Monday, February 20, 2012

137> Playing Monopoly

PRT is a confusing soup of construction, computing, vehicle manufacture, public works, system maintenance, and a bunch of other stuff. It is a complicated business model and it is difficult to estimate costs. What is worse, there is a tendency for would-be providers to estimate costs on a per mile/kilometer basis, which further confuses things. After all, vehicles, track, stations and control each have their own costs, lifecycles, and logic. Building vehicles, for example, has very little in common with building track or stations. The “per/mile” estimate is especially problematic in view of the landscape we face in the US. Here, our cities have grown into a suburban sprawl that has little consistency in terms of the placement of destination-rich areas.

To help the reader understand how this came to be, let me recount what was told to me, back in the seventies, about a little “up-and-coming” real estate developer, Trammel Crow. His formula for success, I was told, was really quite simple. First, find a growing city. Take the main road out of town until the land is sold by the acre, instead of by lot, and by some frontage. Build a tilt-slab office/warehouse on it, put it up for rent, and wait. It reminds me of playing Monopoly. Instead of using all of your money to buy Boardwalk, you can buy the cheaper Baltic Avenue, and “develop” it with houses instead.  In either case, though, you buy and hold, while investing as much as you can afford to make it into an income producer in the meantime. This is instructive in understanding how cities develop such dysfunctional layouts. Land developers are, underneath it all, land holders, and often simply have something minimal on that land to pay the taxes and generate a little income while they wait for it to appreciate. This also helps to explain the vast parking lots that take up so much of our cities. Only at Christmas are they anywhere near filled. The land owners simply don’t have the money to build and maintain anything more ambitious. Even if they have the cash they are more likely to buy additional land and do the same thing elsewhere with those funds instead.  Anyway, this has helped contribute to retail outlets that are big and far from the road, and getting from one such store to another is often not a walkable distance. This effect is also coupled with the effect of freeways, since freeway frontage offers an ideal place to exercise that Trammel Crow model, except the lease is to “big box” retailers, who have discovered that economies of scale are more easily exercised away from the expensive downtown areas.

When it comes to walkable, integrated urban/suburban environments, it is usually the old parts of town that shine. Little towns that get absorbed by cities usually retain their main streets and hopefully a bit of their charm. But these, too, are “destination islands” in a sea of sprawl.

For a PRT system to be a viable way to get around, it has to go to these important destinations. If they are in clusters separated by substantial distances then this strongly suggests a PRT design that allows more than a single speed. It suggests a design that is similar to how freeways work, where there is a slower feeder that runs parallel to the faster main highway. This is because the deceleration and turning are disruptive to the faster, distance-oriented thru-traffic. In the case of PRT serving disparate, clustered destinations, it is possible that the best design might be for the local, feeder system to be quite slow. This would allow an absolute minimum of the double track required for off-line stations and allow for extremely tight turns. Faster track would connect these destination clusters.

This ties in with the discussion of land development and PRT cost estimation in the following way: First, it is important to get some kind of handle on the costs of running track alone. The downtown PRT models tend to assume blanket coverage for a pedestrian rich area. In the case of destination-rich freeway frontage, the proportion of stations and vehicles per track distance would seem similar, being based on walkable distances, although it would be linear, rather than based on loops covering city blocks. The “old town” destinations, referenced above, are classic downtown loops, just scaled back to one or two. But connecting all of these (and the actual central business district) is fast track. So when figures like ten to twenty million dollars per mile get tossed around, it is highly misleading. I would guess this stationless track, on public easements, would be more like 2-3 million per mile. This fast track could easily pay for itself, it would seem, by virtue of the fact that it would cut through so much traffic.

Those “big-box” destinations tend to have that extra parking lot space, and this would seem to be a great place to put PRT stations. But the land is rarely owned by the retailer, and therefore it makes little difference if that retailer wants to provide PRT access or not. The land owner has his own agenda, and understanding his wants and needs is what counts.

The first thing to consider is that he won’t want to give up sovereignty over a single square inch. That means anything permanent is problematic. At present, the advantage of hosting a PRT station is only theoretical and it cannot be expected that this is all a landholder would want at this time. This is especially true if there are excessive requirements for utilities, permits, digging and the like. On the other hand, the land in question is often of very little immediate value to the landowner, since building anything large on it would block their main tenants’ visibility from the street. So it comes down to that ultimate grease to getting things done – profit. How can the land owner directly make money by having a station? It is pretty obvious that in an ideal world, the pedestrian traffic generated by the station would be coveted by the tenants to the point that they would pay greater amounts to lease their stores. The landowner would therefore willingly give up the rights to enough space for a station. Let’s keep our optimism in check, at least for now, and say we have to sweeten the deal.

In theory he could get a small piece of every transaction, and there is also the potential for parking revenue. The amount charged couldn’t be much, because otherwise people would park for free in the guise of being store customers, a potential problem with almost any good transit system that is close to retail outlets. But if the parking is right next to the station, and the station is easily removable, and there is nothing for the landowner to do but collect money, I think there is potential. Electronic payment means can integrate parking fees and PRT fares into a single transaction.

What is good, though, is that this helps solve a major problem of transitioning from a car-based to a pedestrian based cityscape. Absent a real “last mile solution,” PRT could still make a huge contribution to traffic reduction. If the system doesn’t get to your door, it should at least get to your grocery store.

This all begs a bunch of questions that need to be addressed in a future post. Specifically, we need to examine further the issue of decoupling the stations from the track, from the vehicles, from the control. Each has its own rate initial costs, rates of depreciation, etc. Varying the ratios between them greatly influence the profitability of any venture. Can a bunch small stations be used in place of a bigger one? If we are going to bridge a bunch of destination clusters, how far apart is too far? Can the business models for each be separated and therefore simplified to be made more attractive to investors? We’ll get back to this. Stay tuned!


Asko K. said...

Thanks, Dan.

It feels good to have you peek out of the technical bench for a while! Yes, the operating scenario for regular U.S. cities looks daunting, but the solution you draw up is plausible. The only down side I see is that it calls for a rather big initial network, in order to be useful for those from-home-to-mall drivers, so they could leave the car at the mall parking lot while doing "business" over PRT. Did I get your thoughts right?

Europe has it potentially so much easier. We already have "park-and-ride" places at major subway or local train stations, for people to leave their car there and carry on by public. I personally like it. There should be more of that (the parking cost for a whole day is reasonable 2 eur = 3 usd).

The pricing confusion you state in the beginning is on my mind, too. On the one hand, "customers" want to have a clean, comparable pricing. On the other hand, such cannot be given due to the complexities you mention. Leasing the service (i.e. not selling the track, vehicles or stations) shuffles it even further. So my end result is pretty much to just compare the costs to a) bus transport b) trams / light rail c) subway or whatever the particular person is comfortable with. Comparing PRT a with PRT b at the current stage price-wise is a non-trivial exercise.

Then again, comparing mobile phone deals is equally hard. We *must* make the pricing simpler, eventually! :)

Thanks for the write.

Dan said...

Hi Asko,

I have to confess that I enjoy design work more than blogging, and sometimes quite a bit of time passes where the only things I think about, PRT-wise, are aspects of the designs I am working on… I know that the technical stuff is not going to get a lot of readership, but I think I’m going to have a pretty exceptional finished product. Be true to yourself, right?

We have the “park and rides” more than the means to get around once we take them. They are popular for suburbanites who would otherwise face a half hour or more of freeway gridlock, since there are often HOV (High Occupancy Vehicle) lanes for much of the journey. In the thick of things, however, the buses get just as mired in traffic as anything else, so it’s a much better deal if you are getting off at one of the first stops. Here in Houston we have separate P&R buses that go to various other important areas besides downtown. What we lack is a way to get from important area to important area in any sort of timely way, especially without a car.

As far as pricing, I am going to be posting about that next, I think, and what I, personally, am wrestling with is how best to prorate the initial equipment cost and its depreciation so that investors get some attractive return all along, starting at day one. Presumably this can be done at conventional fare prices or prices competitive with driving. (especially considering the time factor) I think vehicles, track and stations need to each get their fair share of the fare based on that. I confess that I really don’t know what percentage return a passive investor currently expects, but I also confess I have done very little work on how to mitigate risks for that investor, and I suppose that comes first. Luckily for me I don’t have any rush to get something on the table! Personally, I’d rather deal with it in the abstract; I sleep better that way!

Andrew F said...

Dan, did you see that analysis done by Koren that argued for ride-sharing or GRT to shave off peak demand? I thought it was very interesting, because it assumed that it was uneconomical to have double the number of vehicles to address peak demand vs what was required when GRT was used to supplement peak demand. It's not clear to me that it is uneconomical. You could charge a slight premium during peak hours which would serve two purposes: first, it would shift some peak demand outside of the peak. Price-sensitive commuters might try to shift their working hours. Discretionary trips would be encouraged to move off-peak. Second, the premium would cover the additional capital cost of the marginal vehicles required.

This isn't to say that GRT is a bad idea. But it seems like it may not be as necessary as Koren suggests. I think it would make sense as a way to connect nodes and trip generators, but that means the GRT network would be substantially sparser than the PRT network.

Dan said...

Andrew, I think it’s hardly worth contemplating the value of GRT unless the context under which it is to be considered is completely known. There are so many variables. I suppose one place to start might be to list what those variables could be, and to try to figure out how likely it would be that they would come into play. For example, one variable is distance. Airports, for example, or often so far from downtown they hardly operate under the same logic as the urban system. Another variable is how many destinations are actually covered. If you only have four downtown stations and they are in a row, there is less need for individualized travel, if you are, say, coming from that airport. Yet another variable is the size of the GRT vehicle. If there is only room for 8 passengers, there is no chance of having to go to nine stations before you can get out. Also larger GRT vehicles would require special, heavier track. Another variable would be the degree to which the traffic flow is one-way. If the node is a remote park&ride, and the destinations are limited to downtown, it might be disadvantageous to squander PRT vehicles by having them take long return trips while empty. That brings up the variable of how new (and therefore incomplete) the system is. Perhaps GRT is better at certain stages of development. Then there are the yet-to-be-established variables of speed and ability to platoon. From a system/hardware design perspective, I can only try not to preclude any possibilities that might prove useful.
Your point about nodes and trip generators hits the nail on the head. If you want to, for example, control traffic generated CO2 and ozone, then a way to get around downtown misses the point. You want to take longer distance commuters off of the road. This kind of far-flung system is probably more GRT friendly than a close-in one. Taking the distance/node concept to the extreme would be the case of travel to adjacent cities. Does the neighboring city get to borrow a PRT vehicle to run around its own downtown if the need arises?
A variable rate fare structure is certainly one tool that can be used to help maximize whatever system attributes you are working with. Together with intelligent vehicle deployment, you could, for instance, find that a GRT shows up for you and seven others because you are all going to the same destination. (It could have an electric destination sign – or even call you by name!) Such none-private service would be an option that you could pick when purchasing a ticket or season pass – presumably at a discount. In other words, perhaps the GRT vs. PRT debate can be played out in real time by the system itself!

cmf-seattle said...

Funny you guys should mention the GRT thing...

All those truck trailers in the background make it look more like a distro/warehouse than a retail destination. I read somewhere that most pallets can handle about 2,000 pounds; would track that can handle that kind of weight (Taxi 2000 quoted about 1,800 pounds for its vehicle, empty) make the vehicle choice less of a factor?

Another way of thinking about this: how many trips, per direction, could possibly be generated in an area, per square-mile? Is there a transport system that can handle that and then some?

Dan said...

Hi cmf…
I just “shopped” together a couple of pics I found Image-searching Target stores, and the one that had the right perspective happened to be a distribution center, not a store, like the closer building. The front truck was imported from Sketchup’s “3D warehouse.” I wanted it to put it under the track in my model to demonstrate the proper track height for clearance. That did not come out in this angle, however. Anyway, I agree about moving merchandise. I have always thought of this as a point-to-point transportation device – be it people or product, baggage handling, warehousing, or even working an assembly line.

Since the essential track dimensions are internal, there is absolutely no reason that it couldn’t take 2000 lb. payloads, simply by beefing up the track size. Also, the difference between closely spaced PRT vehicles full of portly people and a GRT vehicle that has front and rear bogies is negligible anyway. It would greatly narrow the range of station designs though, although not so much for the kind of high throughput stations that are liable to be on trunk lines.

Rather than think about how many people per square mile want to get moved, let’s ask the questions of origin and destinations, and of pipeline size. For example, it would be hard to beat a stadium letting out for pure volume. Do all of the people want to go to one, giant remote parking lot? If so, we need a train. If it is 20 remote parking lots, probably GRT. If it is right to your car, PRT. The question is how many “lanes” are acceptable? It’s easy enough to design a massively parallel station that can spit out 10 vehicles a second. The problem is the spaghetti that comes next as the tracks all converge or diverge. It leads to the prospect of sectioning off large stations by destination. (i.e. all points north, etc.) By the way, I did a rendering of a four track system with a capacity of over 17,000 per hour back in post 111, and pondered high capacity stations back in 49. Boy, I sure need to work on my indexing system. Stuff is getting impossible to find!

cmf-seattle said...

The dynamics of a GRT vehicle with 20 standees are very different from a vehicle with 3 passengers, all sitting and facing airbags.

Since we're not trying to completely replace autos and roads, and most of what people buy can fit into a 4-foot cube (and weighs less than 1000 lbs), does it make sense to build heavier infrastructure?

Dan said...

Dan the Blogger likes links that add context to a discussion! Thanks cmf-seattle. (Sorry, everybody! I really don’t know what I hope to gain by hyping that “pen name.” …Probably just a bad habit!)

First, I certainly agree that infrastructure enabling a vehicle for twenty is too big except in exceptional (rare geographic?) circumstances. Normally PRT would seem much, much better. I can imagine GRT for up to eight, tops.

Anderson’s paper makes my point about the maturity of the system being relevant to the vehicle size. Anderson’s example of a square mile with 12,000 people is principally geared toward establishing the fact that 4 PRT stations per square mile can handle that traffic load. What is interesting, though, is that he states that the average trip by automobile in the US is 7 to 10 miles. This makes one problem very clear. His little one square mile is by no means self-sufficient. For the max ridership to be reached the system would have to extend outward from that square mile by those seven to ten miles - in each direction! Now we are talking about a grid over a hundred times as big. In other words, ridership is dependent on how well life’s needs, including employment, are met within the bounds of the system. Even a PRT system that serves a stadium doesn’t need to be high capacity unless it goes to enough important destinations from there to draw people to use it.

The point I would make is that for GRT to make sense it should travel on the PRT track and travel from one high density node to another, so that most or all passengers share a common destination. This probably means a trunk line, and it really doesn’t matter if GRT can’t go everywhere anyway. A trunk line would have gentler, structurally stronger curves (for higher speeds) and be designed for denser traffic than track in a suburban setting. Because of the extra weight, GRT would need to have extra headway, and probably should have somewhat slower speeds, possibly presenting a conflict. However I can imagine a system that is continually growing, where GRT is phased out of routes as the station density starts producing more competing high speed PRT traffic. The GRT would sort of be a trail blazer, always serving important destinations that are on the edge of the ever expanding network – maybe even on a schedule. I can imagine, for example, the following situation: At the very most distant park&ride lot that has a PRT station, the PRT vehicles are not coming back fast enough to meet demand. Let’s say the station is a full 3 miles further out (along a major freeway) than the next station. The PRT operators could buy more vehicles, but the shortage only occurs half the time, and only at morning rush hour. Now what would happen if along came the old 8 o’clock GRT? Eight passengers get to go downtown and the PRT company saves having to buy 7 or 8 vehicles. A remaining few people who want to go to oddball places might have to wait a few minutes. True, this is might offend PRT purists, but a good system should be adaptable to all circumstances – including not having the money in the budget to do everything perfectly for everybody all of the time.

ARS (Architectural Refinishing Services) said...

We have the “park and rides” more than the means to get around once we take them. thanks

Suspended ceiling