Hacker News Reader

Comment by dredmorbius | Hacker News Reader

It's rail wheels which are beveled, with the radius decreasing to the outside of the wheel. Your top image has the wheels reversed. There's also a flange which extends below the rail depth on the inner edge of the wheel

Should be, showing from the axel to the rail:

  _   __________________________    _   
  _   __________________________    _   
   |  |                         |  |
   |  |                         |  |
   |  |                         |  |
   \  |                         |  /
    \_|                         |_/
      |                         |

As the car rounds turns, the inner wheel rides down and the outer wheel rides up, increasing the relative rotational radius of the outer wheel. This compensates for the fact that rail wheelsets rotate on a single fixed axis, not independently, and without a differential.

For a non-ASCII image see: https://steemitimages.com/DQmUjS7PbhzAYpkTi5A3c2QQQa7jn1c52s...

Longer explanation: https://www.popularmechanics.com/science/a25581/science-behi...

Wikipedia also discusses geometry: https://en.wikipedia.org/wiki/Train_wheel

View on HN

Yes, it is the wheels that are beveled. However, the BART wheels are not beveled, they are flat (cylindrical) at the part where the cars usually ride. See https://www.reddit.com/r/bayarea/comments/2t7x49/the_bart_wh....

This is what causes a lot of noise and wear and tear. I actually researched this and it seems that it was not as stupid of a decision as a lot of people say it is. There is a problem with beveled wheels -- it is called hunting. This is when the car oscillates from side to side. Hunting is unpleasant, also causes wear and tear and noise and creates a danger of derailing.

Hunting becomes more likely if the car is lighter and if it is going high speed. The bart was designed to be lighter than most existing metro systems and to go much faster than existing metro systems in certain corridors. So they wanted to prevent hunting and thus made a portion of the wheel cylindrical.

So it seems that they just made another engineering trade off. And I am sure they had justification. But I do wish that when engineers make these trade offs they take human comfort into account. I does not seem that they gave sound much of a weight when making their trade offs.

I think the solution is to keep the wheels cylindrical but just decouple them. I.e., not to have the right and left wheel connected by an axle. If they are independent they can move at different speeds on the turns and will not cause grinding and noise. And the wheels that are driven may be driven by different electric motors that drive the wheels by the necessary different speeds on turns. This is called vectoring and is already being done for some electric cars.

I was explaining the shape of traditional rail wheels.

BART's since adopted a very-slightly-beveled wheel design pictured here:

https://sf.streetsblog.org/wp-content/uploads/sites/3/2016/0...

Story:

https://sf.streetsblog.org/2016/04/07/new-bart-cars-show-age...

Note that that's as of 2016, six years ago.

The hunting phenomenon is real and has some interesting consequences. For freight trains, empty boxcars have a lower speed limit than fully-laden ones, as the hunting behaviour is greater with less weight.

High-speed rail systems required understanding and managing hunting action. Getting a case of the wobblies at 300 kph is an exciting but short ride.

Of anyone wants to see what hunting looks like, then watch a video of the London DLR.

The DLR has very tight corners, and was originally built on very tight budget. As a result it uses very steeply bevelled wheels to handle the corners.

The wheels do a great job on the corners, but on the straight they suffer from significant hunting, resulting in the trains wobbling side-to-side (and slightly up-and-down), at a rather alarming frequency and amplitude. Not quite enough to make you sea sick, but it’s pretty close.

The DLR is a low-speed network with tight turning radii. I haven't found any footage clearly showing hunting (in part because it's difficult to distinguish other lateral-movement-inducing elements such as turns and switches), but the basic operating parameters are fairly clear.

I suspect the wheels have high conicity. The rolling stock is also likely lighter than a heavy-rail system.

https://yewtu.be/watch?v=i1pAfKsL1bk

Oh trust me, the hunting on the DLR is pretty bad. You’ll notice the second the trains get up any speed on a straight. The problem is bad enough for it appear in its own section of the DLR Rolling stock wiki page.

https://en.m.wikipedia.org/wiki/Docklands_Light_Railway_roll...

Here’s good video about the DLR time stamped to point that demonstrates the hunting well.

https://youtu.be/_WosYRVYMEM?t=907

Thanks.

The video confirms what I'd surmised: that tight turns were the reason for more conical wheels. The sections I watched didn't actually show the wheel geometry, though, and I've not been able to find any photos online (DDG and Google Image search, various combinations of "DLR | Docklands Light Railroad | (conical|cone|conicity) wheel"). The WIkipeidia DLR Rolling Stock article mentions the conicity and hunting issues, but doesn't show the wheels either:

https://en.wikipedia.org/wiki/Docklands_Light_Railway_rollin...

Well, there's a reason they named it the Docklands Light Railway!

(Large stretches of the original network built in the 1980s ran on elevated track, supported on pylons over ground that's a mixture of chalk and clay and tends to waterlog. Heavy rail can be built over and under that part of the Thames, but it has its own special challenges and the DLR was built as the cheapest way of getting commuters in and out of the Canary Wharf office towers, from Tower Gateway and the actual London Underground network to the west. Then from the 1990s onwards mission creep set in ...)

its like the lifeline of east london, and it connected large aread of land that used to be borderline slums. Now there are highriae apartments and the fact that they have access to piblic trandport has increased their valuation by billions.

Providing improved transport and transit is amongst the best way of increasing property values. A stellar example is Denver, Colorado, which grew in size tenfold in a decade after a branch line connected it to the Transcontinental Railroad in 1870, becoming the dominant Rocky Mountain region city, a title which might otherwise have gone to Chyenne, Wyoming.

It's also something of a two-edged sword: the property value boost makes further transit rights of way far more expensive to acquire. I've noticed in particular that much rail development happens either as greenfield projects within countries as they're first industrialising (UK and US in the 19th century, China in the late 20th), or following a devastating war or other economic setback --- Japan's shinkansen and to a somewhat lesser degree France's TGV are both legacies of WWII.

In the US, aside from political opposition from other parts of the transportation sector, it's wealthy landowners who are the most significant obstacle to new development.

Ha! Absoltuely correct and I'm honoured to be Strossed.

Thanks for the, er, bedrock explanation as well.

The Shinkansen also use cylindrical wheels on flat rails. The solution is to spend a large amount of effort and money into maintaining both the wheels and the rails. The amount of maintenance that BART train wheels and tracks get is much lower than what is done to keep Shinkansen running.

Hunting in modern rail vehicles is largely a solved problem using axle yaw dampers and other systems. I struggle to see why BART doesnt use a standard wheel profile and buy from a rolling stock supplier with experience of designing for good ride quality.

They do now, if you search for it there is a presentation on the transition to conic wheels somewhere out there.

https://news.ycombinator.com/item?id=32033094

Links a 2016 article.

is there a reason why they don't have the wheels spin separately?

Does it make it more complicated? Or there is a disadvantage?

The fixed wheels, plus the bevel create a self-centering effect.

As the wheels move off centre, the bevel makes one wheel have a larger effective diameter and the other wheel a smaller effective diameter. Because the wheels have a solid axel, that forces one wheel to start travelling slightly faster than the other, causing the pair of wheels to turn.

With the bevel making the wheels larger on the inside, it means that as a pair of wheels drifts right relative to the tracks, the right hand wheel rides up, starts behaving as if it’s larger, and then starts to overtake the left wheel (which has ridden down, and behaves as if it’s smaller), creating a turning effect to the left, correcting the rightward drift.

When setup correctly, and travelling at a reasonable speed, the big metal lip on train wheels should never touch the track. The self-centering effect does all the work, the lip is just a backup to deal with overly tight turns, points and extremely slow running.

I get the principle of how it works - though it seems imperfect bc it would vary depending on the weight/load. But why favor such a mechanism to just independent wheels?

Or are you saying the whole setup actually tilts the carriage into the turn - helping to prevent it from tipping?

> though it seems imperfect bc it would vary depending on the weight/load.

Weight and load have no impact of the effect. The only requirement is that the wheels remain adhered to the track, and don’t slip. Which is expected anyway, because it would be tricky to accelerate and decelerate a train with slipping wheels.

With independent wheels you have no self-centering effect, so only think keeping your wheels point in the right direction would be inner flange on the wheels colliding with the track. Such collisions obvious involve quite a bit of friction, noise and track/wheel wear.

Not to mention independent wheels are more complex to build. Two wheels directly attached to a single fixed axel is about a simple as you can get. You only need two couplings between your single axel and non-rotating part of your train. With independent wheels you need four, or more complex wheel designs to allow bearings at the wheel/axel coupling.

There no need to tip carriages into turns to prevent tipping, gravity, track banking, and speed limits already assure that tipping isn’t an issue. Tilting trains only exist to increase passenger comfort, not safety. Tilting trains are prefect capable of operating a full line speed with their tilting mechanism disabled or jammed, it’ll just be uncomfortable for the passengers.

One minor correction. Tilt trains exist to allow higher speed service on older lines where tipping would be a safety problem. They can run just fine with the tilt mechanism disabled, provided they go at the lower speed of the older trains.

> Tilt trains exist to allow higher speed service on older lines where tipping would be a safety problem.

That isn’t true. No train network is going to allow train to operate at speed, if the safety of those trains depends on non-failure of a complex tilting mechanism. Having a tilting mechanism fail or jam before a corner (which has happened) can not result in a high speed derailment. That risk is simply unacceptable on modern railway.

Tilting trains make it possible to run at higher speeds on older lines by allowing older lines to have the camber on corners increased to accommodate the higher speeds. But the camber can’t increased enough to hit optimal amount of camber without make the rail unusable to older trains.

Tilting trains provide a useful compromise. Increase the camber enough for stability (but not comfort), while keeping the camber acceptable for older trains, then make up the rest of the tilt with carriage tilting mechanism. The total amount of tilt is then comfortable for passengers. There is a minor safety element provided by tilting trains, in that lateral forces in cornering are lower, thus passengers are less likely to fall, and items won’t come of tables. But the tilting mechanism certainly is a safety critical element to running faster on old lines.

It’s also worth mentioning that carriages on passenger trains have a pretty low center of gravity, as all the heavy equipment (motors, bogies, breaks, suspension) sits below the passenger compartment. As a result tilting a passenger train won’t significantly move its center of gravity, so impact on total train stability is going to pretty minor.

I'd always understood tilt trains to be a comfort factor rather than safety.

Tilting a carriage isn't going to adjust the CoG by any appreciable amount. It will tend to make the net centripetal moment more virtical from the reference frame of passengers, however.

Wikipedia's article corroborates this:

As a train (or other vehicle) rounds a curve at speed, objects inside the train experience centrifugal force. This can cause packages to slide about or seated passengers to feel squashed by the outboard armrest, and standing passengers to lose their balance. Tilting trains are designed to counteract this by tilting the carriages towards the inside of the curve, thus compensating for the g-force.

https://en.wikipedia.org/wiki/Tilting_train

If the wheels were independent the lips on the wheels would constantly ride on the inner sides of the rails. That increases rolling resistance and wear.

There are some trains which use independent wheels, but they need complicated mechanical centering mechanisms to replace the self-centering effect, and these are more expensive to maintain.

Any examples / explanations of which you're aware?

It's a matter of balancing straightaway and turning performance.

For a good explainer see: https://yewtu.be/watch?v=SRsm7mv0Oh8

With separted axles, you wouldn't get the climbing behaviour of the outside wheel. With a fixed axle, the idealised shape of the wheels is of a double-ended cone (see the video for an example) which can ride higher on the outside of a turn. Even with conical wheels, with independent axels there's no way to impart a turning force on the carriage. I'm not entirely sure what the end result would be, but I suspect it would be a far harsher ride, and much more track and wheel wear. Probably a much higher probability of derailments as well.

Cone angle varies by type of train and speeds. The video shows both freight and high-speed rail designs and discusses characteristics of both.

My guess is at a minimum maintaining the bearings would be more difficult.

A solid axel allows placing an external bearings between the truck and the axel.

Hub bearings are internal and service would require dealing with the nontrivial weight of a loose wheel.

See https://en.m.wikipedia.org/wiki/List_of_railroad_truck_parts

It makes it more complicated. The wheels need to be fixed to the axel for breaks to function, so you can’t just stick a bearing in the wheel and call it a day.

Don’t rail brakes generally act on the rolling surface of the wheel? Why would it be any different with independent wheels?

Not regenerative brakes. Only friction brakes work that way.

Regen braking would operate through the same electic motor which powers the wheels in the first place.

https://connectorsupplier.com/regenerative-braking-systems-r...

I'm imagining two half length axels. I guess this creates some torque but the breaks and bearing would be identical

Then you'd still need bearings in the middle of the axle somewhere.

This seems unlikely.

Reply on news.ycombinator.com