Personal Rapid Transit Revisited
Recently I received a comment on my post regarding Personal Rapid Transit, back in October.

The anonymous poster, who appears to have been from an organization called SoundPRT, which appears to be an advocacy group for bringing a PRT system to the Seattle area.

The commenter addressed some thoughts I had (all relatively minor) about the SkyWeb PRT system. The comments sparked a renewed interest in me and I spent a couple hours digging around to find out more about the broader range of PRT solutions and criticisms leveraged against it.

What follows is, in the tradition of the original article I posted, a mostly wandering discussion of random shiny things that caught my eye as I was cruising the various sites, as well as responses to the points the commenter made. Let's start with the latter.

Call and Response
In my original post, I had yammered on about how it would seem like making each vehicle stopped at a station wait for the first vehicle in the station queue to depart might be irritating. To address this, I proposed a Y-shaped section of track for each berth. To which the commenter said:

It is not anticipated that individual delays will be a problem. First, most stations will have multiple bays. Second, boarding is made as simple and fast as possible--payment and route selection occurs before boarding, and the design must make ADA/wheelchair access as easy as it is for the fully-able-bodied.

My response:
The waiting issue I mentioned was not necessarily a result of limited berths. Rather it was my thinking that people who are boarding in the first berth may take an inordinate amount of time boarding for reasons entirely unrelated to system design. People holding a vehicle so that another member of their party could board; because they have many things to load into the vehicle; or because they are taking a while to get situated. In a linear station design, all vehicles behind the one in question would be required to wait until that vehicle was completely boarded. This is more likely to be an irritation rather than a real inconvenience, but it's worth considering that the impact of such a scenario may be greater in systems where station access is non-sequential as people expect their travel experience to be unrelated to that of other people traveling simultaneously.

Likely, developing a solution for such an irritant is over-engineering. Most solutions would require slightly larger stations to accommodate and likely increase the boarding/deboarding process as an individual vehicle takes itself offline in a station, however having both the station offline from the main guideway, and each birth offline from the station guideway might be worth looking at in certain station configurations.

Although, having given it a bit more thought, I could envision an alternative to the Y approach I mentioned in the original article that would have less of an impact on vehicle "docking" times (though would probably have the same relative impact on station sizes). Consider a station guideway that departs from the main guideway. In the station, there would be a main guideway and individual berth guideways leading off of it. The berth guideways could intersect each other (with the outbound guideway intersecting the inbound guideway of the next berth) and the central system could manage vehicles through these intersections. This would mean that individual vehicles would not have to make sharp turns (as in the Y configuration), would not have to back up (again, as in the Y configuration), and each berth could have as long a deboarding/boarding time as needed without interrupting the flow of other vehicles through the station. An added benefit might be the ability to store two (or more) vehicles in each berthing track when demand is low, providing additional vehicle storage to the system.

Next, I wondered aloud about what happens when an individual PRT vehicle broke down on the guideway.

To which the commenter said:
This is pretty much spot-on. In addition, under the Skyweb system a breakdown is pushed to the repair center by the next vehicle. Generally, Skyweb is designed with few moving parts to break, as well as redundancy throughout, making the odds of unplanned stoppage literally astronomically low.

My response:
In the scenario where an individual vehicle becomes non-functional, it makes sense that another vehicle can come along and push the disabled vehicle. But in a system where guideway switching is dependent on the individual vehicle and is not a component of the guideway itself, how does one ensure that the vehicle can be pushed to a suitable destination if the switching mechanism is disabled? Also, what happens when a vehicle is disabled due to a problem directly relating to it's ability to move (e.g. one or more wheels jam)? In such a case, it would seem that PRT critics would have a valid point about debarking in emergencies. This problem would seem to be compounded if the traveler were disabled, insomuch as one could not simply deploy a cherry picker to recover the occupants.

I went on to puzzle on whether or not long lines might form behind a station, eventually backing up on to the main guideway.

To which the commenter said:
Actually, there are primarily three factors: the demand for each station in terms of arrivals AND departures, the number of berths per station, and the size of each "loop" or circuit of elevated track.
Demand: PRT spreads-out demand per station by having many small stations close together (ideally 1/2 mile apart).
Berths: Most stations will have 3 berths. In busy places (downtowns, stadiums, airports, intermodal transit stations) stations would be scaled-up by adding more berths (major station cost is the elevator and trip selection "ATM", marginal cost of each additional berth is relatively low).
Loop size: In the rare (see Demand) event of a station and its approach siding being full, a car can be "waved off" and directed to go around the local loop and re-approach. The ideal size of a loop is 1/2 mile on a side for a total of two miles; traveling this at 40 mph takes only 3 minutes.

My response:
The impact of the unidirectional nature of the guideway might be interesting to understand more. It would seem that, for some station-to-station trips, the trip length might be significantly longer than one might expect when the trip needs to be routed through a series of one way guidelines (I'm assuming here that there would typically only be one guideway per street (or equivalent) and that the illustrations showing stations as being single direction only are correct). I'm sure this is something that is minimized during initial planning stages, but this, among other things, makes me want to sit at a simulator with someone who knows all about this and see different scenarios play out.

Would this, then, mean that, in a scenario where an approaching vehicle has to be "waved off" that the trips might be significantly longer than traversing a 2 mile square? Also interesting to know is how likely it would be to achieve a regular 1/2 mile per side loop. What are the system performance impacts if the per-side lengths expand beyond that, and what is the likely upper-threshold over which performance becomes unacceptable (I'm guessing that this is heavily dependent on the particular installation. More reason to sit in from of a model simulator.

I pondered the question of whether PRT would eventually be outmoded by computer controlled cars.

To which the commenter said:
It depends what effect you think Time might have on PRT's effectiveness. Urban densification and suburban expansion are not a problem for PRT, as the system scales-up well.

My response:
Yeah, I must have been high or something when I wrote this. Even though I have visions of computer controlled cars dropping you off at work and then driving themselves home or some such, the only near-term way automated cars would compete with pretty much any transit system, would be for long-haul scenarios where individual cars could travel as fleets (though I'd be interested in hearing why I might have been more right than I expected ;).

I said I thought it'd be cool if you could access the Internet from your laptop while in the car.

To which the commenter said:

No reason a community couldn't specify this as an amenity.

My response:
Yep, again...high. Looks like some metropolitan wireless schemes may well fill in this need, though I'm not counting on them quite yet. And if they fail, perhaps the cellular network will soon be good enough to fit the bill. In any case, it'd be nice to have in the PRT vehicles, but certainly no point of emphasis for them.

At some point I used the phrase "Doesn't meet commuter throughput claims".

To which the commenter said:
How? Remember, station demand is distributed/spread-out. In addition, routing is spread-out too. The "mesh" configuration of the network means there is more than one way to get from A to B, especially if they are far apart. If one PRT track can handle 7200 vehicles per hour (30 mph, 22 ft headway), imagine what multiple parallel tracks can carry.

My response:
And again...high. I'm not even sure what I meant here (I didn't think anyone would be looking ;). I'm pretty sure I meant it would be a problem IF throughput claims couldn't be met, but your guess is as good as mine at this point.

No Call, Just Response
Now, more ramblings on PRT and the like in no particular order:

PRT as a Complementary Mode
Most places I visited seemed to indicate that PRT was not necessarily direct competition to light rail (et al). Rather it was a supplement. While I can see this for certain applications, like the one I mentioned in my original post about going from Vancouver to Portland, I do wonder what the designers see the specific limits of PRT as being, and where it makes sense to employ other modes of transportation. Specifically, I'd be interested in seeing some cost scenarios (or technical scenarios) where PRT would break down. I think this kind of information helps to provide much needed context to the discussions I see about PRT and other forms of transportation (such as the fascinating PRTDebate page).

A PRT Wiki
I ran across a fledgling wiki site for PRT that was useful as a jumping off point to other PRT info.

Problems of Placement
In terms of getting PRT guideway everywhere it would need to be, no doubt the challenges of getting the guideway approved in enough neighborhoods to grant the 1/2 mile-per-station distance is non-trivial. Much like other forms of public transportation, and, indeed, many other public works projects, people may be in favor of the idea while not wanting it to run by their homes. Unlike busses or light rail, however, PRT will have to fight this battle in many more neighborhoods if the penetration targets are to be achieved.

It seems likely that individual stations would have a relatively small footprint, however providing access to this station (stairs and elevators; the latter required for ADA compliance) might be non-trivial in all the locations where a station would be required.

It would be interesting if a PRT service could replace existing utility poles so that they served a dual purpose: carrying the existing utility infrastructure, as well as the PRT guideway. My guess is that this would help lessen certain arguments about invading infrastructure and would mean that no additional right of way need be granted. Even better, perhaps, if the utility lines could be tucked into the guideway in some way to lessen the negative impact of existing infrastructure. In fact, it appears that UniModal is proposing exactly that .

Stopping the Train (and the PRT)
Arguments about the stopping distance of a PRT seem a bit odd in certain respects, especially when compared to light rail. Assuming the worst case scenario of a vehicle collision with an unexpected, ungiving object, it would seem that a fair comparison would be to compare the effects of a PRT system versus a light rail system encountering the same object, and run the comparison at different passenger loading levels. At first blush, one would assume that in a properly functioning PRT system (and this may be where the failure lies), that the impact vehicle, and perhaps a vehicle or two behind it would be involved, accounting for injuries to 12-16 people (fully loaded). Under the same circumstances, one would think that a fully loaded light rail car (or coupled cars) would sustain injuries to on the order of 50 individuals in such a collision (100 if coupled) as later portions of the rail car cannot brake to avoid collisions with the earlier portion of the car.

Getting it On (and Off)
It seems that discussions of passenger load times are primarily oriented around total time to load N passengers. It would be more interesting to see what the average wait time of individual passengers would be given normal passenger arrival patterns. I would guess this would lean towards PRT. The inter-vehicle wait time should also be thrown into the mix (and, no doubt, there are studies of this). I presume that in an optimal PRT system, the inter-vehicle wait time would be substantially smaller than the inter-vehicle wait time in an optimal LRT system, adding further disparity to each mode's average passenger wait time. Again, I'd love to see some simulations of both modes to see how things work out.

UniModal
The commenter mentioned a system by UniModal. Here are my comments about that:
The UniModal system seems like it's still in the initial stages of design (unlike some of the other PRT systems which seem well conceived at a fairly detailed level), an idea further reinforced by the large number of "coming soon" pages where vital information should be and a copyright date the belies the "soon" portion of that statement. And, c'mon, their Kiddie Train example is filled with holes and serves primarily as flame bait for proponents of alternate transportation modes.

As mentioned, it is not currently ADA compliant (both in the stairway access that is displayed in their pictures, and, presumably, in the nature of their two-seat cars where riders are seated one behind the other), the kludgy drop off and pick up system notwithstanding.

Similarly, in all of the images on their site (e.g.) the cars are shown waiting in queue for riders, but I imagine this would be problematic if an arriving rider ends up at the end of the queue. If there were not enough new passengers, all queued vehicles would have to be sent elsewhere. Other systems appear to have this handled a bit more elegantly by having stations with multiple berths allowing some cars to be stored while still providing room for arriving passengers to depart. There is, however, a strong attractiveness in the simplicity and small scale of the UniModal station design.

The UniModal model appears to be a two station approach with a disembarking station and a separate boarding station.

Feeding the Beast
I thought that the point the UniModal site made about feeder services not being accounted for in light rail travel was interesting. Probably only some of the trips on such systems could be added to the cost of light rail given that at least some of such ridership may start and end their trips on such a system. Nonetheless, to the extent that all such feeder traffic would be accounted for on a PRT system, it should be included as part of the comparative costs.

The End...Or IS it?
Anyway, that's enough rambling for now. Comments are welcome, but they may spin me into hours of research so use them wisely (that is, make sure you have plenty of popcorn to watch me spin).

Thanks mysterious commenter!
- posted @ 2:51 PM