Approximately 35% of the land area of Seattle constitutes public right-of-way. Today, most of this space is dedicated to general purpose lanes and parking and is utilized mostly by single occupancy vehicles. Yet, as the city becomes denser, we are finding this configuration to be inadequate in serving its changing needs. The transportation supply it provides no longer meets the demand as evident by growing congestion. This is only the result of a good thing—strong job growth and record-low unemployment. But it is also a sign that we should consider alternative street configurations.

The many uses of a right of way

Fundamentally, right-of-way is just public land. And besides providing a pathway between buildings, it has historically also served as space for markets, social gathering, children’s play areas, and been critical for community building. The crowding out of these functions by transportation is in part natural, as fueled by growing population, but mostly a function of decades of deliberate car-centric policies. To understand this problem better, below we will calculate how many people actually benefit from the same 10 feet of right-of-way depending on its use: parking lanes, general purpose travel lanes, bike lanes, bus lanes, and even dedicated space for light rail.

Vehicle Travel Lane

The most common use of right-of-way, a vehicle travel lane can move between 500 and 900 cars per hour. It’s interesting to see that the lower the speed of travel, the more vehicles it can move per hour due to lower following distances:

Roger P. Roess, Elena S. Prassas and William R. McShane (1998), Traffic Engineering, Prentice Hall.
Traffic volume depending on travel speed (Source: Roger P. Roess, Elena S. Prassas and William R. McShane (1998), Traffic Engineering, Prentice Hall)

So at 25 mph, lane capacity is maximized at 900 vehicles per hour. Besides vastly increased safety, this is yet another argument for 25 mph speed limits. Given the average occupancy of vehicles at 1.6 persons (based on the PSRC Transportation 2040 Final Environmental Impact Statement), the throughput is:

Vehicle Travel Lane: 1,440 people per hour

Vehicle Parking Lane

Parking, the topic of never-ending neighborhood feuds and controversies… How many people do actually benefit from a parking lane? This is not as easy to evaluate as measuring the throughput of movement. Parking only makes sense as a whole corridor, as opposed to as a point.

To make sense of this, we start with the average non-work trip length of 4.5 miles based on the PSRC Transportation 2040 Final EIS. We use non-work trips, as work trips likely have a significant highway component where there is no parking regardless.

How many parking spots do we have on 4.5 miles of road (per direction)? Given space for crosswalks and intersections, parking can take up about three-quarters of a corridor, or in this case 3.38 miles. Assuming a standard length of 18 feet per parking spot, we have 990 spots. How many people can benefit from them depends on the average parking duration and average occupancy of vehicles.

A Seattle Department of Transportation (SDOT) parking study from 2002 shows an average parking duration across all zones at 2.3 hours. After discussing this with people in the know, I understand the average rate may have gone down a bit, but not significantly, so we can assume it is 2 hours.

So, the 990 spots can be used by roughly 495 cars per hour and given the average occupancy of vehicles at 1.6 persons per vehicle we get:

Vehicle Parking Lane: 792 people / hour

Bike Lane

Bicycle lane throughput varies based on width. A study from Davis, California reports the capacity as 2,600 bicycles per hour per 3.3 feet of lane while most other studies have actually found higher throughputs, so we’ll use this as a lower bound.

Regardless of whether bike lanes are installed in place of 8-foot parking lanes or up to 11-foot travel lanes, we can assume at least two bike lanes per original lane. This gives us:

Bike Lane: 5,200 people / hour

Bus Lane

While we already have a number for cars per hour that can use a lane, we can’t just apply it to buses. Buses are longer (60 feet for an articulated bus versus 15 feet for the average car) and more importantly, make frequent and long stops. The Transit Capacity and Quality of Service Manual has guidance on the average number of buses that can use a lane with everything considered.

Source: Transit Capacity and Quality Service Manual
Bus flow volume at various service quality levels (Source: Transit Capacity and Quality Service Manual)

Assuming “Stable Flow, Interference” to avoid bus bunching and too much of a speed decrease, the lower bound for CBD streets is 50 buses per hour per lane. It should be noted that the SDOT-operated bus lanes in Downtown Seattle carry up to 190 buses per hour, but they do so with significant delay through that section. At rush hour, it can easily take 3 times longer to go through Third Avenue due to the heavy bus congestion.

A 60-foot articulated low-floor bus typically has 65 seats and can carry 55 standees (per manufacturer specification) meaning that we get a throughput of:

Bus Lane: 6,000 people / hour

So a bus lane easily serves 4 times more people than a general purpose vehicle travel lane!

Surface Light Rail Lane

Light rail lanes are similar to bus lanes, but allow for even longer vehicles yet fewer of them per hour. Train length is limited by the length of blocks as otherwise trains would stick out and block intersections. Blocks in Seattle vary from 200 to 450 feet and they vary by route taken.

So using 95-foot Link Light Rail trains for the example, a short block can fit two of them. They fit 200 people each (74 seated, 126 standing per specifications).

The highest frequency at which any US light rail line operates is 2.5 minutes or 24 trains per hour. So with that we get:

Surface Light Rail Lane: 9,600 people / hour

Note how that is more than a bus lane despite running half the number of vehicles per hour. That’s because of passenger density: a platoon of buses is less dense due to required following distance. It’s also important to note that for some corridors in Seattle, one could achieve 19,200 people per hour by running 4-car trains with no degradation of service quality.

Grade-Separated Light Rail Line

I am adding this for comparison purposes only, especially as Sound Transit 3 is approaching. A fully grade-separated rail line is one that is either underground or elevated (or otherwise has no grade crossings). Frequency can be brought down to 1.5 minutes and trains can be of any length desired. Sound Transit typically designs for 4-car trains, which are 380 feet long.

Grade-Separated Light Rail Line: 32,000 people / hour

Whoa. So a light rail tunnel/viaduct can move as many people as 22 car lanes! This is really why it’s important to approve ST3 regardless of the status of bus rapid transit (BRT) projects in Seattle. A BRT line will never come close to moving this amount of people. In fact, the busiest bus line between the US and Canada is the 99 B-Line in Vancouver, which moves about 50,000 people per day (not hour!) and Vancouver is planning a subway along its length because it will actually be cheaper to operate.

At the same time, approving bus lanes for a BRT line can deliver a benefit in a couple of years as opposed to 20 to 30 years for light rail and can be done on far more corridors than we can afford light rail on.

A few words about pedestrian space

While sidewalks are really efficient at moving people (11 feet would move up to 15,000 people per hour assuming 3x3ft space per person and 3 mph), we shouldn’t think of them as just that.

Pedestrian space forms the public realm—the places where people meet, share experiences, and identify with. It’s the base connective tissue that enables the kinds of interactions to happen and enable a city’s economic productivity to grow superlinearly by 130% when population only doubles.

A parklet is a great example of a public realm element. When used by a restaurant (so called streatery), for example, it can seat 5 to 15 people depending on configuration. Assuming that restaurant patrons spend an average of 2 hours at the restaurant, the same as the average parking duration, we can immediately see that a parklet with 10 people is 10 times more efficient than a parking spot.

What’s better than a disconnected set of parklets abutting traffic is simply wider sidewalks. They enable amenities like outdoor seating, art, greenery to be built right into the sidewalk, and greatly improve the walkability of an area.

The tally

How many people benefit from different street configurations? Grade-separated light rail line: 32000 people per hour. Surface light rail-lane: 9600 people per hour. Bus lane: 6000 people per hour. Bike lane: 5200 people per hour. Vehicle travel lane: 1440 people per hour. Parking lane: 792 people per hour.
How many people benefit from different street configurations?

No calculation error, or technological improvement can boost the space efficiency of car-based usage so it catches up with the lead of other modes (and I am using conservative estimates for these other modes). It is clearly the least space-efficient and one must really desire to be part of the flat-earth society to argue this point.

Furthermore, the ever-controversial parking is at the very bottom of the list. Street space dedicated to parking serves almost 8 times fewer people than a bus lane and nearly 7 times fewer than a bicycle lane. Preserving parking in-lieu of a transit or bike lanes is really an attempt to keep our transportation system as inefficient and gridlocked as possible.

Why not more roads

Whenever you see congestion on the street, it is a sign of only one thing—that the transportation demand on that street is exceeding the supply and that more supply needs to be provided. Now we could build more supply off-street, but it is extremely expensive:

  • Tunneling: Seattle has 1,540 miles of arterials and for $1.55 billion per mile (SR-99 cost) we can tunnel under each street for a bargain-basement price of $2.4 trillion. I’m sure the required 2,400-year tax levy will pass just fine.
  • Building viaducts: a lot cheaper, but still exorbitantly expensive at $500 billion or so—for a brutalist sun-blocking concrete monstrosity on top of every arterial.
  • Or we could, take a (parking) lane and switch it to a denser mode…

More on parking

The other aspect of parking is that it really doesn’t have to be on the street. A multi-floor garage stacks cars on top of each other to bring the land utilization of parking more in tune with the denser transportation modes. This frees up space that can then be provided to increase the transportation capacity of the corridor.

But how do we come up with the garages?

First, in Central Seattle (Downtown, Denny Triangle, Belltown, Pioneer Square, Chinatown) we have 237 parking garages with 44,000 parking spaces. In comparison, we have around 4,000 on-street parking meters. Parking garages are on average 71% full, meaning we have more than 12,700 empty spots, more than sufficient to cover for all street parking if repurposed.

Within neighborhoods, most new construction apartments build more off-street parking than required. According to a recent city analysis, 75% of the 219 new developments built between 2012 and 2015 in areas where no parking is required provide parking anyway. These unused spots can be put on the market and some apartment management companies (e.g., Equity Residential) are already doing that.

Switching a lane to a denser mode

So how does this work? Let’s say we have 2 lanes per direction and supply equals demand. There’s no congestion and we have some buses mixed in. We move 1,800 vehicles split as 1,785 cars + 15 buses for a total of 4,581 people per hour.

But as demand increases beyond this point, we simply get congestion. We know that with buses we can move 4 times more people per lane, but why would people switch from a car to a bus, if they get no benefit? Why should I give up my car when somebody else could do that?

So we take one lane and switch that to a bus-only or HOV 3+ lane. If we move  900 cars + 27 buses per hour we’ll move the same 4,581 people with no congestion. But we now have the freedom to scale up to 50 buses per hour and move 7,440 people per hour with still no congestion.

In other words, we increased the corridor’s capacity by 62%, equivalent to adding another lane, by spending money only on paint and funding increased bus service.

Reasons to switch

The reason why such a switch makes sense is nothing more but supply and demand. You can only fit so many vehicles in a stretch of road and nobody wants to pay for an underground highway under every arterial.

A switch is not being made for environmental reasons, for social justice reasons, or because somebody has an evil  agenda. It’s made because of math, plain and simple.

On the flip side, when supply exceeds demand, not switching a lane over to a denser mode is nothing short of undemocratic tyranny. And that tyranny, when also applied to housing, is a key limiter of economic growth and the top generator of inequality in our region.

Where transit lanes are no-brainers

If the theory above doesn’t persuade you, let’s look at actual practical examples from Seattle. The following is a list of streets where the volume of transit riders exceeds the volume of non-transit riders:

Street Vehicles / lane / day Transit ridership / day Transit Lines Transit Lane Status
Aurora Ave N,
south of Fremont
16,750 28,300 E, 5, 16, 26x, 28x Peak-only, none on Ship Canal bridge
West Seattle Bridge 17,883 23,900 C, 120, 21x, 37, 55, 56, 57, 116, 116x, 118, 118x, 119, 119x 1/4 eastbound only
Rainier Ave N 9,150 15,900 7, 9 None
Elliot Ave W 12,225 13,600 D, 15X, 17X, 18X 7-9a south, 7-9a + 3-7p north
Denny Way 8,525 10,300 8 None

(Data is from 2014, latest available.)

So in every one of these cases, a permanent transit-only lane would move more people than a general purpose lane. Moreover, as buses become more reliable and faster, it is expected that more people will switch to transit, resulting in an overall expansion of people moved across the corridor, equivalent to widening the road, but with almost none of the cost as exemplified earlier.

Ultimately, it seems like a no-brainer—SDOT should study every one of these corridors for permanent 24-hour transit or HOV 3+ lanes and implement them as soon as financially possible. By approving Move Seattle, people clearly indicated their support for this approach. If you’re tired of waiting for this to happen, here’s a few things you can do:

Previous articleSeattle’s Agenda For Sound Transit 3
Next articleThe Man Game And How To Beat It
Anton has been living in the Pacific Northwest since 2005 and in Seattle since 2011. While building technology products during the day, his passion for urban planning and transportation is no less and stems from a childhood of growing up in the urban core of a small European city. As a member of the Belltown Community Council he has worked to improve the transportation options and public realm in the neighborhood.

21 COMMENTS

  1. Your main point that increasing the # of buses to move more people will improve congestion is flawed. Not everyone wants to or can take a bus to where they want to go. For some people, the time it takes to take a Metro bus is too long. For some people, buses don’t go where they need to go. Adding more buses slows down traffic all over town.

    • My point is not about reducing congestion at all. My point is about moving more people. We have an increase in the number of people who want to move through the same stretch of road. The only way to handle this without building tunnels, viaducts or acquiring expensive land and razing neighborhoods is by increasing the density of people in vehicles.

      Also bus transit covers the entire city and travel times will only continue increasing with SOV travel as demand continues to grow and supply remains the same. The only way to have good travel times going forward is by using high passenger density vehicles in dedicated right of way.

      • so you want more people in the same amount of congestion? Need to find ways to keep the flow of traffic moving, speed up the through put for all vehicles, go farther than just traffic light to traffic light.
        “The only way to have good travel times going forward is by using high passenger density vehicles in dedicated right of way.” This cannot be true, need more solutions/options than high passenger density vehicles. This solution does not work.

        • Congestion is not a good target for transportation policy. If you can move more people, while congestion remains the same, then you improve welfare.

          But if you care about congestion, then the best remedy is road tolling.

  2. I support adding bus lanes to arterials, but I am not sure that the data presented in the last table supports that conclusion.

    Please correct me if I am wrong, but the first column of “Vehicles/lane/day” should be multiplied by the ratio of 1.6 to get “people/lane/day”. And “transit ridership/day” should be divided into 2 lanes, 1 for each direction (unless you are proposing a single transit lane for buses running in both directions. Yikes!)

    With these two corrections, the people/lane/day in general vehicles is more than the transit riders/lane/day in all five cases.

    Maybe you already did these calculations on the data, and the column titles don’t precisely reflect the numbers provided. Please check.

    • There are a few omissions actually. You are right that average occupancy was not taken into account. However, the traffic volumes used were not corridor averages, but were chokepoint maximums. Moreover, the Elliot Ave stretch omitted three bus lines – 24, 32 and 33.

      With everything taken together I got the attached table. On Aurora we still have more transit ridership per mile than car “ridership” per mile. We’re also very close on Elliot Ave/15th.

      However, the main point of the article remains – as transportation demand exceeds supply on all of those corridors, the only way to move the same amount of people without congestion is by using vehicles with a higher density of passengers than cars. That would also allow us to scale the capacity of those corridors to an order of magnitude higher values than what we have today and given that demand only continues to increase, we basically have no other viable choices.

      • Thanks for updating the table!!

        I support dedicating transit lanes on arterials, but as you note, except on Aurora, the decision would be based on managing future demand and type of city we want to have. As it stands now, converting one lane each direction to transit only would result in more people/hr moving in the general purpose lanes compared to the transit lanes *assuming nothing else changes*.

        But people would respond, as they always do, to the changed conditions by driving less and taking transit more. Over time, transit service would be increased and eventually the transit lanes would move more people than the GP lanes, resulting in a higher-capacity corridor.

        In addition, the overall pedestrian environment would improve due to less traffic noise, pedestrian counts would rise along with transit use, and local shops would see more business. Transit lanes can start a virtuous process that can convert a “car sewer” into an actual place for people.

        • Chad, take a look at my answer to EHS’s question. If you look at it from the perspective of buses per hour, at peak, all corridors have more people on buses than in cars except Denny Way.

  3. Great explanation. A really useful column in your table of Seattle streets and bus routes would be the number of buses per hour. That would allow for comparison to your calculated capacities (obviously time of day poses a challenge), and to other corridors where bus lanes could be added and service increased or consolidated from nearby streets.

    • Hi, thanks for the tip. So I looked at the existing transit capacity in the peak in terms of buses per hour. There are two assumptions:

      1) Peak time buses are mostly artics – this is usually true for Seattle with very small exceptions

      2) Peak time buses are actually full – this is also usually true with many peak-runs being over capacity (I’ve seen values up to 115% capacity)

      When I ran the numbers on that I see that all of the above corridors move more people in buses than in cars per lane (cars cap out at 1440 people/hour) except for Denny Way:

  4. An alternative to bus/hov lines is flexible road pricing. If it is possible to implement it well, we don’t need separate lines for buses, all vehicles can move with little congestion on the same lanes. I think it is a solution one should seriously in a context where it is hard to add additional lines (like bridges), or to grab existing roadspace for sole use of buses.

  5. I loved the thesis of this article. However I noticed that you used average occupancy when calculating personal vehicle capacity (1.6) but used maximum occupancy when calculating transit capacity. Any reason for that?

    • Actually rush hour occupancy for cars is likely not much higher than 1.6 either. Here’s why:

      2014 Center City Commute mode share for cars breaks down as 30.1% SOV and 8.3% HOV. If we assume 4 people/car for the HOVs the average occupancy is still 1.65.

      Buses, on the other hand, are often crush loaded over manufacturer capacity during rush hour.

      Report used: http://commuteseattle.com/wp-content/uploads/2015/02/14-5390-Commuter-Mode-Split-Survey-Report-2-23.pdf

      • I understand and generally agree with your choice of methodology, but there’s something off about comparing the different modes on an equal basis. I agree that a transit lane can hold 6,000 bus passengers per hour but actually finding that many people to use the lane is another challenge. Yes, buses are crushed during rush hour on major routes but this is not uniformly true and a more accurate figure is average bus capacity instead of just examining downtown rush hour occupancy. Some cities will do better than others but the average bus capacity in the United States is a grim 9.2 based on the best evidence I was able to dig up. Some of this low statistic is based on deadhead and maintenance trips but they end up taking up road space regardless of their purpose.

        You argue that BRT adoption would increase ridership and I’d agree with that but then you’re stepping into the realm of hypotheticals instead of comparing actual concrete figures. If you start to compare the theoretical maximum capacity of a transit lane then it seems fair to apply the same standard to the theoretical maximum of HOV lanes, which would multiply your travel capacity to 4,500 (assuming 5 seat capacity per car). That’s clearly absurd, but assuming maximum bus capacity all the time is not that much more reasonable.

        • I think there are two points:
          1. In Seattle, on bus routes connecting neighborhoods (urban villages or centers), the average peak time ridership is the same as the maximum ridership for the entire length of the line between the centers. Seattle never built a grade-separated rail system, so this is the result.
          2. It’s totally fine to have HOV instead of bus lanes, as long as the minimum number of people per vehicle required is set to an amount guaranteeing free movement. On Seattle freeways only HOV 3+ lanes move freely, HOV 2+ usually do not.

  6. Your skewed vision for Seattle traffiic doesn’t consider facts, such as buses not being full, displacement of cars, travel freedom and that your traffic studies were from a distant, European city?? This is Seattle. Listen to Seattlites, not others about OUR home, traffic, plans and likely/probable solutions. Helsinki… really? Yes-I quite agree, the vivion is a “no-brainer.”

    • All studies are from the US, and in fact mostly from Seattle. When there is more demand for movement on a road than the road can handle, the only way to attain freedom is to increase the capacity of the road. And the only way we can afford to do so at scale is by using higher-capacity vehicles in dedicated right-of-way as explained under “Switching a lane to a denser mode”.

  7. Tweak: the 16 is now the 62.

    Request: add in pedestrian lanes (sidewalks), and let’s see how they compare on human throughput.

    I must disagree on ST3. The MLK street-running segment is a bottleneck that’s regularly disrupted. About 9 days ago we had a 3 hour, 20 minute disruption caused by a car crash. ST3 needs to grade-separate this but won’t–especially since it doubles down on the single long line rather than building a network with redundancies and multiple transfer points. Also, East Link includes another street-running segment on Bel-Red Road. We need to fix MLK and change the Bel-Red alignment before we start increasing reliance on them. Then we need to use driverless trains which can run more often than the existing ones (which led to the idiotic $5 billion tunnel under downtown to benefit Everett and Tacoma).

  8. Great article. One thing – because you are focused on ‘lanes’ (space), you seem to be skipping over BRT systems that use multiple lanes to increase throughput; e.g., 20,000-40,000 passengers per hour per direction (Curitiba, Guangzhou, Bogota). Dividing by lane number would bring things back down somewhat; but even then high-capacity BRT could be in the 10-20k range [per direction], so at or exceeding your surface LRT number and approaching your grade-separated LRT number, even without the grade separation (which you can do with BRT anyway – see, e.g., Xiamen). And high-capacity BRT doesn’t need multiple lanes the whole length – just at stations, and turns, giving space back to other uses (e.g., sidewalk, including non-moving uses).

    Thoughts?

Comments are closed.