Monday, November 25, 2013

Could Baltimore Benefit from a Frequent Transit Grid?

As a follow-up to the recent post on frequency mapping, I thought it'd be worthwhile to discuss the concept of a frequent transit grid. The MTA is currently soliciting ideas for the Bus Network Improvement Project, and the most common suggestions seem to be (1) improving schedule adherence by reducing bus stops, boarding times, traffic delays, and bus bunching, (2) improving service frequency, and (3) reducing overcrowding.

Not only could a frequent transit grid address many of these issues, but I think it'd serve as a sorely-needed update to the current “radial” transit network. I hasten to add that my suggestions are presented from the perspective of a bus rider and amateur “transit geek,” so I encourage those with professional transit expertise to correct any errors or oversights.

A diagram of a typical radial transit network.
Baltimore's transit network is currently organized on a “radial” or “spider web” grid (see right). A radial network is primarily composed of two kinds of transit routes: radial routes that run directly from a city's downtown to its outlying neighborhoods, and crosstown routes that run around downtown in concentric circles to connect the various radial routes.

So what's wrong with radial networks? I've often heard comments along the lines of “Baltimore's transit network is actually very good... for 1950.” These comments reveal the underlying weakness of radial networks: many tend to provide good service to the core at the expense of good anywhere-to-anywhere service outside the core. Before WWII this arrangement made sense: at that time Baltimore's downtown and adjacent industrial waterfront were the dominant employment and commercial destinations in the region, so there was little need to provide extensive crosstown service between the neighborhoods outside this core.

For better or worse, this centralized pattern of urbanization has been considerably diluted since then: not only have numerous employment and commercial nodes sprouted outside Baltimore (BWI, Hunt Valley, Owings Mills, Towson, White Marsh, Woodlawn), but the same pattern has unfolded within the city (Canton Crossing, Harbor East, Johns Hopkins Hospital, Montgomery Park, State Center, Tide Point, and eventually perhaps even Harbor Point, Port Covington, and Westport). While all these destinations are already served by metro, light rail, and/or bus lines, almost all these lines converge on downtown, requiring anyone traveling from one outlying area to another to pass through downtown.

As Jarrett Walker noted when analyzing my frequent transit map, only one crosstown route – the 13 bus – currently meets the frequent transit threshold. So if you live, work, and shop outside downtown Baltimore, it's easier to drive to various destinations than to wait for a spartan, infrequent crosstown bus. Furthermore, downtown Baltimore itself is slowly transitioning into a mixed-use neighborhood, so the transit network increasingly needs to accommodate “reverse commuting” as well.

I think a frequent transit grid would be a good solution to this predicament. Since Baltimore's urban fabric has shifted from the “downtown is the center of everything” pattern to the “diffuse nodes” pattern, so too could its transit network shift from the “spider web” pattern to a diffuse grid that accommodated anywhere-to-anywhere travel.

A diagram of a typical frequent transit grid.
Jarrett describes the geometry of frequent transit grids better than I ever could, but the concept is fairly straightforward: as its name implies, a frequent transit grid is composed of multiple parallel north-south and east-west routes that intersect to form a grid.

Of course, most frequent transit grids are not nearly as orthogonal as the diagram on the left suggests; usually geographic obstacles induce numerous deviations. But even in these cases, the underlying grid structure remains and is very useful:

“In an ideal grid system everyone is within walking distance of one north-south line and one east-west line. So you can get from anywhere to anywhere with one connection while following a reasonably direct L-shaped path. For this trip to be attractive, the lines have to be frequent so you don't have to wait long for the connection.”

That last point is important: transit grids require easy transfers which in turn require frequent service. Many bus riders express an aversion to transfers, which is understandable considering that transfers are often timed to connect routes with relatively spartan service: if your first bus is late (a common occurrence), you'll miss your connecting bus and consequently have to wait a long time for the next one. But in a frequent transit grid you don't have to rely on timed connections; if you miss one connecting bus, another one will come along shortly.

So how do you create a frequent transit grid? In most cases, existing radial and crosstown routes are realigned to form new north-south and east-west routes. Jarrett describes how this process occurred in Portland in such detail that it's worth interrupting this post to recommend reading his post first!

One possible configuration for a frequent transit grid; click for full PDF.
In the above post Jarrett points out, “The old network was wasteful, as many overlapping lines converged on downtown. The new network was efficient, with little overlap between lines, and with lines spaced further apart. This is how the resources were found to increase frequency so much.”

I think a similar strategy could work in Baltimore: MTA and Charm City Circulator resources could be pooled and bus routes realigned to eliminate redundancies. Bus routes can't be altered on a whim – there's quite understandably a lengthy public process – but establishing a grid of frequent buses would still probably be much easier, faster, and cheaper than dedicating considerable resources to far-off metro, light rail, or even streetcar expansions. These expansions all are worthwhile, but the city can't afford to wait decades for them.

So I proposed the following changes to existing MTA and CCC bus routes. As with my frequent transit map, I decided to focus only on central Baltimore to keep the project manageable, but the grid realignment I proposed for this area could just as easily be applied to bus routes throughout the region. Furthermore, the map's intention is not to propose a finalized grid but merely to suggest one possible grid iteration out of many.

The distinctive CCC brand, fleet, and fare structure (free!) would be retained, but the service would be integrated into MTA service to eliminate redundancies. CCC routes would be restructured to accommodate several long-distance feeder routes by serving as central city trunk lines.

To ease transfers and minimize overcrowding – they would, after all, be replacing numerous overlapping local bus routes – CCC routes would offer daily 5-minute (or less) service and 30-minute overnight service. The C1 (currently known as the Purple Route) could be replaced with the Charles Street streetcar in the future, but the streetcar would still function as a free circulator.


Local bus routes would be realigned, split, and/or combined into a grid of routes wherever necessary; many would also feed into transfer hubs. Most of the segments within this grid would offer daily 15-minute (or less) service and 60-minute overnight service. Riders could therefore expect uniform service on all gridded routes regardless of route number. Outside the grid, the frequencies on long-distance local buses would be adjusted to demand as necessary. To facilitate the most evenly-distributed grid possible, several minor one-way street segments might need to be converted to two-way traffic.

Quick buses would be restructured to serve as long-distance extensions of the CCC trunk routes by connecting central Baltimore to regional nodes like Annapolis, Columbia, White Marsh, and Woodlawn (though, unlike the CCC, they wouldn't be free).

Quick buses would continue to offer daily 15-minute service but would also offer 60-minute overnight service. They would retain limited-stop service as well, but the stops would be adjusted to coincide with the intersections on the transit grid (i.e. they'd skip the local stops in between).

At night, certain quick bus routes could deviate from their daytime routing to serve important metro and light rail destinations. The Q1, for example, could run through BWI at night to offer service to downtown Baltimore when the light rail isn't operating, thus replacing Route 17's overnight service to downtown Baltimore.

Commuter buses would be restructured to function more like commuter rail lines by feeding into two transfer hubs: one at Lexington Market with connections to the metro and light rail, and a second at Camden Station with connections to the light rail, MARC, and a new intercity bus terminal (more on this below). Routing to Charles Center, State Center, and Johns Hopkins Hospital would be discontinued since commuters would be able to transfer from the aforementioned hubs to reach these points.

Independent express buses (i.e. Routes 104, 120, 150, and 160) would be discontinued and folded into quick bus service. However, the express buses that supplement local buses would be retained and adjusted as needed. Each supplemental express would retain its branding as a limited-stop, 15-minute, peak hours version of the corresponding local bus (i.e. “19X is a faster version of 19”). Supplemental expresses would be spun off as quick bus routes only if demand warranted full-time express service.

Several other redundant routes would be discontinued, freeing up resources for new routes:
  • The CCC's Banner and Green Routes would be folded into various other routes.
  • Route 12 would be truncated to a transfer hub at Towson Town Center and the remaining segment from Stella Maris to Towson would be converted into a neighborhood shuttle (S5) with improved service frequency.
  • Route 14 would be replaced by the Q1 quick bus; the proposed routing outside central Baltimore would be along Hanover Street to Ritchie Highway, after which the Q1 would follow the former Route 14 routing to Glen Burnie and Annapolis. 
  • Route 30 would be folded into Route 20; the combined route would offer frequent service from West Baltimore to Woodlawn.
  • Quick Buses 46 and 47, which ultimately couldn't support full-time express service, would be demoted back to supplemental expresses (5X and 10X for QB46, 15X and 35X for QB47). 
  • Express Bus 104 would be folded into Route 36 as a supplemental express (36X). 
  • Express Bus 120 would be replaced by the Q2 quick bus; the proposed routing outside central Baltimore would be along Monument Street, Pulaski Highway, I-95, and White Marsh Boulevard.
  • Express Bus 150 would be replaced by the Q3 quick bus; the proposed routing outside central Baltimore would be along Baltimore Street, Hilton Parkway, Frederick Avenue, US 29, and Little Patuxent Parkway (thus serving Catonsville and Ellicott City on the way to Columbia). 
  • Express Bus 160 would also be replaced by the Q3 quick bus; the proposed routing outside central Baltimore would be along Lombard Street, Bayview Boulevard, Eastern Avenue, and Eastern Boulevard.
  • Commuter Bus 310 would be folded into the K1 (formerly 320) commuter bus as an improved extension from the Snowden River Park and Ride to Columbia Mall. Commuter Buses 410 and 411 would be combined into a single K2 commuter bus serving various points in the Bel Air area.

To facilitate easy transfers both on the fringes of the grid and within it, I proposed a series of transfer hubs that would be built in a manner similar to the existing Mondawmin bus/metro hub. The number of bus bays per hub would, of course, vary according to the anticipated number of connecting routes.

  • In many instances, as at Bayview Medical Center, Canton Crossing, Lexington Market, Port Covington, State Center, and West Baltimore, transfer hubs could simply be built into existing parking lots and incorporated into future TOD. In other instances, as at Orangeville, Walbrook Junction, and Westport, transfer hubs could be built atop vacant land and likewise incorporated into future TOD. Certain hubs could even connect to future Red and Purple Line stations.
  • In instances where sufficient open land might not be available, as at Johns Hopkins University and Hospital, transfer hubs could be built into the parking lanes of existing streets, such as 33rd Street and Broadway. At Park Circle, a transfer hub could be built into a reconstructed roundabout.
  • At Camden Station, a bus terminal could be built above the parking lot and MARC trackage immediately to the east of the B&O warehouse. The terminal could accommodate both intercity buses (Boltbus, Greyhound, Megabus, Peter Pan) and MTA commuter buses; it could even contain dedicated bus onramps/offramps to/from I-395. I still think the parking lot north of Penn Station and/or the airspace above the JFX/NEC would have been the best location for an intercity bus terminal, but given the local opposition to that proposal, Camden Station might be a good compromise: it's sufficiently buffered from residential neighborhoods by Camden Yards, the Convention Center, and the Federal Reserve branch (not without ill effect!), but at the same time it offers excellent access to downtown Baltimore. To maintain a decent connection to Penn Station (Amtrak), the frequency of the light rail's Penn Station/Camden Station shuttle could be improved.

These proposed changes – particularly those to local buses – seem unsettling at first glance. To expand upon just one proposed change, a restructured 27 bus would no longer give Hampdenites a direct route downtown. But the current 27 bus only offers a 40-minute off-peak frequency and is notoriously unreliable. If the route was realigned along the lines proposed on the map, Hampdenites would have to transfer to the C1 circulator to go downtown. However, since the new 27 and the C1 would offer much better service frequency (15 minutes and 5 minutes respectively), a trip downtown with a transfer at Old Goucher would ultimately be faster and more convenient than waiting around for the current 27 bus! The other proposed route realignments would likewise offer similar advantages over current radial routes.

A snapshot of Philadelphia's transit grid.
Some might also argue that Portland's transit needs are different from Baltimore's, and that a grid network that worked well in one urban context might not work well in another. But this argument, as well as the accompanying argument that frequent transit grids facilitate crime, can be dispelled by analyzing Philadelphia's frequent transit grid (official map available here).

Philadelphia closely resembles Baltimore economically, demographically, and physically, yet the frequent transit grid has become an integral part of that city's transit system. Moreover, this transit grid connects neighborhoods of widely-varying demographics, and there's no evidence that this connectivity has increased crime – in fact, Philadelphia does slightly better than Baltimore in most categories of crime. This suggests that a frequent transit grid might very well be viable in Baltimore, and that it wouldn't necessarily increase crime.

To address several other common suggestions on the BNIP website, I proposed supplementing the frequent transit grid with the following changes:

  • The light rail would finally be given signal priority on Howard Street. Numerous suggestions for grade-separating this segment of the light rail – such as elevating it above Howard Street or repurposing the Howard Street Tunnel once (if) CSX relocates to a new double-stack-friendly tunnel – have been floated, but the likelihood of these solutions ever becoming reality is so slim that, in my opinion, signal prioritization is the only practical near-term solution.
  • The light rail's Sunday operating hours would be extended to match the metro's hours (6am to 12am), upon which Route 14's Sunday morning/evening service to downtown Baltimore would be discontinued.
  • The Yellow Line's peak service to Timonium would be extended to Hunt Valley.
  • To improve Hampden's access to the light rail, I proposed an infill station at the recently-renovated Mill No. 1. The mill's disused freight platform could be converted into a light rail station with a pedestrian bridge across the Jones Falls to Falls Road and Chestnut Avenue. The station could even contain a pedestrian walkway/stairway to the Wyman Park Drive bridge to connect to Druid Hill Park.

  • Weekend service is finally being introduced on the Penn Line, but I think weekday service on both the Penn and Camden Lines should be further improved.
  • The Penn Line's Perryville service would be extended to Wilmington to connect to SEPTA's Regional Rail.

Perhaps the most-discussed topic on the BNIP website is what the “ideal” spacing between bus stops should be. (On most bus routes, stops are generally spaced a block apart.) As the MTA noted, stop elimination requires the careful evaluation of trade-offs.

I agree that the number of bus stops should be reduced, but perhaps not as drastically as some advocate. I think eliminating every other stop or two (i.e. spacing stops two to three blocks apart) would be sufficient, and that resources from the closed stops could be used to improve the remaining stops. Otherwise, I proposed reconfiguring bus stops as follows:

  • To serve the proposed frequent transit grid effectively, bus stops would be hierarchically organized into three classes: transfer hubs, intermediate stops (intersections on the transit grid), and local stops (stops in between the intersections; not shown on the map). All buses would stop at transfer hubs and at intermediate stops, but quick buses, commuter buses, and supplemental express buses would skip the local stops in between.
  • Eventually each stop would contain a shelter, system map, and comprehensive information placard. This would admittedly be a difficult feat to pull off all at once, so I anticipate introducing these amenities incrementally: they would be added to transfer hubs first, then to the busiest intermediate stops, then to the remaining intermediate stops, then eventually to as many local stops as possible.
  • I actually don't think bus trackers would be necessary at most stops, at least not at those within the frequent transit grid. They could be added to stops serving less-frequent routes first, where they'd be more useful.

Most bus riders have experienced the aggravating phenomenon of bus bunching: you're waiting for a late bus, and suddenly two or more buses on the same route show up! The technical explanation of bus bunching is beyond the scope of this post, but the Walking Bostonian offers a thorough summary here.

I proposed reducing bunching by (1) reducing the number of stops in the manner discussed above, (2) adding curb extensions to most stops, and (3) introducing layovers at strategic stops to stabilize uneven headways.

An illustration of a typical proposed intermediate stop.
As depicted in the illustration on the right, I think the MTA, CCC, and city DOT could cooperate to build a series of bus/crosswalk bulbs (aka curb extensions) at bus stops. These would be introduced in the manner discussed earlier: transfer hubs first, then intermediate stops, then local stops.

Dedicated bus lanes would certainly speed up buses, but on many streets there'd probably be considerable spatial and political obstacles to their implementation. Curb extensions are a good compromise with numerous benefits: (1) they reduce pedestrians' crossing distance at intersections, (2) they improve access for the elderly and disabled, (3) they provide additional waiting space for passengers, (4) they increase curbside parking space by eliminating “transition zones” (the clearance space a bus needs to pull out of the travel lane and into the bus stop), and most importantly, (5) they eliminate delays caused by buses repeatedly merging back into the travel lane.

Regular bus riders will note that, on many routes, buses are already stopping in the travel lane and treating stops as if they had curb extensions. On busier routes bus drivers are under intense pressure to do everything possible to adhere to the schedules, so sometimes they're compelled to stop in the travel lane. In other cases, bus stops don't even have the aforementioned transition zones, so they already function with informal curb extensions. In still other cases, the transition zones – or even the stops themselves – are blocked by illegally-parked cars, likewise forcing buses to stop in the travel lane. So in all these cases, adding curb extensions merely facilitates an already-common boarding practice.

I also proposed a system of strategic layovers at designated stops: buses running ahead would pause at these stops to balance uneven headways. CCC riders will note that this practice is already common on that system, though its implementation could probably be improved. Again, the Walking Bostonian discusses this practice in greater detail, but also notes the practice is incompatible with routes running on schedules.

All-door boarding – which requires a proof-of-payment fare system akin to the one used on the light rail – could also reduce bunching.

I hope this map spurs a discussion on the viability of a frequent transit grid in Baltimore! Next time we'll return to the discussion on border vacuums by examining those around hospital and university campuses, so stay tuned!

- Marc Szarkowski


  1. I generally agree with this. I'm not sure radials are a problem per se, especially for people coming from the corners that may be the best option. But I think you are right in that more balance between the radials and N/S E/W routes would improve mobility. Neighborhood to neighborhood transit is always trickier. I think the answer that works best for me are some circular (or semicircular) routes (potentially using one of the tunnels to cross the bay) starting beyond downtown that would provide for maximum transfer possibilities.

  2. Yeah, I wouldn't argue that radial routes are inherently "bad" - there are many instances in which they work fine and are quite necessary - but it does seem like they increasingly need to be supplemented by useful crosstown routes.

    Unless I'm mistaken, the only crosstown route in Baltimore that currently meets the "frequent service" threshold is the 13 bus. Several other crosstown routes offer frequent peak service, but then cut back to 20 minutes midday and even less frequent service on evenings and weekends. These routes don't really offer convenient transfers outside rush hour, which means they're not as useful for non-commuting trips as they potentially could be. So a good first step might be to simply boost service frequency on the crosstown routes that have the best potential to form a polar grid.

    I suppose this gets to a chicken-egg scenario, though: Are the off-peak frequencies on these crosstown routes poor because there is little demand, or is there little demand because the service frequencies are poor?

  3. My feelings have always been that the best solution is a hybrid between a grid and radial pattern. A reorganized high frequency grid-like bus route pattern would serve Baltimore's "urban core" very well for trip within 3-5 miles of downtown. This is the area roughly developed up until WWII. It is largely an interconnected street grid anyway. routes within this area would operate much like streetcars. Longer distance routes, however, between 5 and 13 miles that extend beyond the core into inner and outer suburban areas, mostly developed since WWII would still best be served by radial lines which would operate like regional rapid transit lines, designed with fewer intermediate stops and integrated with urban routes to provide easy transfers. This is how most European cities have designed their modern transit networks and it works beautifully.

    1. I agree; this is essentially what I was aiming for in my proposal since a "pure" grid probably wouldn't be feasible. It seems the current network is already moving in this direction, albeit in sluggish fits and starts - the quick buses are a good example of long-distance, limited-stop radial routes designed to supplement local service.


Related Posts Plugin for WordPress, Blogger...