Oil Train

[rdamon]

Looks like part of the case to say they need more testing ...

BNSF lobbies the government on a range of issues, and crude-by-rail represents a small part of those efforts, spokesman Michael Trevino said. He also said the company supports the study and testing of ECP brake technology before implementation.  

On another note ..

Judge OK’s $430-million settlement fund for Lac-Megantic victims and creditors

http://medicinehatnews.com/business/2015/07/13/judge-oks-430million-settlement-fund-for-lacmegantic-victims-and-creditors/

[Euclid]

That is a very interesting development.  Here’s my question:  What does the crude-by-rail industry mean when they say that ECP brakes could “potentially jeopardize safety,” as the article says?      

[rdamon]

Buffett may benefit as train lobby bids to weaken safety rule

http://www.reuters.com/article/2015/07/14/us-usa-train-regulations-insight-idUSKCN0PO0A320150714?feedType=RSS&feedName=businessNews

 

Reads like one of our threads ... :)

[Wizlish]

Euclid

I have been thinking about dynamic braking and mid-train derailments that do not involve collisions. As I understand it, dynamic braking potential of a train must be limited according to the number of axles, etc. because too much dynamic braking buff (compressive) force will jackknife cars or pop them out of the train.

It is restricted according to the number of axles on the power because there is only so much train weight you can run through a given number of braking locomotive wheelsets before you risk starting to slide them.

There is a limit on how much dynamic you want to apply to a short train, but that's an issue for an engineer's train handling skill, not a technological limitation.

 

So with enough dynamic braking on line, the buff force can be as high as possible without causing the train to derail.

The 'buff force' can only be as high as the countertorque on the locomotive wheels can deal with.  And as you have already noted, it is only 'high' as a function of deceleration unless the train weight is grossly in excess of the dynamic capability (as on the overloaded train in one of the Duffy's Curve wrecks).  Again, good train handling involves modulating the dynamic to minimize any 'peaks' in buff force as the train passes into a 'steeper' downgrade section.

 

Other than a hard slack run-in, I would conclude that the typical buff force generated by conventional air braking is not nearly as high as it is with the maximum degree of dynamic braking that is commonly used. [/quote]

Of course it isn't.  Conventional braking is distributed through the train, so as the brakes go on progressively, the maximum observed buff force location moves aft in the consist and will depend on the inertia of the (decreasing) consist that is still relatively unbraked.  On the other hand, dynamic when used with the air not set will cause buff at the rear of the locomotive consist to be the highest of any measurable buff force in the train, and this will change first when more (or less) dynamic is provided.

[Euclid]

Dynamic Braking and Derailments:

I have been thinking about dynamic braking and mid-train derailments that do not involve collisions. As I understand it, dynamic braking potential of a train must be limited according to the number of axles, etc. because too much dynamic braking buff (compressive) force will jackknife cars or pop them out of the train. 

So with enough dynamic braking on line, the buff force can be as high as possible without causing the train to derail.  Other than a hard slack run-in, I would conclude that the typical buff force generated by conventional air braking is not nearly as high as it is with the maximum degree of dynamic braking that is commonly used. 

Derailments begin with the first wheelset leaving the rails.  From there, they often progress to a massive pileup of cars.  Yet, as seen in the recent Marytown, TN derailment, the derailment never progressed beyond the first wheelset leaving the rails, which is relatively unusual.  This tank car had its leading truck derailed at one axle.  Apparently the second axle of that truck was still carrying the load and guiding it too.  

I conclude that many, perhaps most, high speed derailments have some amount of time/distance between the first wheelset leaving the rails and start of a pileup.  It may only be 50 feet.  It could be 500 feet.  It could be miles.  In the case of a rail breaking on a curve, the derailment might instantly start a pileup with practically no derailed-dragging phase. Some derailments drag for some distance and then re-rail themselves.  It would be interesting to know the actual statistical distribution of the occurrence of the various derailed-dragging events associated with high-speed derailments.

A derailed-dragging event can go on for a long distance, such as over five miles.  It either ends by the onset of a pileup or by the stopping of the train.  The derailed-dragging phase is a delicate process that can easily be perturbed, thus causing the onset of a pileup. This can be triggered by just entering a curve, or snagging something like a switch or crossing; or it could be just caused by the gradual disintegration of the derailed truck due to the dragging friction.  The pileup can also be suddenly triggered by changes slack condition due to power increase or braking.

I conclude that a high buff force of a train in dynamic braking would pose the greatest potential disruption to the stability that allows a derailed car to drag without causing a pileup.  Conventional air braking would also pose a disruption, but not as severe as dynamic braking.  By the same token, being under power with no braking and slack stretched would pose the least threat of disrupting the delicate derailed-dragging phase. 

[Electroliner 1935]

The plumbing analogy is suddenly closing a faucet while the water is flowing. They install anti-hammer stubs on the pipes as shock absorbers but in some cases pipes have burst from the hammer effect. Old DC locomotives had "blow out coils" on the contactors to handle the arcs that occured when opening a circuit with heavy current flows.

[BaltACD]

Euclid

 

I understand that current practice requires graduated release to address both the train handling and the electrical effect.  But I am considering this instant release as part of a new proposal which would require instantly releasing the dynamics in order for the proposal to work.  So I want to find out if instant release is possible with this new proposal.

Everybody loves a technological catastrophy [/sarcasm]- any proposal to 'instantly' eliminate dynamic braking from a train that is operating with dynamics applied is a recepie for a train handling disaster on top of any other situations that may be taking place.

[Euclid]

Wizlish,

On the previous page, I asked two questions related to the effect of releasing dynamic brakes instantly.  One question focused on the train handling effect, and the other question related to the electrical effect. 

Based on what you and Jeff said, I have my answer to the part of the question related to the train handling effect.  I acknowledged that in my last comment on the previous page.  You did address the electrical effect, referring to high current interruption, etc.  Randy Stahl also addressed this. 

I understand that current practice requires graduated release to address both the train handling and the electrical effect.  But I am considering this instant release as part of a new proposal which would require instantly releasing the dynamics in order for the proposal to work.  So I want to find out if instant release is possible with this new proposal.

The only part of the question that I am re-asking pertains to the electrical effect.  The reason for re-asking it is to clarify if the electrical problems from instant release when using existing equipment could be overcome with a design change in the switching and control apparatus.  I will assume that the answer to that is yes.  So I have the answers to my two-part question.  I will follow up with a description of the new proposal.     

[Wizlish]

Euclid

So, my only other question is whether it is electrically possible to instantly, fully release dynamics from their maximum braking level. If not, how quickly could they be fully released?

Is there some reason we keep answering this and you keep right on asking it again?

Technically you can release dynamics quickly -- as quickly as you can safely interrupt high currents with significant amounts of inductive magnetic-field energy storage.  I would assume, to be safe, that this would be in the 2 to 5 second range, might be a bit longer.

But I don't think there are many cases in practical train handling where you'd cut the dynamic instantly to zero rather than modulate it under control (remember that the dynamic works as graduated release).  Just as you would seldom try to go directly from Run 8 to idle ... or vice versa without a computer interceding for you.

There are two effects here, and in the range of mass and inertia represented by trains, they may seem paradoxical.  With the mass of locomotives and cars, the rate of acceleration (let alone jerk) will usually be relatively slow.  But the force represented by that acceleration builds up to very large levels quickly.  That means that even a little differential acceleration, over even a short distance, can result in very severe force when contacting something, especially something with high inertia, moving at a different speed or stopped. 

So, in train handling, you want to avoid fast or hard control 'inputs'  -- from the throttle and brake, from the flange contact, from vertical curves or track components.  i think most everyone here that has extensive experience running locomotives will agree, if you ask them, that it would be better in almost any circumstance to adjust the dynamic slowly.

Paradoxically, you may have the one instance where quick reduction of dynamic would be appropriate: if differential braking during a derailment were required and the front end of the train were decelerating under DB with slack actively running in.  But that's going to be something of an unusual situation - and anyone who gets in an accident and has to say on the witness stand that they abruptly cut the dynamic brake while the train brakes were only partially applied is likely to be given a hard time, even if they think they can prove it was the safest alternative. 

[Randy Stahl]

Yes it is possible, just flip the isolation switch. Not a good idea though unless you are ready to change some contactors.

[Euclid]

Jeff and Wizlish,

I can see how suddenly releasing dynamics while going down grade could result in a slack runout just from gravity suddenly pulling the head end down the grade.  I can also see how the draft gear would not store enough energy to cause hard slack runout.  In order to do that, it would have to quickly accelerate all the cars in the runout. 

So therefore, I assume that there will be no runout sufficient to part the train if dynamics were suddenly fully released while decelerating with only dynamics on level track.  On level track, the head end would not be pulled ahead by gravity if the dynamics were released. 

If dynamics were suddenly fully released while holding back against a train descending a grade, I can see how many complications could cause severe slack run out, as Jeff says.  I suppose that if all those other complications could be controlled by air brakes or other logistical decisions, then theoretically the dynamics could be suddenly released while holding back a train on a down grade without causing a break in two.  

So, my only other question is whether it is electrically possible to instantly, fully release dynamics from their maximum braking level.  If not, how quickly could they be fully released?

[Wizlish]

Euclid

I assume that the runout of slack is due to the sudden releasing of buff force on the car draft gear units. This release allows the draft gear to rebound from its compressed or buff position, and return to the neutral position. Since this return to neutral of the draft gear would be so abrupt, the draft gear would continue rebounding past the neutral and into the stretched or draft position. Then this abrupt stretch can break a knuckle or pull a drawbar.

No.  The only things that can give you runout culminating in a shock will be momentum and gravity.  Any force developed by the draft=gear springs extending is going to be a relatively slow push, achieving little acceleration by the time 'neutral' extension is reached.  And there is quite a bit of damping in the draft-gear mechanism, too, to prevent loose run-out and run-in (of the type, for example, experienced in trains of British four-wheel waggons separated by relatively long-travel springy buffers).  Any 'rebound stretch' beyond the neutral point will have to overcome the substantial inertia of the cars on either side before acceleration could build up to a snatch or shock that would break metal.

Slack run-out if the front end is accelerating with respect to the 'rest' of the train, will be propagating down a long line of cars, each one being made to accelerate a little faster as the draft gear is pulled out with All The Momentum of the part of the train that is accelerating.  This is constant acceleration over much greater time and much greater distance than a few little springs and rubber blocks could possibly provide.

Note very carefully what the proper implications of 'keeping the dynamic applied' are -- for example in terms of avoiding too fast a release when the head end can start accelerating unbraked or substantially so.

There have been a few instances where running gear has been given undamped, or severely underdamped, control.  One significant example was a Reading locomotive, I remember the story as involving a 2-10-0, on which a clever master mechanic had provided two reasonably stiff coil rprings acting on the yaw of the engine truck.  The farther they went, the more critical oscillating frequencies of the spring system they began to discover.  After about 70 miles, probably the majority of which involved track-wrecking lateral banging, someone was called out with a torch, who selved the spring arrangement tightly then and there.  Note that the key word involved here is resonance -- something that may not be seen between large masses at small travel between hard mechanical stops.

[BaltACD]

Getting broken knuckles and/or pulled out drawbars normally end up being caused by the engineers inability to control slack within the train.  Slack run in and slack run out can easily generate stresses that are far beyond the designed strength of knuckles, couplers, coupler shanks and draft gear, especially when the slack situation instantly changes from one state to the other.  As trains get longer and longer - there is more free running slack in trains.

I am amazed that today's engineers can move such 'oversized' (10000 to 12000 feet and 15-20 thousand ton) trains over territories that were surveyed and built in the 19th Century by Civil Engineers that at the start had next to no knowledge of what a railroad was, and in later decades figured 20 cars of 20 to 30 feet long and 600/700 tons was large train and taxed the abilities of the motive power of the age.

[jeffhergert]

Euclid

 

jeffhergert

The dynamic brake handle has positions, but not fixed notches like the throttle.

Getting into dynamics too fast can cause a car to "pop" out under the right ( or maybe more correctly, the wrong) situations.  Getting out of dynamics too fast, or even at all on some heavy grades can cause the headend to run out and break a knuckle.  Or pull out a draw bar. 

Jeff

 

 

 

 

Thanks Jeff,

 

That is exactly what I wanted to know about release dynamics too fast.  What exactly would cause the knuckle to break?  When you mention this happening when using dynamics on a heavy grade, what does the heavy grade have to do with the possibility of breaking a knuckle when releasing dynamics? 

I should also clarify that my questions are based on the effect of releasing dynamics when no air braking is applied.  If knuckle breaking from releasing dynamics can result, what causes the run-out that would break the knuckle? 

 

Gravity.  The head end picks up speed.  If there is no air set, any heavy cars start to pick up speed, too.  The slack that was bunched starts coming out.  Even with air set, sometimes the weight of the engines can be enough to break something. When the faster moving portion finds the right spot where it meets the slower moving portion, snap goes the train. 

Granted it may not happen every time, but the possiblity is there.  You have to remember there can be a lot of variables.  Grades aren't always uniform down a hill.  Subtle changes can affect the way cars roll.  Cars don't all roll the same.  Curves can slow the train.  Loads and empties will pick up speed differently.  Knuckles and draft gear can have defects or the beginning of defects.  (They always ask for a broken knuckle, the percentage of old/new break and where the remnants are located.  A good conductor will say 50% old break, you can tell by how shiny the jagged steel is.  Then he'll do something to make all the shiny steel of your 100% new break look rusty.  Actually, everyone talks about it but I don't know that anyone has ever done this.  Anyone want to guess what this process is?)  A knuckle that has already began the failure process needs less of a jolt to break completely.

If Im going to get out of dynamics on fairly good grade, I like to do it when the headend is going through a curve.  The curvature helps to keep things from picking up too fast.  Or when the head end bottoms out and is starting back up in a sag.  I take into account the grade, curvature and what kind of train I have as to when and where to get out of them.  Not just loads and empties, but type of cars.  I watch for the position of cars in my train with cushioned/extended travel draw bars.  Today I might do something that I won't do tomorrow, it just depends. 

I also take into account the kind of engines I have.  The newer AC engines can brake to a stop in most places with dynamics.  They are easier, unless they aren't working right, to ease out of dynamics.  (You can still get run out with AC.)  DC engines seem to be harder to ease out of dynamic.  (Some are also harder to ease into dynos.  You have to move the lever farther in to get them to respond.  UP's SD70m units are a good example.  When new they were able to ease in and out of dynos.  Now it's like they are all or nothing to begin and reach a point when easing out where they just drop out.) 

There have been times when I gave away speed because I didn't think it prudent to get completely out of dynamics until conditions changed.  I would rather be a little slower than filling out a break in two form.

Jeff 

[Euclid]

Wizlish and NorthWest,

Thanks for your comments.  This is what I was looking for.  Also, please understand, that I am only asking questions at this point.  I am not suggesting anything (yet). 

Wizlish said:

With respect to the other question: Even if the dynamic came off "instantaneously", you would not see a rebound effect in the draft gear, whether or not the 'braking' locomotive were ahead of or behind the node. The cars have tremendous inertia, and there is inherent damping and friction in the draft gear, so there is no 'recoil action' that will cause any kind of 'violent' runout. I would expect other effects, wind and grade in particular, to be exerting more force tending to bunch or separate cars than draft-gear runout. (Other opinions, of course, welcome, provided they are based on science or experience.)

Wizlish,

The above quote goes to the main point of my questions.  Remember, in my example, there is no grade, and the point of the dynamic brake application is only to slow the train down.  I framed the questions to include the application of dynamic brakes, but my basic concern is about releasing the dynamic brakes after decelerating a train on level track.  How long would a full release take if done through the normal controls?  If there were a reason for a release to happen instantly, would this be possible if the controlling components were modified to achieve that goal? 

Regarding the runout of slack that can break a knuckle upon getting out of dynamic too fast as Jeff mentioned:  I assume that the runout of slack is due to the sudden releasing of buff force on the car draft gear units.  This release allows the draft gear to rebound from its compressed or buff position, and return to the neutral position.  Since this return to neutral of the draft gear would be so abrupt, the draft gear would continue rebounding past the neutral and into the stretched or draft position. Then this abrupt stretch can break a knuckle or pull a drawbar.   

[BOB WITHORN]

Yup, ever heard the term 'beating a dead horse'? (I've gotta find a bigger rock)

[rrnut282]

Wow, still going?  [crawl back under my rock]

[Euclid]

jeffhergert

The dynamic brake handle has positions, but not fixed notches like the throttle.

Getting into dynamics too fast can cause a car to "pop" out under the right ( or maybe more correctly, the wrong) situations.  Getting out of dynamics too fast, or even at all on some heavy grades can cause the headend to run out and break a knuckle.  Or pull out a draw bar. 

Jeff

 

Thanks Jeff,

That is exactly what I wanted to know about release dynamics too fast.  What exactly would cause the knuckle to break?  When you mention this happening when using dynamics on a heavy grade, what does the heavy grade have to do with the possibility of breaking a knuckle when releasing dynamics? 

I should also clarify that my questions are based on the effect of releasing dynamics when no air braking is applied.  If knuckle breaking from releasing dynamics can result, what causes the run-out that would break the knuckle? 

[jeffhergert]

The dynamic brake handle has positions, but not fixed notches like the throttle.

Getting into dynamics too fast can cause a car to "pop" out under the right ( or maybe more correctly, the wrong) situations.  Getting out of dynamics too fast, or even at all on some heavy grades can cause the headend to run out and break a knuckle.  Or pull out a draw bar. 

Jeff

[NorthWest]

Euclid

When dynamic brakes are applied, is the application developed in stages of progressively increasing retardation? Is it possible to apply them to the full intended force instantly? If it is possible, is there any reason why this should not be done?

Dynamic brake handles have notches, just like the throttles. You could probably hack the computer to get full instant dynamic braking, but you cannot do it from the control stand in normal operation. There simply is no reason for it, as trains begin to descend hills gradually and not all at once.

Euclid

Similarly, what is the method of releasing the dynamic brakes? Can dynamic brakes be fully released from their maximum application by simply opening the circuit so the brakes instantly release as if turning off a light switch?

Once again, the engineer must notch the dynamic brake handle/combined power handle to reduce braking. There is no need to immediatly stop braking.

[Wizlish]

Euclid

When dynamic brakes are applied, is the application developed in stages of progressively increasing retardation?

Yes.  I expect a number of people here can and will provide you with detailed information on the right use of dynamic brakes, links to manuals and other information, the history of this form of braking, etc.

In general, you would want to apply dynamic slowly at first, to get the slack bunched properly, then with proper modulation to prevent wheelslide.

 

Is it possible to apply them to the full intended force instantly?

Obviously not 'instantly': there will be some time constant while the magnetic field builds up.  Obviously not nearly instantly; the magnitude both of the peak current and of the effective risetime should be controlled (to prevent damage to the equipment).  Obviously not so quickly as to produce even momentary wheelslide.  Dynamic (or regenerative) is not an emergency brake, and to my knowledge even where it is a component of blended braking, it would be applied relatively slowly compared to the pneumatic brake.

If it is possible, is there any reason why this should not be done?

There is EVERY reason why that should not be done.  The potential gains from a couple of seconds' worth of high counter EMF would not justify the safety issues, or the wear and tear on the equipment, or the very real likelihood that something in the dynamic/regenerative brake would fail or degrade unespectedly and remove its contribution to safe stopping just at the moment you would put especial reliance on it.

This made me think of one of the responses that was tried during the wreck of the Federal at the time of Eisenhower's inaugural.  A rather obvious thing to do in the absence of air braking would be to try applying motor power 'in reverse' (Dave Klepper will explain why GG1s did not use regeneration for braking).  All that happened was that the overload breakers tripped as soon as any power was applied.

Can dynamic brakes be fully released from their maximum application by simply opening the circuit so the brakes instantly release as if turning off a light switch?

I gently suggest that you read up on magnetic induction before you hypothesize about this any further.  The answer to your question about 'opening a switch' is pretty much NO.  That's as in "HELL, NO!"  I could be considerably more emphatic, but will use restraint instead.

When you have a large current being developed through magnetic interaction, interrupting that current also involves interrupting the outrush current developed by collapse of the magnetic fields -- the time constant of which is remarkably short.  That will put very high momentary current across your switch contacts while they are still close enough together to develop an arc.  That would be very poor engineering.

With respect to the other question:  Even if the dynamic came off "instantaneously", you would not see a rebound effect in the draft gear, whether or not the 'braking' locomotive were ahead of or behind the node.  The cars have tremendous inertia, and there is inherent damping and friction in the draft gear, so there is no 'recoil action' that will cause any kind of 'violent' runout.  I would expect other effects, wind and grade in particular, to be exerting more force tending to bunch or separate cars than draft-gear runout.  (Other opinions, of course, welcome, provided they are based on science or experience.)

[Euclid]

Dynamic Braking Questions:

When dynamic brakes are applied, is the application developed in stages of progressively increasing retardation?  Is it possible to apply them to the full intended force instantly?  If it is possible, is there any reason why this should not be done?

Similarly, what is the method of releasing the dynamic brakes?  Can dynamic brakes be fully released from their maximum application by simply opening the circuit so the brakes instantly release as if turning off a light switch?

Here is what I am getting at:

Say a train is decelerating due to dynamic brakes applied on the one locomotive that is on the head end of the train.  Say the track is level.  Naturally, the slack will be bunched against the braking locomotive.  Say the dynamic brakes are suddenly fully released, while there continues to be no power being applied to pulling the train. 

If the dynamic brakes were suddenly fully released, would the energy stored in the compressed draft gear of the cars create a recoil action that would cause the slack to run out from the head end even though there is no pulling underway?  If so, could this runout be violent enough to part the train?    

[Wizlish]

dehusman

The end of the axle burning off due to a failed journal is something else, that's a journal failure. Two completely different things

Keep reading one more sentence - I edited some of that this morning after checking the link schlimm provided about how the statistics were tabulated.

 

A frozen bearing does not normally cause the axle end to "kick out of the side frame" (not sure what that would even be).

Neither am I, come to think about it.

I thought there were incidents where a seizing bearing would cause the wheel to slide, or 'flat' it enough that the vibration would cock the piece between the bearing and sideframe or actually knock the axle out of the seats; that's unlikely, for reasons you indicate.

[dehusman]

Wizlish

I would think that there's a difference between an axle end burning off due to bearing failure, and a wheelset seizing and sliding (with substantial heating of the axle but without the axle actually separating) due to bearing failure. So yes, I'd rank a burned-off axle failure as an axle failure (and in part this is because the consequences of this, in an accident, are those of axle failure rather than, say, a frozen bearing causing a wheelset axle end to kick out of the sideframe.)

An axle failure is when the axle itself breaks.

The end of the axle burning off due to a failed journal is something else, that's a journal failure.

Two completely different things.

A frozen bearing does normally cause the axle end to "kick out of the side frame" (not sure what that would even be).  If the bearing freezes then it still turns  in the side frame until the friction with teh side frame burns through the bearing race, then it burns through the axle end until to side frame drops.  That's how a bearing burns off.  Nothing "kicks out".

[schlimm]

http://safetydata.fra.dot.gov/OfficeofSafety/publicsite/Query/inccaus.aspx

FRA, all causes of various rail accidents, 4/2014 to 4/2015: 

Very few (4-7) caused by axle problems.  More from bearing failures (26).  Human errors of many types are the largest cause (736), the worst being switches not properly lined (101). Track problems are fewer in number, but caused greater dollar damage ($92 mil. out of a total of $300+ mil.).

[Wizlish]

Euclid

This has gotten me interested in broken axles, and it raises some questions that have been mentioned in the comments above by Wizlish.  It seems that broken axles are uncommon, but are often cited as the cause of derailments.  First of all, I would like to ask if the term “broken axle” include axles with ends burned off due to bearing failure.  If it does, how frequently do axles break due to that cause versus just snapping cold?   

Other than bearing failures, what causes an axle to break?  What can be done to prevent it?  If an axle breaks, are there any indications ahead of time?  What are the axle failure modes besides bearing failure?

Other than overloading, is there anything in operation that can damage an axle?  Can rough track cause overloading of axles?Axles

 

Axles are, I believe, heavily overdesigned precisely because they see such high and relatively unpredictable stresses in service, and because the result of many failure modes can be essentially (or quickly) catastrophic

Following are my opinions.  Don't trust 'em without more evidence, proof, and actual experts' opinions..

I would think that there's a difference between an axle end burning off due to bearing failure, and a wheelset seizing and sliding (with substantial heating of the axle but without the axle actually separating) due to bearing failure.  So yes, I'd rank a burned-off axle failure as an axle failure (and in part this is because the consequences of this, in an accident, are those of axle failure  rather than, say, a frozen bearing causing a wheelset axle end to kick out of the sideframe.)   But it's not up to me, it's up to whoever compiles the statistics (in the United States, the FRA).

EDIT -- they consider journal failure from bearing overheating its own failure code E53C; note that for 2014 there was no corresponding E53L reported for locomotives) and provide it smack-dab in the middle of other axle failures in the coding and in the reports from the very useful tool schlimm provides below.

I'd find the risk of an axle in otherwise good condition 'snapping' under any load applicable either in service or an accident to be relatively low.  Most of the failures will result from some form of crack propagation, probably associated with the reversing stress in parts of the axle as it rotates under load.  There are various factors that would make cracking start, and others that would make cracks propagate dangerously.  Various kinds of lateral and vertical road shock might, for example, be contributing factors.  Personally I think the likeliest source will be repetitive shock from wheeltread damage -- and I expect the current emphasis on keeping wheels in heavy service turned and well-profiled will have a great deal to do with limiting axle cracking.

Someone in the industry can discuss relative approaches concerning corrosion, particularly in the outside and inside of the 'fit' between the axle and a wheel.  You will notice the careful radius/fillet that is provided in good wheelset designs precisely to limit cracking in these regions. 

It's possible that there are some 'levered' loads due to lateral flange force or impact.  Again, you'll want an expert to explain precisely what the magnitude of peak force here is, and how it propagates up through the wheel; I can't do it for you.  I can only note that it was observed a while back that the recorded peak shock force in g recorded on tested Amtrak wheelsets could be in the 180g range, which even if of relatively short duration seemed surprising to some of the better heads on this forum.

I'm going to sit back and watch as more experts flesh this subtopic out.

[Euclid]

This has gotten me interested in broken axles, and it raises some questions that have been mentioned in the comments above by Wizlish.  It seems that broken axles are uncommon, but are often cited as the cause of derailments.  First of all, I would like to ask if the term “broken axle” include axles with ends burned off due to bearing failure.  If it does, how frequently do axles break due to that cause versus just snapping cold?   

Other than bearing failures, what causes an axle to break?  What can be done to prevent it?  If an axle breaks, are there any indications ahead of time?  What are the axle failure modes besides bearing failure?

Other than overloading, is there anything in operation that can damage an axle?  Can rough track cause overloading of axles?

[Wizlish]

schlimm

How are "axle problems" detected currently?

Mostly, as he said, by the method in the old engineering chestnut "Don't exceed maximum loading" -- what's that?  "Oops, it broke"

I'm sure at least some sections of axles are NDT tested when a wheelset is taken out of the truck for some reason, perhaps each time the wheels are turned.  As noted, there's presently no good way to perform 'direct' testing with the axle in the train, and a visual inspection is probably not likely to show anything in many of the areas where axles are likely to break.

I recall seeing at least one paper that talks about vibration signatures in the wheelset that can indicate incipient axle failure from progressive crack propagation.  I don't have the reference but I suspect buslist would know.  My suspicion is that in the practical world, the situation is normally taken care of by good quality control and by taking wheelsets out of service before axle cracking can start (or at least become critical).  

Whether or not there are nonlinear causes of stress raising in HAL wheelsets, I don't know, but would be interested to find out about them if so.

[schlimm]

How are "axle problems" detected currently?

 

[jeffhergert]

I've had to stop many times to do bearing checks requested by the "back office" Dave speaks of.  Both as a conductor doing the actual inspection and as an engineer, just stopping to allow the conductor to go back.  The requests from the hot-bearing desk come, like Dave has said, because a specific bearing(s) has/have been hotter than the rest of the train over consecutive detectors.  That doesn't mean the bearings set off any of the detectors, the heat difference wasn't enough to do so.  That's why they request a field check before it gets bad enough to se off the detector.

Often, the check of the car doesn't find a failing bearing, but a hand brake not fully released, an air brake that doesn't fully release due to defect or a retainer some how set to a retaining position.  If those conditions are corrected, the car is good to go.

The last time I remember having to stop to check a car, we had to interupt the dispatcher for a moment.  We were about to clear a detector and get the exit message.  It was "no defects" and then the dispatcher told us to stop and inspect a car.  The conductor found a slight hand brake.

Jeff