Connection Points

62 Replies to “Connection Points”

1. The challenge is to get ST to objectively study the idea — as well as to get ST to inform the public about measuring transfer difficulty in general.

   ST does not publish the estimated rail-rail transfer numbers of riders nor the number of steps required to transfer — vertically or horizontally. And they won’t even promise two escalators in each direction at heavy transfer points (in case one breaks down).

   So there’s this basic structural approach to the overall ST planning approach that must be changed before any design changes would be considered.

2. This is a critical topic. The issue is reciprocal; yes, bus routes have to be well-designed; but Link stations also have to be well-designed. All our agencies (e.g., CT, ST, Metro, SDOT) have made or are making both good and poor decisions about connection points. Transfer points are where the transit trip slows down; they include waiting, that riders like least; they include walking transfers; the average pedestrian travels at three mph (264 feet per minute); seniors or those with disabilities travel the transfer distances more slowly. Crossing busy arterials to make transfers add danger (e.g., Rainier Avenue South, Fremont Avenue North, International Boulevard South, Montlake Boulevard NE ). Transfer points can be hindered by the clutter of dockless e-devises. Transit has seams; it cannot be seamless; agencies decide how wide the seams will be; seams are of time, distance, information, and price.

   1. Sound Transit is incompetent with transfers. My pet peeve is the horribly designed UW station with all the 520 route transfers as well as the 48 and NE Seattle buses. It’s ridiculous for a major transfer point to have transfers requiring crossing 2 busy streets at-grade and requiring about 6-8 minutes to male the connection.

      They are going to make this mistake again but even worse with the horrific Dearborn Station. Here it’s so bad I hope the entire ST3 gets killed, this is catastrophic location that will compromise the entire system. Key transfers will be 6 blocks away. The transfer situation really isn’t great anywhere in the system except the 1990 era stations Downtown when they put the stations under the street with multiple entries.

3. With buses, there’s an inherent tension, where the best possible connection experience requires detouring every bus into a transit center, which adds considerable travel time to every trip through the area. So, it is often necessary to compromise on transfer quality and accept that connecting riders might have to wait for a stoplight or cross a street for the of keeping buses moving.

   Trains are different, though. When the entire system is separated from traffic, there is no reason not to make the connections as seamless as possible, except to penny pinch on construction costs.

   1. The detours are deviations. If to/from a Link station, they are almost always worth the time of the network and its riders. It is best not to built transit centers as deviations. Kirkland has one of each: the KTC is along the way on 3rd Street; the TLTC is a deviation, requiring the routes to spend minutes for the connectivity; the TLTC is more a monument than a useful transit tool. It is best to build the Link stations so that they do not require the bus routes to deviate or to go through congestion. For the latter, consider the two Shoreline stations. North Shoreline is on NE 185th Street, an arterial without congestion that crosses I-5. South Shoreline is next to a full I-5 interchange with congestion that Shoreline (roundabouts and bike-ped bridge) and ST (Stride) are spending millions to correct.

      So, in ST3, the Executive and Mayor want the split station in the CID to add much longer transfer times between the east and south lines?

   2. With buses, there’s an inherent tension, where the best possible connection experience requires detouring every bus into a transit center, which adds considerable travel time to every trip through the area.

      It gets more complicated than that. With bus-to-train transfers a lot depends on the station design and placement. If a station straddles the street with entrances on both sides, then buses running perpendicular don’t have to detour at all. Same goes for same direction buses — if the station is in the right location then the buses can drop people off right by the station while the bus can keep going the same direction. Of course rarely do all of those things work out.

      The opposite is when the station simply doesn’t exist. The bus may go above (or under) the train line while largely ignoring it. Or it is forced into a big detour to make the connection.

      With bus-to-bus transfers you can sometimes have same direction transfers without a detour. This happens all of the time downtown, especially on Third.

      With other type of bus-to-bus some compromise has to be made. You inevitably end up with some people walking farther for their transfer than others. Even with a transit center this exists, although typically the rider doesn’t have to wait for a light. The best option is to consider what transfer is most likely and favor that one. For example the RapidRide G will intersect the 48. The most common transfers will be outbound G to northbound 48 or the reverse (southbound 48 to a bus headed downtown). Thus you want the bus stop on the corners that don’t require crossing a street for those combinations.

      With train-to-train transfers there tends to be more flexibility because you are spending a lot more money and are often not restrained by the roadway. Crossing routes can be designed for that and have short stair/escalator trips connecting each combination. Same direction transfers can occur at the same stop (if two different lines overlap) or with a cross-platform transfer (when they don’t). If riders are likely to reverse directions (with merging lines) then a center platform is ideal. It does get complicated, but with enough work and thought it can be done quite well.

      1. The G Line will improve the network, but could have been better.
         RossB mentions the transfer opportunity at 23rd Avenue; the G Line will not have an eastbound stop proximate to 23rd Avenue. The G Line will not be close to the Capitol Hill Link; in the U Link Metro P2, Route 49M was suggested; it would have connected the U District and 1st Avenue via Capitol Hill and Madison Street. The G Line will have a fairly long walk to/from USS under 3rd Avenue (one or two blocks, Marion-Spring streets). The G Line will make the walk to/from the Marion Street causeway and Colman dock longer by one block.

         RossB also discusses common stop transfers on 3rd Avenue. There are other examples. Route 330 has common stop transfers with the E Line at North 155th/160th streets; Lynnwood Link is giving that one up. Swift could have had common stop transfers at North 192nd Street; CT is giving that one up, at least for a while (they say they are waiting for a Shoreline project on North 185th Street. Routes 31, 32, 40, and 62 have common stop transfers on Fremont Avenue North between North 34th and 35th streets; but SDOT wants to close the northbound stop and impose walks and arterial crossings on those transferring to/from the northbound Route 40.

         SDOT also wants to lengthen transfers for northbound Eastlake riders with the shift of J Line pathway to the Roosevelt couplet from the Route 70 pathway; the last outbound stop will be on NE 43rd Street nearside 12th Avenue NE; riders oriented to Link will go one block uphill; riders oriented to the Ave will go two blocks; riders oriented to 15th Avenue NE will go three blocks.

      2. Detours to connection points introduce structural slowness to the route. That is, they guarantee that every trip you make on that bus is going to be much slower than driving, even if the bus has no riders and shows up with no wait time. This is bad. Ideally, any time difference between bus travel time vs. drive time should result from actual passengers getting on a off the bus.

         Also in the bus to bus world, when service tends to be infrequent, avoiding a street crossing to make a connection buys you nothing if the connection isn’t going to show up for another 10-15 minutes anyway. You need very good frequency before the overhead of the connection starts to actually impact a connecting passenger’s travel time.

         Of course, some will say that not needing to cross a street to make a connection is an end in and of itself, even if it doesn’t save the rider time. But, the problem is that it imposes a time penalty on other riders, and it feels unfair to ask other rides to sit through a detour simply to avoid having to move your feet during what’s going to be a 10+ minute wait anyway.

         My opinion is that, in general, a bus detour into a transit center is justified only when the majority of riders on the bus are making connections there, that is, more ons/offs than thru riders. When the detour isn’t justified, the bus can still stop on the street near the connection point, so connections can still happen, just with a bit more walking.

      3. “RossB mentions the transfer opportunity at 23rd Avenue; the G Line will not have an eastbound stop proximate to 23rd Avenue. The G Line will not be close to the Capitol Hill Link”

         That’s been a disappointment to me ever since U-Link. When the 43 was reduced to peak-only, it was assumed that transferring from the 8 to the to the 11 or 2 could replace it. But going south+west the stops are a block or more apart and the routes are too infrequent to make it work. The 48 reached 10-minute frequency but that’s now gone; and the 11 and 2 drop to half-hourly evenings. I’ve sometimes taken it when there’s an outage at UW Station, but inevitably I find it better to find an excuse to stopover at Trader Joe’s (walking to 17th & Madison) rather than waiting for a transfer. The 8 is a bit more frequent (15-20 minutes) but the stop is also two blocks away; it doesn’t go downtown but it can work for southwest Capitol Hill trips.

         “in the U Link Metro P2, Route 49M was suggested; it would have connected the U District and 1st Avenue via Capitol Hill and Madison Street.”

         I’ve always been skeptical of that routing. It serves a bit more of lower Broadway but still not all of it, and it dumps you onto lower Madison, when all the retail is on Pike-Pine. It’s good for the library, medical clinics, office buildings, and maybe the ferry, but little else.

      4. The G Line will not be close to the Capitol Hill Link

         The problem is more that Link won’t be close to the G Line. Link simply didn’t add enough stations between downtown and the UW. If they added a station at 23rd & Madison then the connection would be very good.

         The lack of a station creates the type of situation that adsf mentions. The bus would zig-zag to reach the station (treating it essentially as a transit center) then continue on its way.

         It makes way more sense to simply have the buses form a grid. Run buses along Broadway. Run buses along Thomas/John/Denny. Run the G straight on Madison. Stop trying to weave this way and that try and find some magic combination, because it simply won’t work. You end up with a failed system: buses running infrequently while straightforward trips along the same corridor (like Boren) require going way out of your way. It is really the worst of both worlds. Trust the grid.

      5. The lame route design choices made for RapidRide G will be a textbook example of bad decisions one day. Just a few ….

         – Bad connectivity to transfer points.
         – A few left-side stops meaning buying different vehicles.
         – Steep slopes creating problems holding bicycles, bags, strollers and small children in a vehicle with fewer seats and nearby straps.
         – Several years of construction required (years later it still is a mess).
         – High cost to build because the project’s unspoken primary objective was to rebuild Madison Street rather than improve transit service.

         And I hope that a fully loaded battery bus can climb that hill at a reasonable speed.

      6. The G Line will not have an eastbound stop proximate to 23rd Avenue.

         Yes, the transfer is bad. Here is what it looks like: https://www.seattle.gov/documents/Departments/SDOT/TransitProgram/RapidRide/Madison/2021_DennyTo24th.pdf. Just to back up here, these are the most common transfers:

         1) Outbound G to northbound 48 (e. g. Downtown to Montlake). The stop for the G should be just east of 23rd, while the stop for the 48 should be just south of Madison. That way a rider walks a short distance (without crossing a street). Unfortunately the outbound stop is close to Denny. This means the rider has to walk along Madison and cross 23rd (https://maps.app.goo.gl/wN9tcSC5F1T1NJX78). This is a degradation, as the 11 has a stop right where you want it. Making things worse, there is no stop for the 48 nearby (to the south). The closest connection is beyond John. This may explain why they didn’t just reuse the stop at John (the connection wasn’t that good) but that suggests they aren’t looking at the big picture. They should simply add a stop for the 48, closer to Madison.

         The best option is probably to stay on the bus until 24th and walk to the stop at John (https://maps.app.goo.gl/tYjC2tSQoLS5FgA47). This isn’t too far, but it requires crossing Madison and requires walking up hill. Riders can avoid the hill, but that walk requires crossing three intersections (https://maps.app.goo.gl/nQeJp3KTd3vTUWsr9).

         2) Southbound 48 to inbound G (e. g. Montlake to downtown). This isn’t quite so bad, but it still isn’t ideal. The 48 bus stop is in a good spot (just a bit north of the 23rd). The problem is the stop for the G should be on the other side of the street.

         Overall this is a big missed opportunity. It seems like this transfer could be at fairly close to ideal without a lot of work:

         1) Use the existing bus stop for the 11 (just east of 23rd on Madison).
         2) Add a northbound bus stop for the 48 just south of Denny. (You can’t squeeze a stop between Denny and 23rd). This would mean riders going from downtown to Montlake would walk across Denny (https://maps.app.goo.gl/nQeJp3KTd3vTUWsr9) where there is a stop sign (and no traffic light).
         3) Move the inbound stop for the G across 23rd. There seems to be enough space there — at most you would get rid of a few feet of parking. The current setup does have the advantage in terms of the bus merging — there is that. But losing that seems like a small price to pay for an ideal transfer that is likely to be common. But again, that transfer isn’t nearly as bad as the other one.

         I think it has more to do with the stop spacing. For whatever reason an outbound bus stops at the east side of Denny. This would mean a stop just east of 23rd would be fairly close. The solution is to just move the stop on Denny to the west (on the other side of the street). Better yet you could move that Denny stop further west, closer to 20th/Olive. That would actually be better stop spacing. The current approach has a bus stop right by the Safeway (which is good) but it appears it set up a cascade of bad stop decisions.

      7. “ Yes, the transfer is bad. Here is what it looks like:”

         The aerials don’t show how truly steep Madison is east of 23rd. It makes the transfer effort more of a problem than it appears even here.

      8. – Bad connectivity to transfer points.

         I agree, but most are not that bad. For example it stops on Third, which is where most transfers will occur. It stops on First, which means that if they ever add a bus (or the streetcar there) it will connect well. At first glance the connection to Broadway buses/streetcar looks bad, but that has as much to do with the streetcar as anything. The streetcar (and buses) don’t stop close to Madison — they stop down by Boylston. Thus a stop there is about as good as you are going to get (https://maps.app.goo.gl/UkRLqkqam1qJuHRw5). Just like with the 48 you would have to move the other bus stop to make it work better. Hopefully they do that at some point.

         – A few left-side stops meaning buying different vehicles.

         You think it is better to have buses stuck in traffic?

         Steep slopes creating problems holding bicycles, bags, strollers and small children in a vehicle with fewer seats and nearby straps.

         The only alternative is to regrade Madison, which is really not an option. Same goes for Queen Anne as well. We have hills (we didn’t get rid of all of them).

         Several years of construction required (years later it still is a mess).

         Which they would have to do anyway because:

         High cost to build because the project’s unspoken primary objective was to rebuild Madison Street rather than improve transit service.

         Nonsense. The primary objective was to improve transit service. Holy cow, we are talking about buses running in their own lane every six minutes! Not a BAT lane (that is often clogged up while cars turn right) but their own lane. This will definitely be an improvement.

         But yes, much of the cost was to do utility work. Unfortunately this is common. They use the money from the feds to pay for things that aren’t really transit related (fixing pipes). But they eventually would have to do the work anyway. Not all of the work was of this nature though — they widened the street in a few places in order to squeeze in the buses. This often means additional utility work (e. g. moving a utility pole). The overall cost of the system is not really that high for what we’ll get (in terms of ridership per dollar, ridership time saved per dollar, overall service time saved per dollar, etc.).

         The biggest problem is not the line itself, but how Metro responded to it. The network largely ignores it. Buses will basically overlap it for no good reason. It will be ridiculously frequent compared to many of the other routes mainly because the restructure is so bad. You are going to have this bus running every six minutes while other buses run every twenty minutes at best. At best! So instead of complementing the route, the buses will be making redundant trips or worse yet making a bunch of unnecessarily turns for some antiquated reason. In comparison the poor transfers are a minor issue. Someone who makes a transfer from the G to the 48 has every reason to complain — but it is quite likely they will complain about the 48 being infrequent as much as they complain about the bad transfer.

      9. The aerials don’t show how truly steep Madison is east of 23rd.

         No, but the links I included do. If you expand the Google Directions it includes an elevation profile (e. g. 52 feet gain in about 400 feet of distance). Getting off at the stop to the west (Denny/22nd) is fairly flat. I would probably walk 22nd just because it is more pleasant. The big drawback is that you have to cross two intersections (along with walking a ways).

         What this shows more than anything is that projects like these should be an excuse to look at the greater network. I’m sure moving the stop for the 48 (let alone the stop for the streetcar) were not part of the scope, but they should be. To get it right in the future will require moving more stops as a result.

      10. “ – A few left-side stops meaning buying different vehicles.

          You think it is better to have buses stuck in traffic?”

          The only difference between a center and side bus lane is any delay from right turning vehicles in front of buses. There are plenty of on-street solutions that could have been developed to minimize delay.

          The real way to reduce travel time is with signal priority and off-board fare payment along with an exclusive bus lane — but that lane does not have to be in the median.

          When they could not find a vendor for the articulated ETBs with left doors, the median station idea should have been nixed.

          It needs to be understood that RapidRide G was developed by the City and not Metro — and the City has completed some pretty bad transit project designs, like the snail-like FHSC. Had Metro led RapidRide G design, it would seemingly not have required a different vehicle type.

4. Running the West Seattle Link into the existing tunnel is a no-brainer and the simplest solution. With each of the three lines running at every 6 minutes, this would mean 2 minute headways in the tunnel which is technically feasible, and arguable, desirable (after all, frequency is freedom). I think there is some minor engineering work but this is small in comparison with the overall cost of the West Seattle Link.

   1. Agreed. The best transfer is no transfer. If the trains all go to the UW (and beyond) you eliminate most of the transfers. Meanwhile, same direction transfers (e. g. Everett to SeaTac ) would be ideal. You simply get off the train and wait for a different train at the same platform. Reverse direction transfers would require going up and over at CID, but that will be the case soon anyway. It is still much better than every option that has been considered seriously, let alone the option they are favoring now. It would be ideal for riders, while likely saving the agency quite a bit of money.

      1. And if you add a center platform, then even the reverse direction is easy.

      2. But, that would require the ST board to admit to making a mistake about the DSTT capacity. I have zero faith of board members ever being willing to swallow their pride to do that.

      3. Two lines in SODO with one center platform transfer effort: Less than 15 feet of walking for riders. No more than 10 seconds walking and waiting at one of 16 doors. No level change escalator or elevator.

         ST preferred plan transfer in SODO ( note Westlake and CID-N/ Pioneer Square much more time-consuming with at least 4-6 level changes): Two level changes with no plans for down escalators, or about 80-90 stair steps Walk up to 400 feet. Forced onto single path for making the level change. Without crowds at least two minutes and even longer if crowds are heavy. (Only exception is going northbound from West Seattle to southbound Rainier Vally and SeaTac.)

         Center platform in SODO would facilitate all transfers. If the desire is to accommodate two trains, just have a double train length platform like Boston Green Line has!

         How ST is so comfortable with two lines from CID to Lynnwood but not two lines in SODO puts the arbitrariness of ST station planning decisions in full view.

      4. “But, that would require the ST board to admit to making a mistake about the DSTT capacity. I have zero faith of board members ever being willing to swallow their pride to do that.”

         Other countries would recognize the best solution and make it happen. If DSTT1 needs capital improvements to accommodate three lines, make it happen. Passengers need it. Balducci’s constituents going from the Eastside to the airport need it.

      5. I think Balducci is one of the few board members who would listen to this idea, and change the current approach. Most seem unable to grasp the various issues, or they simply can’t be bothered. They see this as a neighborhood battle (folks not wanting the station in a particular location) rather than a fundamental transit weakness. This is certainly the case with Harrell and Constantine.

      6. “They see this as a neighborhood battle (folks not wanting the station in a particular location) rather than a fundamental transit weakness.”

         That shows they don’t understand the impact on passengers. The pro-CID/N-CID/S neighborhood activists don’t either.

   2. The demand from West Seattle will not be high enough to warrant 6 minute frequencies, and 10 minutes will be more than enough. The reason for the 20 trains promised in ST 3 was to meet the demand north of Downtown and a promise to make no link rider wait for more than 6 minutes at peak.

      East Link EIS even assumed an 8 minute peak, and the 6 minute promise in ST3 has not been technically explained.

      The only possible overcrowding on a branch is under Beacon Hill.

      In trains per hour per direction, that’s 10 for Rainier Valley/Seatac, 8 for the Eastside and 6 for West Seattle. That’s 24 or an average of 2.5 minutes. It may be that 8 for Rainier Valley, and 6 for the Eastside and 6 for West Seattle will be enough for each branch in a post-Covid office worker peaking situation.

      At SODO, it would be 16 trains an hour in each direction. That’s the same number of trains promised between CID and Lynnwood by the end of next year! That’s also less than the 20 promised in ST3.

      At most platform doors may be desired for safety at stations — but that’s not required. . But that’s a heck of a lot cheaper than building a giant station that requires two level changes. .

      1. “In trains per hour per direction, that’s 10 for Rainier Valley/Seatac, 8 for the Eastside and 6 for West Seattle. That’s 24 or an average of 2.5 minutes.”

         ST says the limit with the current DSTT specs is 20 trains per hour (3-minute reliable frequency). So drop 2 and it would fit in what ST said was the capacity.

         With capital improvements, DSTT1 could go up to 40 trains per hour (1.5 minute frequency). It already does that after ballgames, but ST doesn’t want to do it every day. (And to achieve ballgame service, the schedule goes out the window, and southbound trips at Roosevelt and Capitol Hill fit in whenever they can.

      2. It isn’t really about West Seattle. If you merged all three lines from the south (West Seattle, SeaTac and Bellevue) then it is about the maximum demand for one of the branches. Which one is hard to say, but 6 minutes is probably more than enough. 7.5 minutes is probably OK and more about providing extra value for riders (not having to wait that long) than it is dealing with crowding.

         There are other things that can be done for crowding. One is to change the type of trains. This is relatively cheap (https://seattletransitblog.com/2016/12/20/will-link-waste-its-capacity-for-the-sake-of-operational-convenience/). Another is to improve the alternatives. For example Sounder could run more often during peak. Buses could provide alternatives that would be popular with riders.

         This issue deserves a bit of an explanation. It is assumed that riders prefer Link over a bus and all other things being equal, that is a safe assumption. But all other things aren’t equal. For many riders, taking Link requires a transfer. An express bus is often faster than Link. This is especially true for the trains that would merge. Rainier Valley has the 7, which is gradually improving and is closer to more people than the Link stations. Tacoma has Sounder, which is faster and more comfortable than Link. Express buses are significantly faster from Federal Way to downtown. The same is true from Redmond to downtown. Riders from Issaquah wouldn’t save a huge amount of time with an express, but they may just prefer the one-seat ride to downtown. The same is true for the vast majority of riders in West Seattle.

         These are all basically “worst case scenarios” and they really aren’t bad. It is common for agencies to simply endure crowding and even bunching without focusing on it — even while they expand in other areas. The Green Line in Boston is notorious for its bunching. It is probably one of the “bunchiest” trains in North America, if not the world (if you don’t count trams stuck in traffic like Toronto). Yet eliminating this bunching is not a priority. The Canada Line in Vancouver is a huge success — despite the issues with crowding. Again, there has been little work done, simply because it isn’t a priority. By all means if you can avoid crowding by spending a bit extra then do it. But we aren’t talking about spending a bit extra. We are talking about spending a lot and making things worse.

         There is simply no good reason for the agency to be fixated on crowding issues that likely won’t even occur. It is completely backwards. The Canada Line has to be considered a huge success, and various agencies in North America are jealous. It is like a restaurant that is crowded within weeks of opening — most restaurant owners wish they had that problem. We run the risk of the far more common and far worse problem: not enough people ride the trains despite the high cost. The system just doesn’t work well, and yet trying to improve it becomes costly. You can’t spend money running the buses (or even the trains) more often because you are still spending money on the trains.

      3. “ But we aren’t talking about spending a bit extra. We are talking about spending a lot and making things worse.

         There is simply no good reason for the agency to be fixated on crowding issues that likely won’t even occur. It is completely backwards. ”

         Yes I fully agree with Ross here. ST is predicting a problem with many good alternative solutions before spending billions on a parallel deep subway in Downtown. Create more capacity in each vehicle by removing seats or driver cabs (or buy replacement vehicles with more efficient layouts)! Install platform screen doors to speed up trains pulling in and out of stations!

         And nowhere does ST have any similar concern for when to install more escalators to handle heavy transfer loads. Not only is changing levels a hassle, but escalators and elevators and even stairs can back up when lots of riders get off a train at once. If overcrowding was a concern, why isn’t there isn’t similar focus on vertical conveyance capacity? It seems to be an arbitrary concern to justify spending more money.

         In West Seattle’s case, ST released ridership demand forecasts 20-30% lower in the EIS than they said in 2020. All of the peaking assumptions were developed pre-Covid and work from home popularity. And the place with the biggest overcrowding worry between Symphony and CID Stations is paralleled by the Third Ave transit mall literally on top!

         And as I’ve explained before, if one train is too crowded it gets slower and slower (doors stay open longer at stations), so the following train gets closer and closer and picks up fewer and fewer riders. Unless there is a massive service disruption, a rider will almost never have to wait for more than one or two trains even when overcrowding occurs.

      4. “There are other things that can be done for crowding. One is to change the type of trains. This is relatively cheap (https://seattletransitblog.com/2016/12/20/will-link-waste-its-capacity-for-the-sake-of-operational-convenience/).”

         We had an article on this in 2016? Then ST is talking out of both sides of its mouth, and there’s no excuse for it. ST can’t complain about overcrowding and needing DSTT2 when at the same time it ignores low-hanging fruit to get 20% more capacity out of the existing number of trains. And that’s just for open-gangway trains. Switching the rest of the seats to side-facing like on the NYC subway would also add capacity.

      5. “The Canada Line has to be considered a huge success, and various agencies in North America are jealous”

         If only Sound Transit, the three counties, local cities, and WSDOT and the legislature were jealous. But they either don’t know the Canada Line exists or don’t bother to look at how effective it is.

      6. To be fair the Honolulu skyline is supposed to be similar to the Vancouver Canada line and American construction methods have belabored it as well with infinite environmental delays and high construction costs

5. Thanks for the article.

   Though I was kinda surprised no discussion about perhaps building a cross platform interchange at sodo. (aka similar to macarthur bart station)
   https://en.wikipedia.org/wiki/MacArthur_station_(BART)

   1. If ST determines that they want two transfer platforms, the original alternative of a second elevated double track (no Lander overcrossing) would work. All ST would have to do is to simply put the northbound tracks on the ground and southbound tracks in the air. (They instead proposed leaving the Rainier Valley beach on the ground and West Seattle in the air.)

      Of course, it would require shifting at least one track to create center platforms on each of the two levels.

      It also would allow easier same-direction track switches.

   2. yes, many ways to improve the transfer experience!
      Cross platform interchange helps tremendously with same direction transfers, but not with reverse direction while Al’s elevated track also helps with reverse direction as it requires only one escalator rather than two.

6. Now that West Seattle Link is separated from DSTT2, there is no reason to have the giant rebuilt SODO station as part of the West Seattle Link project. ST can simply move the platform slightly northward (something that ST has already planned to do) and build it as a single center platform.

   The construction phasing could be to build the new track just outside of the existing track. Then close the line for a few weeks to drop in the track connections (something the current plan will also require). Once trains are on the new tracks, rip out the old track to then build a center platform. Then build the new stairs and elevators to connect the SODO Station back to the neighborhood.

   I think the biggest engineering challenge to the SODO situation is actually the crossing of southbound tracks towards Beacon Hill with the northbound track from West Seattle — south of SODO Station. However that’s not a fatal flaw. The split doesn’t have to be in the same place northbound and southbound. ST has to connect West Seattle to the Central OMF no matter what. So ST could build a new connection in many, many locations between Lander and the Beacon Hill tunnel portal.

   1. Yes, the construction phasing could help!
      I’m not sure the OMF connection will help much. The current design calls for the OMF connection to be far further South. I’m not sure why they build a new connection and don’t reuse the current connection.
      I think you could lower the northbound track from West Seattle a bit into the ground so that the southbound RV line can pass above.

7. I’ve been making this argument for years. SODO will be a critical transfer station, especially if there isn’t a CID hub. There should be easier same direction “spine” transfers – UW to/from the airport.

8. It is worth noting that some transfers are more common than others. It is unfortunate that CID was not rebuilt with a center platform, since it will be common place for a reverse direction transfer. But for now I don’t see that many people switching trains there. With Beacon Hill to the East Side the transfer is a given. Rainier Valley to the East Side is unlikely, as riders will just transfer to the 7 (which is quite frequent and saves time for most trips). East Side to the airport riders riders have the infrequent (but handy) 560. This will likely go away when Stride 1 starts. This will be more frequent than the 560, although it will require a transfer (that may get worse over time). Some may prefer a train-to-train transfer, but the bus should save some time. Same goes for trips like Federal Way to Bellevue.

   As Link is expanded the combinations increase. If ST continues on the (misguided) path it is on, there are a number of extra combinations. I won’t list them all, but give examples. For example “UW to SeaTac” also applies to UW to Rainier Valley, Rainier Valley to Northgate, etc. Here are the combinations as I see it:

   1) Same direction UW to SeaTac. This is a new transfer that will inconvenience plenty of riders. None of the transfers look good.

   2) Same direction Ballard to West Seattle. Same type of transfer, but likely involving fewer people (even when you consider SLU or Uptown).

   3) Same direction Ballard to East Side. Similar to 2 (except riders can’t transfer at SoDo).

   4) Reverse direction Ballard to UW. Some combinations will be common, but depending on the quality of the transfer, alternatives may be better. For example from Ballard itself to the U-District, the 44 is probably better. From The Seattle Center to the UW the monorail may make more sense as the first leg. The quality of the transfer and the frequency of the lines will have a big impact on ridership for these combinations.

   5) Reverse direction SeaTac to East Side. Mentioned up above.

   6) Reverse direction West Seattle to Eastside. Fairly common as this will be the main way this trip is taken. Not especially difficult, but again it is too bad the CID station doesn’t have center platforms.

   7) Reverse direction West Seattle to SeaTac. I don’t see this as being that common, simply because the 50 offers this right now and it is not that popular.

   If I was focused on transfers I would do the following:

   1) Rebuild CID with center platforms. This would pay dividends almost immediately, and improve things in the future.

   2) Run the three southern lines (from West Seattle, SeaTac and the East Side) to the UW (and beyond). Not all of them would go to Lynnwood (let alone Everett) but same direction transfer would be easy and fairly frequent. You would lose some direct connections (Ballard to the south end of downtown) but retain others (UW to SeaTac). This approach would offer plenty of other advantages as explained earlier — smaller stations, more frequent trains. Better frequency improves the transfer experience.

   3) Focus on the Westlake transfer. This is critical with either approach, but if the Ballard train is running frequently and the other trains are running frequently until they branch at CID, then this transfer would become quite popular if it was a good one.

   1. 3) Focus on the Westlake transfer.

      You are absolutely correct, Ross. But remember that the only way to improve it significantly is to make it shallower.

      Since we at the Blog –except Lazarus I’m sure– would prefer that the tunnel turn uphill toward Boren and Madison someday, making it deep enough to match up with The Chairman’s and Hizzoner’s deep “New Pionere Square” proposal is no longer required.

      But there is still some pain to be suffered by the City, because the only way to make the new platform(s) significantly shallower is to cut-and-cover it/them in the Fifth or Sixth Avenue street right-of-way.

      That will not be popular with those operating current businesses in the immediate area, but very popular with their grandchildren.

9. Thanks, Martin, for an excellent write-up of this, the only reasonable plan for SoDo in the current economic and ridership environment. I’m glad you did it, AJ and the other people in the “ST is doing a good job” caucus listen to you.

   You might want to “blow the diagram up” some, because it looks like the busway is being consumed by the trackway. Your write-up makes it clear that you are proposing taking (at most) one lane for the long block south of Lander, but the diagram might put some people off.

   Given current and likely future bus volumes on the busway, having a one-block single-lane section is entirely acceptable. I do think you might want to add a caveat that if this is done both Lander and Holgate would have to overpass the trackway/busway and (most likely) Lower Royal Brougham Way will have to be severed at the rail line. The busway would subsume the street west of the tracks with the Greyhound station taking the eastern end.

   1. Yes, it gets tight where the 4 lines merge, doing a short one-way section or taking some land south of the station might help. It also depends on the width of the center platform and whether you could shift it east a bit.
      Yes, the Lander and Holgate overpasses would still need to be built but timeline may need to be accelerated.

      1. Thank you very much for listening to refinement ideas. You are a ery clear writer, and the very soul of “comity”.

      2. Thanks, Tom!

10. Martin,

    I’d like to note that it is possible to do this without changing the existing flyover. There is (barely) enough room between the descending northbound RV track and the backs of the buildings along the bikeway that the track could just be placed on the bikeway. Yes, this would be very tight, and it might make sense to buy the buildings, build new back walls six feet farther toward Sixth South and demolish the western six feet.

    The merge turnout would be right in the Lander right of way (remember, it’s now overpassing).

    The southbound alignment would look like your diagram except that the track rising would still be RV. West Seattle would stay on the ground next to it, taking the northbound busway lane. The single-lane section would still be required.

    The nice thing about this is that it allows the construction of the new southbound track in the location of the current platform without interrupting existing operations. Since no new elevated construction is needed, the only time lost would be relatively brief. The existing northbound trackway would be cut in Lander to add the frog for the merge turnout. The southbound track would have to be cut at mid-Lander and just north of the existing platforms in order to connect the new southboud trackway at both ends of the to-be platform.

    These are quick and easy things that railroads do all the time.

    The station would have to be out of service for a while, though, because the southbound platform would have to be take out of service befoe the new southbound track could replace it, and would stay out of service during the removal of the old southbound track and its replacement with the new center platform.

    But it is would save some of ST’s precious “esteem in the community”, doing it your suggested way would work just fine.

    1. Whew, two bad sections of spelling late at night:

       “have to be taken out of service before” in the next-to-the-last paragraph

       “But if it would” in the last one

    2. If only ST would just study the possibility of building SODO Station differently. This is the solid wall of denying that needs to be busted! Then the suitable engineering can be determined.

       Another minor musing: Would ST benefit from having a temporary platform surface quickly available when needed? That’s gets into the general lack of investing in contingency planning (crossover tracks , single track operations, partial platform closures, rider messaging). I could see how a temporary platform could be put into place very quickly. A quick Google search yields several products and case studies in other countries where temporary platforms have been successfully used.

       1. Excellent point, Al. Again “Not invented here-itis” bites Skycastle in the butt.

    3. “If only ST would just study the possibility of building SODO Station differently. This is the solid wall of denying that needs to be busted!”

       Both at SODO and at CID.

       “That’s gets into the general lack of investing in contingency planning”

       ST needs to realize it’s a light rail operator now and plan for high availability. It has been doing some things in the maintenance periods — I don’t want to discount that — but there are a lot of other things it should be doing or preparing for their eventuality. Starting with operations management when it currently single-tracks and lets frequency fall to 20-30 minutes. Of course it has to do that sometimes, but it seems to be the first resort rather than the last, and not enough attention is paid to shortening the span, keeping it to late evenings, or maintaining normal frequency on at least part of the line (i.e., the northern half where demand is highest).

11. I think it is also worth considering the difference between bad connections and a bad network. A bad connection does not show up on a transit map. In fact it looks very good. For example, look at Oran’s excellent map: https://seattletransitmap.com/app/. Consider a couple of trips involving a transfer:

    1) Wallingford to Capitol Hill (44 to Link).
    2) Central District to the airport (48 to Link).

    In both cases the bus goes straight and meets the train without deviating. Both of these look great on the map. As it turns out, the transfer itself varies (the first is much better than the second) but both are good from a network standpoint. Now consider a third transfer:

    3) Bitter Lake to Capitol Hill (345 to Link). In this case it is clear that the network is not good. The 345 makes several turns and ultimately loops around before reaching the station. Even though the transfer itself is good, the network is not. It takes riders a very long time to get to the station, and the good transfer can’t make up for it. Furthermore, the extra time taken by this (and other) buses means that the bus is infrequent. So folks in Bitter Lake have an infrequent and slow connection to Link. This means that the network connection is so bad that for this trip (and plenty of others) the riders are better off just ignoring Link and taking more frequent buses (the E and 8).

    Like so many weaknesses in the network it is easy to blame Metro, but in this case that really isn’t the problem. Northgate Station — for all its strengths — is terrible as a transfer node. Buses can not go east-west (which explains why the 345 loops around). The north-south corridor (1st) has limited value, since it is adjacent to the freeway has a lot of traffic north of the station and ends abruptly to the south. It is basically the opposite of the Roosevelt Station, where buses are bound to go by the station whether there is a station there or not. From a network standpoint the only significant advantage of Northgate Transit Center is that it has easy access to the southbound express lanes — a feature that is now obsolete.

    It is too late to do anything about Link, but Metro can adjust the network accordingly. It can’t be ideal, but it can be better. It is essential that Northgate have buses, but it doesn’t need to have as many buses serving it. As much as possible we should send the buses towards stations that are easier to get to, like Roosevelt and 130th (when they build it).

    1. Where would you put Northgate Station instead? Or could the 345 transfer at another station? In the future the northern half could transfer at 130th Station. The southern half would still have to go to Northgate. And it would have to go around the button-hook or it would lose access to the college. The pedestrian bridge is moderately doable but it’s rather long to expect everybody to use it. Or you could move the station to the college I guess, but Metro hasn’t contemplated a network based on that, and the college hasn’t contemplated giving up a significant chunk of its land or bulldozing its wetland. Some route has to serve the college, and it may be arbitrary that the 40/345/346 were chosen to, but those are the established routes, and some people in those neighborhoods may be counting on them serving College Way.

       1. Where would you put Northgate Station instead?

          Roosevelt & Northgate Way. Better yet go to the other side of the freeway and put it close to the college, and put the next station at 105th & Aurora. The 40 would then go “straight across” towards Lake City. You would probably still have buses going north-south on either side. To the east the bus would go along Meridian to College Way and then join the 45 in connecting to Roosevelt Station (and perhaps the UW). The 348 would go straight down Roosevelt to the UW. It isn’t clear where a bus traveling on 5th NE would start but it would head down 5th until Weeden and 65th (to connect to the Roosevelt Station).

          Or could the 345 transfer at another station? In the future the northern half could transfer at 130th Station. The southern half would still have to go to Northgate.

          Yes, that is what I’m getting at. Because the stations are awkwardly placed, the buses are forced into an awkward network. That is why I wrote “It can’t be ideal.” But it can still be better. For example, the 348 should go straight to Roosevelt (and then the UW) instead of curving around on its way to Northgate. The 75 should not head towards the 130th station (and come within five blocks) only to follow the same squiggly pathway to Link. The connection for riders along 5th to Link should be better.

       2. Mike, don’t you often argue that Link stations should be in the heart or center of a community? Where’s the heart of Northgate? It’s certainly not at 1st & 103rd.

       3. The heart is probably on 5th. As for locations at Roosevelt & Northgate Way or next to the college, those are good ideas but I’ve never heard of them until now; nobody raised them in the 1990s or 2000s when they could have influenced the alignment.

       4. As for locations at Roosevelt & Northgate Way or next to the college, those are good ideas but I’ve never heard of them until now; nobody raised them in the 1990s or 2000s when they could have influenced the alignment.

          It shouldn’t be up to random individuals to design the mass transit system. A handful of people in the neighborhood should not be the ones that enable a station at 130th that every transit expert would consider obvious. A transit advocate — with no actual training — should not be the one who gets Sound Transit to consider running the Ballard Line on a pathway that even they admit would carry the most people (through Queen Anne). We shouldn’t have to form a movement just to get Sound Transit to consider merging the various lines downtown — something that would save money AND provide for a better transit experience.

          As for the center of the Northgate neighborhood, the area has evolved, largely because it was allowed to evolve. There is nothing really special about the places that have big buildings other than the fact that Seattle allowed them to have big buildings. Thus if they allowed growth around 5th & Roosevelt there would be development around 5th & Roosevelt. Unlike Ballard or Capitol Hill, the area doesn’t really have an historic center or large institutions, unless you count the mall, which is probably the worst possible place to put a station. Malls have historically been focused on the automobile — that is their very nature. Many have evolved, but you would basically roll the dice and hope that one developer manages to build something as urban as what exists on 5th.

          The college is a different matter, and complicates things. The college is pretty big (17,000 students) so it makes sense as a station. It is also awkward to reach, as it sits along the Meridian/College Way corridor midway between 85th and Northgate Way. Thus it can only be served either by a north-south bus (e. g. Roosevelt to Haller Lake) or what will be the 61 (Lake City to Greenwood bus). Thus serving it with Link adds a lot of value (riders get there a lot faster).

          The point is, either approach works. Either approach would result in a bus network that involves fewer twists and turns to integrate with Link. The key is to put (at least some) of the stations at major intersections. One of the big weaknesses with Link is that there are just too few stations. If I had to choose one station location between Westlake and the UW, I would probably pick exactly what they chose. But I shouldn’t have to pick only one! Even just a second station at 23rd & Madison would greatly improve the bus network. A stop at 15th and you’ve enabled easy bus connectivity for the entire area. The other problem is of course, the freeway. There is definite value in having the terminus be close to the freeway (for buses that travel along it). But every station north of Roosevelt is by the freeway. Roads by the freeway tend to start and stop. This means they basically just work for east-west connections. That can be fine if you have enough stations (and cross streets) but Link (and the city) doesn’t. This explains why we talk about “L” shape routes (i. e. a bus going down Aurora and then turning to serve one of the stations) instead of just having east-west connections every half mile or so (like the Canada Line). Even if we had good stop spacing (and line spacing) the intersections by the freeway are often congested and difficult to serve with buses. But even an ideal north-south route does not offer as much connectivity as a diagonal route (assuming a cardinal street grid) like the Expo Line. Link actually does go diagonally (between downtown and the UW) but they just didn’t add the station(s). If Link cut over to Aurora after serving Roosevelt it would be moving diagonally, offering good north-south and east-west connectivity without much effort.

          Despite an outstanding example to the north, folks weren’t interested in building a transit network. Too many of the folks see buses as a separate entity, or something that Link replaces, not complements. Connectivity was never a priority. Completing the spine was.

    2. Route 345 in 2026: between Four Freedoms and NW Hospital via North/NE 130th Street, 5th Avenue NE, NSC, and Meridian Avenue North, serving two Link stations, Northgate center, NSC, multifamily housing, and medical clinics. Look ma, no deviation; the awkward activity centers become the termini.

       1. The addition of 130th Station will eliminate one of the big connectivity problems. It basically solves the Bitter Lake problem (while making the connection from Lake City and Pinehurst much better). Four Freedoms and Northwest Hospital remain awkward to serve (since they aren’t “on the way”) but making them end points of a route (as you suggest) essentially solves that problem. I think this is an important idea. Out of the way places are best served at the end of a route, or at least close to the end, so that fewer people (or no one) are hurt by the deviation.

12. I have always prefered our light rail stations being built with a center platform. They seem more efficient to me and easier to add transfers to in the future. This may only be beneficial here. It may have made sense to the New York Subway. There center area was for express trains that did not stop. At least that is how I understood it.

    In 2003 when Sound Teansit decided to build a modern system, without express tracks, why did they build side platform stations? I understand the reason for the original bus tunnel. It also makes sense for an infill station, like 130th. Tracks are already there.

    I do not understand the benefits of side platform stations in Sound Tramsit’s network.

    1. Center platforms are better for passengers because it allows you to transfer or turn around at the same platform. Eastside riders will soon have an opposite-direction transfer to the airport. Sometimes I miss my stop and have to turn around. Sometimes the travel time or wait for my original trip is so long I change my destination en route or cancel the trip and need to go the opposite direction. Center platforms allow you to do all this, while side platforms don’t. So it’s an important passenger-experience issue.

       The reason DSTT1 has side platforms is the agency balked at widening the tunnel at stations so that trains buses could stop at the sides. They said it would be too difficult to make the bulges or something.

       The initial Link segment in 2008 followed suit with side platforms, although SeaTac is center platform. Later in U-Link design in the 2010s, ST realized that center platforms were better, and installed them from Capitol Hill northwards. But it couldn’t bring itself to do so for the most important train-to-train transfers in the system, at CID and Westlake. It talked about renovating CID Station and possibly making it center-platform, but it never did.

       1. I understand that and the history. So why are there side platform stations in East Link? More than 15 years after the initial segment?

          Also, Stadium, Beacon Hill, and I think Rainier Station are center platform. All part of the 2009 segment. So they built both styles along the first segment.

       2. In general a center platform is easier to build as you can also share escalators/elevators for both directions, but you need to pull the track apart before/after the station. That requires space you may not always have. For ground level side platform stations, you can walk up to one side without crossing any track. In-fill stations usually are side-platform, too.

       3. All they had to do was adopt left-hand running. Both ends had complex ramps, so it would have been easy.

       4. The bus tunnel was designed to handle a huge number of buses. Buses had the ability to pass buses, which basically gets you the kind of busway capacity found in Bogota, which can get 49,000 riders each direction per hour (and 2.5 million riders a day). My guess is they didn’t want the buses to weave at either end of the tunnel, as this would delay them.

          This is one of the fundamental issues with trying to predict the future. If we doubled down on the bus tunnel (or built a new train tunnel downtown) then the bus tunnel would have been fine as it was designed. But we leveraged it for rail, and it isn’t great for rail. That being said, it is common to retrofit stations all over the world. CID had to be rebuilt to handle East Link trains. That would have been a good time to add a center platform. I could even see having both in this case (although that might require a wider station). You would basically only need an emergency set of stairs for that center platform (since it would only be used by people reversing directions).

          Then again maybe it was too expensive. I’m not sure that many people will do a reverse transfer at CID. It doesn’t work well for East Side to Rainier Valley (might as well take the 7). TIBS to Downtown Bellevue will have an express bus. Beacon Hill or SoDo to East Side is a clear case where reversing directions makes sense — I just don’t think that many people will do that. Southern suburbs to East Side sounds reasonable, but again, the bus might work out better for a lot of riders. I doubt there are that many people on the East Side who go to SeaTac on a regular basis (unlike people who live to the south). I suppose it makes sense with West Seattle to the East Side — I could definitely see that.

          The biggest rail-to-rail transfers are likely to be same direction (assuming the build a second downtown tunnel) and reverse direction at Westlake (which will be two different stations).

Comments are closed.