Daily Archives: October 14, 2013

Make Your Light Rail Look Like LA’s

Regular readers know that this blog doesn’t have a devotion to any particular transportation technology. I’m all about efficiency. The best options are the ones that move the most people and goods as fast as possible. Now you can drift off into daydreams about “slow transit” but fact is, people usually want to get where they’re going quickly and reliably. People vote with their feet and if you want their feet on your transit vehicle instead of on the gas pedal, your transit better be competitive. People don’t want slow transit any more than they want slow freeways.

So, if you’re investing money in a light rail network, you ought to make it look like ours in Los Angeles. Chances are your city doesn’t have the density of Manhattan or narrow streets of downtown Boston that make subways the only practical option. If you look at LA’s light rail network, you’ll see a combination of pragmatic decisions that gets a lot out of the money put into the system, and generates good ridership. At the other end of the spectrum are cities where decisions seem to be made based on an infatuation with trains as they worked 100 years ago, and everything else flows from there.

Let me say at the outset that I don’t want to come off as trying to beat up on Portland. I’ve never even been there, and I’m sure that planners there have good reasons behind their choices. Those choices are often driven by regional land use planning rather than just transportation considerations. However, the “Portland model” is inescapable these days. You read about it everywhere, and many cities in the US cite Portland as an example when promoting their own streetcar or light rail plans. So as far as I’m concerned, Portland’s network design in general is fair game. If you want your city’s plans to be successful, I think you’re better off trying to emulate LA than Portland.

There are several ways in which LA’s network design is superior; in order from broad planning down to engineering details, they are: service area, overall route configuration, station spacing, grade separation, and route geometry.

Service Area

The first, biggest planning question is what part of your region you’re going to serve with your light rail system. Odds are you don’t have enough money to crisscross your city with rail lines in any time frame other than the region’s long-range (30-year) transportation plan. For the uninitiated, this plan is where politicians and planners stow the projects that you want but the region has no plausible way to finance. Any project that’s currently on the books to be done by 2040? Yeah, we’re about five years away from it being pushed back to 2050.

That means that you have to make hard choices about where you’re going to build your first line or two. Don’t screw it up, because a failing line will be fodder for a Randal O’Toole blog post and might turn the electorate against you. Well, at least if you’re not VTA, it might.

Many people, from developers to planners to politicians, often see LRT lines as corridors for potential development. This can backfire if the development doesn’t actually occur, or if it does occur but fails to generate appreciable ridership. Instead, it’s better to focus on places where there’s already a lot of travel demand and additional development potential. Since most American cities have relatively low density, there’s plenty of these places around. The takeaway here is that it’s better for transit to be reactionary – that is, serving travel demand that already exists – than it is for it to be anticipatory – that is, serving travel demand that may theoretically exist in the future.

Now take a look at the Los Angeles LRT network.


The first line, the Blue Line, connected downtown Los Angeles and Long Beach, the second largest city in the region. It’s in an area where there’s already a crapload of travel demand, as indicated by the congestion on the 110 and the 710. And there’s virtually unlimited upzoning potential all over the place, even if it hasn’t really been capitalized on just yet. Is it any surprise that the Blue Line is closing in on 100,000 riders per day, while the entire MAX network generates just 130,000? LA’s system generates about 2,800 boardings per mile (with the Blue Line at about 4,000 per mile), about 1.5 times MAX, and that’s without a critical piece of the network even being built yet (Regional Connector).

LA’s other light rail lines follow the same pattern. They serve parts of the city that are already built up, have a lot of destinations, and have plenty of growth potential. The Gold Line north connects downtown LA to Pasadena parallels the 110. The Gold Line east serves East LA and isn’t far from the 60; this line suffers from poor connectivity to other lines but that will be rectified by Regional Connector. The Expo Line serves downtown LA, USC, Culver City, and Santa Monica, more or less near the 10 on the Westside. You could even throw the Orange Line BRT in here, since it serves built up areas in the Valley and is in the same corridor as the 101. All of these freeways are among the most congested in the country, and there’s a ton of demand for more development in these areas.
The Green Line, which runs from El Segundo to Norwalk in the median of the 105, is perhaps the most enigmatic of LA’s LRT lines; it doesn’t follow the same development pattern as the others. Nevertheless, it did better than expected for ridership, and I’m pretty confident it will become more useful as the network is built out. The upcoming Crenshaw/LAX LRT project is underrated as relief for travel in the La Cienega and La Brea corridors, and at any rate, the obvious intent there is to eventually extend the line north and south, making it very useful.

Now on the other hand, PDX built their system with some intent that development would follow the rail lines. They’ve managed to encourage some pretty impressive growth in the Pearl District and South Waterfront, but O’Toole’s been keeping the bills paid for years writing about vacant TOD lots. But here, ain’t no way he could ride the Blue Line and say that no one wants to use it, and Reason has already made a laughingstock of itself in about a year and a half of Expo Line operations.

Despite that, on the whole, I don’t really have a problem with Portland’s service area. They serve the central city and connect it to surrounding nodes of density – such as they are, since Portland on the whole is pretty low density. If there’s a failure of the service area, it seems to me that it’s a failure of overplanning, of trying to force development into certain parcels while protecting most of the city from redevelopment. Others have speculated on this as well.

Overall Route Configuration

Once you’ve figured out what areas of your city you’re going to serve, you need to lay out your routes. In general, you should follow Jarrett Walker’s route design guidelines for bus service: make ‘em straight, make ‘em long enough to aggregate demand, don’t deviate to serve specific points, and try to put something worth going to at the ends. A good yardstick for this is actually to pretend that the LRT network is a freeway network, and ask yourself if the highway department would propose it with a straight face.

LA’s Blue Line has a dream alignment once it turns south off of Washington. In addition to stations in the middle that generate a lot more demand than TOD-types would expect, it’s anchored by downtown LA and Long Beach. Expo Line has a few medium speed curves, but it’s pretty straight, and anchored by downtown LA and Santa Monica. Gold Line has downtown LA and Pasadena. The big knock on the Gold Line in this regard is the really slow curves on both sides of Union Station, which could have been avoided by a straight alignment on Alameda. Now the benefit to this nasty routing is considerable: much shorter transfers to Red/Purple Lines, buses, Amtrak, and Metrolink at Union Station. I leave it to you to decide if the penalty is worth it.


The Green Line, again, is the lone man out. It’s a high-speed alignment the whole way, and has solid demand at El Segundo, but it doesn’t get close enough to LAX to be a real option, and on the east end it just sort of fades away in Norwalk. A short extension east to Metrolink might be useful in that regard, and long-term an extension along Imperial should shore things up. (More on that soon.) Crenshaw’s route is decent, and eventual extension to the north will make it a practical line between LAX and Hollywood.

Again, for the most part, I don’t have a problem with Portland at this level of design. The routes appear to be reasonably straight and have logical destinations. My real issues with their network are going to manifest themselves further down the chain regarding station spacing, grade separation, and local route geometry. However, I do have some questions about some of the outlying route terminals, which look like they’ve been built in a very anticipatory manner.
For example, here’s the southern end of the Milwaukie line, now under construction.


With the exception of Milwaukie’s tiny downtown, the area is surrounded by low-density SFR development. There’s a few apartment complexes off to the east and southwest, but it’s hard to see how this area is going to generate that much ridership.

Here’s the north end of the Red Line, ending at Portland International Airport.


We have a station in no-man’s-land in the median of the 205, bordered by low-density residential, an apartment complex or two, and vacant land. The next two stations serve, plausibly, a mall and big box retail, some airport-related industry, and airport hotels, along with vacant land and parking lots far greater than the developed area. The mall might generate some ridership, but you have to think that the hotels offer shuttles, since most people don’t want to carry their luggage from a station to the hotel. Hotel and retail employees would be potential riders.

Here’s the south end of the Green Line, at Clackamas Town Center.


I have to say, this one is really puzzling. The line ends at a big parking garage at the forlorn edge of an enormous mall. The next station north, Fuller Rd, serves a big parking lot, a few SFRs, some industrial land, and some big box retail. Obviously, the hope is that someday the mall and its sea of parking will become an urban neighborhood. But in the meantime, in between, there’s a slew of apartment complexes – most of which are just far enough way to encourage people to drive, especially since parking is provided. It looks like there might be enough space to put a station near Causey Av. What gives?

The Yellow Line is even worse.


Its northernmost two stations are literally surrounded by nothing, the southern one being saved only by what appears to be decent park-and-ride usage from people getting off the 5, which is rough with 15 minute headways. The only plausible explanation is that eventually you want to go to Vancouver, but Washington adamantly doesn’t want light rail.

Station Spacing

Now we’re getting into a level of detail where I have real beef with the Portland model. Stations should be spaced about a mile apart, assuming you’ve got enough density along the line to justify them – which you will if you plan your service area and route configuration properly. Closer spacing is acceptable in areas that are very dense with residents and/or employment, or at the very end of the line where slower speeds will affect fewer passengers, but even then, close stop spacing should be used sparingly.

Again, LA’s Blue Line has it right. The stops in downtown LA on the Regional Connector are about half a mile apart, which is appropriate for an area that dense. Things open up to about three-quarter mile spacing from 7th/Flower to San Pedro, and then it’s stations about a mile apart on a beautiful alignment all the way to Long Beach. The only problem is between Imperial and Compton, which is over two miles, and should probably have an infill stop at El Segundo Blvd. The two-mile spacing from Artesia to Del Amo is acceptable because it’s just industrial land in between them. Same goes for Del Amo to Wardlow, where a stop in between would be in the middle of the LA River and only serve a golf course and the Blue Line yard. In downtown Long Beach, we go back to half-mile spacing from Willow to 5th, and the stops are very close on the loop. I could live without 1st and Pacific. However, since it’s at the end of the line, it’s not slowing down all that many riders, so it’s not a huge issue.

Expo Line is pretty much done right too. Stops are about three-quarter mile spaced from Pico to Jefferson/USC, and then more or less mile-spaced all the way to Santa Monica. The exceptions are Expo Park/USC (pointless, only 0.33 miles from Expo/Vermont, shouldn’t have been built) and Farmdale (which splits up the 1.16 miles between Expo/Crenshaw and Expo/La Brea, and wouldn’t have been built except for Damien Goodmon’s shenanigans). Other than that, maybe the Expo/Westwood stop should have been at Expo/Overland to better split up the distance between National/Palms and Expo/Sepulveda.

The Crenshaw/LAX Line is also well planned in this regard. With the addition of Florence/Hindry and Leimert Park, the entire 8.5 mile line will have 9 stations, all on spacing of between 0.7 and 1.3 miles.

The Gold Line to Pasadena goes through older parts of the region and hilly terrain, which doesn’t have the typical LA style half-mile arterial grid of roadways. This resulted in irregular station spacing between about half a mile and two miles, but the locations are logical (you could question the utility of Del Mar and suggest an infill station near Altadena Dr or San Gabriel Blvd). The Foothill Extension to Azusa is mostly two-mile spacing; you could argue for infill stations if the area gets denser. The station spacing for the Gold Line to East LA doesn’t make sense to me; I could do without Pico Aliso (only 0.3 miles from Mariachi Plaza), and Maravilla should have been at Eastern.

The western end of the Green Line, between Aviation and Redondo Beach has spacing a little under a mile, with Mariposa and El Segundo only half a mile apart. This section is reasonable due to the density of development in El Segundo and the development potential in the area. On the section in the median of the 105, there are several locations where the spacing is too large. There should be stations at Western and Bellflower, and a couple stations in the four miles between Lakewood and Long Beach. (Again, more on the Green Line soon.)

In contrast, the station spacing on TriMet is just terrible. For the most part, we can ignore the Red Line and Green Line, which are almost just spurs. Suffice to say that Cascades and Mt Hood Av on the Red Line are too close for the development in the area, and I’ve got to question the spacing of Division, Powell, and Holgate on the Green Line. The real offenders are the Blue Line and Yellow Line.

At the north end, the Yellow Line has two stations 0.70 miles apart, in an area where there is no development to speak of. The stops are then spaced every half mile through North Portland, despite the area being mostly SFR development. But the really crazy section is in downtown Portland, where there are seven stops, all at most a quarter of a mile apart. Stops that close together will inevitably cannibalize each other’s ridership, and the frequent stopping ensures low average operating speeds, which make transit less competitive.

The Blue Line is even worse. Nominally, the average spacing on the Blue Line is about 0.75 mile, but this is deceptive. Take out the four station spacings greater than 1.5 miles, and the average drops to 0.60 mile. Take out an additional four station spacings greater than 1.3 miles, and the average drops to 0.50 miles. That’s 40 stations in 20 miles of track, about the same length as LA’s Blue Line. In other words, Portland’s Blue Line has twice as many stations, for a city that’s not even half as dense.

The station spacing on the Blue Line drops to about a third of a mile in Hillsboro (4 stations), about half a mile in Beaverton (6 stations), about half a mile on the east side (4 stations), and less than half a mile in Gresham (3 stations). Hillsboro and Gresham aren’t terrible, since they are the end of the line, but the short spacing in the middle of the line is bad, because it drives up travel time. None of these places are dense enough to deserve stations so close together, and many of the stations were obviously built in an anticipatory plan.

But even those station spacings aren’t that bad compared to downtown Portland, where the Blue Line has 10 stations in less than 2 miles. Some stations aren’t even 600 feet apart, so close that an NYC subway train would straddle them. They’re equivalent to taking a train from one end of a subway platform to the other. Across the Steel Bridge, it’s the same thing in the Lloyd District, where there are four stations in 0.54 miles. This spacing is awful even by the lowly standards of US bus stop spacing. In effect, it makes transit almost useless for trips going through downtown and the Lloyd District because the time penalty is so high (see, for example, the previously linked Keep Houston Houston piece where the author describes using a bike to bypass the downtown light rail).

TriMet’s schedules suggest that the Blue Line averages about 6 or 7 mph in this area. That was acceptable in 1890, when traveling at 10 mph through an urban environment was revolutionary (and someone said, the dream of the 1890s is alive in Portland). But good god, in 2013, you need to be competitive with driving.

Grade Separation

Another one of the ways you stay competitive on travel time is grade separation, so that traffic doesn’t interfere. In other words, you have to have an aesthetic appreciation for concrete. This means going underground in very dense areas, and judicious use of overpasses and viaducts elsewhere so that your service stays fast and reliable. Where you’re at grade, you want your own ROW or at least an in-street reservation. Mixed-traffic running should be avoided.

LA’s LRT network does an excellent job of this. You get tunnels in downtown LA (including the upcoming Regional Connector), East LA, and a short one in Pasadena. You get grade separations of the major streets, so the Blue Line goes over Slauson, Firestone, and Del Amo, and the Expo Line has a bunch of grade separations so that it doesn’t get stuck in that famous Westside traffic. Crenshaw is going to be grade separated at the major streets. In fact, when people in LA complain, they don’t complain about the visual impacts of overpasses – they complain because you’re not grade separating enough.

At the same time, LA isn’t afraid to go at-grade to save money where conditions don’t warrant grade separation. So the Blue Line crosses many streets at-grade, and the Expo Line crosses Vermont, Normandie, Western, and Crenshaw – all major arterials – at-grade as well. But even when LA is running the line at-grade, it usually has its own ROW. In the few places where lines run in the street, there’s always a physical demarcation, like a curb, to keep cars out of the train’s space.

Portland does have some places where the lines are completely grade separated, where following freeways. The Blue Line, Red Line, and Green Line all have long sections that follow freeways and are grade separated. But in the town centers and downtown Portland, it’s all at grade. In many places, nothing separates the rail ROW from traffic other than striping or pavement textures, which allows cars to enter the train’s space and cause delays. This, combined with the close station spacing, increases travel time and decreases reliability.


In downtown Portland, there are also many places where the rail lines cross each other at grade. The Yellow/Green Lines cross the Blue/Red Lines at Pioneer Square, and all four lines merge to cross the Steel Bridge. The streetcar crosses both the Blue/Red Lines and the Yellow/Green Lines at other locations downtown, and crosses itself at-grade in several locations. In addition to affecting travel time and reliability, these decisions will constrain the ability to increase service in the future. Then again, with 15 minute headways on MAX and the streetcar, it’s not like demand is that high now.

Route Geometry

At the finest level of detail, you can hurt your transit line by making individual curves too sharp. These may only cost you seconds at any specific location, but over the whole line they can add up to minutes. (Amtrak, substandard rolling stock aside, spends millions to eliminate speed restrictions on the Northeast Corridor that only cost seconds.)

Yet again, LA does this right. On the Expo Line, you get a 35 mph underpass at Flower and Exposition that would otherwise be a very low-speed curve requiring a three-phase traffic light. The Blue Line has sharp curves at Washington and Grand, but other than that, it’s straight. The Green Line is a dream alignment. Crenshaw/LAX will have an underpass where it turns from the Harbor Subdivision onto Crenshaw Blvd, avoiding a very low-speed curve. The worst line is the Gold Line, and even most of it is pretty good. It has the aforementioned sharp curves near Union Station (again, up to you to decide if it’s worth it), and it’s got a nasty curve at Little Tokyo, but that will be less of an issue when Regional Connector is done. The curves at Indiana are bad, but they’re right at a station, which reduces their impact.

The decision to go at-grade in downtown Portland, along with the at-grade rail-rail crossings, has resulted in many sharp curves and special track work. In fact, Portland is probably a track engineer’s dream – where else would you get to work on so many special turnouts and skewed diamonds? But these alignments result in the need for trains to “hang a left” (or right) at intersections that were designed for cars, or at best, streetcars. This results in low-speed operations that cost your passengers time.

Further out from downtown Portland, there’s some other questionable geometry. For example, here’s the Red Line diverging from the Blue Line.


That curve has to be close to the absolute minimum the vehicle can negotiate. It’s 5-10 mph track the whole way, and it’s single track at that. Here’s the Blue/Red Lines at the junction of the 26 and the 217.


In my humble opinion, that type of geometry is just a giant eff you to riders. It basically says that we don’t care about your time enough to spend a little more money and give you a much better service. In the former case, it’s a stark contrast to the freeway, which gets high-speed semi-direct ramps in all directions. (At least at the 26 and the 217, drivers are getting crappy geometry too.)

You’re Doing It Wrong

If you keep all of these considerations in mind, it’s not hard to see why a system like VTA is struggling. The Blue Line is three-quarters of an ellipse, which means that for many of the trips that it could theoretically serve, you’d be much better off taking a more direct bus route. The tortured route of the Green Line to Mountain View, both overall and in local geometry, is never going to be competitive with driving or even with a direct transit service.

The station spacing, averaging about two-thirds of a mile over the network, is too close for the type of development found in Santa Clara County. The Green Line to Mountain View has closely spaced stops, which combined with sharp curves, ensures that this route is always going to offer low average speeds. In downtown San Jose, the stops spacing is like downtown Portland, and the Green Line has another section of terrible geometry between St James and Diridon. Honestly, I would love to know what possessed light rail planners nationwide and made them think that downtowns needed surface running LRT with streetcar-tight geometry and stop spacing only marginally bigger than local bus service. In an ironic twist, we probably have LADOT’s concern for auto traffic to thank for the fact that downtown LA escaped the same fate.

You could say that VTA would have been a much better comparison to LA-style LRT than Portland, but I wanted to use a system that’s widely known and respected. Pretty much everybody concedes that things have not gone as planned on VTA, so no one is going to show up and model their system after VTA. But Portland is a reasonably successful system that demonstrates the network that results from making different decisions on system characteristics.

Apologies to Portland

Again, I apologize if it seems like I’m trying to beat up on Portland. They’re coming at things from a different place than LA. Portland is building the rail network they want, and trying to coerce land use patterns to follow the rail. In North America, only Vancouver is probably doing a better job with that approach. LA is building a rail network in an existing dense city, with existing land use patterns that have proven to be more apt for rail transit than many expected.

In this regard, perhaps the real point here is that you can have a successful system without following the Portland model. Obviously, given this blog’s relentless promotion of LA-style density, I’m going to prefer the LA model. I’ll leave it to others with better knowledge of Portland to say how well their model is working for their goals. But if you’re looking to build or expand an LRT system, you should consider the LA model as well, and decide if it’s a better match for your goals.