The Seattle Alignments and the Hidden Costs of West Seattle Rail

The Sound Transit 3 proposal package can roughly be divided into two parts, the suburban projects and the urban projects (the suburban projects could in turn be further divided into completing the “spine” and the eastside). The urban projects are the projects located in Seattle that are intended to serve Seattle residents. Of course, these projects will also serve suburban residents (at least much more so than suburban projects) because many Seattle neighborhoods are high-density employment centers that are not readily accessible by car, at least during rush hour.

My previous posts have mostly concerned some of the issues around the suburban projects, namely their poor cost-effectiveness and issues around freeway based alignments. This post intends to look at the issues around the urban projects, as well as propose a (in my view) superior alternative.

Let’s start by getting the facts straight. By the basic numbers, Ballard has by far the best ridership per dollar. Using the most recent numbers from Sound Transit, and basic assumptions of upfront payment of costs, 30 years of operation and a 20 year wait before service opens, Ballard has an extremely low cost per rider and is over 2.3 times more cost-effective by this metric as compared to West Seattle link as seen in Table 1 below.

Table 1:

Ballard WS Table 1

However, these numbers are misleading. The 129,500 figure for Ballard ridership includes ridership on South Link that will use the new tunnel, but only to go to/from the downtown stations to points further south. This is effectively ridership that would use the old tunnel under a no-build scenario. In order to compare West Seattle rail and Ballard rail apples to apples we need to use ST’s estimate of 67,000 riders that actually board or alight from the stations north of Westlake.

We also need to update the costs for both Ballard and West Seattle. Sound Transit maintains that the existing tunnel cannot have three separate lines (South Link, East Link and West Seattle Link) all sharing the same tunnel, as it would water down each branch’s headway to an unacceptably low 9 minutes. For similar reasons, Ballard Link also can’t use the existing tunnel. So in effective terms both lines require building the second tunnel and so the cost of said tunnel should be split evenly between the two lines for accounting purposes. Table 2 below shows these numbers under the assumption that the downtown tunnel is 1.25 miles at $750 million per subway mile.

Table 2:

Ballard WS Table 22

As you can see, the Ballard segment still grades out as more useful than the West Seattle segment. But the difference is closer to 33% better not 130% better. Combining these two columns we can derive the cost effectiveness of the Seattle rail projects as shown in Table 3.

Table 3:

Ballard WS Table 32

This $5.27 figure is ok, but these numbers can be improved upon with better routing decisions. I noted earlier that the West Seattle and Ballard lines required construction of a second Downtown tunnel. But this isn’t exactly true. Building a West Seattle line requires a downtown tunnel because only two of the three lines going south from the International district can use the existing tunnel. But the Ballard line doesn’t have to go all the way through Downtown. It can instead head east towards First Hill and the Central District. Because Sound Transit obstinately refused to ever study real transit to First Hill in this round, we don’t have exact counterfactuals, but let’s consider the Green Line in the image below, serving Madison, First Hill near Broadway and Jefferson, the Central District near 23rd and Jackson and Judkins Park station.

Green Line


For ridership, if we assume that Interbay gets about 7,000 boardings then on average an urban station for Ballard link is worth 12,000 boardings. So First Hill station would account for an additional 12,000 riders. The Central District and Judkins park are somewhat less dense so let’s estimate that combined they add an additional 12,000 riders bringing the total to 24,000 riders.

For costs, the net cost of a First Hill alignment over building a downtown tunnel would be about 1 mile of additional tunneling plus 0.6 miles of additional elevated track. At costs of $750 million per mile tunneling and $250 million per mile elevated (in line with the West Seattle rail alignment), the total net cost would be about $900 million, or about $1 billion dollars less than West Seattle rail. That leaves $1 billion dollars in Seattle’s budget, enough money to build direct access ramps between the West Seattle bridge and the SODO busway ($200 million) and completely grade separate the Interbay section ($400 million), with potential additional funds for infill stations, improved amenities around stations, a ship canal tunnel or other short-term improvements.

The alignment change also has an additional cost/benefit advantage. With the Ballard line not interlining with an existing Link line, the Ballard line can be automated (it could also have smaller station footprints and hence cheaper capital costs but I am ignoring that complexity for now). Automation comes with substantial operational efficiencies, as well as making it cheaper to maintain high frequency at off hours. Based on Vancouver Skytrain operational costs of $1.94 million per mile and Ballard and West Seattle link operational costs of $5.35 and 4.68 dollars respectively, automation should reduce operational costs by at least 50%, a substantial value improvement long-term.

Table 4 below outlines three scenarios. Ballard and West Seattle as proposed (Table 3), Ballard to Judkins Park automated with grade separation in Interbay and Ballard to Judkins Park light rail at grade through Interbay. Note that the automated scenario assumes that grade separation through Interbay will not add any additional riders, even though speed improvements through that section should increase ridership by a non-trivial amount.

Ballard WS Table 42

Based on these assumptions the automated line would increase ridership per dollar by 12%. Of course without better ridership estimates for First Hill, the Central District, grade separated Ballard and frequency improvements with automation, this figure could be too low or too high. For instance, if the First Hill segment drew 30,000 riders instead of 24,000 we’d see a 22% improvement over the current proposal with a $4.32 subsidy per rider. This analysis also ignores the cost/benefit of bus transit improvements for West Seattle* or how the remaining ~=$400 million dollars budgeted get spent. The analysis also ignores the costs associated with the disruption associated with building at grade along 15th ave. All these factors indicate that the Green Line would be substantially more cost-effective than what is proposed.

If your disappointed about the at grade segment in Ballard or the lack of service to First Hill, note that the fundamental problem is that service to West Seattle requires using limited resources inefficiently, as a large portion of the West Seattle alignment is redundant and West Seattle. though dense, isn’t dense enough to overcome that issue. Moreover, if you want to find funding for complete grade separation or a tunnel under the Ship Canal, the way to do it is prioritizing First Hill over West Seattle. I’d also add here that if Sound Transit is a regional organization then First Hill, with high employment density and a key regional asset in Harbor View Medical Center, represents a regional destination in a way that West Seattle, with mostly residential density and no regional destinations doesn’t.


*At most times of day bus service to the portions of West Seattle is extremely fast. During midday Rapid Ride C is scheduled to get between Alaska Junction (California and Alaska) and Downtown in 20 minutes on average. West Seattle rail would get between the Junction and Westlake in about 17 minutes, with most of the time savings coming between Alaska Junction and Delridge, where the C line makes several additional stops. But at peak, the West Seattle bus routes are prone to crowding and congestion on SR-99 and/or the West Seattle bridge. The simplest fix to that problem is to bypass that congestion by making the left lanes of West Seattle bridge HOV 3+ and then building off ramps from those HOV lanes to the SODO busway. The buses would then connect with Link at SODO and potentially continue north to Downtown.


Freeway Transit and Aesthetics

In my previous post, I discussed the transit productivity of freeways and showed the benefits, in terms of efficiently moving lots of people, of ensuring free flowing bus transit on freeways. But in terms of rail transit in freeway corridors, the post only critiqued rail transit in so far as reliable bus transit on freeways is often far more efficient per dollar. However, there is a second common critique of freeway based transit, which is that it can’t readily generate long-term high ridership through transit-oriented development around stations. Theoretically, I believe this claim is largely bunk and this post will address why.

The first problem with the claim is that we have counterfactuals of high-density urban development adjacent to freeways. In Seattle, for example, the section of I-5 between Lakeview Blvd. and Yesler Way (i.e South Lake Union, Capitol Hill, Downtown and First Hill) are generally covered in dense buildings that are at least five stories tall. Moreover, move one more block away from I-5 and it is hard to tell that you are near a freeway, given noise levels and development patterns. Indeed, Seattle’s tallest building is situated in just such a location. Dense development can occur near freeways and dense areas near freeways are generally walkable up to one block away from the freeway itself.

Another potential downside with freeways is the amount of space they take up. Freeways are about a city block’s worth wide and this does cut into the walkshed around a potential station. But assuming that a rail station’s walkshed is 2,600 ft. (0.5 miles) and assuming a city block takes 260 feet, a freeway will take up about 1,352,000 sq ft. (260 * 5200) out of a total walkshed of 13,520,000* or about 10% of the walkshed. That’s a significant number but hardly a deal breaker. Indeed, the 3rd avenue stations of the Downtown Seattle Transit Tunnel lose about 12.5% of their walkshed to Elliot Bay alone. And yet these stations are projected to have some of the highest ridership in the entire light rail system. In short, in terms of developable land area, freeways only have a marginal impact on the potential utility of a transit station.

Indeed, it seems the principal issues around freeway transit concern the micro issues around walkability. It is generally very unpleasant to be near a freeway and in many instances frontage roads and off ramps around freeways make the area extremely difficult to navigate on foot. If people feel like it is an ordeal to get from the station to their destination, they will be far less likely to use the service. This aesthetic argument represents the core of the transit-oriented development argument against freeway rail alignments.

But making the walk to and from that station is a problem that can largely be fixed with two key design elements. The first key design choice is to build the station at an overpass where there are no on or off ramps. This way street design near the station can be oriented principally around pedestrians who may use the station (or just walk through the area).

The second key design choice is to lid the freeway with retail on either side of the overpass, like this lid the city of Columbus built over I-670. Adding human-centered uses where pedestrians exit the station can largely eliminate the discomfort associated with walking near a freeway. The overpass feels just like a normal street and not a freeway overpass. In addition, retail lids can be built fairly cheaply with cost estimates around $10 to $20 million, due to their minimal size and their being utilized by rent-paying businesses. With this relatively small additional capital investment (stations typically cost $80 to $100 million), the transit station can integrate more or less seamlessly into the urban environment. It’s also worth noting that smart lids’ have benefits for the neighborhood as a whole by providing better means for crossing the freeway. Overall, with smart station location and effective design, accessing a freeway based rail line can feel like accessing a non-freeway based line.


Freeway-based transit alignments have the benefit of readily available right of way and minimal neighborhood pushback due to noise pollution. Thus, they can be a very effective way to provide high capacity transit in a city that lacks it. But integrating freeway based transit into the city fabric and creating useful alternatives to car dependence requires effective design. Citizens and transit advocates should push hard for the addition of these design elements to freeway stations. In order to maximize ridership and rider experience, it is very important to ensure that planned freeway stations come with meaningful investment to maximize walkability in the area.

*13,520,000 = 2 * (2,600 ^2) (the area of a square with cross sections of length 5200).

Freeway Transit and Lane Efficiency

On Saturday, the Seattle Times editorial board wrote an editorial raising questions about the freeway-oriented nature of Sound Transit’s draft proposal. The editorial is rife with misleading or just plain old wrong information. But the board does raise a worthwhile point, which is that buses on freeways can be extremely useful transit that moves tons of people effectively. Let’s look at some numbers on freeway based bus transporpation.

Typically, transportation planners estimate that the average freeway lane can carry 1,900 vehicles per hour. Given an average vehicle occupancy of 1.2 persons per vehicle, that means 2,280 people can use a given lane in a given hour. (Note that this is in freeflow conditions. When congested, freeway lanes move less vehicles per hour). Thus, speaking abstractly, a freeway lane dedicated to transit would only need to carry about 2,280 people per hour for that transit lane to be a fairly efficient use of resources. All this is during peak hours when freeway space is scarce. At times when the freeway is well below capacity, the transit lane wouldn’t effect travel times in other lanes.

Fortunately, WSDOT has travel data on key freeway corridors so we can see just how many people utilize freeway based bus transit during peak hours. For example, on I-5 southbound at Northgate during the peak morning period (5am to 10am) the HOV lane carries 48% of all persons traveling. Of this 48%, at least 73.5% are transit riders meaning that the number of transit riders through the area is about 35% of total travelers during peak period. Given that there are five lanes southbound at that location, transit is carrying about 2.7 (35 / (52 / 4))  times as many people per hour as each general purpose lane.

Transit moves people along I-5 far more efficiently despite the fact that the existing HOV lanes that transit uses, though 11 minutes faster on average between Everett and Seattle, are still congested during rush hour, with about 12 to 15 minutes of delay over freeflow conditions. In short, there are enough HOV2+ car drivers to substantially congest the HOV lane buses use. Transit ridership would probably be at least somewhat higher if the lanes were converted to transit only lanes, ensuring reliable 60mph commuting for buses in the corridor and 12 to 15 minute travel time savings. Because transit moves so many more people then the general purpose means, converting HOV lanes to transit only lanes, at least during peak, would be a very effective use of limited freeway space.

Here is a more detailed cost/benefit analysis. At 35% (conservatively) of total I-5 users during morning peak, the benefit to the average transit rider would have to be about double the cost to the average non-transit rider from the conversion of the HOV lane to a transit lane. If the average transit rider is delayed by 15 minutes, then as long as the average driver was delayed by less than 7.5 minutes with the removal of the HOV lane society would unambiguously benefit from the switch. Unfortunately it is difficult to make definitive claims about how much additional delay drivers would face with the change, except to say that the 13% of non-transit HOV users would have an additional 11 minutes of delay, which means the remaining four lanes would have to have delays of less than 5.3 additional minutes for the transit lanes to be socially preferable.

It is worth noting here that the previous paragraph undersells the benefits of transit lanes on I-5. Reliable, time competitive transit tends to develop strong ridership and 12 to 15-minute improvements in travel time along with substantial improvements in reliability should convert many more riders to transit. Indeed, speed and reliability improvements are key reasons the Everett link extension would generate 35,000 to 43,000 daily riders or about 5 to 6 times as many riders as ride the current equivalent Sound Transit express bus routes.

This analysis just looked at I-5 southbound near Northgate during morning rush. But similar claims can be made about other HOV corridors in the region with only 3 of 14 HOV corridors meeting Washington State’s reliability standard of: “an average speed of at least 45 mph, 90% of the time during the peak hour of travel.”

The capital and economic justification exists today to implement fast reliable transit covering large swaths of the region by raising the standards of HOV lanes or simply converting them to transit only lanes. And if transit becomes extremely fast and reliable in the region’s freeway corridors, then the added value of building rail lines along I-5 to Everett and Tacoma, I-90 to Issaquah and the West Seattle Freeway to West Seattle pales in comparison to the Billions and Billions of dollars that rail would cost.

It is worth noting that there is some value from building light rail over having a bus rapid transit corridor, namely operational efficiencies at scale (trains are much longer than buses) and rail bias. But these benefits, particularly capacity, are not very important for the freeway rail corridors in the ST3 draft plan. Ultimately spending billions of dollars for additional rail along freeways in lieu of converting HOV2+ lanes into transit only lanes (or at least HOV 3+) represents an inefficient use of resources, incentivizes auto-usage over sustainable mobility, and ensures that excellent transit will reach these corridors 15 to 20 years later than it could, at great cost to travelers.