Wednesday, February 25, 2009

LRT Number Crunching

I got into a debate on Streetsblog about passenger miles and capacity and saw a calculation I had never seen used as a reason that light rail sucked. It's an interesting calculation but it tells me nothing. GaryG states:
You simply multiply vehicle revenue miles by vehicle capacity in passengers to get total revenue capacity in passenger-miles. You then divide actual passenger-miles by revenue capacity passenger-miles to calculate the share of capacity that was actually used to transport passengers in revenue service. The calculation works out to an average occupancy of about 14 passengers per light rail vehicle. Each vehicle has a capacity of 140 passengers or more.
This makes no sense what so ever to me. First off, averaging anything over the whole day seems silly to me. Do we talk about roads being a waste when they don't carry cars at midnight? Another thing is that there's no filter for the peak period which is what of these lines relieve pressure on. The telling thing is that when we looked at the New York Subway, it was only 28 passengers per car. It reminds me of that calculation done a few years ago that all transit systems except for BART are worthless.

Anyways, my argument was that we should look at cost per passenger mile to see the efficiency of each mode. There was a fight about capital costs as there always are from people who think buses can do any job rail does but it led me to the 2007 National Transit Database numbers in easily readable format. Check out some of these numbers:

Cost Per Passenger Mile 2007

34 cents Light Rail
67 cents Bus

39 cents Light Rail
93 cents Bus

San Diego
27 cents Light Rail
71 cents Bus

53 cents Light Rail
55 cents Bus

42 cents Light Rail
72 cents Bus

San Jose
103 cents Light Rail
155 cents Bus

1.63 cents Light Rail
1.24 cents Bus

116 cents Light Rail
101 cents Bus

Perhaps light rail in San Francisco is a waste. We should just rip out the tracks, no one likes riding on rails anyways...right? Wouldn't it be nice to toss a three car train on the N or KLM? Perhaps making Muni less like a bus stopping at every stop sign in the Avenues and on Church would help too.


crzwdjk said...

You have to be really careful here. Almost all bus systems have all sorts of relatively lightly-used lines, which are of course less cost-effective, while light rail lines are almost always only on the busiest corridors. What you're seeing in SF is the crush-loaded 1, 14, 30, etc. outweighing the outer ends of the J and M. And I assume that rail maintenance counts toward operating costs, whereas road maintenance does not.

And if you want to fix Muni rail service, the first thing would be to get decent dispatching in place so that they can keep the service running well through the inevitable disruptions. Also, get rid of as many stop signs as possible, and optimize running speed. Now, the problem is congestion in the Market Street tunnel. One thing that can definitely help here is coupling trains at the portals, which would free up some capacity in the tunnel. And they really need to throw out the schedule and start over, and think hard about how to balance core Market St service against the outer sections of the lines, how to balance service frequency against reliability, and how to best serve east end termini (Caltrain, Third St, Embarcadero). And maybe look into some capital investments, like more ticket machines at busy stations, and maybe some more high-level platforms.

Pantograph Trolleypole said...

Yeah it would be nice if we could separate out bus lines in the NTD or if transit agencies would post that type of information. They should do it so we can see how things are being accounted for. Just as they should post ridership as well.

Unknown said...

The metric used (space-miles vs passenger-miles) does have some validity, but definitely not for off-peak service planning. It is appropriate to use when planning the balance of PEAK service between a number of lines that feeds into a congested and constrained area like the Market St subway.

Robert said...

The delta between San Diego's bus and rail costs per passenger are very surprising. The lack of a delta between Houston's costs are similarly surprising.

Thomas said...

The lack of a major difference between bus and rail operating expenses per passenger mile in Houston might be low due to the relatively short trips that the Red Line, as a 7.5-mile-long collection and distribution facility, carries. If you transfer from a bus to the train at Wheeler Station to finish your trip into downtown, your trip on the train is going to be relatively short. Also, a sizable portion of the Red Line’s ridership consists of Texas Medical Center employees who travel from remote parking at Smithlands to their destinations at Dryden/TMC or Memorial Hospital/Houston Zoo; this is a trip of about 1 to 2 miles max.

As for the Streetsblog debate: I read that thread, and I thought about responding but I’m not sure I want to get into another useless and tedious quibblefest with a rail opponent. Anyway, what garyg is doing in with his “capacity utilization” calculation is attempting to create a transit equivalent of what the airlines refer to as “load factor:” multiply the number of seats available by the number of miles traveled to get available seat miles and then divide by actual passenger miles to get a volume-to-capacity ratio.

This is a useful performance indicator in the aviation world, where specific point-to-point trips are flown a few times a day, but it is not a standard metric in the transit industry. I’ve never seen this type of calculation or its result used by anyone, from the FTA to local transit agencies to engineering or operations consultants. This is simply because this particular metric is meaningless relative to the purpose and intent of public transportation.

It has already been pointed out, for example, that garyg's metric completely ignores a transit system's performance during peak periods, when it is used the most and provides the most benefit. Nor does it take into consideration trip lengths, which as my Houston example above shows can have a profound effect on passenger miles. Finally, this same type of metric could just as easily be applied to highways to prove that they are severely underutilized. (Take a 24-hour vehicle count at any point along a freeway, divide it by the number of lanes on that freeway multiplied by 2000 veh/hr and 24 hr/day, and the resulting percentage will probably turn out to be rather low even though there's bumper-to-bumper traffic at that point during the AM and PM peaks).

Garyg himself comes up with a capacity utilization of 16% for the entire NYC subway system! That alone should indicate that this performance indicator is of dubious value.

Adam said...

This is why I don't like streetcars; they basically are buses on rails. It's kind of like the Newark Light Rail's Broad Street line (which is VERY slow), and the Hudson-Bergen Light Rail within downtown Jersey City. Things work much better whan trains get their own dedicated rights of way, and if they have to cross a street, use traditional grade crossings and not traffic lights (but grade separation is of course the best of them all).