Here is a little deep dive into how bus network designers think, with an example from very close to (my) home.
Over the past year, our firm has been working with the Portland area transit agency, TriMet, on the “Forward Together” project, which developed a plan for the next several years of bus service improvements. (The plan will eventually represent about 10% more service than the agency ran in 2019, but TriMet is still recovering from post-Covid workforce shortages, so it can’t do everything at once.) The plan went through major public outreach and was adopted earlier this year, and a few items have already been implemented. Now, (August 27, 2023) they are rolling out what will probably be the most controversial part of the package, a series of big changes in Portland’s southwest hills.
The Forward Together plan was motivated by twin goals of ridership and equity. In general, the plan retains and expands services that either:
- … will attract high ridership, because they provide useful service to lots of people, OR
- … focus on areas with low-income and/or minority populations where the need for transit is greatest.
The plan includes a major expansion of the Frequent Network in high demand areas, and new local routes in underserved suburban areas with large low-income populations. Where TriMet was running services that are justified neither by ridership nor by equity, the plan reduces or even eliminates those services. That’s because even though the total service budget is growing, TriMet is also trying to align its services with current goals, rather than just continuing to run services designed around the goals of the past. I wouldn’t recommend this in every city, but this is the answer to the specific problem posed by TriMet’s limited resources and the goals they have adopted.
Let me give you a quick tour of SW Portland, using TriMet’s map of the old (2022) network.
TriMet network in 2022, before the changes. Green number shields and wider lines denote high frequency. Dashed lines run only briefly (mostly rush hour only). Map by TriMet.
Downtown Portland is in the NE corner of the map, where all of the colorful light rail lines converge. Immediately west of downtown is a bank of steep hills, running north-south. Just southwest of downtown, where you see Line 8 ending in a loop, is a huge hilltop complex of medical destinations, including the Veterans Hospital and the old part of the Oregon Health and Science University campus. This area, called Marquam Hill, is one of Portland’s biggest transit destinations outside of downtown.
Along the west and south edges of the map are the inner-ring suburban cities of Beaverton, Tigard, and Lake Oswego along with the Washington Square Mall. That ring, linked by orbital lines 76 and 78, is typical older suburban fabric with some dense centers but heavy car orientation.
TriMet’s mapping style makes all the bus routes blue but distinguishes the Frequent Service Network (every 15 minutes all day) with a slightly wider line and a green number sheld. You can see that there are two big frequent corridors in SW Portland: 54 along Beaverton-Hillsdale Hwy and, further south, the 12 along Barbur Blvd. Except for those two corridors, inner SW Portland (excluding the inner-ring suburbs) is mostly low-density and relatively affluent.
With a focus on ridership and equity, and the drop in downtown commuting due to ongoing work-from-home, this area just had to be rethought. Here is the new network as of August 27, 2023:
New TriMet network, same color scheme as the map above. Map by TriMet.
As I mentioned, the plan does delete some services that meet neither ridership nor equity goals, mostly all-day local routes in affluent low-density areas that had little ridership except at school times. Some of these are reduced to school-hour trips only, and a few, such as the 50 in the NW corner of the map, are dropped entirely.
In addition, this redesign seeks ridership and equity with two other big moves:
- Shift in focus from peak to all day.
- Shift in focus from office commute destinations to destinations of diverse workers and visitors.
The old network also had a number of services geared to bringing office workers downtown. As in most US cities, there are fewer of these commuters than there were, and like many agencies, TriMet has heard that it needed to provide better service to a more diverse audience, not just office workers but also everyone else traveling all the time for all kinds of purposes. So some specialized downtown-oriented services were removed. In the old network, for example, Line 94 in the southwest corner of the map approached from suburbs beyond Tigard and went all the way into downtown duplicating Line 12. That was no longer justifiable, so in the new network those riders will need to use Line 12 to complete their trips.
But the biggest and most complicated reworking, encompassing all of these issues, happened at Marquam Hill. At Portland’s largest medical destination, the network had become too focused on rush hour, even though medical workers and patients come and go all day.
The hill has long had Line 8, a Frequent Service bus to downtown, and it also has a cute aerial tram to another medical campus nearby. It also had a big network of rush hour express routes from all directions, numbered in the 60s. These routes duplicated all-day services and mostly helped people avoid transferring. That can be justified only if the buses were full, and they no longer were. They ran only at rush hour, which meant they were not useful to many of the lower-income people working shifts all around the clock at the hospitals.
We needed to rework this to remove some of the duplicative rush hour service, and to provide a more useful all-day pattern of service from the south. In the old network, you could get to Marquam Hill from the south only at rush hour. To do this, we revised two of the lower-ridership southwest lines (43 and 56) so that instead of going downtown, they’ll go to Marquam Hill. Where they cross Line 54 in the Hillsdale district, just south of Marquam Hill, people coming off of outer part of these lines will need to transfer if they are going downtown, but people from all over the southwest will be able to use these lines to get to Marquam Hill much more directly.
This is an inconvenience to some existing downtown riders, so we selected routes where that would affect as few people as possible. We chose Line 56 for this role because much of this line duplicates the Frequent Line 54, which is being upgraded to be Frequent all the way to Beaverton. The segment along Beaverton Hillsdale Highway served by both 54 and 56 is where most of Line 56’s ridership is. Only riders from the Scholls Ferry Road segment of 56 will need to transfer to reach downtown, and this is not a high-demand area. Meanwhile, the segment of Beavertion-Hillsdale Highway where 54 and 56 run together has many low-income apartment areas where people depend on transit. Instead of giving them duplicative routes to downtown, the new arrangement gives them direct Frequent service to downtown (54) and a direct bus to Marquam Hill. There’s no avoiding some duplication here, but this change makes it less wasteful and more useful.
As for Line 43, this issue is rich with memories, as I lived near the intersection of Terwilliger & Taylor’s Ferry as a teenager. I was at TriMet as a teenage intern when the current shape of the line was designed, and while I used it I knew it would always be a poor performer. Much of the line is close to the more frequent Line 12 on Barbur, and if you are west of where the two lines cross, the 43 is so slow to downtown that it makes sense to transfer to the 12 for a faster trip. That’s one reason why we felt that only a small area (between Barbur Blvd. Transit Center and Macadam Ave) would be affected if we reoriented this route to Marquam Hill, an area that had neither high ridership nor a high priority on equity grounds. (It does have one business district, which we served another way; read on.)
Just south of there, the plan removes Line 39 from Lewis & Clark College, prompting the biggest outcry in the public outreach. Lewis & Clark is a liberal arts college located on what is effectively a cul-de-sac for transit.[1] It’s not on the way to any other destinations, so it has to justify its service all by itself. Ridership was poor, and the campus was running its own shuttles to downtown. The final design here divides Line 35, which runs along Macadam Avenue just west of the river, and sends half the service via Terwilliger & Taylor’s Ferry — not directly to Lewis & Clark’s front door anymore, but within a short walk. This change restored direct all-day downtown service from the business district at Terwilliger & Taylor’s Ferry, which was losing it with the removal of Line 43. I’m normally very resistant to this kind of splitting of service because it reduces frequency. Here, however, it was the right solution because the segment of Riverside Drive that loses half of its Line 35 service is very low-density, very affluent, and has almost no ridership.
There’s a nice side effect of this change: new all-day service to the main gate of Tryon Creek State Natural Area, one of Portland’s most spectacular wilderness parks. A great new opportunity for people to get out into nature on transit.
You can explore the whole plan, and the justification for each part of it, in our Forward Together final report, which is online. There you’ll find both an explanation of each change, and also a description of the how the plan was revised in response to public outreach. Not everyone will like these changes, but they will help TriMet create a more efficient network that’s useful to more people, and more people who really need it.
[1] The neighborhood beyond Lewis & Clark, called Dunthorpe, is very, very low-density and affluent. Before the big redesigns of 1979-82 it had its own bus route from downtown, evidence of how completely oriented toward coverage (rather than ridership) the system was in those days. Today, nobody would think of running buses into Dunthorpe, probably least of all the fortunate folks who live there.
It looks like in the old network, the 54 only provided frequent service combined with the 56, meaning it was only frequent as far as Scholls Ferry. The new network makes the 54 frequent all the way to Beaverton, and I would imagine it was always more popular, but your description of the 56 as though it was a wasteful, duplicative route glosses over how it was the reason that corridor had frequent service to begin with.
You also make it sound like you could have justifiably sent the 35 down Taylors Ferry and Terwilliger full-time and provided Riverside Drive with part-time service only. I’m guessing the fear was that a bunch of wealthy people with too much time on their hands would put up an even bigger s**tshow than the removal of the 39 was creating.
One other thing: it looks to me like Terwilliger is largely isolated from anything else from Palater/the turn-off to Lewis and Clark to where it meets Riverside, while there is *nothing* along Macadam/Riverside from a little south of Taylors Ferry to where it reaches Dunthorpe, so I could make the argument that you could send the 35 through Dunthorpe and lose very little coverage compared to the split routing. Problem is that it would take longer and might be too busy for the affluent people along the route to put up with.
That segment of Terwilliger is the only way to get a reasonable travel time to Lake Oswego, as all of the roads thru Dunthorpe are very very slow. Our routing also hits a big destination of civic importance: Tryon Creek park. While we don’t expect big ridersihp to it, access to nature by transit is a valuable thing.
No one thought about the seniors living in that neighborhood served by the 39 did they? They can just learn the ropes of Uber and Lyft. At 80 urs old it should be as easy as mail on voting and gender rights…. Come on!
I haven’t looked at the recent ridership stats, but in old times the 61 was the most productive route when measured by passenger miles per revenue mile (as in how full a bus is on average). I’m guessing many rode all the way from/to the other end. Of course there was a lot of deadheading involved, and it cherry picked the peak times.
Also, canceling the 94 and 96 is really going to increase some people’s travel time. They had both been turned into all-day services. I can maybe see going back to peak only and longer headways, but cancelling them completely seems harsh.
And the 43 (at least the Corbett portion) went where it did because that’s where the streetcar went. But it does indeed duplicate the 35.
You have never run a bus down terwilliger…now only one stop…for homeless people to camp at tryon creek…worried abou t more crime and homeless.. more noise pollution…more exhaust pollution..good for you but not helpful for residents…this road has had no maintenence for the 40 years i lived here
About OHSU, I hope the scheduling department knew to interline the 43/56 to the 8.
I remember pre-tram that layover space was limited (reminds me of the Beacon Hill VA loop).
I hope that SMART has a way to serve Willamette townsite after the 154 goes away.
We debated making 43 and 56 into branches of 8. It’s awkward because the 8 ends in a loop, so you can’t extend it while keeping two way service to all the stops. It also creates a legibility/numbering challenge, but I wouldn’t rule it out in the future.
Exactly what is the problem here? Just run along the 8’s loop in both directions. For a detailed view of the area, I recommend Trimet’s start page, this is one of the few “interactive” maps that are actually useful. There are no stops along the eastern part of the loop so you don’t miss any stops there. The only stops you miss are the two stops that only 43 and 56 serve but they are very close to the 8.
For the numbering issue, I think you should make the 8 a “spine” and name the branches 8A and 8B, like you did in Madison for example.
Unfortunately, 45 is not a frequent bus and only comes every hour and I am unable to connect with 44 or 43 and get to work on time. I don’t drive. At least I have my husband to drive me to the bus stop to catch 56.
One thing transit planners/agencies seem to forget/ignore is station amenities at transit centres. The Beaverton TC in the 1990s had a little shop that sold snacks and reading material. Last time I was there, it was closed and the original space was unused. Things like this make the commuting experience way more pleasant. The 4th and Townsend CalTrain Station in San Francisco even sells canned cocktails (yes, with real alcohol) during the PM rush hour. Now, maybe that’s not “appropriate” for local buses/light rail, but it’s an example of making the mass transit experience more pleasant. Unfortunately, the commuting white-collar worker is a slowly dying breed, but pleasantries like this would really do a lot to “sell” transit to more people. Portland and Seattle are coffee towns: maybe coffee carts could be encouraged to take up shop at their transit centers from about 6 to 9 AM and maybe snack vendors from about 3 to 6:30 PM. Also, the “no food or drink on buses/light rail” rule should be modified to “no MESSY or SMELLY food or drink”. This would enhance transit’s time advantage, especially in the AM peak. Boston and New York commuters would never tolerate a “no food or drink” rule….
Totally agree on the food and drink comments. Here in the UK eating and drinking is largely allowed / tolerated on train and bus. Alcohol was barred in London a few years ago but other than that one can generally make good use of one’s time in eating and drinking. It’s surprising how annoying / disruptive a ban is on one’s journey, as I discovered one day when about to board a train in Brisbane having just innocently purchased a coffee
John Charles. Transit agencies can create these business opportunities, but they have to work as businesses and sometimes they don’t. Agencies also have mixed feelings about selling food/drink at stations because (unlike Caltrain) they don’t want to encourage food/drink on board.
The loss of downtown express service on the 94 is pretty bad for anyone traveling NE/SW to/from downtown, but will be felt most acutely by people traveling to/from places SW of Tigard. The 94 was about 10 minutes faster than the 12 from Tigard to Portland, and factoring in transfer times it’s approaching 20 minutes longer for people who need to transfer to the 12. That’s a big loss, and while I understand it I’m still sad about it. It was also the bus I took to work (from Portland to Tigard).
Concerning the 43 – removing it on Corbett is honestly pretty frustrating as well. Yes, Barbur is close as the crow flies, but there are no good places to cross Naito between Hamilton and the Ross Island Bridge landing. People who live in that area of Lair Hill will have functionally no real transit access within a short walk. This is also made much worse by the recent rerouting of the 19 over the Tilikum. Yes, it’s a relatively small area but it’s reasonably dense and has a lot of interesting history. Plus, everytime a bus is removed from a historic streetcar routing we lose a bit of our past. And it sort of duplicates the 35, but I’d argue that could have been remedied by a routing over the Sellwood Bridge (to replace the soon-to-be dead 99) – which would maybe be an entirely different line. I do like the SW section of the new line.
And lastly, the 35. Is there any reason to have it on Riverside Drive at all? Other than travel time between Lake O and Downtown, the Terwilliger-Taylors Ferry segment is superior in every way. Especially considering that TriMet is extremely bad at serving our regions extremely wonderful park system, access to Tryon Creek is a slam dunk. Have you ever tried to do a weekend hike in Forest Park via public transit? It’s pretty frustrating (made worse by the lack of service in NW Portland – especially Thurman).
However, the 56 going to Marquam is definitely a huge boon. Access to the hospital complex from the west is extremely lacking (just take one look at the number of people getting onto 26 at Broadway during rush hour from Terwilliger!), and hopefully it helps make transit a more viable option for more people.
And lastly, the 51. I really wish there were more service to Council Crest. I realize that it’s not a big equity area, or even a huge ridership draw, but Council Crest and the history of Portland transit are genuinely important. A bus from NW to OHSU via Council Crest (well close to it at least) could fill two big gaps at once, it would provide a faster trip than currently can be managed from 23rd-ish to OHSU (streetcar to the #8 is a bit funky) while also providing access to yet another great natural space (and one with a rich transit history to boot!)
Andrew. I don’t really disagree with any of these comments. On 43 we looked a lot at that one gap that is opened up at the north end of Corbett, but decided to tolerate it for lack of a good alternative. I think it will be interesting to watch the 35 evolve. Ridership on Riverside Dr is very low but not totally zero, so there are impacts to removing it. I think the bigger issue is the travel time between Oregon City, Lake Oswego and Portland, which will definitely be longer. (TriMet made a guess in its schedules but they’ll refine that based on experience.) Remember that the 35 is the direct link to downtown Portland for Oregon City, not just Lake Oswego, so that’s a consideration.
Any chance of putting a separate number on the two branches of the 35? It’d make it a bit easier to use.
This type of thing nailed me once in Bellingham as they have at least one route that ends at the Amtrak station only on Sunday, but all other days ends in a completely different area.
We debated this. Giving two numbers makes the whole line more confusing for everyone, so it makes sense to do it only when the two variants of a line are more different than similar. The fewer people are affected by a variant, the less prominent it should be in the numbering.
I appreciate the justification behind the changes – this helps very much in understanding the thought processes behind some of the decisions. I have to say, though, that even though I live in SW Portland, and work at OHSU, I cannot take advantage of the change to direct service with line 43. There is NO safe way to go on foot from my neighborhood to any of the stops on the 43 line. Until foot traffic becomes safe, all these upgrades service little purpose.
Teresa. That’s a problem throughout SW Portland and one of the reasons the area is not a better market for TriMet. Pedestrian infrastructure is the City’s responsibility and their backlog of sidewalk needs is truly overwhelming.
Thanks for blogging about this! I intend to link to this on Nextdoor, where there are a few SW transit users. :^) One anecdote – Last evening, on a lark, I boarded #43 in Hillsdale at Ida B Wells westbound thinking I could ride 43 up to OHSU, and then continue along its scenic route, which would take me back through Hillsdale, Bertha, Barbur, over the Terwilliger bridge, and I would get off at Taylors Ferry near where I live. Little did I know that at OHSU the 43 morphed into 56.. I didn’t hear any announcements as I waited for the bus to continue on its way. Long story shortened – after the driver didn’t turn at Bertha, I panicked and bailed at Shattuck, and backtracked. I won’t Digress. TriMet customer service apologized and said the driver should have been more assertive about interlining. (I told the driver what I planned to do when I boarded.) Anyway, this is kind of insidious from a rider’s viewpoint — the route map shows the bus going around in a loop, etc, and one might assume that if they stay on the bus, they’d also stay on the Line.
There are some large gaps in coverage, and the goal of only prioritizing low income areas does not set the foundation for the building of a great transit system.
What I got from the overview text on these changes was that Trimet and Jarrett were basically saying that if you are middle class or rich, and live in a single family home, then Trimet does not want to serve you.
This is a very “American” way of thinking, and it does not produce good results.
People with means do take transit, and I think it is a slap in the face to residents in those neighbourhoods losing service, that you say they are too rich to take transit. Only when all income levels are using transit do you have a successful system.
If ridership was low in those areas, maybe (I can guarantee) the service was not that good. Some of the areas listed as losing service and having low density would have buses every 15 minutes running through them in some North American cities like Toronto or Vancouver.
Some of the changes are nice. But the system is only as good as the weakest links in the system. And cutting people off from transit will not produce ridership. You are basically telling entire neighbourhoods to keep driving, and telling their kids to not give Trimet a try.
I have been in low density, super rich neighbourhoods with buses running every 15 or 20 minutes, and they are used by more than the nannies. But the key is the service.