November 2001. When transit planners and lobbyists for train manufacturers want to bring a light rail or monorail system to a city, it seems like the answer to a commuter's prayers. The proposed trains will seat x-numbers of people, and make y-number of runs per hour. Z-thousands of passengers per hour are predicted. Can packed trains and open roads be far behind? So then why has the performance of new rail systems been so disappointing? The reason is that in the effort to sell the public on trains, mathematical capacity is being equated with ridership. Obviously, the two are not the same-- just because a 300-seat monorail runs every 3 minutes doesn't mean 12,000 people (300x[60/3]=6,000, in each direction) will be riding it every hour.

PRT capacity: Total Riders Served

When transit planners look at PRT, one of the first things they want to know is how many passengers can it carry per hour, per direction? So you tell them 3600 cars, with 1-3 passengers each, assuming 1 sec between vehicles. And they reply, oh, but a train could carry 6,000. And 1 sec. headways don't sound safe, they continue, assuming it's even technically possible let's make it 5 seconds. Hey, that's only 1,200 per hour per direction. PRT is low capacity, they conclude.

But passengers per hour per direction is a nonsensical way to evaluate PRT capacity. Because PRT is a network, extending north, south, east or west, there is no representative section of guideway reflecting ridership throughout the system. Unlike a train system, where even brief rides can be averaged to the whole system because all riders are forced to ride in the same corridor, PRT riders can be going in any direction, on trips of various lengths, and using different routes. Moreover, when a PRT trip is concluded each vehicle becomes available to another rider. Thus, in an hour the capacity of a single PRT vehicle is many times its number of seats.

PRT is High Capacity Transit

Understanding that PRT is a network makes it clear that measurement of passengers per hour per direction becomes irrelevant. What counts is the total number of people served by the entire system, no matter the route they take or length of trip. Instead of "passengers per hour, per direction", the correct measure applicable to PRT is "passengers per hour, ANY direction". Let's hypothesize:

Number of PRT vehicles in system	5,000
Average passengers per trip	1.2
Average trip duration	10 minutes, Footnote
Average time between trips	1 minute (a generous amount of time to enter/exit a car)
Average trips per vehicle	5.4 per hour (60/[10+1])
System Capacity per hour	27,000 trips, 32,400 riders ([5.4 x 5000]x 1.2)

Footnote: Yes, only 10 minutes. Remember, "forget everything you thought you knew about mass transit." In this case, PRT trips take less time because they are non-stop. At 35mph, a 10 minute PRT trip covers 5.8 miles, not a bad estimate for an average.

Pretty impressive, eh? But that's vehicle capacity-- What about station throughput? Can the network actually board and unboard that number of passengers per hour? In the essay Why PRT is Less Expensive, Pt. II a PRT network with 5,000 vehicles and 218 stops is hypothesized. Let's assume the above system has 218 stops too; some stops would have one berth, most would have three, others at malls, stadiums, colleges or hotels would have more. Let's pick 3 as an average. How many boarding-deboarding pairs (since all trips have a beginning and an end) can the system handle?

Number of PRT stops in system	218
Average #berths per stop	3
Total #berths in system	654
Time between arrival or departure	1 minute
#arrivals and departures per hour per berth	60
Arrival:departure split per berth per hour	30:30
Trips per hour	19,620 trips (654 x 30)
System Capacity per hour	23,392 riders (19,620 x 1.2)

This can be increased if the time between trips is reduced reasonably. How long can it really take to go through a turnstile, stick a destination-encoded ticket into a reader, walk into a PRT car, sit, and go? 15 seconds? 20 seconds? (Real video | Win video) Let's be conservative again and use 30 seconds-- the number of trips doubles to 39,240 per hour, any direction. So we can conclude that not only do the network's cars have the carrying capacity, but also that the stations have sufficient throughput. In short, line-haul, 75-200 seat trains are not necessary to achieve MASS transit.

Of course this is all theoretical so far. But let's overlay a real-world number-- King County Metro Transit, average weekday trips per hour: 13,078 (313,882 average per weekday, Source). The demands on the above PRT system come out as:

Average passengers per trip	1.2
Average #trips per hour needed	10,898 trips (13,078/1.2)
Number of PRT vehicles in system	5,000
Average trip duration	10 minutes
Average trips per vehicle	2.18 per hour (10,898/5,000)
Average time between trips	38.2 minutes (60-[2.18 x 10])

So PRT capacity exceeds metropolitan Seattle's existing weekday transit demand on an average basis. Would the system approach capacity at rush hour? Well, if we figure that a single hour of 'rush hour' accounts for 10% of total daily trips, then the demand between 8AM and 9AM is 31,388. And that's for Metro Transit's ENTIRE service area; according to the "ITR Tunnel Team Report to King County Council Transportation Committee", the peak hour 'load' in the Downtown Seattle transit tunnel (over 2010-2020) will be 18,500 people (55,500 over 3 hours) on buses or 11,000 on the light rail alternative. And either way there's going to be buses entering Downtown on surface streets as well. So let's make it tough and assume 100,000 in-bound Downtown PRT commuters over 3 hours of morning rush: this works out to--

Average #commuters arriving per hour	33,333 (100,000/3)
Average time to deboard	15 seconds
Average #commuters arriving every 15 seconds	138.8 ([33,333/60]/4)
Number of 6-berth stations	23

So PRT could handle Downtown rush hour, probably even with an allowance for reverse-commuting by Downtown residents employed in other parts of the city. Q: But can the guideway handle this volume, considering a range of 1,200-3,600 cars per hour on a segment of guideway? A: Oh, you mean 1,200 at 5 sec. headway, 3,600 at 1 sec.? Q: Yes. A: What would be a reasonable number of PRT lines entering Downtown Seattle? How about 3 from the north, 3 from the south, and 3 from the east-- a total of 9, Footnote. 3,600 per line per hour equals a total of 32,400 per hour entering the Downtown grid, with the demand per station being 1,408.6/hour, 234.7/berth/hr, and 3.9/berth/minute (about one arrival every 15 seconds-- or a 660 ft spacing between vehicles travelling at 30mph).

Footnote: Think about it in terms of Downtown being bordered by 9 loops of the network, each loop passing PRT traffic into the loops covering Downtown.

But what about crowds?

Let's conclude by examining the subject of crowds and transit. In most cases there won't be crowds at PRT stations like you'd find at train stations. Because PRT is on-demand, riders come and go all the time and crowds don't accumulate. System efficiency does not depend on forcing people to wait (and ride with strangers who have different destinations anyway).

Another crowd situation: it is often opined that PRT couldn't possibly handle the crowds of people attending major public events such as Seattle Mariners games at Safeco Field.

The first thing to realize is that people don't arrive or depart from a stadium (or a mall, or a university, or a civic festival, or an office building) all at the same time. No, it happens over a period of time. Fans may start arriving an hour before the first pitch, and even after the game starts there are still people arriving during the 1st and 2nd innings. A typical game is about 3 hours long, but again, the fans don't exit precisely at the end of the 9th inning: depending on the closeness of the game, departures may start in the 7th or 8th. After the game, many fans stay to seek autographs, or watch the retractable roof close.

But the bulk of the spectators do start leaving the stadium after the final out-- again, not simultaneously, but in what might best be described as a wave. It is the challenge of transit to dissipate the leading edge of that wave. It is unquestioned that a train station could handle this, but it would do so in 'clumps': people arrive to board the train, and there is a wait for the train to fill and the scheduled departure time to arrive. Then the train leaves. BUT the people still at the station have to wait for the next train, so a crowd builds, and therefore a large station must be built to hold them. And where does the train take people? To another station. It merely takes the crowds from Safeco Field and relocates them. People still have to get home.

So what could PRT do? First, in this situation you WOULD build a large PRT station, but not one with a huge waiting area-- it would have many berths to continually board and carry away a stream of rabid baseball fans. It could be a long, sidewalk-wide platform with a line of berths and ticket dispensers; buy a ticket and immediately board and depart. The leading edge of the 'wave' of people is continually dissipated. Could PRT handle a sellout crowd, 47,000 fans? Well no. But the goal of transit is not to handle all of them, just a significant portion. Remember, there is a garage and paid lots at Safeco, most fans would still drive their cars. Let's set a target of 30%, or 14,100, that most are in groups of 2 or 3, and let's say two 15-berth PRT stations are built at opposite corners of the stadium. The demand on the PRT system works out as:

Average passengers per trip	2.5
Number of vehicle trips needed	5,640 trips (14,100/2.5)
Number of PRT vehicles in entire system	5,000
Number of vehicles temporarily assigned to stadium-only service	2,982 cars (5000-[10,898/5.4]), Footnote
Average time between PRT departures	0.25 minutes (15 sec. to get into car)
Number of PRT departures per minute	120 per min. (30 x [1/.25])
Time needed for 5,640 vehicle departures	47 minutes (5,640/120)
Average trips per vehicle	1.89

Footnote: Start by calculating how many vehicles could satisfy the average hourly weekday demand if running at capacity of 5.4 trips per hour-- 10,898/5.4=2018, this is the number of vehicles remaining in general service. Subtract 2018 from the total number in the fleet (5,000) and you get 2982 vehicles temporarily assigned to stadium-only service. Note that in actual practice these would not be the SAME 2982 cars throughout the 47 minute period, but rather the number assigned at any one time. Cars will not necessarily make a trip and then deadhead all the way back to Safeco to pick up more people. Instead what would happen is that a car might make a 7 mile run to Ballard; it would then be released to general service and replaced on stadium duty by a car that is close by Safeco Field. On the other hand, it would be reasonable for a car to deadhead back from a run to Georgetown.

47 minutes is not unreasonable. And PRT takes travelers directly to a station within walking distance of their destinations. So doesn't it make more sense to...

Get on board! Personal Rapid Transit?

The author has a degree in Policy Analysis from the University of Washington Graduate School of Public Affairs (now known as The Evans School).