A Comparison of "Subways" and "Rapid Transit"

d.w.rowlands [at] gmail.com

Visiting Cleveland reminded me of this post on a DC area urban planning blog I like and its distinction between pre-auto-age "subways" and the more recent heavy rail "rapid transit" systems. Since the blog post identifies the Cleveland Rapid Red Line as the last of the "subways", despite it being separated by about twenty years each from the last of the other subways and the first of the other "rapid transit" systems, I decided to do a numerical comparison of the subway and rapid transit systems in the US to see how clear a distinction I could make.

Half of the stations on the Cleveland Red Line have parking, compared to 9 out of 22 on the Boston Red Line (although the only pre-1980 station with parking is the downtown intercity rail station) and half the stations on the Los Angeles Red Line have parking. The other comparison that seemed worthwhile was to calculate the number of stations per mile for each of the fifteen heavy rail transit systems in the US (ignoring downtown circulators like the Detroit People Mover). Transit planners generally assume that riders-of-choice will walk half a mile to or from a rapid transit station: any longer walk will make them decide to drive instead. To provide transit service to a contiguous area larger than a circle with a half-mile radius, stations need to be spaced close enough that these circles will touch or overlap. On the other hand, park-and-ride stations intended for suburban commuters to drive to do not need to be so closely spaced, and wider stop spacing is beneficial for decreasing travel times to compete with freeway driving.

The following table shows the year each system opened (although some of the older ones have had significant extensions much more recently), the length of rapid transit lines in miles, the number of stations, and the number of stations per mile. The LACMTA Metro and MBTA T are special cases in that the systems contain, as well as heavy rail, lines that would be considered "light metro" in Europe: grade separated and with their own rights of way. In the MBTA case, I have included the D-Riverside branch of the Green Line along with the Blue, Red, and Orange subway lines. In the LACMTA case, I have included the whole system, since the Green Line is totally grade-separated, the Gold Line has absolute priority with gates at its grade crossings (which are minimal) and most of the Blue Line is grade separated or has gates at its grade crossings. (I didn't go to the trouble of figuring out exactly how long the non-controlled ROW is.)

The pre-1955 systems are all in the four cities that had subways before World War II: the 'L' in Chicago, the T in Boston, the Staten Island Railway, Subway, and PATH in New York City, and the SEPTA Market and Broad lines and PATCO Speedline in Philadelphia. The post-1955 systems are the multi-line networks in the San Francisco Bay Area (BART), Washington, DC (WMATA), and Atlanta (MARTA), single lines in Baltimore, Miami, and San Juan, and the Los Angeles subway and light rail network (LACMTA).

A pattern can be observed here. The pre-War systems mostly have between one and two stations per mile, with the exception of PATH and the PATCO Speedline, which are both short services that exist partly to cross wide rivers (the Hudson and the Delaware). PATH also connects Newark and Hoboken/Jersey City, which are separated by marshes and the Hackensack and Passaic Rivers. The New Jersey portion of the PATCO speedline mostly serves suburban park-and-ride stations, and is more similar to a modern rapid transit system than an urban subway. Even with these large over-water components, PATH and the PATCO Speedline still manage nearly one station per mile in their fairly short lengths.

As for the post-1955 systems, BART is an extreme in the other direction, with less than one station per two miles, a consequence of it being much more of a regional rail network disguised as a subway than an actual urban mass transit system. (San Francisco has its own light rail / streetcar network with a subway under Market Street that provides access to more of the city.) Other than that, the majority of the systems have 0.8 to 0.9 stations per mile, less even than PATH and PATCO. The more recent Miami and San Juan systems do seem to have somewhat higher station densities, possibly evidence of a tendency to move back towards denser lines.

From this, it does seem that the RTA Rapid in Cleveland is sort of an intermediate case. It has a lower station density that most--but not all--of the pre-War subways. On the other hand, it has a slightly higher station density than most--but once again, not all--of the more modern rapid transit systems.

System Opened Length
Stations Stations
Staten Island Railway 1860 14 22 1.57
CTA 'L' 1893 107.5 144 1.34
MBTA T (grade-separated only) 1897 51 71 1.39
New York City Subway 1904 229 468 2.04
SEPTA Market and Broad 1907 25 50 2.0
PATH 1908 13.8 13 0.94
PATCO Speedline 1936 14.2 13 0.92
RTA Rapid Red Line 1955 19 18 0.95
BART 1972 104 43 0.41
WMATA Metro 1976 106.3 86 0.81
MARTA 1979 47.6 38 0.80
Baltimore Metro Subway 1983 15.5 14 0.90
Miami Metrorail 1984 22 22 1.00
LACMTA Metro 1990 79.1 70 0.88
Tren Urbano 2004 10.7 16 1.50

So, given we have this data, what is the optimal station density for a rapid transit line? Having stations closer together slows down travel, but also increases the number of people who are close to a transit station. If one is running a line through car-dependent suburbia and the stations will only really be accessed from adjacent park-and-rides, of course, one can have stations further apart than in a walkable urban area. And, as noted above, stations will logically be further apart when a line connects neighborhoods on different sides of geographic obstacles such as bodies of water or mountains (i.e. the Hollywood Hills in Los Angeles).

In an urban area, though, the usual station spacing should be a function of how far pedestrians can be expected to walk to or from a transit station. Transit planners often assume that about a half mile---a ten minute walk for most people---is a reasonable distance, indicating one station per mile. This makes sense if you only care about your transit system serving people who live right along the street it runs under. To increase the area served, it may make sense to have the stations rather closer together, even if this results in overlap in their coverage areas.

This slows down trains and makes transit less convenient for longer trips, but it makes it more convenient for short trips, where the walk to and from the station is a larger part of the total trip time, especially if trains run frequently. If you're going to have long trips, though---as needed in a city like New York or Los Angeles---with dense stop spacing, you're probably going to want to put in express tracks. Unsurprisingly, the network with the densest stops--the New York Subway--also has express tracks on most lines. Unfortunately, given how hard it is to get funding to build a subway in the US at all, it is unlikely that a new system anywhere will get funding to be built with express tracks, and once a line is built, it's generally impossible to add express tracks as ridership justifies it.