Wednesday, September 30, 2009

How Fast is Fast?

In my first rail-related post I defined three different terms: high-speed rail (HSR), medium-speed rail (MSR), and standard-speed rail (SSR). I've been reviewing the Federal Railroad Administration's rules about train speeds, and I now see that I need to add another definition.

There are two criteria that the FRA uses to set limits: quality of the track, and type of signaling system. For the track, there are 10 classes each of which has increasingly strict tolerance, maintenance, and inspection requirements. For signaling, there are three levels. Below 60mph (49mph for freight), trains can operate in "dark territory." There are no signals on such lines, and train movements are authorized either by the almost obsolete method of pre-written train orders, or by track warrants received by radio from a dispatcher. For trains to operate above 60mph, a railway line must have some kind of electronic trackside signaling. This is usually some form of colored indicator lights, though in the past it also included mechanical semaphores. The signals can operate automatically based on how they are wired, or they can be controlled from a central location, or both. Above 80mph, the rail line and the trains operating on it above 80mph must have some kind of system that will automatically protect a train (usually by stopping it) under certain conditions. Unfortunately, how the protection is implemented and under what circumstances varies quite a bit between protection systems, of which there are several in the United States alone. But they usually combine in-cab signaling and a way of applying the train brakes without assistance from the train operator. In-cab signaling is present because the status of a signal becomes harder to determine at higher speeds, and even a brief moment of distraction can cause an engineer to miss a signal. In most instances the cab signaling is a supplement to trackside signals. However, on dedicated TGV lines SNCF has dispensed with all external indicators, as there is no need to support legacy locomotives without cab signaling.

Back to my definitions. My SSR standard would require Class 4 track and trackside signaling. Both are what can be found on most major rail lines, like BNSF's TransCon. Amtrak currently operates on some Class 3 and even Class 2 track, which is really somewhat pointless. The average speed for a train running on low-quality track is well below what can be achieved in a car, which most people would opt for. Even if all the bright new plans for HSR networks come to naught, Amtrak should make an effort to upgrade all of its routes to SSR standards.

The next level I defined was MSR, which topped out at 110mph. This corresponds to FRA Class 6 track equipped with automatic train protection. And 110mph just happens to be the maximum speed of Amtrak's two intercity diesel locomotives, the F59PHI and the P42DC. Various bits of track maintained and operated by Amtrak meet the Class 6 standards, as does a very limited amount of track owned by various state rail authorities. In the immediate future, most passenger rail projects will be building to MSR standards.

Finally, I defined HSR as trains operating over 150mph. That means I left a big gap between 110 and 150 mph. Amtrak's operations on Northeast Corridor fall right into that gap. I'm going to call a new standard for that range of speed ISR, where "I" stands for intermediate, or improved, or impressive, or maybe something else that sounds catchier. This standard corresponds to Class 8 track, again with automatic train protection. The reason the ISR standard is set at 10mph less than FRA allows on Class 8 track is because the lower speed is what Amtrak has determined as safe for its equipment and track. "True" HSR rail lines would have to conform to Class 9 standards. No track of this quality has been built in the US to date.

One more thing about the ISR standard: it is - or should be - irrelevant. Away from the Northeast Corridor and the Keystone Corridor, there aren't many existing right-of-ways where triple- or quadruple-tracking is possible. If a passenger-only right-of-way is being built out, it should be done to HSR standards so as to get maximum benefits. For track shared with freight traffic, upgrading beyond the MSR standard is pointless because the much heavier freight trains will throw the track too far out of alignment for higher speeds to be safe. Curve radii are likely to be too small for high speed operation on existing lines, as most were laid out by the 1890s, when expected speeds were much lower.

Wednesday, September 23, 2009

Goin' Down South

Despite the frustration I experience while planning a HSR line across the Bay State, I've attempted to tackle the Big Kahuna: the southern Northeast Corridor. This 225 mile line is the busiest in the country. Naturally, it has many important stations: New York (number 1 busiest in the country), Washington (2), Philadelphia (3), Baltimore (8), Wilmington (11), Newark (13), BWI Airport (15), and Trenton (24). Amtrak's Acela Express trains stop at each of those stations (though not all individual trains do) plus Metropark in New Jersey. Amtrak's Northeast Regional trains stop at an additional 8 stations, and the various commuter rail operations serve at least 40 more. The line has a very long history: part of the line runs over the oldest operating railroad bridge in the world, built in 1836. The most recent major change to the line is the electrification between Washington and Baltimore, which was completed in the late 1930's.

Theoretically, Acela trainsets can reach 165 miles per hour, but due to a number of constraints the average speed for a trip from DC to NYC is just 78mph. The trip currently takes 2hr 47min with 4 intermediate stops, and 2hr 52min with 6 intermediate stops. The magic number people want to achieve for DC-NYC travel time is 2hr 30min, which would mean an average speed of 90mph. To cut the journey by a further 30min to 2hr, the average speed would have to reach 112mph, a 40% increase over the current figure.

I experimented with a few sections of entirely new track, which turned out to be pointless. Building a 25 mile stretch of HSR track between Washington and Baltimore would cut travel time by only 10 minutes. That doesn't seem like a good way to spend $1.25B, which is what the new segment might cost. And trains would still have to crawl through the Baltimore and Potomac Tunnel. Creating a new line east of the Delaware was the only way to lay new track north of Wilmington. It would be very expensive and somewhat useless because it would bypass so many busy stations. Elsewhere, sprawl made a new line completely impossible.

The inability to plan a new line meant that incremental improvements would have to be the means for reducing travel time. Amtrak and its supporters have been arguing for more capital investment in existing NEC infrastructure for decades, but the funds allocated by Congress have been minimal. Below are the improvements I would suggest, in rough priority, based on reading a number of reports, reviewing track diagrams, and eyeballing aerial photographs. The four cost levels correspond to guestimates of <250M, $250M-$1B, $1-3B, and >$3B;.

  1. West Baltimore tunnels - The B&P Tunnel for passengers and the Howard Street Tunnel for freight are the biggest choke point in the entire corridor. They both need to be replaced for reasons that are very well explained here and here. Cost: high. Benefit: high.
  2. Catenary - The power system south of NYC is unique, and not in a good way. It supplies 11kV, 25Hz AC electricity, which is used nowhere else in the world. The configuration of the wires themselves limits top speeds to 135mph. The entire system between NYC and DC needs to be replaced for those two reasons, and general age. Cost: moderate. Benefit: high.
  3. Signaling - The current signaling system is sufficient, but higher densities (meaning shorter blocks) on the entire line would increase capacity. Cost: moderate. Benefit: moderate. Update 9/26: Amtrak has already installed a Positive Train Control system on its equipment, and plans to have all commuter and freight rail upgraded in the next few years.
  4. New Jersey flyovers - Northbound commuter trains entering the corridor from two branch lines in central New Jersey (one is just a stub, really) have to cross the two southbound tracks and the northbound express track before continuing. This creates safety and scheduling issues that should be eliminated. Cost: low. Benefit: low.
  5. Curve straightening - There are a number of areas where curves could be eliminated. In most locations, doing so would reduce travel time a small amount (15-30 seconds or so) and at the same time would increase passenger comfort and eliminate higher wear curved track. Cost: low (each). Benefit: low (each).
  6. West Delaware track and curve - From Wilmington south to Newark there is a section of the line with only three tracks. Newark is the southern terminus of the SEPTA commuter rail system, so adding track would reduce scheduling conflicts. There is also a curve on the section that could be eliminated. This work could be done in conjunction with the bypass below. Cost: low. Benefit: low.
  7. East Baltimore tunnels - The two relatively short bores of the Union Tunnel east of Baltimore's Penn Station have a total of only three tracks, and need to be rebuilt due to age and height issues. Cost: high. Benefit: low.
  8. North New Jersey bridges - Two movable bridges just north of Newark need to replaced. Cost: high. Benefit: moderate.
  9. Northern Maryland bridges and track - Between Baltimore and the Delaware state line there are three two-track bridges in an area with only three tracks. New bridges need to be added for capacity reasons, and the originals eventually replaced, and another track added to help fully utilize the new capacity. Cost: high. Benefit: moderate.
  10. Central Maryland track additions - Most of the segment between Baltimore and DC has only three tracks. Another track would add capacity. Cost: moderate. Benefit: low.
  11. Wilmington bypass - The Wilmington station is in a very constrained area, has only three tracks, and is bounded by sharp curves at both ends. The solution here would be to bypass downtown Wilmington with two high-speed tracks. The existing track would remain, but through traffic would zoom on by. Cost: high. Benefit: low.
  12. North Jersey track additions - A short section of track from just south of Elizabeth north to Newark Airport narrows down from 6 tracks to 4. More tracks would help scheduling in a busy area, but the construction would be difficult due to the surrounding area of dense urban development. Two curves could be eliminated as well. Cost: moderate. Benefit: low.
  13. Northern Jersey viaduct - This 10 mile long structure from south of the Newark airport to near the Hudson tunnels entrances would bypass three stations (Secaucus Junction, Newark Penn, and Newark Airport), two movable bridges, and a rail yard. Cost: extreme. Benefit: low.
  14. 30th Street Station approaches - The track and switches on either side of the main Philadelphia station are a mess. Consensus opinion is that the situation needs to be addressed, but I have no idea about how large of an undertaking it would be. Cost: ? Benefit: ?

I think completing items 1, 2, 5, and 6 would shave 15 minutes off the trip. Items 3, 4, and 10 might save some time, but most likely they would just allow for more trains to be scheduled. But increased service frequency is a worthwhile goal, too, especially for the commuter rail systems. Items 8 and 9 might save 2 minute each, but the main purpose for undertaking them would be end-of-life replacement of the existing infrastructure. Item 7 is another project that is important due to age, though the time benefit would be negligible.

Completing tasks 1-10 would bring travel time to 2hr 25min. Not bad. Amtrak experimented with a limited service Acela train, which had only one intermediate stop at Philadelphia. The journey time on the offering was 10 minutes faster than regular Acela service. Completing item 11 would allow for another intermediate stop at Baltimore without changing the travel time. The rest of the projects - items 12, 13, and 14 - might wring a few more minutes out of the journey time, but 2hrs 15min for a limited service train would be pretty good. Northeast Regional trains would benefit, too, though the amount would be less because the catenary upgrades wouldn't change their top operating speed.

(Updated 9/28) How could I have not posted these links at the beginning?

Tuesday, September 22, 2009

I, For One, Will Welcome Our New French Overlords

According to the Transport Politic, SNCF has submitted proposals for several of the FRA designated HSR corridors. This is an interesting development because the French definitely know a thing or two about high-speed rail. However, even if the inevitable knee-jerk dismissals were surmounted and SNCF won a bid, cultural differences would make the implementation a disaster. Entirely stupid, but that's just the way things are. The French are more likely to win a contract in Canada, which is once again mulling HSR for the Windsor-Québec corridor.

Patterns on the Land

I've created two more maps of rail lines: one for New Hampshire and one for Southern Québec. I find it interesting how different the land-use patterns are between the two areas. Note that I'm working off a variety of old maps, some of which are merely diagrammatic. Thus the exact alignment of some abandoned lines is unknown, and I've certainly left out items.

As a side note, the mapping interface is pretty buggy (at least in my browser). When the number of items on the left side exceeds one page, the list starts doing weird things. The behavior makes sorting the items nicely a real pain in the arse. Hopefully Google will fix this soon.

Update 9/27: I found a very nice repository of old Canadian topo maps, so I'm now pretty confident the routes I've traced are accurate.

Saturday, September 19, 2009

Baked Bean Breakdown

Another day, another high-speed rail analysis. Designing a line for Massachusetts was difficult for two reasons. In the eastern third of the state, sprawl is almost continuous. In the western half, there is again sprawl around Springfield, and then an area of very jumbled geography. None of the mountains are particularly high, but at the same time there are no large, level valleys that can be used to approach the highest peaks. Anyway, in the eastern half or so, the best solution was to parallel I-90. Following the current CSX alignment wasn't possible because it runs through the center of so many towns. In the west, because of the built environment west of downtown Springfield, a detour to the south was necessary. After that, it was a matter of finding the least bad route until the rail line could meet up with I-90 again. To keep speeds up, several tunnels and viaducts were needed.

I think the main takeaway from this exercise is how restrictive the sprawl around American cities can be on new transportation construction. Most existing rail lines were first laid out in the 1800s, and the Interstate Highway System was planned in the early 50s, when the population was half of what it is now, and much less suburban. Projects planned now, unlike those other two, have to deal with a built environment that is fully built up within a 20 mile radius of large city centers. And large city centers are where high speed rail lines need to go.

Tuesday, September 15, 2009

Capital of the Empire

Truth be told, my exercise in planning a high-speed rail station in Burlington was not very challenging. It is a small city in a very rural state, so I found it easy to get a line close to downtown without having to plow under hundreds of homes. So I decided to do another analysis

This time I chose the Albany-Schenectady-Troy, NY Metropolitan Statistical Area, a.k.a. the Capital District. It is a much more populous area than the Burlington, VT MSA, being home to over 1.2 million people by the Census Bureau's estimates. And it is a major rail hub, with seven main lines radiating from the area. Albany is the top of the "Selkirk Hurdle," which is shorthand the detour trains must make due to the lack of freight rail connections across the Hudson below the city. Any freight destined for New York City or southern Connecticut has to either cross the river (which is really a long, narrow fjord below Troy) on CSX's Alfred H. Smith Memorial Bridge just below Albany, or be transferred to truck, usually at the Port of New Jersey, and driven into the city. A freight rail tunnel from New Jersey to Queens is being planned, but its opening will be many years in the future.

Before I started laying out track I created a number of requirements. One was that there would be four HSR lines converging on the area, one from each of New York City, Boston, Buffalo, and Montreal. However, only four of the six possible city pairs needed to be easily completed (BOS-BFX, BOS-MTL, NYP-BFX, NYP-MNT) as there would be better routes for the other two (BOS-NYP, BFX-MNT). Another was that the station(s) should be as close to downtown(s) as possible. A third was that the considerable freight traffic in the region not be permanently impaired. And of course minimizing destruction and disruption would be nice.

It turns out keeping the freight moving wasn't so hard for the five lines south of the city. But getting freight arriving from the north on CP or from the east on PAR to the five southern lines, along with the reverse, was much more problematic. Currently, some of it passes through Albany along the waterfront, but most of it goes through downtown Schenectady. Building a new line to connect the south end of CP's Mohawk Yard to CP's line south to Binghamton was key to untangling the knot of lines in Schenectady. I located this line on the north/west bank of the Mohawk, but a crossing further west could accomplish the same end.

With the Schenectady choke point gone, the rest of the pieces fell into place pretty easily. The most disruptive element would be the new station in Albany. The current station across the river in Rensselaer, while not very far physically, cannot be accessed from Albany on foot. A pedestrian bridge could be built over the Hudson, but I think most people would be uncomfortable using it at night. The station would also have to be reconfigured extensively to accommodate the much longer HSR trains and an extra set of tracks. The original Albany Union Station is in a great location, but the ROW has been mostly obliterated by the waterfront highway, so it can't be reopened. An alternative would be to excavate a tunnel, but that would be very expensive and disruptive. Plus, it could only be accessed by electricity-powered trains. I thought the best solution would be build a new station on the Albany side of the river in an area that consists mostly of warehouses and small businesses. I am not a fan of "urban renewal" projects that destroy acres of existing city fabric, but at least this project would allow expansion of the downtown area northwards once the existing lines were removed.

As for the rest, hopefully the map explains itself. Enjoy!

Tuesday, September 8, 2009

Just Passing Through

I was looking at my HSR network again and started wondering about having the NYC-Montreal line pass through Burlington. Here is a quick analysis of the area. I'll have more about it later.

Update: My mapping has succumbed to feature creep.

I've developed four "alignments" for where a HSR line might be run near Burlington. From west to east they are:

  1. Downtown Alignment - Of course, my first instinct was to put the line through downtown. It turned out to be the worst choice by far. It would cut the city off completely from it's small but developing waterfront, there would be no room for parking, it would create a lot of traffic through the city, it would have to share track with the local shortline freight railroad, and there would be no room for run-through tracks. The only upside is that downtown (which is up the hill from the station) would be within easy walking distance.
  2. UVM Alignment - Running the line near I-89 might be the best choice. The station would have good highway access, it would create little additional traffic through residential neighborhoods, it would be within walking distance of UVM, and it would be closest to the major hotels clustered along Williston Road. The downside is that it would be the most disruptive to existing buildings, and it would displace UVM's working farm. It might also be too close to Fletcher Allen, disrupting patients with noise and vibration, and it would not be within easy walking distance of downtown.
  3. Airport Alignment - Running the line near the Burlington International Airport would be the least disruptive to the existing built environment. Passengers would be close to currently operating rental car agencies, but not much else.
  4. Suburban Alignment - Locating the line and station out in the 'burbs (such as they are in Vermont) would put tourists a few miles closer to the ski resorts, but at the same time draw development well away from downtown. There would be plenty of room for parking.

Of those 4 alignments, my choices, in order, are: 2, 3, 4, 1. I wish that the downtown alignment could work, but it wouldn't. It would be so damaging that I'd rather see the line on the other side of the lake.

Just for kicks, I extended the downtown and suburban alignments to the Canadian border and all the way to Schenectady. The placement of each is strictly guesswork; I did it mostly to get a sense of what kind of disruption a line would bring to the Vermont countryside. I also populated the map with lots of other tidbits that only a railfan would find interesting. I'll probably move a lot of these to a separate map at some point.

Update: I've cleaned up the Burlington map. See the post above for railfan stuff. I've also made a high level map of the Federal High-Speed Rail Corridors for reference.

Sunday, September 6, 2009

Need for Speed

I can't stop myself. Srsly.

After reviewing my first try at designing a HSR network, I realized I was somewhat over-eager in drawing lines. I decided to represent the top cities visually instead of working off a list.

Cities are colored as follows:

  • Red - top 20 metro areas, 37% of the population, cutoff at 2.67m
  • Cyan - next 20 metro areas, 50% cumulative total of the population, cutoff at 1.32m
  • Green - next 20 metro areas, 56% cumulative total of the population, cutoff at 830k
  • Magenta - next 40 metro areas, 65% cumulative total of the population, cutoff at 500k
  • Brown - next 100 metro areas, 75% cumulative total of the population, cutoff at 200k
  • Orange - next 82 metro areas, 80% cumulative total of the population, cutoff at 132k
  • Yellow stars - state capitals that are not top 100 cities
  • Cities in Mexico and Canada are classified according to where they would fit into the American rankings.

Rail lines are colored as follows:

  • Red - core national lines, track built to full HSR standards (180mph)
  • Cyan - important national lines, track to HSR standards wherever possible, MSR standards (110mph) otherwise (changed from blue)
  • Green - regionally important lines, track to MSR standards wherever possible, SSR standards (79mph) otherwise
  • Magenta - spur or infill lines, track to MSR or SSR standards

This map shows only the top 100 cities and the national lines. Most of the changes are in the Southeast, where one core line was relocated and several national lines were downgraded to the the third tier.

Here is a closeup of the Northwest from a map with all 282 metro areas and all four classes of lines. The main change in this region was the upgrade of the eastern Washington line to tier 4.

The main changes in the Southwest were the addition of a spur to Reno and an upgrade of the central coast line. There were several priority changes in the mountain region, but the segments remain isolated.

The relocation of the core route through Spartanburg instead of Columbia and Augusta was the biggest change in the Southeast. The Piedmont region saw a lot of changes, and it still remains in flux as I have more to learn about the physical geography in the area.

Routing in the Northeast was little changed aside from some priority alterations.

For comparison, I overlaid a map of the US national lines with a map of France at the same scale. French HSR lines are shown in dark blue, along with the Channel Tunnel and HSR1 in the UK. The comparison really underlines the scale of the undertaking that would be needed to build out a HSR network in the US. The distance from Lille to Marseilles is roughly the same as from Sacramento to San Diego, and Raleigh to Boston compares to the journey from Marseilles to London.

Thursday, September 3, 2009

High Speed Followup

Here are a few more maps to flesh out yesterday's post.

The network I developed previously is overlaid on a population distribution map in this image. Now, the map is from 1970, but as far as I know no entirely new cities have been built since then. Growth since then has been greater in the South and Southwest, so the sizes of the circles (urban population) and density of the dots (rural population) is somewhat wrong. But, in general, Americans are living in the same locations today that they did back then. As you can see, I've hit nearly every major city with a red or blue line, excepting those in the Mountain states. The green or yellow lines hit most of the rest. One area I need to investigate is the western Virgina/western North Carolina/eastern Tennessee region. There are a lot of medium-sized cities there, and I think I missed an important line or two.

Here is the network plus a whole slew of routes that might be considered for passenger rail if the country was to go all out on upgrading and electrifying the rail system. I've included every route that seems plausible. There are a few intentional omissions because of physical geography and size (Grand Junction, Medford), because of isolation and size (Roswell, Bismark), and because of geography and density (north of Lake Superior). I have also excluded the dense commuter rail networks that exist around Northeast Corridor cities and Chicago, which would get even denser if rail had a higher priority. There's probably a bunch of unintentional omissions, too. I tried to avoid filling the Big Empty Spaces out west with extra lines, as the area really is empty. Nonetheless, I ended up with a several routes that would be well-utilized only in a world where flying or driving was prohibitively expensive.

I also compared my plan to this plan from The Transport Politic. The blogger there did a very detailed analysis using a mathematical model to help decide on where to run the HSR lines. Nonetheless, I have a number of minor quibbles (marked in green) about gaps and spurs in the secondary network, and four larger questions (marked in blue) about the HSR lines.

Working west to east, the first is about the Los Angeles to Phoenix connection, where I think it might be a mistake to bypass Yuma. There are about 300,000 more people on a southern route, the geography looks easier, and there could be a connection to San Diego via two decent-size cities in Mexico (it's remotely possible). The diversion does cause a 30 minute time penalty, which pushes the duration of the downtown-to-downtown trip to about the limit. But LA is such a vast area that many of the trips would originate east of downtown, and those journeys would stay well below the limit. Next, the gap south of Macon seems like a missed opportunity. The key here is not to go straight to either Savannah or Jacksonville, but to hit a point between them and fork. Farther south, running the main line on the west side of the Florida peninsula would substantially increase the time to Orlando and Jacksonville from Miami. I think the importance of tourism to the state needs to be acknowledged, and running the line on the east side would pass more touristy areas. Plus doing an EIS for trackage across the Everglades is too horrific to contemplate. Finally, the kink to hit all the cities in North Carolina means the line won't really be HSR. The distances between them are just too small. That's why I bypassed Durham, Burlington, Greensboro, Winston-Salem, and High Point with the main HSR line between Raleigh and Charlotte. The branch route could be served with HSR track and trainsets. But station dwell and acceleration/deceleration times start to dominate with that many stops, so the track might be built only to MSR standards. I think there would be about a 40 minute time penalty for the western route over a direct connection.

That's it for now. I'm all HSRed-out.

Update: One last gasp. Here's the extended system overlaid on the population map.

Wednesday, September 2, 2009

Just Lines on a Map

One of the most pleasing parts of the economic stimulus bill (aka the clumsy acronym EGTRRA) - at least to us liberals in our decadent coastal enclaves - was the inclusion of several billion dollars of money allocated for high speed rail (HSR). This has prompted an outburst of proposals for HSR networks, conveniently presented in graphic form. Here's some of them:

  • The Federal Railroad Administration's designated HSR corridors with some criticism.
  • A proposal from The Transport Politic, with a followup.
  • US HSR has a plan that projects build-out through 2050.
  • America 2050 also has a plan that is based on the concept of megaregions.
  • The Infrastucturist has mashup of several different proposals.
  • The VHSR group has a nice map of existing Amtrak routes and population estimates of another set of megaregions.
I have my own proposal, but before talking about it let me back up a bit and cover some basics.

First, what is HSR? Internationally, it's defined as a fixed guide-way transportation system that operates regularly at speeds above 240 kph (150 mph). In the US it has been defined by the Federal Railroad Administration (FRA) as railroad transportation that can travel in excess of 110 mph (175 kph). The international definition covers both conventional rail and magnetic levitation systems. For the rest of this post I'm going to define HSR as conventional steel-on-steel trains and track that allow for regular operation above 150 mph. I'm also going to add two other definitions that are unique to me: MSR as conventional rail that travels at speeds up to 110 mph, and SSR as conventional rail that that travels at speeds up to 80 mph.

Second, what kind of equipment does HSR require? All currently implemented HSR networks use trainsets powered by electricity provided by an overhead catenary system. This combination isn't the only possibility, but it's definitely the best one. Electric power is better for this application than a diesel-electric combination because hauling around a large prime mover (a railroad term for the engine, which is separate from the motors powering each axle) and its fuel supply isn't necessary. And with diesel power, more than one locomotive would be necessary to meet the required power. Electric power is better than a gas turbine-electric combination because turbines are very inefficient in this application. An overhead catenary is better for providing electricity than the only other option, third rail, for safety reasons. A trainset is better than a separate locomotive pulling independent cars because a trainset can be more easily made aerodynamic, access between cars is better, and they are more attractive. (Some HSR designs have two semi-permanent, integrated locomotives that push and pull a set of articulated cars, instead of being part of the articulated set.) For MSR and SSR, diesel becomes viable, though electricity is still better in many situations. A dedicated trainset is still best at all speeds, but legacy locomotives and cars work as long as they are well maintained.

Third, what kind of track does HSR require? A HSR trainset can operate on any electrified track provided the voltage is correct, but requires a separate, fenced-off, dedicated set of tracks free of grade crossings in order to operate at maximum speeds. HSR needs to be separate because the track has to be built to a much higher standard than for MSR or SSR, because mixing speeds slows down traffic substantially, and because heavy freight trains cause much more wear and tear than lighter passenger trains. A dedicated line can also have grades that are too steep for freight traffic. In most HSR networks the trainsets operate on dedicated track outside of city centers and merge with general traffic for the last few miles. MSR and SSR do not have to be segregated, but the track needs to be in better shape than for freight-only traffic. MSR needs to have an in-cab signaling system installed on the line, as does HSR, but SSR can operate on traditional trackside indicators. MSR also requires all grade crossings that can't be eliminated be protected with large, complex gates.

Finally, where should HSR go? Because of the high cost of building dedicated track, HSR is only viable between major population centers. Additionally, when compared to aircraft, HSR is competitive only for trips of up to 3 hours, or 300-400 miles. The number of stops must be limited in order for the average speed to stay up, as does the amount of time spent on low-speed lines, in order to have an advantage over cars.

So, there's a lot of stuff to consider when planning a HSR network even before tackling issues like geography and existing conditions. Let's look at a potential line.

Semi-randomly, I've selected Boston to Buffalo for this exercise. It's not an obvious route, but it does terminate in a top 10 city. From west to east, the cities along the route are (populations are for metro area, mileage is from Google Maps):

  • Buffalo, pop. 1,170,000, MP 0, MSA rank 47, Possible connections: Toronto, Erie, Cleveland
  • Batavia, pop. 16,000, MP 36
  • Rochester, pop. 1,037,000, MP 67, MSA rank 51
  • Seneca Falls, pop. 9,347, MP 117
  • Auburn, pop. 28,574, MP ?
  • Syracuse, pop. 732,117, MP 155, MSA rank 80, Possible connections: Binghamton, Philadelphia
  • Oneida, pop. 10,987, MP 180
  • Rome, pop. 294,862 (in Utica MSA), MP 196
  • Utica, pop. 294,862, MSA rank 155, MP 212
  • Herkimer, pop. 7,498, MP ?
  • Johnstown, pop. 8,100, MP 262
  • Amsterdam, pop. 18,355, MP 273
  • Schenectady, pop. 853,919 (in Albany MSA), MP 290
  • Albany, pop. 853,919, MP 303, MSA rank 57, Possible connections: NYC, Montreal
  • Rensselaer, pop. 853,919 (in Albany MSA), MP ?
  • Pittsfield, pop. 42,931, MP 338
  • Springfield, pop. 682,657, MP 386, MSA rank 74, Possible connections: Hartford, New Haven
  • Worcester, pop. 749,973, MSA rank 65, MP 434
  • Framingham, pop. 66,910 (in Boston MSA), MP 453
  • Newton, pop. 83,829 (in Boston MSA), MP 466
  • Boston, pop. 4,411,835, MSA rank 10, MP 476, Possible connections: Providence, Concord, Portland
When set out like this, most of the destinations are obvious (Buffalo, Rochester, Syracuse, Utica, Albany, Springfield, Worcester, and Boston). Others aren't. Schenectady is too close to Albany to be added for a stop on its own merits, but existing conditions mean a stop would have a low penalty. I would also include Newton because Boston is such a big metropolitan area that traveling into the city center can be prohibitive. A similar suburban spot could be added near Buffalo, but the majority of the population lies to the north or south of downtown, making the added utility smaller.

But wait, you say, the line being discussed is much longer than 400 miles. That's true, and somebody going from Buffalo to Boston would most likely fly (at current prices). However, there are so many intermediate destinations with considerable populations that the entire line would be heavily utilized. The line also intersects with several others: HSR lines at Buffalo, Albany, and Boston; and MSR lines at Buffalo, Syracuse, Springfield, and Boston. Many trains could run on the corridor for only part of their routes before branching off, which would increase utilization even more. So, using my rather imprecise technique, this looks like a very good line to include in a HSR network.

Where else should HSR go? To answer that in a semi-methodical way, we could look at the US metro area rankings, and see if any groupings emerge. I've done that, and here is what I saw. I've used the Census Bureau's MSA definitions, so a few larger megalopolises have more than one entry. Marginally attached cities are in parenthesis, and logical extensions into Canada and Mexico are in brackets.

  1. Northeast
    • Top 20: NYC, Philadelphia, Washington, Boston, Baltimore
    • Top 40: Providence, Virginia Beach
    • Top 60: Richmond, Hartford, Bridgeport, New Haven, (Buffalo, Rochester, Albany)
  2. Southwest
    • Top 20: Los Angeles, Phoenix, San Francisco, Riverside, San Diego
    • Top 40: Sacramento, Las Vegas, San Jose
    • Top 60: Tucson, Fresno, [Tijuana]
  3. Midwest
    • Top 20: Chicago, Detroit, Minneapolis, St. Louis
    • Top 40: Pittsburgh, Cincinnati, Cleveland, Kansas City, Columbus, Indianapolis, Milwaukee
    • Top 60: Louisville, (Omaha)
  4. Texas
    • Top 20: Dallas-Fort Worth, Houston
    • Top 40: San Antonio, Austin
    • Top 60: (Oklahoma City, Tulsa), (New Orleans)
  5. Florida
    • Top 20: Miami, Tampa
    • Top 40: Orlando, Jacksonville
    • Top 60:
  6. Piedmont
    • Top 20: Atlanta
    • Top 40: Charlotte
    • Top 60: Birmingham, Raleigh
  7. Northwest
    • Top 20: Seattle, [Vancouver]
    • Top 40: Portland
    • Top 60:
  8. Non-corridor
    • Top 20:
    • Top 40: Nashville, Denver
    • Top 60: Memphis, Salt Lake City, Honolulu, Albuquerque
The results are fairly encouraging, because 38 of the top 40 cities in America, which have populations of 1.3 million or more, can be organized into 7 different regions. The three strongest regions are the Northeast (where the MSR-ish Northeast Corridor already exists), the Southwest, and the Midwest. The groupings get a bit shakier over 40 due to distance and intermediate density. Four top 60 cities will never be on a national priority network due to geographic isolation: Honolulu (obviously), Denver, Salt Lake City, and Albuquerque. Some people like connecting the last two, but there are very few residents between Pueblo, CO, and Albuquerque, and the geography is punishing. Nashville and Memphis are just a little too far from everything else to fall into any region.

Using the groupings above, a list of the top 200 cities, and various bits of specific knowledge that have lodged in my brain over time, I've come up with the network below. A few more comments are in order. Unlike most plans, New Orleans is a stub, not a hub. This is because I think the Gulf Coast line should be well away from the ocean whenever possible. In some places I have bent the route from straight in order to build a triangle that could connect more routes. One in Texas has already been proposed, and I added a small one in Illinois and a large one in Tennessee. I included some state capitals that are marginal, but that was for inevitable political reasons. Finally, the colors. Red is for the core national lines, all of which should be built to HSR standards. Blue is for nationally important lines, which would be best built to HSR standards, but could work as MSR lines. Green is for lines that are significant for a state or region, but probably merit only MSR standards at best. And yellow is for examples of MSR or SSR infill or spur lines, and is not meant to be comprehensive.

Kinda crappy looking, eh? I hope to find a better base map to work off.

Okay, time to step back again. The underlying assumption for this plan and most other plans is that we will be operating in a more-or-less business-as-usual (BAU) mode in 2050 or 2100. Most of our cities will still be car-centric and car dependent, gas (or other fuel) will be cheap, and airplanes will be affordable enough to dominate city-to-city traffic over 400 miles. That assumption could be entirely wrong. A lot of peak oil adherents believe that transportation will be entirely different by 2100, and that a program to electrify every existing rail line should be started immediately. Only after that should HSR be built. That plan does better reflect the reality of existing North American rail transportation: it is freight centric. As such, it operates quite well - if you are on a main corridor and need hundreds of carloadings per year. Most HSR plans, including mine, fudge or ignore the issue of freight. That's a big problem, because most MSR and SSR routes would run on existing right-of-ways, albeit with much improved track. If freight can't be kept moving at a good pace, a lot of it will end up on the highways, which would be a big negative.

Another major problem with most HSR proposals, again including mine, is relatively weak treatment of Canada and Mexico. That might sound like an odd criticism when the starting point of the whole discussion was funding from the US government. But America and Canada are heavily integrated economically and culturally, so the border there should not weigh too heavily on route planning. The US-Mexico border is much more problematic, to say the least. At this point, future HSR or MSR connections should be planned only at the high level. As for specific lines, the Windsor-Quebec City corridor is obvious and everyone includes it, but other lines should be considered as well. For instance, a small isolated MSR network might be viable in Nova Scotia and New Brunswick. An isolated HSR line is being planned for the Edmonton-Calgary corridor, and there is potential for MSR connections to other western cities. Mexico is in desperate need of passenger rail, as very little exists. There are seven potential locations for crossing the border: San Diego, Yuma, Nogales, El Paso, Laredo, McAllen, and Brownsville. Beyond that I don't know much about Mexican cities or geography, but I hope to add lines for that country soon.

Finally, by their nature, no national plan gets down into the weeds. Some regional plans do; for instance, the California HSR proposal is quite detailed, and environmental reviews are already underway for some segments. Elsewhere, though, millions of devilish details remain undiscovered. The biggest problem will be acquiring the right of way. The endless sprawl that covers the country east of the Mississippi might lead to hundreds of billions of dollars being spent in order to develop reasonable routings. NIMBYism will reach new heights once a specific route is proposed, as there is no denying that a train blasting through at 150 mph makes quite a racket, even if it is brief. Every mile of every segment will have to go through an arduous approval process before a single tie can be laid.

So enjoy the pretty maps you see popping up all over. You won't be seeing actual HSR in the US until at least 2020.