Physics of derailments

[FJ and G]

The wonder of it all is not that our model trains sometimes derail, but that the real ones derail far less.

The thought occurred coming home last night on the VRE. As the train pulled into the Alexandria Station I observed the modified SD-45 (yes it is still being used) swaying from side to side, its trucks "wandering." Later, we crossed a bridge over a deep chasm and as the train entered the trestle, it leaped up into the air and after crossing it sunk back down quite abruptly.

Some of you very serious modelers have gone Proto 87 (or proto 48). Your wheel flanges are very tiny indeed, approximating the 1 inch real ones. The risk of derailment is even greater for you if the trackwork isn’t exactly right on. Yet, even non "proto" modelers experience derailments on the best of layouts.

Why then, don’t real trains derail a lot more than they do?

Obviously, there is more weight on them to hold them down and the track is less curvy.

But even on the most perfectly laid model railroad track, the risk of a derailment is still greater than on the real thing.

Any opinions or comments?

[Anonymous]

It has to do with newton's law of motion, or inertia. The reals ones are very massive and require much more energy to sway them from the straight line they want to travel in. Our models are scale size, but not scale weight. If a real loco, say an sd45 weighs 390,000 lbs our 1/87 models would need weigh over 4000 lbs to be scale weight. How can this be? Well, when something is twice as big it's really 8 times heavier. Take 1 block and make it twice as big. So now its two wide, two long, and... 2 high. That's 8 blocks! So if you model sd45 weighed 4000 lbs it would stay on the rails better. Also, if its spinning wheels were as heavy as they should be they would also track better (just like bicycles with bigger wheels) due to centrifugal or gyroscopic effects. So the mass of the real loco and train is what allows them to do things models can't. In fact, most minor derailments occure at low speeds where the above effects are less. A friend who worked for the RR says the wheels wobble and hunt the most at 15 mph which is also the yard speed limit. I also didn't mention springs which the real ones have and our's generally don't. Springs push down against the truck-track-ground as well as up against the load. So if the track dips the wheels will drop much faster due to their lesser mass than the load. This keeps them in contact better with the track than our solid units can. FRED

[Puckdropper]

I have a couple cars with sprung trucks... I'll have to watch them to see if the action is the same.

[ndbprr]

The weight is the primary factor. A 100 ton engine (probably very underweight) should weigh 1/87th in HO or 1.15 tons if the wieght was scaled. 1.15 pounds is more like it and it does affect tracking along with no suspension like a real engine has which is designed to keep all wheels on the rails regardless of what the other axles are doing.

[Anonymous]

Yes,200,000 lbs is way light, I looked it up and 390,000 lbs is the correct weight for an sd35...FRED

[FJ and G]

Very interesting! So, I would need to build my layout on Neptune so that my tiny locomotive would be the correct weight. I then might have to strengthen my track and trestles as well, and get a more powerful motor!

[Anonymous]

The HO trains are 1/87th in each of 3 (three) dimensions. Therefore the weight should be 1/87^3 (power of three) or 1/(87x87x87) or 1/658503. A 400,000lb engine should be an approximately 0.61 lb model or about 9 1/2 ounces.

However, since weight and dimension are different measures, this may not be exactly correct because they don't scale the same way, but you get the idea. It takes 658,000+ models to fill the same volume as a real engine, not 87.

Andrew

[FJ and G]

Andrew,

The plot thickens then. 9 1/2 oz. would put our little locomotives in the correct size/weight proportions as their bigger siblings. Another explanation then would be needed to explain the derailment phenomenon.

Maybe this falls in the "unsolved mysteries" file?

[Anonymous]

I stand corrected. FRED

[nfmisso]

On real trains, on good track, the flanges do not touch the rails. The wheels are conical. As they wander towards one side, the diameter of the wheel closer to the rail is increasing, and the one moving away from the rail (towards the center of the two rails) gets smaller. This causes a correcting force that pushs the axle back so that both wheels are running on the same diameter.

On curves, similar happens.

Nothing to do with weight, inertia, etc. compared to out models. The problem with our models is size, if you could keep the track proportionally clean, and smooth, and had conical wheels, model trains would run as well as the real ones. P87 etc, is an attempt to do that.

[FJ and G]

Nigel,

Your idea makes sense. However, I've seen stretches of atrocious looking track with flat spots and if you stand near a moving train you can also hear flat spots on wheels and they repeatedly hit. As for clean track that is not always the case on the prototype either. I've watched a locomotive cut a tire in two that some kids had placed on the New Haven tracks in The Bronx in the 60s. But, there, you have weight also being a factor in crushing it (actually it exploded the tire when it hit, so great was the force). Overall, though, you may have something.

Interesting would be to compare causes of derailments of real vs model trains.

For instance, heat is sometimes a factor in real train derailments (rail expansion) but probably not so much in the model ones as we are usually spoiled with climate control.

Also, sharp S curves might result in more model trains stringlining than in the real (though a recent case out West proved that stringlining really occurs in the prototype as well).

Also, model trains probably get banged a lot harder during coupling (proportionately) than the real ones, unless you are running very slow when coupling.

However, let us return to the original premise of well-layed model railroad track so that this topic does not broaden too much from the original thought on the physics of derailments.

If P87 track/wheels were perfectly contoured I'm betting that there would still be more derailments than on the real track.

[Puckdropper]

I remember reading a question in a magazine somewhere that the prototype railroads purposely lift the outer edge of the rail so the wheels make better contact. It was explained better in the article. (Which was part of a mail-answering thing...)

[dehusman]

One difference between and real cars is that real cars have much more flexible trucks so they can adjust to track irregularities more easily.
The other is that when you really come down to it, speed for speed, real track has much tighter standards (gauge, surface, crosslevel, line) than track does.

Regarding "lifting the outer edge of the rail", prototype tie plates have a 40:1 cant to the bearing surface, matching the conical taper of the wheels. Tie plates in switches are generally flat.

[Puckdropper]

More comments (Hey! You don't get two stars by just reading!)

I believe most of the weight in an empty car is in the trucks. Using a heavier truck may help avoid derailments.

I wonder if putting in that 40:1 cant to match the MSI's layout would help it's longevity... Most other layouts don't get run enough to make it worthwhile...

[Anonymous]

A few things factor in here:

Real cars and locomotives have a MUCH greater percentage of their weight in their trucks than our models do, and therefore their center of gravity is lower. The weight of a plastic truck is no where near "scale weight"

Our Turnouts are a LONG way from the dimensions of the RL ones. It would take turnouts 2 feet long to begin approximating prototype turnout lengths

Our turn radii are MUCH smaller than they would be IRL in which a radius could be measured in miles not inches.

No matter how hard we try, our rail will likely never be laid to the same standards. A bump in the rail of mere millimeters we'd likely never notice but would be scale inches in size and the prototype would likely fix it.

Conversely, things the prototype would never notice or care about will derail one of our trains in a heartbeat -- like ballast. A real train would roll over ballast and blow it to pieces without a second thought. Try that with your model, it will likely snag and derail when it hits a little piece of ballast. This is again a factor of weight.

And lastly, trains derail a LOT more often then most people realize. Most just aren't big derailments that make the news -- and even big ones won't make the news if it's not near a neighborhood (and most mainline rail isn't). Switchers in yards and locals will often derail every single day due largely to the condition of track they operate on.

[IRONROOSTER]

Derailments?? I don't have derailments. I weight all my cars to NMRA recommendations and replace all plastic wheels with metal ones. Oh and I run in S scale on code 100 Shinora track 36" minimum radius with easements and no. 6 turnouts. I think the extra weight of S helps. I can even back a 17 car train around the layout without a problem.
Enjoy
Paul

[Anonymous]

Talk to Newton! You cannot scale gravity or mass effects. Our track is no model of the real rail in size or dimensions. Our spirals are only a few scale feet long. Our grades are frequently steep. The operators may or maynot be formally qualified. All things considered the real rails should get the performance demonstrated in the model world.

Lindsay Smith

[Anonymous]

I think that hyperion hit a few good points. I think it is mosly attributed to four things. First, you can't take our curvature out of the question. Our sharp curves and turnouts make a big difference. Second, our tolerances are no where near the prototype values after you account for scale. A half inch depression at one tie is a defect. But in HO scale that turns out to be less than one third of a millimeter. I suspect each of our rail joints qualifies as a defect then. Similar tolerances effect truck hunting and other performance factors. Third, dirt on our trucks and track make a huge difference but would not have a chance to build up on the prototype as it does on our model stuff. Scale again has a lot to do with this. Dirt doesn't scale, it is always prototypical size. Fourth and finally, operations of our trains do not scale. We are much too rough with our models if protypical standards are to be scaled as well. We bang 'em and shove 'em and jerk 'em until it is surprising we don't jerk the couplers out more often. Bottom line, we can try and do a better job, but unless you are a fanatic, you will always experience more derailments than the prototype for the same number of ton-miles (oz-feet?). Oh, and there realy are a lot more derailments than you may think on the real thing. - Ed

[Anonymous]

Our scale trains are much more ridgid with no flex in the track as they pass over. Real trucks are sloppy in their tracking compared to the stiffness of our scaled-down plastic ones. If my N scale track was as flexable as prototype rail, it would be limp as a wet noodle( just watch the prototype as they unload or pick up quarter-mile pieces of rail). The full size track and trucks are just more forgiving thanks to the flexability of their larger mass.