Saturday, June 2, 2012

142> Dodging Bullets


Well, as I write this, it is (for those of us who are US citizens) Memorial Day, or as we say here in rural New Hampshire, “Better-shoot-off-our-guns” Day. And so I am “sheltering in place,” in my cabin, trying my best to ignore the gunfire and musing over the state motto, “Live free or die!”. 

In my last post I outlined a design that I am hoping ends the long quest for the perfect framework for a ultra-high performance PRT system, and I want to add a couple of additional comments. First of all, I would point out that my depictions illustrate both a track and bogie wheels, but not just for any bogie. It’s a high-speed one, and that is a very important point. You see, if we were just talking about the speeds normally associated with PRT, any one of a number of wheel designs would suffice. But I am looking for a design than can potentially well exceed highway speeds, and that explains my fixation with larger wheels. You see, I believe that the corridors created by highways are perfect for PRT, and that commercial development has tended to be along these areas anyway. In most US cities, almost all of one’s needs may be met within a block or two of a highway. I also don’t mean ONLY high speed. It must also be highly maneuverable to get everywhere you might want to go in a dense suburban environment. 

Giving PRT longer-haul, faster capabilities has certain design ramifications, and those larger wheels are meant to address the wear multiplied by both higher speeds and greater distances. This does not mean that the larger wheeled system is the only one that can run on the track, but rather that the track profile is equally well suited for high speeds as low. The use of off-the-shelf steel makes the running surfaces  extremely cheap and easy to fabricate, and can be housed in a variety of ways, such as trusses, a box-beam, or even open, such as in an roofed enclosure.  

The profile is advantagious because it does not force the wheels to run on round pipe, something that concentrates wear on the center of the tires or solid wheels, yet the square tubing is easy to bend using a conventional 3 roll pipe bender, as shown below. (Needs V-grooved rolls)


Of course running PRT wheels on flat surfaces is nothing new, but established designs mandate that the beam that houses the running surfaces be of sufficient size to house those various guide wheels – generally oriented sideways. This limits those wheel sizes to less than one half of the interior of the truss or box beam. This has meant a trade-off between what you had to look at overhead, and the life of those wheels. Whereas that track girth can otherwise be justified for structural reasons if the spans are to be large, if it is convenient and cost effective to space track support poles more closely, then the track itself can be thinner, cheaper, and more attractive. Turning those flat running surfaces diagonally allows for either longer-lasting guide wheels, thinner track, or a little of each. Of course placing the main guide wheels outside of the track doesn’t hurt either!

That brings up the point by alert reader Rick, who noted that there has been an alarming departure from previous designs in that there is no failsafe for the possibility of a steering guide wheel failing to respond. This is true, but I stand by my design. A feature of many early designs, such as the one depicted in this patent drawing, is that they always had one or the other the of steering guide wheels engaged. That way if the something failed, the vehicle would still go one way or the other, and would never crash by trying to go both ways or somewhere between. 

That is fine, and an excellent feature, but those wheels that stay engaged for safety sake must, in a high speed long distance system, do so for thousands of miles per week at high RPM. What if the system is to include an express shuttle that bypasses many areas that have no PRT service, so there are very long stretches with no off ramps? Would we want to be wearing out the steering guide wheels for that whole time?  

I would note that there has been somewhat of a culture change underway in recent years in regard to the trust we put in our computer controlled equipment. When Anderson patented this mechanism, meant to “snap” the steering into full left or right positions, it was in a time when making the system centrally controlled was not a choice but rather mandated by the size and cost of even limited computational power. I am on record as advocating a more autonomous control architecture than many proposed systems, and autonomy certainly plays a part in any discussion on the safety of a system that requires an action at every junction. Google’s robocar, now licensed to drive in Nevada, could, on a twisty road, certainly veer into oncoming traffic by a similar failure to actively steer. Luckily the vehicle is not driving by Google Maps alone, but by an array of sensors and onboard computers all working in concert to create a vehicle that can act and react autonomously. Now, in the days of “cloud” computing (many seperate computers drawing from, and collaborating through, a central computer) and supercomputers made of dispersed computers sharing a common program, (such as SETI) the lines have been totally blurred. With today’s technology, steering gear deployment can be assured through a number of cooperative means, including autonomously from within the vehicle, from sensors within the track, from communication with a central computer and/or any combination therein.  

Another, separate point that needs to made is that the track and bogie combination has been carefully designed to allow extremely tight turns, including changes of pitch, and the combination of this attribute with high speeds and switching is particularly challenging.

Anyway, to get back to a point I was making earlier, by dialing back the speed a bit, many other bogie designs are possible, and may be more practical than what I have shown for those speeds. My focus is to not PRECLUDE high speed (or other valuable attributes) in the track first, and then to design practical bogies for that track. I am really not sure what a downtown-use-only bogie for this track would look like at this point. I'll have to work on it between the bullets.

13 comments:

Vehicle Wraps Portland said...

Cool blog. More power to you!

Juho Laatu said...

First I just give support the design principle that speed and cheap tracks are important. If your target is a general purpose track that will be deployed widely, people will certainly soon be interested in those two properties.

Then some comments on the failsafe mechanism. I think your latest design and the idea of the bogie being always in either left or right turning mode are not in conflict with each others. The mechanical design could well be such that when the bogie runs into a switch by surprise, and the left and right halves of the track strat to move away from each others, the bogie would be either in a mode where the right side must lose its grip from the track, or the left side must lose its grip.

Although digital components are more reliable every day, we may still assume that sometimes they fail. In that situation it is a better solution if in the example one of the sides loses its grip, instead of both keeping their grip as long as they can. That mechanism might not be expensive - just some clever mechanical design.

We would of course have similar features also when the bogie enters a switch from the reverse direction.

Nothing is 100% foolproof, but if we are able to keep (without major extra cost or weight) the idea of having a mechanical design that is safe also when the more advanced parts of the vehicle fail, probably we should keep it. People probably find the thought of falling down from the track disturbing. Being able to trust on the fact that there are extra safety mechanisms in place, may help them to trust the system more (irrespective of the level of probability of the failure).

Andrew F said...

It's a good point. Humans are very bad at assessing relative risk. They fret more about flying than driving, even though flying is much safer.

Dan said...

Dan the blogger is back, almost...
Thanks for giving us all something to ponder, gentlemen, in my absence.
I just got back to town and should be back to it in a day or two.

cmf-seattle said...

Anderson's design didn't call for switch rails everywhere, just at merge/diverge approaches.

Another advantage is the ability to throw it by hand if necessary. It'll work in both forward and reverse travel directions. And I counted 11 parts (check the Taxi 2000 Vehicle Weight Estimate PDF from the Skyloop CALS Rebuttal to PB docs).

Dan said...

I’m having a bit of a time trying to come up with a mechanical safety release that is, in itself, absolutely safe. After all, the mechanism that holds the wheels in engagement had better be very strong. But here’s my take… It seems like the safest tack is to have the switches preceded by the absence of something that is otherwise there. That way redundant sensors can be kept in an “on” state unless there is either a switch coming or there is a malfunction. If the expected switch architecture (signal or physical clues) is not then encountered, then there has been a malfunction, and the vehicle must reduce speed and take the prescribed measures to ensure passenger safety. Such a system has similar logic to the air brakes on a truck, where it is the proper functioning of the engine and compressor that allows the brakes to NOT lock up. . In other words, it is only the continuous proper functioning of all systems that allow both steering guide wheels to remain engaged in the first place. Then it becomes a matter ensuring mechanical ability to disengage. I did a post back a while, (42) showing redundant servos for the steering guide wheels connected by a ratcheting mechanism. That way either servo could do the job without stopping the other. One idea might be to have one servo run by the battery backup and one from line power. Each can have independent position sensors, so if they both run at once and both sensors are in agreement, then the system is working properly. If only one is working then the vehicle needs repair, even though it would still work fine. The odds of both servos quitting at the same switch, if they are truly independent, would be astronomically small. Anyway, I like your idea, if it is indeed possible, without introducing weakness. It would need to be some kind of release that would only engage just before parts start snapping.
Andrew, I agree. There is a question, thought, whether a safety system that they do not understand will be of psychological help. Who would have the patience to learn how the switching happens in the first place? Some would be terrified if they even thought about it. I think simply making the bogie not fit through the track is probably best psychologically, since everyone can understand that.

cmfseattle, I did not know that, though it makes perfect sense. Still, there were other sideways facing guide wheels that did get continuous use. What I have tried to do is to make all continuous use wheels out of rubber with just enough compressibility to allow zero clearance. (no dimensional slop rattling, less noise and vibration in general) The steering guide wheels, being only engaged for switching, are not really problematic as he conceived them, but for the fact that I have given mine multiple uses and made sure that they could do extremely tight radii, both up and down and sideways. Being able to switch manually would be nice though...

Juho Laatu said...

How about having physical "ridges" before the switch? They would physically force the bogie to disconnect one side (coming to a switch from the "reverse" side) or diconnect whichever side or one selected side (coming to a switch from the "front" side) if the bogie has failed to do that itself for some reason.

Dan said...

Dan the Blogger is Not Happy …
Juho, I wrote a detailed answer to this and thought it posted, so I deleted it. Now it’s nearly midnight. Here are the highlights, anyway. I think you would need 3-state logic, like, perhaps, a lever that is up for left, down for right, and in the middle for both. That way the “ridge” would only keep you from having both guide wheels down. Otherwise the vehicle would have no freedom of choice. That is on a diverging switch. For a converging switch the system, of course, has only one safe setting, and the ridge would enforce it.

The system you describe seems difficult to implement with the preferred screw-jack type activating mechanism, but is much easier for cam/ratchet system like the one I described but with separate cams for each side. That is because the cam system only “locks” in the engaged position so turning the cam even a fraction of a turn means it can slip out. A full half turn and it is as would happen by servo. It is also undamaging and probably easy to reset on the fly. Pity cams are otherwise kind of a lame approach to lifting those heavy wheel sets.

It may be that the best thing to do is just assume it is a problem that is solvable without huge changes to overall design and kick the can down the road for a while. I already have preliminary versions of both screw-jack and cam systems on the drawing board. I’m kind of interested in how each affects logical placement of emergency brakes and the swingarm to bogie connection, since neither are well defined in a real-world, “you-can-build-this” sense. We need to fit all of that stuff in the same approximate location plus possible radiators if we want liquid cooled drive motors. That’s what I have been working on lately.

Juho Laatu said...

Hers's one possible logic for the "ridges". In this scenario there are three different ridge types ("no left", "no right", "decide"), and the bogie has two modes ("left", "right"). The bogie is in either of the two modes also when both left and right wheels are connected.

- "no left" ridge disconnects the left wheels (and forces the bogie to "right" mode)
- "no right" ridge disconnects the right wheels (and forces the bogie to "left" mode)
- "decide" ridge disconnects the left wheels if the bogie is in "right" mode
- "decide" ridge disconnects the right wheels if the bogie is in "left" mode

Rigdes "no left" and "no right" are typically used to disconnect the "wrong side wheels" in a converging switch. But they can be used also e.g. when a full track becomes a half-track, or when some section of the track is closed for maintenance.

The "decide" ridge is typically present before entering a (diverging) swich. The bogie can not change its "left"/"right" mode while the track has a "decide" ridge. The "decide" ridge also forces all those wheels that are needed for half-track driving to be connected.

Before the start of a ridged segement there is a soft mark (electronic or physical) that the bogie can read and make its decision before the physical ridges force it to make the decision. For example when entering a converging switch there can be first a soft "next no left" mark, followed by a "no left" ridge, and then a "decide ridge" (that starts before the "no left" ridge ends). After the switch the "decide" ridge will end, and it is followed by a "next no ridges" soft mark. The soft marks are different (or are read differently) for "forward" and "reverse" traffic (I guess in most tracks this is an option although the tracks are mostly used in one direction only).

Dan said...

Juho, before we get carried away here, this is all just backup, right? Because as far as I am concerned the vehicle will know exactly where it is at all times within inches – and how far away the next switch is with equal precision. (as well as the number and location of all switches for the whole route)

Also, are you suggesting switching via a movable ridge in the track?
I would prefer no physical contact in ordinary operations. By the way, did you note that on tight radii the lower guide wheels can all remain engaged by lifting or lowering by various amounts? If the vehicles are parking themselves in a dense 3D array and doing so quickly, or if the system was being used as a robot, such stability would be welcome. It does, however, suggest controls that are more complex than for simply putting those guide wheels up or down.

Juho Laatu said...

Yes, just a backup. Although it would be ok to have also simple bogies that do not have any intelligence but that rely on the physical ridges.

In normal operation I expect especially fast bogies to know where they are and know when they approach a swich. They may never touch the ridges since they will make all the necessary actions already earlier. That probably leads to much smoother operation and less noise. The isea is just to have a doubled safety mechanism to stop bogies making any bad mistakes.

For normal smooth riding vehicles the ridges would be there only to make sure that everything is safe also is special situations, like when some user manages to disconnect the computers, or decides to change the direction from left to right manually in the middle of a switch, or when the cenrtral coputers fail, or when the soft information about the tracks is incorrect, or when the track is a left half-track and the bogie must not detach its left wheels.

I'm not sure what you meant with "movable ridge", but I expect typical ridges to be static (e.g. welded).

I agree that when the curves get tight the wheels need to make some tricks. Maybe you can draw more pictures like he ones in the Hot Rod article (http://openprtspecs.blogspot.fi/2012/02/hot-rod.html) to study and point out what the wheels should do in such tight turns. There are many tools like adjusted wheels, adjusted wheel packs, different shape of the tracks, half-tracks, turning wheels, pivoting bogies, and even moving track related solutions like turntables and elevators. In some garages the bogies might also be passive (moved by the storage system). One interesting question is the positioning of the steering wheels with respect to the main wheels on the front-back axis.

Alexis Read said...

Your new design looks nice, I appreciate the larger stabilising wheels compared to previous in-beam designs. I am wondering whether stabilising wheels are needed though.
With 2 LIMs (built in to the bogie) you should be able to stabilise and switch the bogie, you'd still need 2 'switch guide' wheels, but they would then be used to stabilise the LIM track height through the switch.

2 LIMs in the bogie makes for a slightly different power/propulsion setup compared to one big LIM. I've not thought it through, but that may be advantageous overall wrt. redundency (you can always switch/move in one direction and stop),
wheel friction vs magnetic stability,
modularity (ie. 2 smaller LIMs are cheaper than one big one?),
pickup switching,
location sensing at the diverge point
etc.

Overall, an electric LIM switch mech would be much simpler compared to a more complicated mechanical design.

Of course that means using a LIM with passive track - you'd get good location awareness for free though thanks to the passive coils (and you only need one coil for both LIMs). The track expense vs extra weight/complexity/maintenance in each bogie if you don't use a LIM?

Additionally, I'd suggest using nylon (PTFE skateboard) wheels over rubber ones as you're not using them to grip, merely support, and they provide a modicum of give (for comfort/noise) compared to steel.

Dan said...

Dan the Blogger Returns!

Alexis, I agree that magnetic means can go a long way toward solving many problems related to high speed switching, centering, and propulsion. Unfortunately these approaches are rather unwieldy for close-quarters maneuvers. I believe gradual turns and long ramps, in an urban setting, are unnecessarily impractical, intrusive and expensive.

LIMs loose a great deal of efficiency with anything but extremely close proximity between the track and vehicle. The ones I have looked at are quite heavy for their performance. If I had to use them, I might look into some kind of servo mechanism to dynamically maintain that small gap. Otherwise it’s like having a car that rides so low a cockroach can barely fit under it. If you increase that gap to the size of your pinky finger, you lose 60% of your power. At least that is my understanding. So you can see how that limits 3D maneuverability. Frankly, the various configurations I have been playing with are the result of wanting the best of all worlds – small track, very fast speeds, and super maneuverability. Any one of these can be had, and quite easily, but at the expense of the others. I see, from your other posting, that you favor the winch approach to getting to ground level, so that frees you from the need to maneuver vertically. I still favor very steep (up to vertical) ramp capability instead.

As for wheel materials, I would just like to note that there are many differing chemistries for rubber as well as the other compounds in use for load bearing wheels. Most are urethane based. (I think many nylon compounds might soften, warp or melt) Anyway, (and I am stealing this from my upcoming post) the urethanes (such as roller coasters use) can be formulated to be soft or hard. There is one, Dupont’s
Hylene PPDI that has gained some acclaim. Anyway, though, your premise was that traction is unnecessary if you use LIMs, but I think hub motors are better because of the gap/maneuverability/efficiency issues. But I never say never….

Thanks for commenting, and sorry for the delay getting back to you!