Side Pressure on Rails in a Curve

[Northtowne]

A yard job was taking a string of cars on a branch line to an interchange yard close by. It had to pull past a switch on the branch and push the cars through the switch on to a spur to the interchange yard (all tracks there were dead end). It derailed on the curve just past the switch. I went to see it and about 40 or so ln ft of outside rail lay on its side, about 8/10 car lengths from the switch; about 8 cars were still on the tracks past the overturned rail. Those cars were some tanks and normal open hoppers, empties, plus one centerbeam car loaded with lumber. The next car(going toward the engine), also cen bm, loaded, was on its side along side the overturned rail. The next car, also cen bm loaded, was upright but one end of its trucks on the ballast. The next car, also a cen bm loaded was on good tracks still coupled to the upright one. Then there were about 5 mixed cars and then the engine all coupled on undisturbed track. My question to all is; could the loaded cen bm cars (81 ft inside length), at a point where two or more are coulpled together, being pushed on a curve, exert enough pressure to overturn the outside rail? As I remember, it was welded rail, I could be wrong. The track is well maintained ;I used to work at a plant along side this spur for many yrs. 

Northtowne 

[mudchicken]

Short answer is yes.... all it takes is one centerbound truck under load putting the rails in rotation, a long car/short car scenario or limited throw couplers (think newer locos)in the mix. (in compression / slack run in)

 

How sharp is the curve? (over six degree cv?)

[ChuckCobleigh]

mudchicken

Short answer is yes.... all it takes is one centerbound truck under load putting the rails in rotation, a long car/short car scenario or limited throw couplers (think newer locos)in the mix. (in compression / slack run in)

Wasn't there some discussion recently regarding outer rail rolling being more likely on a shove? Or would the inner rail be more at risk on a pulling move?

Another random question coming to mind is whether Pandrol clips are more or less susceptible to roll than spikes. Given that spikes are much more likely in yards and industrial leads, that's maybe not much of an issue?

[Northtowne]

Gonna look at the curve again and make a guess as to the radius. I recall that it was not too severe (degree) but could not help but imagine the pressure that the long loaded lunber cars would put on the outside rail; two sets of trucks close together at the coupler trying to go straight. And you gave me a yes; I am a neophite on such things.

Northtowne

[Northtowne]

Could have been a similiar discussion before; I have been off the forum for a good while.

Northtowne

[7j43k]

Northtowne

Gonna look at the curve again and make a guess as to the radius.

 

 

Don't need to guess.  You can probably scale off of an overhead shot on Google Satellite.

It might be interesting to check the frog number in person, though.

 

 

Ed

[mudchicken]

ChuckCobleigh

Another random question coming to mind is whether Pandrol clips are more or less susceptible to roll than spikes. Given that spikes are much more likely in yards and industrial leads, that's maybe not much of an issue?

 

Pandrols and D-E Clips can behave like springs. They can break or work loose under heavy tonnage wear. The issue often becomes what type of fasteners are holding the tie plate down and does the rail stay seated. Spikes often ratchet as will hairpins. Screws, depending on the type, have their own weirdisms depending on the tie material. 

 

Your yards and backtracks often have lighter rail weights that are not suitable for pandrols, have anchors to worn to be effective and have ties less able to hold a fastener ... so guess what happens...

[Overmod]

ChuckCobleigh

Wasn't there some discussion recently regarding outer rail rolling being more likely on a shove? Or would the inner rail be more at risk on a pulling move?

Perhaps a more significant question is whether the degree of 'rail rolling' (or lateral displacement of the structure causing geometry distortion) is greater than the tendency of the wheelset(s) causing the lateral force to climb or rock into derailment or tipover.  My impression is that you'd be much more likely to get the latter in a consist either pulled or shoved around a curve, well before good rail securement would lift enough for significant rail rolling.

Another random question coming to mind is whether Pandrol clips are more or less susceptible to roll than spikes.

The answer is a bit more complicated than this: of course the rail is more 'susceptible' to being rolled outward as the clips spring a bit, but of course the clamping force is then re-established consistently once the force is no longer being applied, and the rail will then 'roll back' to summarily the same position in the tieplates that it should.  This is of a piece with the action of Pandrol clips under severe vibration (which would tend to lift spikes or even 'work' screw fasteners loose) which can lift substantially, perhaps enough to let the rail move or work longitudinally over time, but then return the rail to proper alignment on the tieplate and tie soon afterward.

I do think the practical effect on 'rail rolling' as a gross failure mechanism is much less for Pandrol systems, as the effective spring resistance will rise sharply as the clips are compressed, up to the point the metal in the clips distorts beyond the yield point, with only a few degrees of effective rotation around the 'fulcrum' of the opposite edge of the base in the tieplate groove.  I doubt this is anywhere near enough to facilitate permanent rail misalignment, but it would increase the 'angle' of the railhead assisting effective cam-out of the flange and fillet (and consequent derailment).

[Erik_Mag]

Northtowne

Could have been a similiar discussion before; I have been off the forum for a good while.

Northtowne

 

Your question has provoked some very interesting discussion, so I'm glad you asked it.

[gregc]

Overmod

Perhaps a more significant question is whether the degree of 'rail rolling' (or lateral displacement of the structure causing geometry distortion) is greater than the tendency of the wheelset(s) causing the lateral force to climb or rock into derailment or tipover.

i don't know if the above is saying that ... won't a rail that rolled, have to lift up the wheel?

or did the rail just get pushed out of  position?

were the spike loose (or loosened)?

[Overmod]

gregc

Overmod

Perhaps a more significant question is whether the degree of 'rail rolling' (or lateral displacement of the structure causing geometry distortion) is greater than the tendency of the wheelset(s) causing the lateral force to climb or rock into derailment or tipover.

I don't know if the above is saying that ... won't a rail that rolled, have to lift up the wheel? ... or did the rail just get pushed out of  position? were the spike loose (or loosened)?

None of the above: the flanges are only 3/4" high, with a helpful fillet and taper to their 'action', so I expect it's much more likely that they will be lifted enough to jump over the railhead (into derailment) or will serve as an effective enough fulcrum that the car tips over before the force acting laterally will permanently move a rail.  And these two effects are what I predominantly see (or think I see) as the proximate cause of stringline accidents on modern and well-maintained track.

Even in the bad old days of harmonic rock, I think far more derailments occurred through wheels leaving the correct 'relationship' between fillet and gauge corner than by the harmonic force buildup actually moving or turning over rails.  I'm prepared to be corrected on that point if wrong.

Now, after a good derailment, I expect to see damage to the track geometry, perhaps substantial damage.  This has to be distinguished from damage caused strictly by wheelset interaction up to the point of derailment.

 

Something else we might want to shove in -- no pun intended -- at this point is the kind of failure that caused the oil-train accident in Oregon a couple of years ago: any sort of transient or progressive gauge failure that allows a wheelset or truck pair to drop into the gauge.  At that point lateral force inducing massive failure of one or the other rail's attachment, and then progressive between-the-rails derailment, becomes likely, no matter what form of rail securement is in use...

[Northtowne]

To Overmod;

You got me thinking, would be interesting if we could somehow calculate the forces involved on a rail in a curve. We could come close to quantifying the vertical force, which you therorize is lifted enough to derail. The lateral force would be almost indeterminate. It surely is a function of car length and weight, degree of curature, shove or pull, rail size, and speed, at least. I still think, in my mind, that it would take a tremendous force to lift the wheels up over the head, even with the short distance (3/4 in), and the taper at the flange (is the taper the fulcrum effect you speak of?). Weight (vertical force) is what makes the railroads work.

Northtowne

[Overmod]

Northtowne

You got me thinking, would be interesting if we could somehow calculate the forces involved on a rail in a curve.

I think it's not difficult to do this, as it's the reaction vector to what the car experiences.  A starting place might be to use Don Oltmann's formula for stringlining accidents, and consider that any moment tending to lift or rotate the car will have equal and opposite vector into the rail, moment by moment, at each point of contact.  You could then treat the sections of 'beam' rail in between, as constrained by the securement system to the ties, in torsion.

A little more complicated, and possibly empirical, is how the force on the rail deforms or strains the securement means, rather than moving the ties relative to the grade.  Some of the formulas for sun kink may provide the order of magnitude of force needed to shift trackwork from longitudinal rail stress that results in lateral force expression.  These can then be compared with the vector force applied in a stringline or similar situation.

[PNWRMNM]

Computer programs to calculate sideways pressure on rails and wheels have been around since the 1970's under the heading 'Train/Track Dynamics'. Southern Pacific pioneered this work and the AAR picked it up shortly after. That was mainframe at the time, but I am sure there is at least one program that will run on a PC.

[Overmod]

PNWRMNM

Computer programs to calculate sideways pressure on rails and wheels have been around since the 1970s

The French had some very sophisticated track models by the early '70s for the design of the LGV (that was the first place I saw sixth-order differential equations used in mechanics!) and those were reduced to code, as you might expect, as quickly as practical.

I had thought he wanted a relatively simple model that he could get results from 'by hand' rather than multibody simulation, but you are very correct that computer support exists and may be at least relatively available to him, particularly if he's in an 'academic environment'.

Here is an interesting PDF from last year

that contains much of the information to get him 'up to speed' on the alternatives.

[tree68]

PNWRMNM

That was mainframe at the time, but I am sure there is at least one program that will run on a PC.

Given the growth of computing power, etc, the same program will probably run on your smart phone or tablet today.  Our IBM 4361 had hard drives of ~792Mb mounted in dishwasher sized cabinets - I think there were at least a dozen of them, maybe more.  You can get that much memory now on a chip the size of a dime.

[Overmod]

tree68

Given the growth of computing power, etc, the same program will probably run on your smart phone or tablet today.

Assuming somebody ported it to run on iOS or Android, or has an emulator that runs on that hardware...

Our IBM 4361 had [at least a dozen] hard drives of ~792Mb mounted in dishwasher sized cabinets ... You can get that much memory now on a chip the size of a dime.

But to use a '4361-compatible' program, you'd need to port an emulator like Hercules, then deal with licensing issues for the "IBM OS" to run in emulation, then go through all the rigmarole of creating a virtual DASD drive and getting the program loaded on it ... all to get cryptic and possibly primitive results*.  Not only hardware has improved by leaps and bounds!

Meanwhile, here is the ADAMS program (apparently now not all upper-case!) student edition for easy download to your computer, ready to go.  Interesting to see if anyone has done the port from laptop-compatible to phone or minimal tablet.

Resonate says you can get an academic license for VAMPIRE Pro; contact them at [e-mail] or contact ENSCO Rail (their channel partner in the USA).

 

*I am not now, or in future, pretending to be a mainframe guy, and I'm sure those who are will be chuckling at some of the naivete in this example as given.  It's supposed to be an illustration of some of the issues, not an attempt at polymath tech sophistication...

[tree68]

Overmod

Assuming somebody ported it to run on iOS or Android, or has an emulator that runs on that hardware...

Overmod

But to use a '4361-compatible' program, you'd need to port an emulator like Hercules,

I was thinking more in terms of the relative computing power than running those specific programs.  I'm sure some enterprising individual could write an app from scratch to do the job, though.