Sunday, April 29, 2012

141> GOT IT!

If you have followed this blog for any length of time, you have witnessed version after version of bogies and track. I started this blog with the na├»ve hope that I could recruit a bunch of mechanical engineers to create an open source PRT design. Such a consensual design would have had a credibility that was especially lacking back in those dark days before Heathrow and Masdar. Well, lacking the participation I had hoped for, progress has been pretty slow, although, in my own defense, I’ve set the bar pretty high, not just in terms of performance capabilities, but in terms of simplicity, cost and practicality. That’s a bit like trying to mix oil and water.

Anyway, folks, get out the Champaign and the party hats, because I am pretty sure the search is basically over, although I’m still studying tight-quarters switching, and a few other details. Let me reiterate, for new readers, what differentiates these designs. The challenge is to go faster, climb steeper, turn tighter, be less expensive, and therefore be more versatile and practical than other transportation alternatives. I seek a fully 3D solution instead of a system that requires raised stations or long ramps. The pictures here show only the bare essentials. Most structure, including the passenger compartment and the track’s support means have been left out for clarity. 

Here are the highlights of the changes. First, I have replaced the round running guides (A) with square tubing tilted to make a diamond profile. This is almost as easy to bend as round pipe, but offers flat running surfaces that won’t wear grooves into the guide wheels. I have inverted the drive wheel flanges (B) to be convex instead of concave. This results in a single point of angular contact that will be very prone to wear, but for the fact is that these flanges only need to come into contact for specific, short term duties, most of them at very low speeds. For switching they are employed for only a few dozen revolutions. The steering guide wheels (C) and what were called “hold-down” wheels in previous iterations have been combined. Now, when ascending or descending very steep or even vertical runs of track, both of these wheels can be raised to serve the hold-down function. The new geometry makes their previous “anti-rock” function unneeded. The running guide wheels (D) and steering guide wheels (C) must now, though, all move independently, instead of in pairs. Here is how switching would generally work in four pictures.

1. The first of the four is the ordinary running configuration. The drive wheel flanges, although close, do not actually make contact, so all wear is on the larger softer, smoother, and quieter hard rubber wheel surfaces.
2. Several steps have taken place here. First, a steering guide wheel is raised. Next the pairs of diamond guides “taper” into a position where contact can be made. (They start spread out.) Meanwhile, the drive wheel guides make contact with the plastic wheel flanges. This will probably be accomplished by raising the guide wheel slightly on the side that isn’t being lowered. This will pull the bogie out of center, clamping the track between the flange and the guide wheel.  
3. With the prior steps taken, the track sides can begin to diverge. Note that one side of the bogie can now be unsupported. This solves the situation often referred to as the “frog problem.” (The “frog” is the little piece of track that, in suspended designs, has no means of support.)
4. The fourth pic shows the resumption of support under both sides of the bogie. Any “frog” would be cantilevered from this point, as the thin wall separating the two tracks is the only means of its support. At this point, the top guides can be discontinued and the drive wheel flanges can cease contact, to stop wear. From here the tracks can diverge into their respective directions. I have included a couple of additional screenshots, below. As for me, well, you’ll find me throwing out a couple of years’ worth of now obsolete designs. Oh, sweet victory!  


Jenny Brien said...

I wonder how a single-axle bogie would work for tight turns. One pair of driving wheels and one pair of running guide wheels. Would it run smoothly enough? If the suspension point. For the cabin were below the rails it would naturally tend to increase traction on acceleration and braking and in turns, where the guide wheel would act as a fulcrum to press the drive wheel on to the rail.

Dan said...

Dan the Blogger says-
Hi Jenny, thanks for stopping in… If I understand you correctly, you are suggesting something like an upside-down Segway. I think such a design is possible, but would be much better suited for vehicles with quite different proportions. Specifically, it would work best where the center of gravity is much lower compared to the track (like with a longer “neck” between cab and bogie) and where there is no worry about shifting weight forward and backward inside the cabin (like someone’s kids running to the nose of the vehicle just as it is docking.) Lowering the center of gravity would mean that it would be difficult to run track into existing buildings, something I would like to be able to do. It was with great reluctance that I finally conceded that the vehicles, (in the desired weight, shape and performance) could not free-hang. You might take a peek at post 118, where I depict a tall skinny PRT “pod” that might utilize your idea.
Anyway stability is always compromised to some degree when you make a vehicle design more concise, so that forces of momentum or imbalance have greater leverage. I’ve fighting this inconvenient truth for years. To specifically address the turning radius issue, this design can certainly turn in under a 5 ft. radius in a “half –track” mode. (only the inside wheels engaged with track) I still haven’t determined the absolute minimum. I’m actually working on that now. Thanks for your comment!

Rick said...

Dan, it looks like an action is required whether you turn off of the mainline or go straight. If you don’t lower one of the guide wheels at every off ramp you will have a problem (the guide wheels will squeeze the spreading diamond guide rails). I can envision fail-safe mechanisms to prevent guide wheels on both sides from retracting at the same time but I don’t see how to prevent the controller from “forgetting” an exit or for an actuator failing to work. I think you need a design that only requires an action if you want to exit the mainline.

If exits and merges are only on the right of the mainline, like a properly designed freeway, a fail-safe switching mechanism should be less complicated. This would require the pods to “go around the block” or have a “fly over” to turn left but I think it would be safer.

Just some thoughts from a mechanical engineer ;)

Dan said...

You make an excellent point, Rick. Here is, perhaps, one way to deal with it. You could consider the “prepared for a switch” state as the default, where a single upper and lower are generally kept engaged. After all, the real advantage to the setting with both lower guide wheels engaged is smooth high-speed travel with minimal noise and wear. In such a scenario presumably there would be time to stop if sensors did not confirm the completion of actions required to approach a switch. Then comes the question of whether the “urban” mode (one set of guide wheels engaged) is robust enough for the lion’s share of the daily grind. I think they would fare reasonably well. I don’t like the single (or even two) point engagement of plastic wheels as the de facto running mode but I could live with it if that design delivers the other capabilities we’re after.

I think it is important to point out that a there are a great number of capabilities and requirements besides simply turning right of left. For example, tight vertical turns require loosened lower guide wheels, so that they will not pinch the rails. Very steep or even vertical travel requires using the upper steering guide wheels to hold the bogie in engagement with a rack or even just for maintaining good traction. Tight horizontal turns require that the outside lower guide wheels be lowered. After creating a truth table of all of the possible maneuvers, I finally concluded that the best approach was to simply allow each of the four guide wheels to operate separately and to rely on electronic, electric, and/or software means for a fail-safe. In return, ultra simplicity. Doing it mechanically is extremely challenging without cutting back capabilities greatly. That is not to say that it can’t be done, but mechanical complexity itself reduces reliability. I know in the old days nothing but mechanical would cut it. But you have to ask yourself, “Where is the Google robocar’s failsafe when a pedestrian walks out into the street?” Why is PRT any different?

Anyway, I am all ears if you have a mechanical idea for a failsafe that wouldn’t restrict performance. If there is such a design, I certainly haven’t been able to find it. God knows I’ve tried… And by the way, Rick, it’s good to know someone is paying attention to the technical stuff. I can use all the help I can get! And one thing is for sure, from what you pointed out - that is that my lead-up, track-wise, must be long enough to allow a confirmed guide wheel maneuver or else come to an emergency stop before the switch, something my drawings grossly understate. Thanks for the help! I, for one, will be continuing to try to address those concerns.

ItsEric said...

Dan, Why wouldn't a rocker arm with two horizontal wheels on them to engage tracks at the top be smarter? It seems your design would put a lot of metal on metal wear and tear with a lot of noise at each turn. Whereas rocker arms at the front and back of the bogey would give guidance and support to a pod that is swinging outward.

The unused rocker arm wheel could also add support if it had a surface area on the outside to provide additional support through the rocker arm assembly.

If we had the lateral support could we make the wheels below vertical instead? They would then be fixed vs having to tilt away.

Finally, as long as the main wheels are twice as wide as the gap we shouldn't see or feel anything as they go over the frog. Am I missing something there?


Juho Laatu said...

Here are some comments on the latest track design.

The upper steering wheels run on two diamond profiles. Also the lower steering wheels could use the same approach. In the latest design the lower steering wheels are in continuous contact with the track. In the solution that I'm thinking of, only the main wheels woud be required to be in contact with the track on straight track segments. Lower or upper steering wheels could be close to contact to the track, just like the flanges are in the latest design. In surprising wind gusts the steering wheels could lean on the nearby diamod profiles. The bogie could lift the steering wheels (and/or "braking pads") (to contact the track) whenever there is a need to do so.

In this approach where the steering wheels would not always be in contact with the track it would make sense to have steerable main wheels. The first approch would be to bend the bogie in the middle to left or right. If you trust on digital technology in general, maybe you could trust also on the computers to steer the bogie. This capability of bogies to bend could help you also in making the track and bogie turn in as tight curves as possible.

One possible benefit (maybe problem too) of using diamond profiles also in the lower steering wheels is that then you could get rid of the flanges. They are not probably ideal from aerodynamical point of view. But maybe more interesting is the fact that without the flanges the main wheels would be closer to the centre of the bogie. This would mean that driving on half track would be easier due to lighter sideways pressure on the steering wheels. Also the overall width of the track may be an important factor (=> two flanges narrower). The possible problem that I mentioned is that if you leave the flanges out, also the width of support of the main wheels is narrower and the bogie is less stable (that is, unless you widen the tires).

On straight track segments (outside the switches) you may not have the upper steering wheel track (like in the latest design), or the lower steering wheel track, or maybe you have both, or neither of them. All this depends on the level of stability that we need. The lower steering wheel track may of course be always present also because it can be seen also as a supporting structure of the main wheen track (are two diamonds better or worse than the current one diamond at the edge of the track?).

(Then some more distant ideas that I just mention briefly. In places where the steering tracks are needed only for emergency purposes or braking the structure of the track could be simpler than in switches where these features are always used. One approach could also be to use one diamond profile and two wheels. The possibility of triangle shaped tracks or ridges under the track I have mentioned also before. Maybe they are nt as easy to manufacture and bend?)

One additional question in my mind is if you expect those steering wheels that are not in continuous contact with the track to rotate at correct speed already before they touch the track. Or is the idea that the wheels are cheap and durable and the traction is so low that it is no problem to let them capture the required speed only when they touch the track?

And one more question. You assumed than in the switches the flange would be in contact with the track. Will the diamod profile be in the same position all the time, and the main wheel and flange will move so that the flange touches the track, or does the profile of the track change? (like the upper diamond profiles change position when they spread out)

Dan said...

Dan the Blogger has some catching up to do!
ItsEric, if you’re still out there, I know I emailed you my response by mistake, and I see now that what I sent missed one of your questions. The reason I do not have the upper guide wheels on a rocker is because these also act as “hold-down” wheels if the bogie is climbing vertically. Whatever holds down “pinion gear” onto the “rack” (I use rack and pinion climbing) has to be pretty solid, and I used to have a whole separate wheel and track just for that. That has been eliminated. I am not contemplating metal to metal, however, but a hard plastic, as is used in rollercoaster wheels. The drive wheels and the angled bottom ones are rubber. A lot of the logic of this design relies on the sponginess or hardness of the various wheels. There are no hard wheels making contact at high speeds, and the soft ones must be pretty big. The tilt of the angled bottom soft wheels helps greatly in keeping them out of the way of the swing-arm, emergency brakes, etc. Actually there’s quite a lot to fit down there already. As for the business about making the wheels wider than the track opening, note the fat wheels I drew back in post 9. Actually, I think aerodynamics will start forcing skinnier wheels, if anything. It is true that having wheel flanges (and the diamond track design) widens that gap way up, and I don’t much like that part, but they greatly simplify very tight turns, among other things.

Juho, I tried a number of pivoting bogie designs and never found one I liked. I think simplicity is a virtue here. Adding parts tends to multiply potential problems. The design as shown can handle very tight radiuses, both vertical and horizontal, just by being a bit stubby. The flanges allow such low speed turns without the steering guide wheels. Also, I would be wary of designs that involve parts “almost” touching unless you have an adjustable mechanical way keep them where you want them. Stuff has a tendency to start swaying or vibrating in unexpected ways. I do not mean to be dismissive. It is perfectly possible that a pivoting bogie design would be a viable way to go. I got hung up on hanging the cabin weight from it in such a way as to still allow tight vertical turns. It was getting too complex for me to predict.
About those flanges – I am figuring at this point that they would be separate, but loosely connected magnetically. (so the rotational speed would be very close, but not perfect, on contact, perfect after that) The top ones would probably use low wattage hub motors. The spacing between the bottom diamond supports would change prior to switches, with the diamonds being slightly closer together, so that both flanges must engage.

Juho Laatu said...

Thanks for the good arguments behind the design choices. Here are few more thoughts still on my mind.

- If the gap (and related wider track) is a problem, another tilted wheel could be used instaed of the flange.

- If the lower diamond moves closer to the flange in switches, I guess also the tilted wheel shall move closer to the flange at the same time. Maybe both sides do this first, and only then one side will be disconnected??

- My draft aimed also at simplification by replacing two wheels, the tilted soft wheel and flange with one hard wheel. It avoided the continuous contact of the tilted wheel (simpler but maybe less stable than having continuous pressure against the track from multiple directions). Wheels or break pads may almost touch just to be prepared for surprises like need to break.

Dan said...

OK, Juho – To bullet point one. I think the tradeoff is between the optimal hanging /leverage advantages of a small gap track (small gap between the two supporting running surfaces) and the very nice and simple way to negotiate very tight turns that anyone with a welder can fabricate. (Which the flanges provide). I would like such simplicity particularly because I would like to see private spurs that might end up saving some company from having to buy a forklift or build a loading dock. Also, indoors, a very simplified, ceiling-hung track might be used, and I don’t really like complicating that from the start either. By itself, the advantages of the larger diameter of the flange are pretty much canceled out by the (less than optimal) angular contact, certainly from a wear perspective, although they might be a bit smoother when they engage at high speeds. Anyway, a diagonal from the center of the bogie wouldn’t be utterly out of the question. It just seems to me that the ultra-simple ultra-maneuverable nature of the flange for low speed use make it sooo easy to store, stage, stockpile, station, and stack vehicles in a crowded world. But I’m open to alternatives that preserve this advantage, since I don’t like cantilevering the vehicle weight either.

Number two- What we have here is a stubby little critter sandwiched between opposing forces. There is zero free play, so the tail doesn’t get to waggin’! (Folksy ain’t I?) Seriously though, if you notice, the hard wheels all push, directly or indirectly, against soft ones. If you look back at that Anderson patent diagram, you will see that he establishes rails within the overall structure that can be adjusted to take up slop. This is very wise, since things have a tendency to warp from the heat of welding, and he has those hard guide wheels that don’t take up slop on their own. I am (perhaps naively) hoping to eliminate all of that cost and weight. The minute we have opposed hard wheels, we had better devise a way to take up slop, because even ma’ stubby little critter could git to a shakin!

Number three – see number two. Thanks Juho. Sometimes I design things a certain way without even consciously knowing why; Most things I design I go on to actually build and follow up on, so after so many years, it’s almost all instinct. It is a rare treat to have someone ask the hard questions. (and make me crystalize my thoughts) And that goes to a bunch of you out there. You know who you are. Thanks.

Juho Laatu said...

The track could be also somewhat different in the tightest turns. One approach is to assume that we have a half-track (that in a way has an infinitely wide gap). Half-track may also be the chepest approach for slow speed and tight space private or private company tracks. We could have a very narrow gap in straight high speed tracks but allow some more freedom in the slow and cheap sections.

I think the current design could work quite well also withouth the flanges in tightly turning half-tarcks. Maybe the tight turn would always be to the same side of the half-track. A long (fast speed) bogie might be clumsy. This is where a pivoting bogie might help. Or maybe other arrangemenst (like steering wheels that are positioned in the optimal place or that can move a bit from their regular position) are sufficient. Maybe long stiff bogies have to have half-tracks that turn towards the "missing half-track side" in the tightest turns.

In point number two I didn't yet fully understand how he details will work. Maybe on tilted wheel will simply tighten its grip (and forces the flange to contact the track) when the other tilted wheel disconnects.

Point number three includes one thought for consideration. Are (some clever) bogies allowed to disconnect the steering wheels if they think they can follow the track also without them?

Dan said...

I had just assumed that the full track would be used basically throughout . The system as is really wasn’t designed for half-track use for extended periods, wear-wise. As I said in finishing the last comment, the steering guide wheels go up and down various amounts under various circumstances. When we talk about tight radii turns, let’s not forget vertical ones. Cresting a slope requires the leading wheels to drop, for example. This is where you really need to know where you are to the inch and the exact nature of what is to come. (highly detailed “route map”, highly accurate track reader) This is also why I would like to make the construction of curved track segments simple and standardized. Right now, the only thing (for horizontal curves at least) that a track builder needs to know is that the gap gets wider on curves that are very tight. How much would be on a column on a reference table for that particular radius. On the very, very tight turns, (A radius perhaps half the length of the cab) the steering guide wheels might have to both be dropped out of engagement, and steering (centering) would be done by flange alone. You are right about the possibility of halftrack turns, but there is also the whole matter of hanging the halftrack, which would be highly leveraged, so what do you really gain?

One the matter of “point two...” What I would do is engage ALL wheels, hard and soft, before letting go of one side to steer into a switch. That would entail the gap in the track (between the diamonds) narrowing to engage both flanges at once. (Instead of none) This is all to make the switching smooth to the rider…There is never a point when there is not full support from all directions, and every support change fades in or out in a long, seamless span, so it can happen a very high speeds.

OH! yeah! You are right. Another bottom guide wheel height change would be called for – a tiny lift. You know, this really is not a 2 or 3 state system. It’s really a dynamic system, and a fault in steering is not a situational binary type problem, but a breakdown in an ongoing process. Maybe the primary system should just be close range proximity sensors and the mapping and track reading be the reference against which the proximity data is compared, and the steering be a suspension of that process. Now that I think about it, I remember coming to that conclusion back around the time of post 136, after rotating those tight curves around in space for about an hour, to see how the wheels would bind or not touch. In fact that was part of what led me to the diamonds, because they are forgiving in this respect compared to the round stock I was playing with in those pictures. Ah yes… 136… My only musical post!

Juho Laatu said...

I also assumed that in your design full track would be the main rule. Maybe half-track can be used in small slow, cheap and tight sections like in my home garage (in addition to switches that have short half-track segments).

In "point two" I was thinking that if the lower diamods move closer to each others before the switch, the tilted wheels will lose their grip and must then retighten it again before they disconnect the other side.

But I now see that you assume that there should be an automatic, maybe computer and sensor controlled mechanism that tightens the grip always when when possible (e.g. when the other side opens its grip). I still wonder how the wheels will find the correct grip after the switch (one side should loosen its grip and the other side tighten its grip until the bogie is in the centre). The sensors and intelligence that you propose are able to to that. If we assume double safety here to, bogies taht have lost their mind should also not cause a catastrophe.

P.S. I assumed also considerable intelligence in the scenario where the steering wheels are not in direct contact (but may be in close contact) with the track.