Last month, Zef covered in detail TriMet’s proposed service changes to both NE and NW Portland. (He also had some additional thoughts on the cuts as a whole over at Portland Afoot, see also here). Zef is unfamiliar with Washington County, so we agreed that I would do the article about the proposed changes affecting Beaverton and Hillsboro instead.

The changes

Here is TriMet’s map of the proposed service changes; click on the thumbnail for a full-sized edition:

The major cuts can be summarized as follows:

• The 67-Jenkins/158th will no longer run to Beaverton Transit Center; instead it will end at the 158th/Merlo MAX station. Some stops on SW Jenkins will lose bus service. (I assume it will be renamed something else, such as the 67-158th/Bethany)
• The 89-Tanasbourne will cease to exist; its function will be taken over by the 48-Cornell and the 47-Baseline/Evergreen, which instead of ending at Willow Creek Transit Center, will continue to Sunset TC. The 48 will take the southern leg of the 89 (staying on Cornell) whereas the 47 will serve NW Bronson and NW Oak Hills. Both routes will run between downtown Hillsboro and Sunset TC, bypassing Willow Creek entirely. The combined 48 will have seven day service (as did the old 89; previously the 48 did not run on Sundays); the 47 will continue to have weekday-only service (in essence, the Bronson and Oak Hills deviations will lose weekend service).

The theme

If there’s a theme that links these proposed changes to the northeast changes, it could be the concept of building the grid. One of the best topologies for transit is the high-frequency grid, where a network of lines in a rectangular grid pattern serve in area in a fashion similar to blocks in a street grid. Assuming that parallel routes are adequately spaced (so that patrons can walk to either a N/S line or an E/W line, assuming the grid is aligned with the cardinal directions), it is theoretically possible to make any trip with only a single transfer.

In areas with sparse transit coverage, however, the best transit topology is the pulse or star topology. In this arrangement, transit lines run between a small set of transfer points (generally corresponding to “transit centers” in TriMet terminology), with timed transfers occurring at these nodes, and bus lines scheduled to converge upon the nodes at the same time. This is the best way to make low-frequency services which require transfers work. One drawback with the pulse network is it produces redundant service around the nodes, limiting the geographical area which can be covered with a certain number of service hours.

In NE Portland, the network is mostly a grid; but Rose Quarter TC contains a few pulse elements. In particular, the 73 and the 70 both deviate west to terminate at the TC; the new routing removes this deviation. As the combined 70 intersects with MAX, which runs at five-minute headways in this part of town, and there are plenty of other frequent-service lines in the inner city, this isn’t really a problem.

In Washington County, the TriMet network is more pulse-like. Other than MAX and the 57 (and unofficially, the 76/78 multiplex between Beaverton TC and Washington Square), none of the routes are frequent service. Most of the bus routes run between transit centers, and some circulator routes (such as the 53) serve the same TC at both ends. The transit centers of note in the Beaverton area are Sunset, Beaverton, Willow Creek, and Washington Square.

Both of the proposed service changes, on the surface, represent a migration away from a pulse network to a grid. In order to save service hours, deviations of bus lines from their nominal route to the nearest TC are being eliminated. Is this a good thing?

It depends.

The analysis

The good news is that the new 48 will be a far stronger line than either the old 48 or the 89. Cornell Road is–or ought to be–an excellent transit corridor; one that by rights should perform better than TV Highway. Among the destinations along the corridor are the county seat, two full-service hospitals (and a third to open next year), several Intel facilities, the county fairgrounds, Hillsboro Airport, the Orenco development, the Tanasbourne development, OHSU’s Washington County complex, numerous major retail destinations, a Krispy Kreme, the Cedar Mill neighborhood (and the Science Park industrial complex), and the Peterkort development. With a line providing seven-day service the length of the corridor, I predict that the line will prosper.

The route of the 47 is not a strong corridor; on the other hand. It’s new route will be rather serpentine. The western half (serving Evergreen Parkway, then dropping down at Orenco to Baseline) isn’t too bad, but east of Tanasbourne, the line is weak–redundant with the 48 other than a few deviations. I wonder if it might have been better to create a new northern line out of this and the 50.

The truncation of the 67, as it stands, is unfortunate. Unlike the 47/48/89 and 70/73 combos, where disconnected feeder routes were combined into stronger corridors, there isn’t a southern line to connect the 67 to–it will simply stop at the Merlo MAX station (a weak stop that gets frequent mention when the subject of closing MAX stations comes up). In the comments for the bus savings article, I suggested a way to extend the 67 down to Tigard, combining it with fragments of the 92 and 45; a simpler combination with just the 92 to Washington Square might be possible as well.

The caveats

There is one major caveat with all this grid-giddiness. In the enthusiastic discussion of grids above, the word “grid” was necessarily preceded by the qualifier “high frequency”. For grids to be an effective transit topology, they have to be frequent–30 minute or so headways at worst. (They also have to be available whenever the transit system is running–for TriMet, that means seven days a week). Timed connections in a grid are pretty much impossible, so the connectivity that makes grids work depends on frequency. A grid of busses that comes once an hour isn’t going to cut it. Unfortunately, if current service levels on the affected lines are an indication, the Washington County grid may leave something to be desired.

If TriMet’s adaptation of a grid topology represents a commitment to improved service frequencies when more operating funds becomes available; this is probably a good time as any to make the change. Nobody will complain if frequency is added in the future; but it’s hard to make these sorts of changes in good times. If, on the other hand, the changes are only about the appearance of quality transit, then not so much. Only time will tell.

If TriMet is committed to building a grid in Beaverton (along with Gresham, Beaverton is probably the most transit-friendly of cities not named Portland in the TriMet service district), there’s a few other changes it might consider:

• Rather than ending the 48 at Sunset TC, continue it along Barnes Road to downtown, replacing the 20. The 20-Burnside/Stark is one of the longest lines in the system, and it shortlines downtown anyway (many mid-day busses turn around at JELD-WEN Field rather than continuing over the West Hills); I think it would be useful to have the western terminus of the 20 be downtown, and have the 48 run between the Transit Mall and Hillsboro.
• Of course, if you do that, than you need a new line to serve Cedar Hills Boulevard. A new line, which leaves BTC heading north on Cedar Hills, serving the Cedar Hllls and Peterkort neighborhoods, and possibly heading west on Burton/Oak Hills or West Union (and possibly replacing the eastern half of the new 47) would be beneficial
• Another option for serving the Oak Hills neighborhood would be for the 62, rather than turning east on Cornell and heading to Sunset TC, to head west on Science Park Drive and then north on NW 143rd for a spell. Or, the 62 (which runs half-hourly) could simply turn around at Cedar Mill, and the service hours currently used to get between Sunset TC and Murray Blvd could be spread out over the entire Cornell corridor to improve frequency on the 48.
• One other suggestion: If TriMet ever finds the budget to upgrade the 59-Walker/Park Way to more than just peak hour service, connecting it to the western half of the 47 makes more sense.

Thoughts?