New downtown tunnel for light rail, current one for buses and streetcars

Both transit advocates and politicians have been discussing the idea of a second downtown tunnel through Seattle. The tunnel would act as the foundation for future rail service to Ballard and West Seattle. Seattle Subway’s proposal currently looks like this and seeks to address a few issues:

  • Mitigates the closing (for buses) of the current Downtown Seattle Transit Tunnel by building the tunnel initially for buses
  • Provides capacity enhancements in the downtown core
  • Can provide exclusive lanes for buses heading to West Seattle and Ballard

The proposed tunnel could also have a few drawbacks:

  • Costs of unused railway infrastructure: If the tunnel were to be built initially with railway infrastructure (tracks, power and communications) as Seattle Subway suggests, these would still require some form of maintenance even if they were not being used. These infrastructure components don’t stand still over time, and will have an aging effect. No one really knows how long it would take for rail services to begin, but it would be a waste to maintain infrastructure that is not being used.
  • There are only 3 stations (Westlake, Madison and International District), fewer than the current DSTT. Although this would give a direct connection to Madison BRT, it would also mean that buses serving the new tunnel have farther stop spacing and potentially shorter travel times than light rail in the DSTT, the opposite of what it should be.
  • Conversion to rail could be a slow process due to institutional and political inertia, which may result in another DSTT situation where buses are “phased out”. The lesson to be learned from the DSTT, is that joint-operations is difficult and results in unreliable service for every mode, even six years into operations.
  • Alignment constraint in the current DSTT are not addressed, such as the Westlake or Chinatown curves

In an attempt to explore how these issues could be resolved, I’ve created another proposal, which is a tunnel on 4th Ave that will accommodate all light rail services from the beginning and in the future. The current DSTT will then serve buses (again) and potentially streetcars. The “Westlake curve” will also be removed, allowing higher speeds in the tunnel. Here are the key advantages of this proposal:

  • Removes infrastructure constraints from the network: The current DSTT has a few limitations on speed and acceleration. One is the horizontal shift in the rails before and after the platforms. The others are the tight curves just south of Westlake and International District. With a new tunnel, Sound Transit can take advantage of the (rare) opportunity to remove those limitations by designing a large-radius curve between University St and Capitol Hill, while using better transitions before/after the stations. This will allow for higher travel speeds through Downtown Seattle. This also removes one of the infrastructure constraints that prevents Sound Transit from purchasing higher-capacity non-articulated vehicles, such as those on standard subway systems.
  • Immediate travel time and capacity improvements for the entire regional light rail network: With infrastructure limitations removed and stops modified, higher travel speeds through Downtown Seattle will not just improve local travel time, but regional travel time. Both Central Link and East Link stand to gain a few minutes, along with reliability improvements. With higher stop spacing, punctuality and faster travel times, the capacity of the tunnel can also be increased to accommodate trains from additional lines. Converging multiple lines in one city center tunnel will provide a very high-frequency service through downtown
  • More leverage for regional funding: The prospects of regional funding for a tunnel that will, at its beginning, only serve Seattle routes will be a tough sell. However, if it can be used by regional rail in the beginning (Central Link and East Link), and provide travel time/frequency improvements on opening day, there could be more leverage to ask for improvements.
  • Separation of operations from day one: Dedicating the new tunnel to rail on opening day will prevent a repeat of DSTT. Joint-operations doesn’t work for a variety of reasons, but the basic concept is that rail operations are completely different from bus operations. Rail involves one large vehicle regulated by signals, occupying a section of the track at a time as it travels through the tunnel. Bus operations involve several vehicles, arriving and departing at various times. This has the potential to occupy multiple sections of track simultaneously, delaying service behind it. Providing a separation from day one will avoid this situation.
  • DSTT can serve both buses and streetcars: DSTT rail infrastructure could be converted to serve streetcars. Joint operations for streetcars and buses would be more compatible than light rail and buses, as their operation philosophy is mostly similar (short vehicles that can operate in mixed traffic without dedicated signalling systems). Some modifications will have to be made. This includes extending the width of the platforms by about 3 inches to accommodate the narrower streetcars, but the floor height is roughly the same (355 mm for Link and 350 for Seattle’s streetcars). Streetcars will also have to be capable of running at 1500 V DC in the tunnel and 750 V DC on the surface, although dual-mode operations are not uncommon elsewhere in the world.

Of course, no proposal is without drawbacks:

  • The most obvious drawback is that Central Link and East Link will skip Westlake Station, which is necessary to provide the large-radius curve between University and Capitol Hill. The good news is that the DSTT will still serve Westlake, and a future light rail extension to Ballard could reintroduce light rail service beneath the current Westlake platforms. Skipping the station also improves travel times for the regional-oriented Central and East Link lines.
  • Construction impacts when service is transitioned to the new tunnel
  • Bus and streetcar services in the DSTT depends on the fate of the Convention Place Station, and whether access to the DSTT will be preserved after its expansion. If not, a new access point will have to be built.

Comments? Ideas? Share your thoughts below!

Transit objectives should be defined before its infrastructure

As Seattle’s transit advocates, we often like to brainstorm about transit infrastructure because it generates discussions. It helps define what the region wants, and these discussions often drive politics. Light rail connections, BRT, and a second downtown tunnel are just some highlights of an ever-growing wish-list.

However, our focus on modes and infrastructure also leads us to overlook the actual objectives of these transit investments. We ask ourselves whether we want light rail and BRT, but rarely do we emphasize, “How quick and reliable should the system to be?” or “What is this system trying to accomplish?”

As a result, transit advocates are often surprised by operational deficiencies late in the process, leading to reactions like this:

We have supported RapidRide and BRT from the beginning but Metro and the Council have let “BRT creep” and politics take over, not what is best for riders. When RapidRide C and D lines open on October 1st we’ll have a glorified shiny new bus that is slower than existing service. – STB

Part of the reason is because specific objectives are not clearly defined before transit is built. There were never defined travel times or reliability requirements from the beginning. We simply said “Build BRT” and assumed that everything would work out. As a result, decision makers have leeway to push for more infill stations to satisfy a few constituents, because why not? There was no legal requirement for what the actual travel time between A and B should have been. It’s easy to backtrack and modify service objectives that were never well-defined in the first place.

On the other hand, if we had defined our objectives as, “The BRT system shall travel between Aurora Village and Downtown Seattle in 35 minutes, at 90% punctuality within 10 minutes of the scheduled time”, and made them project requirements from the beginning, the results may have been different.

I can’t speak for all engineers and planners, but I would suspect that most (myself included) would not only prefer clear objectives to accomplish, but also more defense against backtracking due to political pressure. If decision makers want to satisfy a few constituents with an infill station, clearly-defined project requirements would provide more leverage to say, “Adding this infill station would breach the project conditions, unless there was more funding for a bus-only lane.”

I’ve been using BRT examples, but this applies to rail as well. If we had defined that we wanted a regional rail network with a 40-minute connection between Everett and Seattle, we could say that the currently proposed half hour travel time between Lynnwood and Everett is far too slow. However, we specified little more than the fact that we wanted a train. In the end, we got light rail, which takes 7 minutes to travel 1.3 miles through Downtown Seattle and is proposed to spend an hour between Westlake and Everett (28 minutes to Lynnwood and 30 more to Everett).

This pattern needs to change, and we can start by modifying our planning process: Define specific and realistic service objectives, reach a consensus with agencies, and make the objectives part of the project requirements.

Including specific preconditions into the discussion process, enforcing them and defending them against negative political interests will make the outcomes of our transit investments more predictable. After all, transit should be built to achieve mobility goals, not just for the sake of building it. Let’s take a look at how we can add these objectives to our discussions.

A few examples for defining objectives

Service objectives should be defined before the infrastructure. These should be specific, measurable and realistic, rather than general descriptions such as “improve reliability”.

Here are examples of what specific objectives could look like (with fictitious numbers):

  • Travel times (e.g. West Seattle to Ballard within 30 minutes)
  • Reliability (e.g. 95% punctuality between West Seattle and Ballard, where a vehicle is on-time when it arrives within 5 minutes of scheduled time)
  • Frequency (e.g. Provide the capability to run trains at 90-second frequencies)

By specifying these objectives in our discussions, we can begin a dialogue with agencies to determine whether they are realistic. Once a consensus has been reached and the objectives determined, there should be a method to enforce these objectives, giving agencies more leverage to ask for funding or defend against political interests. Service objectives would then determine the infrastructure, such as whether surface options would even be worth considering.

We would then hopefully be presented with several alternatives that, more or less, satisfy our objectives, rather than receiving a smorgasbord of varying options with questionable operational effectiveness and then collectively wailing. In the end, we may also save ourselves from wondering what the infrastructure is actually capable of.

Moving forward with our discussions

Setting clear quantifiable objectives in cooperation with our agencies will help them better determine what people really want from transit investments. Providing means to enforce these objectives will give agencies more leverage to push back against political influence. With potentially billions in transit investments around the corner and discussions about to begin, we need to refine how we plan transit. After all, why vote for and throw money at projects without fully knowing what they’re trying to accomplish?

As we move forward, discuss and define the travel times that regional rail or urban rail should provide, not whether we should build light rail. Discuss and define the reliability that BRT should provide, not whether we should have bus lanes. The infrastructure (and funding) should be decided only after it is known what will be accomplished. By setting service objectives earlier in the process with agencies and turning those objectives into requirements (this is worth its own discussion), we may actually get the system that we expected to get.

Technical issues facing light rail as a regional service

A few days ago, I posted on why light rail vehicles may not contain the performance necessary for regional transport (i.e. services that bridge suburban rail and low-end intercity rail).

A brief overview of technical limitations was presented, but in order to keep the article short, the technical issues were simplified and details were omitted. There were a lot of comments and questions regarding those missing technical pieces.

If you’re the type who enjoys a discussion about some technical challenges unique to light rail, here’s your article. We’ll discuss design characteristics that separates light rail from conventional heavy rail, and touch a bit on why higher vehicle performance is more complicated than just asking the vehicle manufacturer to tweak their trains for a higher design speed.

Operational conditions of light rail

Light rail vehicles are built for flexibility, literally. Low-floor vehicles provide accessibility benefits and double-articulated bodies allow the train to make sharp turns, letting designers use tighter curves on track alignments. This is useful for street-running, in cases where the alignment is constrained to the roadway grid (e.g. between Westlake Station and University St Station) or in areas where land procurement can be avoided (between Beacon Hill Station and Sodo Station). It offers additional leeway for the infrastructure designer and provides some cost savings.

In order to provide this flexibility, traditional light rail vehicles like those used by Link contain unique design characteristics. These include:

  1. Coned wheel profiles, more so than many modern heavy rail vehicles
  2. Double-articulated, 70% low-floor bodies supported by a low-floor center trailer with independently-rotating wheels (IRW)

These characteristics present challenges when operating the vehicles at higher speeds. Yes, the effects can be mitigated. Yes, some light rail vehicles are capable of higher speeds. However, they require additional mitigation from both the vehicle manufacturer and the infrastructure provider.

Wheel profiles and wheel-rail interaction

Many factors influence the type of wheel profile that an operator may choose to use on their trains. These include the curvature of the infrastructure, the design speed and the type of rail. Generally, for light rail applications, coned wheels are used in order to navigate tight curves. On Link, the wheel profile is a 1:20 taper.

Coned wheels connected by a rigid axle allow rail vehicles to steer around curves by creating a differential in rolling radius. When in a curve, the inside wheel rotates with a smaller radius, and travels a shorter distance, than the outside wheel. This differential reduces friction and flange contact, reducing wear on the wheels and the rails.

Coned wheels also provide lateral forces that keep the rail vehicle centered along the railway and prevents flange contact, reducing both wear and the risk of derailment. If the train shifts to the right, the rolling radius difference steers the train back to the left, then back to the right, etc. The side-to-side movement is eventually damped and the train will become centered again. However, it only does so below its intended design speed.

The lateral forces of coned wheels increases with speed. Above a “critical speed”, the higher lateral forces begin to overcompensate when the train becomes off-center from the tracks. The damping effect that allowed the train to become stable at lower speeds is overcome by the higher lateral forces. This leads to a continuous side-to-side motion called hunting oscillation, which if not mitigated, leads to flange contact, excessive wear and even derailment.

As a result, many trains that operate at speeds higher than light rail vehicles don’t use strictly-coned wheels. While Link vehicles use a 1:20 taper, a German standard wheel profile DIN 25112 Type C used for conventional rail contains a 1:40 taper for most of the contact surface, with 1:20 only towards the outer edge. At higher speeds, this reduces lateral forces, but allows for stabilization when lateral movement is extreme.

However, the situation becomes even more complex. The wheel profile must also be accommodated by the shape of the rail profile themselves. After all, it’s the interaction between the rail and the wheel that determine vehicle dynamics. It is in fact possible to use rail profiles that support higher running speeds with 1:20 wheel tapers. On the other hand, a flatter wheel could also show excessive wear on un-optimized rail profiles.

Amtrak’s high-speed Acela trains, which operated on tracks with more curvature than other high speed systems, showed increased flange wear as a result of wheel profiles that were not optimized for the rail profiles.

The takeaway here is that generally, coned wheels are less stable at high speeds, but railway profiles can be modified to achieve a conicity that allows for higher speeds and lower wear. The trick is to make sure that new infrastructure is designed for the rail vehicle(s) using it, and that existing infrastructure is modified (if necessary) before new vehicles enter service. If high performance is required, then it has to be designed into not just the structure and alignment, but also the wheel-rail interface.

Independently rotating wheels (IRW)

Light rail vehicles like Link contain double-articulation using a short center trailer. To achieve a low-floor design in the middle of the train, the axle is removed, allowing the wheels to turn independently from one another. The combination of short center trailers and independent wheels has a few implications. We’ll discuss a few from this study sponsored by the Federal Transit Administration.

Without an axle, the self-steering mechanism that stabilizes the vehicle (in this case the center trailer) is removed. Coned or not, the wheels are free to rotate independently of one another, which means it no longer self-steers towards the center of the track.

As a result, flange contact becomes more common with IRW systems. In some situations, such as tight curves, turnouts or even when travelling at higher speeds on straight tracks with unevenness, the trailer may be guided by the flanges of the wheels rather than the rolling radius differential. This contact increases friction, wear and sometimes noise.

NJ Transit reported “higher wheel wear on the center trucks of its LFLRV fleet than on the wheels of the motor trucks” and “very high rail wear on sharp curves.” Portland Trimet came to a similar conclusion, reporting that “Portland MAX has experienced higher LFLRV wheel flange wear on the center truck than on the motor trucks.”

At its worst, the forces generated through this increased flange contact may be high enough to let the wheel climb over the tracks and cause a derailment, and the report notes:

Trucks with IRW center trucks are, therefore, fundamentally more susceptible to derailment and, as a result, their behavior can be strongly influenced by other factors, which would normally be of only secondary importance for trucks with solid axles.

As with any engineering issue, there are methods for mitigation, including modification of infrastructure for better wheel-rail interfaces and maintenance. Wear and noise can be mitigated using several methods, including lubrication, which Sound Transit has done. Additional protection through curves could also be provided by restraining rails. Operationally, the report also suggests that curves should be taken at constant speeds, with acceleration only after the curve has passed. Finally, improved vehicle and infrastructure design can also reduce these negative effects. Manufacturers and operators like Sound Transit have surely learned much more since this report was presented (2006).

However, the need for these mitigation measures demonstrate that for light rail, “track standards have to be tighter than might be acceptable with more conventional vehicles. It is also generally recognized that the management of the wheel/rail interface is even more critical.”

Generally, this means that it is more difficult for light rail to achieve the same performance (speed, acceleration, curving capabilities, etc.) compared to a conventional train.

By the way, some of you have mentioned other rolling stock that use IRW systems, yet operate with high speeds, such as the Alstom Citadis and WSDOT’s own Talgo trainsets. Alstom’s Citadis is an example of a 100% low-floor tram that can operate at high speeds, but the wheels are connected by a low-level driveshaft linking the wheels together, which do not make it a true IRW system. Talgo’s trainsets use a passive steering mechanism to overcome the lack of rolling radius differentials supported by an axle.

Implications for Link

So how does this apply to Link? The purpose of this document is to demonstrate the complexities that result from designs specific to light rail and to show that it’s not simply just a vehicle on rails that happens to bend well.

My previous article suggested that light rail may not be the best solution for regional services. On a 60-mile regional line, vehicle performance (speed and acceleration) influences not only travel times, but also travel time reliability.

Many comments pointed out that there are light rail vehicles available to provide higher performance necessary for regional transport. For Sound Transit to order new light rail vehicles with greater performance at this time may turn out to be an extremely complicated process.

First, it would have to determine what kind of performance increases would both reduce travel time and improve reliability. Next, it would need to work with manufacturers to determine if a vehicle is available to provide these needs. Then, they must together determine whether the new vehicle or the track infrastructure requires modifications. Increasing the speed changes dynamic forces, and dynamic forces require mitigation. Finally, if the changes to the infrastructure are necessary (rail grinding, wheel profile modifications, etc.), they will have to be applied and not at a low cost.

Operational requirements should dictate the type of vehicle that is used. The infrastructure is then built to support the vehicle.

In Seattle, low-floor light rail has been selected to provide flexibility in city centers. The infrastructure and the vehicles have been designed to operate below freeway speeds, much slower than cars and buses except during congestion. What our light rail system lacks in performance though, it makes up for in urban flexibility. Urban areas is where its capability shines.

Perhaps Sound Transit is willing to figure this out. They may be willing to work with manufacturers and procure higher-performance light rail vehicles that fit our infrastructure. Or maybe they’ve already considered all of this and the tracks are ready to support the next-gen Link train. That’s something we can’t say yet, but it’s certainly worth having a discussion about, especially when a chunk of a $15 billion package is potentially on the line (pun intended).

For now, however, by pushing our light rail system as it is to operate regional services between Tacoma and Everett, it may not offer the type of performance that regional services need and deserve.

Limitations of Light Rail as Regional Transport (Part 1)

Deutsche S-Bahn

With light rail expansion and planning well on its way, things look positive for rail transport in the Puget Sound Region. The momentum and demand for rail-based transport in the region appears higher than ever, with residents beginning to realize that our current transport network is simply inadequate for the growth rate in this region. When a single fish truck can bring the region to hours of standstill, transport alternatives cannot come soon enough.

However, as with every major project, there is always a time to step back and once again look at the big picture. What type of transport objectives are we trying to accomplish? What kind of connections and services do we need?

But here is the biggest question that we need to answer before Sound Transit 3: What exactly are we building right now?

The simple answer is, of course, light rail. The more complicated answer is that we are building a regional light rail network.

And that could be a problem, because light rail vehicle technology is not intended for regional services. If Sound Transit pushes these vehicles to compete with cars between Everett and Seattle, or Tacoma and Seattle, it will have to find a delicate compromise between competitive travel times and travel time reliability. Let’s discuss why.

Vehicle performance matters in a regional context

Continue reading “Limitations of Light Rail as Regional Transport (Part 1)”

Limitations of light rail as regional transport (Part 2)

In part 1, we discussed how the limitations of our light rail system may impact its performance as a regional service. Now, we will highlight some potential issues within cities that we may face when our light rail capacity is adapted to serve high-demand suburban services.

Light rail’s need to provide high capacity and reliable mobility for the region is supported by its infrastructure. When the infrastructure means “surface running”, such as on MLK, high capacity (i.e. longer trains) is accommodated by large stations and reliability is provided by long stretches of track with uninterrupted travel. Although this means that mobility to and from the neighborhood is improved (regional mobility), accessibility within the neighborhood may be compromised.

Mobility vs accessibility

Transport engineers like to focus on mobility, which characterizes the movement of people or goods. It is a means to an end, in which mobility is required for a person to perform an activity that takes place somewhere else.

On the other hand, land use planners like to focus on accessibility. This has several definitions, but in the context of urban design, it defines the ease at which activities can be performed.

Let’s take a look at Jarrett Walker’s definition:

“Access is how many useful or valuable things you can do.”

And let’s take a look at Litman’s:

“Accessibility (or just access) refers to the ability to reach desired goods, services, activities and destinations (collectively called opportunities).”

Now there’s an important difference between mobility and accessibility. While accessibility focuses on the ability to perform activities, mobility focuses on the means to get there. If an activity is closer, it requires less mobility, and thus the accessibility may be higher. Jarrett Walker also notes this:

“If a new grocery store opens near your house, that doesn’t improve your mobility but it does improve your access.  You can now get your groceries closer to home, so you don’t need as much mobility as you did before…A lot of the work of access is simply about eliminating the need to move your body around the city in order to complete the economic and personal transactions that make up a happy life.”

The ability to perform activities within a neighborhood (accessibility), is fundamental to urban design. How can residents in a neighborhood perform their daily tasks as easily as possible, without the need to travel long distances and without the need for motorized transport?

This is achieved using various ways, among them, higher density developments, narrower streets and more pedestrian access. The solutions are up to the planners, but they often hint towards some kind of compactness in the neighborhood structure that shortens trips and allows activities to be brought closer together. The infrastructure within the neighborhood has to support that.

Limitations to accessibility presented by high-capacity light rail operations

While Link will bring mobility to surface segments (i.e. Rainier Valley) as well as some Transit-Oriented Development, its design places a limit on accessibility when it runs on the surface. It can take people farther away from the neighborhood, and it can generate development around the station, but it doesn’t allow trips to be as compact as possible within the neighborhood. This is notable in surface running applications such as on MLK.

One of the touted capabilities of our light rail system is its capacity, not because of its frequency, but because of its ability to run 4-car trains. This is necessary to support demand elsewhere in the network, but not necessarily in the Rainier Valley.

In grade-separated operations, the ability to run 4-car trains is a plus. However, on surface streets such as MLK or the planned Bel-Red corridor, this may become an issue for accessibility.

A 4-car train totals just over 380 feet (116 m) and the surface stations (such as those on MLK) built to support them meet or exceed 600 feet when ramps and access points are taken into account. Our light rail stations surpass the length of a typical downtown Seattle block length and conflict with components of accessibility. These stations are fenced off on both sides, and prevents access from any point other than the ends of the blocks they sit on. The infrastructure needed to accommodate the train’s sheer size represents a large barrier in the middle of MLK and enforces the distance residents must travel to get to “the other side”. The train itself may be accessible, but a trip starting from mid-block and ending on the opposing mid-block is separated by more than 600 feet, even though it is physically not more than 100. It removes connections across MLK, connections that will become even more important when TOD is built around the stations.

And then there is the long distances between intersections. Between S Orcas St and S Graham St, there is only one pedestrian crossing over a span of nearly 2000 feet.

These compromises are the direct result of Link’s attempt to serve both the region (through a 60-mile corridor) and urban neighborhoods (street running). It needs the reliability provided by uninterrupted movement through the corridor, but on street-level, that requires minimal intersections and pedestrian crossings. It needs the long platforms that can handle 4-car trains to meet demands elsewhere in the network, but occupies large footprints in what is to become the center of TOD. Where Link strives to provide mobility for the region, it does so at the expense of neighborhood accessibility.

If surface rail is aimed at complementing mobility, it should be scaled to support it. But in this case, Link is scaled for the need to provide mobility out of the neighborhood, rather than complementing the neighborhood in which it operates.

So what exactly are we building?

This is a question we need to answer, even if it may already be too late. Are we building a light rail system that satisfies the mobility needs of Seattle and its immediate suburbs? If so, our sizable trains that try to meet regional needs may negate the accessibility benefits when it runs on the surface. It also doesn’t serve enough of Seattle to satisfy urban mobility needs and instead, ventures out into the suburbs with a stop every 2 miles.

Does that mean we are building a regional network to connect cities in the Puget Sound region quickly and reliably? That too is questionable, because in that case, we are pushing the very concept of light rail beyond its intended use, where it may not keep up with the performance demands of regional services.

Urban mobility and regional connections are two completely different concepts that each require their unique solutions. Urban mobility is provided by an urban or suburban network with high capacity, frequency and accessibility to facilitate short trips within an urban area. Regional mobility is provided by high-performance trains that can travel between multiple cities quickly and reliably. These two concepts conflict in many different ways.

Unfortunately, it appears that we are trying to serve urban, suburban and regional needs with one mode, and in the end we may get something that serves none of them particularly well.