed components? What do you do when a train stops for a failed sensor?
The trick is not to create more trouble than you solve.
Suffice to say the failsafe for the sensor system would be to run the car as we do today, as of its not there at all.
Modeling the Cleveland and Pittsburgh during the PennCentral era starting on the Cleveland lakefront and ending in Mingo junction
OK oltmand, a little more information on what I'm thinking. The switches could mount to the center will or bolster near the kingpin, a truck mounted arm would trigger the switches as the truck rotates. I've thought more about the limits I originally proposed, and determined that tighter limits of say 2 degrees off the centerline of the car would work. We only need to know if the truck is pointed in the direction of the cars travel. If it is, then the truck is most certainly on the rails, if not then the truck is like off the rails. The 2degree limit should allow some slack for things like sloppy installation, looseness in the kingpin, truck hunting and such. A timer would be added to prevent false alarms when entering curves, switches, crossovers etc. Say a 30 second delay.
The logic circuit will use readings from 4 switches per car (2 per truck) and compare the position of those switches to each other. Certain combinations indicate all is well, some indicate the cars is in a curve, some can indicate a wheel set off the rails, some can indicate the car is dog tracking or jackknifing, and some would indicate a failed switch.
More to the points you brought up: rate of change would be great for studying what happens in a derailment, but we only need to know a yes or no as to whether the truck is in line relative to the car.
There is absolutely no need to detail a car in testing, jacking the car and rotating the truck proves out the detection circuit.
Since this detection system doesn't tie into the brakes directly, a failed system will not disable the car. Remember I'm proposing a system that will alert the crew to potential wheels off track events. To fully outfit the national fleet will take years to be sure, but starting in dedicated trains makes the most sense. 100 sets of switches,a 100 logic board, and 100 transmitters and just one receiver outfits a unit train.
For proof of concept and prototyping, we can forgo air generators and just use battery power(cheaper in short term). I suspect the batteries to last about 3 months before recharging is needed.
I really need to find a real computer to type this stuff out, maybe this weekend I can expound on the details.
ruderunner2 degrees off the centerline
So, how is that going to work on a 5 degree curve (or tighter)? Not all lines are straight as an arrow.
Larry Resident Microferroequinologist (at least at my house) Everyone goes home; Safety begins with you My Opinion. Standard Disclaimers Apply. No Expiration Date Come ride the rails with me! There's one thing about humility - the moment you think you've got it, you've lost it...
Depending on how you are measuring it, you will need somewhere in the 7-9 degree range of travel. You want it to go through a #9 frog without alerting. A #7 frog has an angle exceeding 8 degrees.
With only 2 degrees you will most likely get a false positive through every diverging route on a swtich with a speed less than 30 mph.
Every false positive will un-necessarily stop and delay the train about 2 hours.
Dave H. Painted side goes up. My website : wnbranch.com
If you are going to introduce a thirty second delay, you might as well not put the sensor on the car in the first place. A lot will happen in thirty seconds. A train travelling at 30 MPH will travel a quarter mile in that time.
Remember that the Lynchburg derailment only involved about 1200 feet of track...
Isn't reality amazing!
Never too old to have a happy childhood!
Taking the oil train derailment under discussion as an example there were about 9 cars that derailed prior to the pile up. At 40 mph a train travels about 60 ft per sec, which is about a car length in this case. So its reasonable to say that there was between 8 and 10 sec between the first wheel hitting the ground and the general pile up.
If the system just notifies the train crew and takes no action, it will take a couple seconds for the trucks to skew enough to trip the sensors, another sec or two for the electronics to communicate to the head end. Then the crew will have to realize they have an alert, do whatever they have to do to acknowledge it, then take whatever action they are going to take. Realistically they are down to maybe 4-5 sec before the pile up occurs. Even if they plugged it and had ECP, it would still take a couple sec for the brakes to actually apply.
In this case the difference between the derailment putting the train in emergency and the crew putting the train in emergency is 2-3 sec. If the system applied a penalty or full service reduction, the train would have gone in emergency 4-6 sec seconds later.
The real benefit of this type of system is with a train like the ones in some of the other videos where the empty cars are drug for long distances without the train going in emergency. As has been pointed out several times before, the situations where a car will travel upright and in line in a train with one end derailed for miles are situations where the car is an empty. If the car is a load, our friend gravity will cause it to dig in and result in a pile up much quicker.
tree68 ruderunner 2 degrees off the centerline So, how is that going to work on a 5 degree curve (or tighter)? Not all lines are straight as an arrow.
ruderunner 2 degrees off the centerline
How much would the trucks need to rotate relative to the carbody to round that curve?
And once both trucks are in the curve, the logic detects that so no false alram.
I really need to find some time to draw some skecthes and do a thorough explanation of what I envision.
tree68 If you are going to introduce a thirty second delay, you might as well not put the sensor on the car in the first place. A lot will happen in thirty seconds. A train travelling at 30 MPH will travel a quarter mile in that time. Remember that the Lynchburg derailment only involved about 1200 feet of track...
And what prey tell could be done in 30 seconds to prevent a derailment? If it happens that fast, nothing can stop it. Ok so maybe we try it with a 5 second delay? How about 10 seconds? Whats your suggestion besides "don't bother"?
BUT, what about situations where a wheelset has been dragging for the last mile? It might be fine till it come to a crossing, switch, bridge or whatever. Wouldn't you like to know you have a truck off the rails before that?
Am I proposing to stop all derailments? No but early detection is the best way to prevent any derailment (or disease or catrastophie or fill in the blank)
dehusman Depending on how you are measuring it, you will need somewhere in the 7-9 degree range of travel. You want it to go through a #9 frog without alerting. A #7 frog has an angle exceeding 8 degrees. With only 2 degrees you will most likely get a false positive through every diverging route on a swtich with a speed less than 30 mph. Every false positive will un-necessarily stop and delay the train about 2 hours.
Very possibly true. Again, how far does the truck need to rotate relative to the car body to negotiate such trackage?
Conversly, how far would a truck have to rotate to drop one wheelset off the rails?
I suppose this would vary according to a cars wheelbase (center to center of kingpins) so adjustments need to be made according to car size. Not a big deal at all to do that.
Euclid I think it would take a couple measures of different factors for a sensor to add them up and conclude that a derailment has occurred. But let me ask this. If a device detected a derailment, what should the proper response be? The first objective would be to get stopped, possibly limiting the destruction by stopping quickly.
Exactly, stop the train and fix the derailed wheelset (if possible) A train doesn't always fly off the tracks as soon as the wheels leave the rails.
Lets define derailment, basically the flanges are no longer both between the rails. It doesn;t have to include a big flaming pileup (though thats what the general public thinks)
Wouldn't it be nice to catch a derailment early? That's what I'd like to accomplish. And frankly if the trucks aren't pointed the same direction that the carbody is, wouldn't that indicate a derailment?
BaltACD Isn't reality amazing!
I haven't heared any better suggestions from you...
Perhaps you have your own ideas about how to prevent big derailments, care to share?
Seriously though, you have some experience in railroading. Do you have any suggestions that may help the situation? You really are coming off as just a grumpy old head rather than being proactive about a problem facing your industry. I'd love to hear some hairbrained idea you might have had no matter how uncoinventional or uneconomic it may be. Who knows, it may just change railroading as we know it...
dehusman Taking the oil train derailment under discussion as an example there were about 9 cars that derailed prior to the pile up. At 40 mph a train travels about 60 ft per sec, which is about a car length in this case. So its reasonable to say that there was between 8 and 10 sec between the first wheel hitting the ground and the general pile up. If the system just notifies the train crew and takes no action, it will take a couple seconds for the trucks to skew enough to trip the sensors, another sec or two for the electronics to communicate to the head end. Then the crew will have to realize they have an alert, do whatever they have to do to acknowledge it, then take whatever action they are going to take. Realistically they are down to maybe 4-5 sec before the pile up occurs. Even if they plugged it and had ECP, it would still take a couple sec for the brakes to actually apply. In this case the difference between the derailment putting the train in emergency and the crew putting the train in emergency is 2-3 sec. If the system applied a penalty or full service reduction, the train would have gone in emergency 4-6 sec seconds later. The real benefit of this type of system is with a train like the ones in some of the other videos where the empty cars are drug for long distances without the train going in emergency. As has been pointed out several times before, the situations where a car will travel upright and in line in a train with one end derailed for miles are situations where the car is an empty. If the car is a load, our friend gravity will cause it to dig in and result in a pile up much quicker.
Yes, every second counts. So some feedback on how many seconds, how many degrees etc willhelp dial in the system.
FWIW electrical siginals travel at nearly the speed of light so knock off 2 seconds from yur math. But even using your example that 9 car pile would have been only 6 (or less) Thats nearly a 30% reduction right there.
Now I'm asking for some clarification on a topic that was touched on earlier, namely braking effort. My understanding is that railcar brakes are designed to affect the highest level of braking that an EMPTY car can stand without sliding wheels. Is that correct?
If so then loaded cars are dramatically underbraked (meaning they don't stop nearly as fast as they could) so the statement that railroaders like to throw out that "throwing the train into emergency is the fastest way to stop a train" isnt always true. It IS the fastest way to stop a trains of empties, but far from the fastest way to stop a loaded train.
If all the brakes are operating at a level that provides maximum braking (without wheelslide) isn't reasonable to assume then that a loaded train can stop much faster than it currently can? No not as fast as an empty train but a dramatc improvement none the less.
I'd also like to remind fourm members that "perceptin is reality" And the percieved reality of the general public is fast becoming that oil trains are rolling firebombs waiting to go off. I think it's in the industrys best interest to be proactive about this, certainly before the government decides to do something.
And doing something, even if its wrong, is a step in the right direction.
ruderunnerAnd doing something, even if its wrong, is a step in the right direction.
So if I say that we should make the oil car wheels square, that is a step in the right direction?
The opinions expressed here represent my own and not those of my employer, any other railroad, company, or person.
For that matter, one could simply point out some feature of the current system, tout it as "improved," and the public would be similarly asuaged. Say, f'rinstance, two-way EOT's.
The problem with doing something "wrong" is that it costs money, and doesn't provide a return on investment. And that money has to come from somewhere.
And if something happens despite having employed that "wrong" solution, then the credibility of all solutions comes under suspicion. Look at the "better solution" 1232 cars...
I think most of the criticism of any of the derailment prevention/minimization systems presented here is based on knowledge of the realities of railroading. There are probably parts of all of them that have merit, or would after some tweaking.
The devil is in the details, however. Power, communications, durability, and a host of other factors all figure in. Like - how much air would all of those air turbines consume on a 100 car train?
The truck deflection detector, for instance - how does one reconcile sharp curves with the potential that a truck might not deflect enough, even though derailed? I'll go back to that example of the car on a western railroad that ran a number of miles before rerailing itself. It may never have reached the deflection limit.
The biggest thing to remember here is that we aren't attacking the person with the ideas - if anything, we're giving them information that they can go back with and refine their idea.
One of my concerns about the system is that there are instances where a longer slowdown period would result in the cars hitting and sticking on switches or grade crossings that they would miss if they slow down as rapidly as possible.
tree68 [snipped - PDN] . . . The devil is in the details, however. . . . The truck deflection detector, for instance - how does one reconcile sharp curves with the potential that a truck might not deflect enough, even though derailed? I'll go back to that example of the car on a western railroad that ran a number of miles before rerailing itself. It may never have reached the deflection limit. . . .
Larry - the train that ran a long distance over a switch machine, etc. and then rerailed itself was a RoadRailer - see:
- Paul North.
Roadrailers at Sandusky Bay
ruderunner If so then loaded cars are dramatically underbraked (meaning they don't stop nearly as fast as they could) so the statement that railroaders like to throw out that "throwing the train into emergency is the fastest way to stop a train" isnt always true. It IS the fastest way to stop a trains of empties, but far from the fastest way to stop a loaded train.
Probably yes.
That is just flat wrong.
Wasting time, wasting money, and not improving anything isn't helping anything.
I would like to point out that all the schemes being proposed here, the derailment detectors, the load/empty adjustors, the differential braking are like seat belts and parachutes. They don't prevent the crash they just provide some possibility of lessening the consequences. None of the schemes presented prevent a derailment. None prevent a pile up. None prevent a car from being breached. None. Not one. They all just make the pile smaller. Many times making the pile smaller is the right answer. Doing random stuff just to be doing stuff is not.
Railroads have determined that joints in rail are a weak spot in the track. So they invented continuous welded rails. They found that field welds (thermite welds) are less reliable than factory welds (electric flash welds). They found that field welds near the ends of bridges are incrementally higher in risk. So many railroads have been actively removing field welds near the ends of bridges and replacing them with "in track welds" which are almost as reliable as factory welds. Factory welds are still a risk so some railroads are ordering 480 ft rails, rolled as a single piece. That required building a ship and developing a loading system to get them on and off the ship without damaging them. Now a 1/4 mile CWR string only has 2 welds in it instead of 15-30 in a conventional string.
Those activities ELIMINATE RISK. The weld can't fail because the weld isn't in the track.
That is how you improve safety. You do stuff that actually works.
ruderunnerNow I'm asking for some clarification on a topic that was touched on earlier, namely braking effort. My understanding is that railcar brakes are designed to affect the highest level of braking that an EMPTY car can stand without sliding wheels. Is that correct? If so then loaded cars are dramatically underbraked (meaning they don't stop nearly as fast as they could) so the statement that railroaders like to throw out that "throwing the train into emergency is the fastest way to stop a train" isnt always true. It IS the fastest way to stop a trains of empties, but far from the fastest way to stop a loaded train.
From what I understand, the load/empty sensors reduce braking effort for empties. So if a train is loaded, then it will still have full braking ability.
But throwing the train into emrgency is still going to yield the most braking force as the brakes are currently configured:
When an emergency application is enabled, the emergency portion of the reservoir is combined with the auxiliary reservoir. That causes the aux.reservoir pressure to be higher than the brake cylinder, and you will get ~ 20% more pressure in the brake cylinder then you would at full service or equalization.
ruderunnerNow I'm asking for some clarification on a topic that was touched on earlier, namely braking effort. My understanding is that railcar brakes are designed to affect the highest level of braking that an EMPTY car can stand without sliding wheels. Is that correct?
Yes. That's correct. The braking force is constant regardless of load.
ruderunnerIf so then loaded cars are dramatically underbraked (meaning they don't stop nearly as fast as they could) so the statement that railroaders like to throw out that "throwing the train into emergency is the fastest way to stop a train" isnt always true. It IS the fastest way to stop a trains of empties, but far from the fastest way to stop a loaded train. If all the brakes are operating at a level that provides maximum braking (without wheelslide) isn't reasonable to assume then that a loaded train can stop much faster than it currently can? No not as fast as an empty train but a dramatc improvement none the less.
For any given, train with existing braking, emergency is the fastest way to stop. It provides the most braking force each car can provide at that moment.
If you mean that at train with load proportional braking could stop faster than one without it, then yes, that's true.
So, when you hear "trains take a long distance to stop because they're heavy", that's not quite right. This is what you're getting at....
But, here's the rub. Load sensing braking systems aren't popular because:
In short, it's more trouble than it's worth - and it's not risk-free.
Now, could you develop ECP and on-car sensors to do a much better job of braking, alow higher speeds, prevent derailments, mitigate derailment damage, etc, etc. Sure. But it will take lots of dollars and a ton of R&D to get there.
Euclid Regarding the comment by Ruderunner: “And doing something, even if its wrong, is a step in the right direction.” There are two ways of interpreting this comment. One way is to interpret “doing something” as doing a full implementation of some untested improvement without knowing whether it will work or not; as though you have to take that much of a gamble. That would be absurd, and the railroad industry is the last institution that would make that mistake. For critics here to highlight the risk of doing a full implementation of something that might not work is a giant red herring to attempt to shoot down suggestions for improvement. The other way to interpret what Ruderunner meant by his comment is that “doing something” means doing research, design, and testing as a developmental process BEFORE any system-wide implementation. In “doing something” as research and development, there is indeed a process of “doing something even if it’s wrong” and having it become a “step in the right direction.” That kind of trial and error is the core of research and development.
Exactly! Doing nothing while potential for accidents looms is almost always a guarantee to fail. Or said differently, to try brings the chance of achieving a goal or not achieving it. But to not try is guaranteed to not achieve the goal.
Zugman, that's just plain old sabatoge...
Larry, I get your point but see the above. As for too much air useage, perhaps time to rethink a truck mounted generator? Something that doesn't require special wheelset or trucks. An add on system if you will.
Paul_D_North_Jr tree68 [snipped - PDN] . . . The devil is in the details, however. . . . The truck deflection detector, for instance - how does one reconcile sharp curves with the potential that a truck might not deflect enough, even though derailed? I'll go back to that example of the car on a western railroad that ran a number of miles before rerailing itself. It may never have reached the deflection limit. . . . Just about all equipment has to be able to negotiate a 12-degree curve (in a 100 ft. chord), and most has to be able to get around a 23-deg. curve (250 ft. radius). For a car with 50 ft. truck centers (roughly 70 ft. long), each truck would be angled from 3 deg. to about 6 deg. from the centerline axis of the car. Larry - the train that ran a long distance over a switch machine, etc. and then rerailed itself was a RoadRailer - see: "The Invisible Derailment - If a RoadRailer trailer derails in the middle of the night and nobody notices, does it still derail?", by Larry Gross, from Trains, November 2007, pg. 52 &etc. ["Magazine Index" 'keywords': derailment prototype roadrailer ] - Paul North.
Just about all equipment has to be able to negotiate a 12-degree curve (in a 100 ft. chord), and most has to be able to get around a 23-deg. curve (250 ft. radius). For a car with 50 ft. truck centers (roughly 70 ft. long), each truck would be angled from 3 deg. to about 6 deg. from the centerline axis of the car.
Paul thank you for that reference. I'm still curious of how much deflection a truck with one wheelset off the rails has.
Anybody know?
Also what would normal track speed be in that 23degree curve?
ruderunnerZugman, that's just plain old sabatoge...
What?
Our community is FREE to join. To participate you must either login or register for an account.