Monday, March 23, 2015

Ban the ban, not the plan

Last week, Minnesota House representative Pat Garofalo (R-Farmington) issued a news release touting a compromise between him and supporters of the Zip Rail line being planned between the Twin Cities and Rochester. Oh good! Compromise! Our representatives must have done their job and avoided petty politics! That's what we always want our representatives to do, isn't it?

Unfortunately, this is no compromise—merely a softening of the harsh and arbitrary position Garofalo set out in a bill he authored in January. The initial bill would ban funding and planning of the Zip Rail line entirely. Now he says he'll drop the language that bans route planning by the state, Metropolitan Council, and regional rail authorities, but would still prevent those entities from funding any construction.

Right now, our transportation system is like a house where all of the money has been spent on windows and walls, but nothing has been put into adding a roof.  We don't sit in our houses asking for money to magically print out from the ceiling to pay back its installation cost, but the benefits of having it in place add up: Better health for the people who live there, lower maintenance costs, and the ability to heat and cool the space without spending a fortune, among many other things.
When highways are built, we don't necessarily expect them to pay back their costs in full. Many of them do, but others don't. For instance, Minnesota 212 in the southwest suburbs isn't covering its costs with revenue from the gas tax, motor vehicle sales tax, or license fees. There are some improvements planned for U.S. Highway 14 in southern Minnesota that almost certainly won't be directly paid off by the automobiles traveling over it.

Benefits from things like improved safety and higher travel speed often covers the financial gap. These are less tangible since they don't translate to a direct revenue stream, but it still means that projects like these can be worthwhile anyway. This is one of the most important roles of government—running projects that help society at large but are too expensive or complex for individuals or small groups to do themselves. The right projects will reduce overall social costs or boost the economy enough to cover the difference.

The frustrating thing is that rail projects are always put under the microscope and scrutinized to a far higher degree than highway expansion, even when the rail lines are expected to show good benefits. State-sponsored studies have looked at passenger rail to and through Rochester for nearly 25 years and have consistently shown it providing a net benefit to the state and region. For good reasons, it has bubbled up to the top tier of routes to be built under MnDOT's state rail plan.

Zip Rail planners have often mentioned that the line is attracting interest from companies willing to help pay its construction costs. I've always assumed that this would be in the form of a public-private partnership (PPP), where a company or consortium would pay for some of the startup cost and would operate the line in exchange for taking back a chunk of the route's annual revenue, but the "compromise" from last week suggests that someone is be willing to pay the entire cost to build the route.

An organization called North American High Speed Rail Group has come forward as a potential backer of the route. That's great, if true, but shouldn't be a reason to restrict the availability of public funds at this point. We don't really know if they have the resources to pull it off at this time, especially since construction and operational costs haven't been determined yet.

Garofalo's bill should just die in committee with no further action taken. Leaving the issue alone will let the Environmental Impact Study phase play out, and the line's backers will have to come back to request funding anyway, if they need it. The bill does nothing to "protect taxpayers", and would be likely to do more harm than good.

Monday, March 9, 2015

The market and the math to make pasenger rail work

An Amtrak train passing a farmhouse in Michigan
Photo of an Amtrak train passing a farmhouse in Michigan (CC-BY Russell Sekeet)

The Twin Cities region sits at the edge of the populous eastern third of the United States. There aren't big million-plus metropolitan areas for long distances to the west, so when looking at plans for passenger rail expansion, there aren't many obvious destinations. However, the lone intercity train to pass through Minnesota, the Empire Builder, is popular enough to suggest that lightly-populated areas can still provide enough ridership to be viable.

That's something positive to consider as our state looks to improve the transportation system for people who can't or don't want to drive everywhere. Development has been proceeding for routes in MnDOT's passenger rail plan, but the pace has been hampered by weak political backing and limited budgets. But what if there's a real business case for these routes? Could that accelerate development?

Airlines as a guide

People have a number of different ways of getting from city to city, but today, if they don't drive, they usually fly. Chicago is one of the first places people think about when contemplating rail projects for Minnesota, but we have fairly frequent and inexpensive flights from Minneapolis–Saint Paul International Airport to both O'Hare and Midway. It's easy to dismiss the idea of improving rail travel when a flight is only $30 more expensive than today's Amtrak service, yet gets you there several hours faster.

But if you look beyond Chicago to the other routes being contemplated, there's a price gap that certainly looks big enough to run a train through. Here's a table of direct, one-way airfares I found for April 2nd, 2015 (including a $20 to $25 fee for one checked bag):

City/AirportMedian priceFlight distancePrice/miAnnual passengers to/from MSPDaily round-tripsAverage per flight*
Chicago O'Hare$130334$0.391,644,0002590
Chicago Midway$108349$0.31902,0008154
Des Moines$391232$1.69194,000553
Duluth$331155$2.14152,000542
Fargo$480223$2.15332,000765
Madison$408228$1.79296,000581
Milwaukee$102296$0.34570,0006130
Rochester$25076$3.29104,000271
Sioux Falls$441195$2.26252,000658
(* Average per flight based on number of flights for April 2nd.)

Passengers can get a discount on the expensive legs if they connect to other flights, though that also effectively raises the fare to the cheaper destinations on the list (digging into the relevant data for Q1 2014, the median MSP–O'Hare airfare worked out to be $0.51 per mile because of that). There are a lot of people paying hundreds of dollars to make hops that would have been far less expensive by ground, and maybe just as fast.

Trains are cheaper and more effective

By contrast, a typical Amtrak fare is about 10 to 20 times cheaper than what the airlines charge for the smaller cities listed, and often half that of the larger destinations. For each passenger, the Empire Builder averages ticket revenue of about $0.17 per mile on its long route. Even at that low fare, it covered about half of its operating cost, and it's a big, expensive train to run. The Empire Builder has sleeping cars and lounge/dining cars that aren't necessary for short routes that only take a few hours to cover.

Across the Amtrak system, the full cost to operate a train tends to land in the range of $0.30 to $0.70 per passenger-mile, though that number is dependent on a lot of things, perhaps the most important being the number of passengers onboard. Nonetheless, all of Amtrak's routes are less expensive per passenger-mile than any of the small-market airline routes listed above. (Amtrak's worst performer by far in 2014 was the Chicago–Indianapolis Hoosier State at $1.14 per passenger-mile).

This shows how much economic benefit there could be from rebuilding proper passenger rail service across the state and country—existing travelers could save money, and new passengers who previously couldn't afford to fly would now have a cheaper option.

Corridors up to about 150 miles in length are the easiest ones to make competitive with air travel, since regular trains traveling up to the normal limit of 80 mph can cover the distance as fast or faster than flying. Airline passengers get bogged down by TSA security lines, more time spent boarding (airliners have just one door while trains can have many), dealing with luggage (trains allow much larger carry-on bags than airlines do), and getting to the departure point.

Railroad stations are often downtown versus airports that are ten to twenty miles away. Trains can also make stops along the route to serve towns that are too small or too close to larger airports to justify air service of their own. Combine that with fares that can be many times cheaper than flying, and ridership could climb several times higher than what the airlines can do, even on routes that extend significantly further than 150 miles.

Finding the magic combination

Unfortunately, Amtrak has failed to properly tap into the short-haul travel market, partly because they've never had great funding. A number of ingredients need to be properly mixed in order to build a train service that can carry people profitably: The train needs to be fast enough and frequent enough to attract passengers—with a price to match—while having low operating and maintenance costs.

I'll start with the last item first:

No matter how much money you initially put into a train service, operating costs are what can really make or break the rail line. If your fares aren't enough to cover the cost of the onboard crew or the day-to-day maintenance of a train, then there isn't really a way to pay off the fixed infrastructure or rolling stock.

I have a recommendation: Run trains with high-capacity passenger cars—as few as you need to carry your passengers.

Long trains don't help

My reasoning is pretty simple—just look at this plot of commuter train operating cost versus length for different providers around the country (data from the 2013 edition of the National Transit Database):



This really surprised me when I plotted it out. I would have expected operating cost to taper off as trains grow in length (some economy of scale), but the linear trend line on this graph shows that each extra car on a commuter train adds an operating cost of almost $600 per hour. Many costs are included in these values, including overhead like management, insurance, and payments to the railroads, so it's remarkable that it fits so well.

This makes it easier to think about what it takes to operate a train at a profit—rather than worrying about the train as a whole, we can just think about whether each individual passenger car can cover its costs or not. It doesn't matter very much whether a train is long or short.

Still, even though the trend is very linear, there's still a big range across the systems in the database. The lowest outlier turns out to be an Amtrak train, the Boston to Maine Downeaster, which is overseen by a regional rail authority that submits their cost data to the NTD. It costs $1,004 per train hour, or about $219 for each car per hour.

The country's most expensive commuter train in terms of cost per car hour is our own Northstar commuter service, which clocks in at a massive $4,590 per train hour, or $1,179 per car hour. (This is bad and should be fixed.)

It's hard to say why these are so different, but it's safe to assume almost any new train would be less expensive per hour than Northstar. And, with the Downeaster's low cost, there's good reason to believe that new intercity trains would fall somewhere below that graph's trend line.

Use bigger cars instead

Commuter trains around the country often use bilevel passenger cars with two floors of seating, and Amtrak also runs bilevel trains on many corridors using their Superliner equipment or some of the related spinoff designs. While they don't double capacity, they do increase the number of seats on a car by 50% or more.

This can help a lot when trying to get enough fare revenue from passengers to cover costs. For instance, a long-distance Amfleet car has 60 seats, compared to 90 or a bit more on Superliner cars. At the Empire Builder fare of $0.17 per mile, an Amfleet car averaging 60 miles per hour would max out at $612 per hour, while the comparable Superliner would bring in $918 if it was fully loaded.

Northstar's commuter coaches pack riders much more tightly, averaging 145 seats per car. A full car could bring in $1,479 per hour at 60 mph, exceeding even its outlandish costs.

However, it's impossible to always run fully-occupied trains, and intermediate stops to let passengers on and off can drag down the load factor. Passenger loads in the 50% to 70% range are much more common. A train with Northstar-level costs would need to raise fares to keep ahead, so it's extremely important to drive down operating expenses as much as possible—if it had Downeaster-level costs, it would manage some major surpluses on the operating budget.

Bilevel trains are probably a bit more expensive than their single-level counterparts, but it's hard to say how much more. Commuter rail services with single-level passenger cars seem to have lower costs than the ones that run bilevel trains, but they are often used in areas that have electrified trains which need to fit underneath the catenary wires that power them. Electric locomotives are less expensive to operate and maintain than their diesel counterparts, which skews the numbers a bit.

Hopefully the cost per car of a bilevel intercity train could land in the range of about $250 to $500 per revenue hour.

Get them at the right price

Trains do have one great advantage over the airlines—they're ten times cheaper to buy than an equivalent aircraft.

A Boeing 767 with 350 seats runs about $180 million, more than $500,000 per seat. In contrast, a fairly standard train with four 90-seat Superliner-style bilevel cars (360 seats total) and a new locomotive would probably run $12 to $18 million, up to about $50,000 per seat.

A Northstar train with 145 seats per coach could bring the cost down to a bit over $20,000 per seat—a whopping 25 times cheaper than a jetliner. That's not quite the right equipment for trips lasting a few hours, though. Some seats would need to be taken out to add space for luggage and a little extra legroom.

Let's say you have something middle-of-the-road with 120 seats per car at a price of $30,000 per seat. If you want to pay off that piece of train equipment within its first million miles (a point when rail vehicles often go in for a mid-life overhaul), you'll need to dedicate about $0.06 per passenger-mile to the cause of paying off the train.

Jumping back a second to operating costs, if that 120-seat car cost $375 per hour to operate (halfway between $250 and $500), about $0.09 would be needed per passenger mile to cover expenses. Add that to the $0.06 to pay for the equipment, and you've got a narrow surplus of $0.02 per passenger mile.

Run them often enough and fast enough

One thing that helps keep flying popular is the ability to choose from multiple flights each day. Frequency of service is a huge failing of Amtrak's current system, perhaps only matched by the extremely skeletal nature of their network. It's rare to find a regularly-scheduled air service that has fewer than four round-trips daily, so it's remarkable that most of Amtrak's stations only see one train per direction per day.

Running multiple daily round-trips is key to attracting passengers to the rails. The Northern Lights Express to Duluth is planned with eight daily round-trips, and the Zip Rail service to Rochester may have even more. Only small parts of the country in the Northeast and California have intercity trains running that often today, which has severely limited the attraction of train travel up to now.

Operating expenses found in Amtrak's statistics and the National Transit Database typically include some payment to the railroad to cover track maintenance and access charges, but that may not be enough to pay for more track capacity. Money is needed upfront to add room for passenger trains, and it often adds up quickly.

The big problem is that it can take tens of millions to hundreds of millions of dollars to build capacity and/or buy the access to run that many trains, with costs climbing to the billions for longer and faster routes. A basic upgrade would include things like signaling upgrades and a few added or expanded sidings for trains to pass each other. That's something that could be done for $1 to $2 million per mile. Adding extensive amounts of double-tracking would probably raise the cost into the $4 to $5 million per mile range, and entirely new track could run upwards of $10 million.

People shouldn't immediately be scared off by the cost, though. If we take the hypothetical train car I described in the last section and charge a fare of $0.30 per mile, the first fifteen cents would go to operating costs and paying off the rail vehicle, while the other fifteen could go to paying off track and do it at a decent rate. Relatively modest travel markets with 500,000 to one million annual passengers over the entire distance—probably within shooting distance of several of the cities I listed at the start of the article—could pay off simpler track improvements within about 15 years.

Annual passengers needed to pay capital cost @ $0.15/mi:
price/mi$1 million$2 million$5 million
5 years1,333,3332,666,6676,666,667
15 years444,444888,8882,222,222
30 years222,222444,4441,111,111
70 years95,238190,476476,190

Even when the price climbs beyond what can be paid off in a reasonable timeframe just by passenger fares, it's important to remember that existing freight operators also benefit from track upgrades. They might split the cost with a passenger operator if it helps them move trains more effectively. Faster trains can pay down their hourly costs more quickly since they cover more distance in that time, but that's just a slice of a slice of the overall budget—higher speeds attract more riders, though.

Some upgrades can't really be justified by a profit motive, though—catenary for electrified trains, for example. Unless you're pushing a huge number of trains down the track and grabbing tons of passengers, it would be a net negative for investment since you just don't save enough on fuel and maintenance costs. But, electric trains do have great environmental benefits, so government grants should be considered to cover that cost for relatively busy lines.

A case for doing something rather than nothing

Rebuilding the passenger rail network for cities around Minnesota and across the country is important for building a stable platform for the future. Our transportation is out of balance, and far too dependent on automobiles which have completely altered the face of our towns cities over the last century. As our region continues to gain population, new growth should be focused in ways that are better for the environment and the people who live there.

To reconstruct our cities on the scale of the pedestrian and cyclist, we have to turn the dial back against car-dependent development patterns. Transit systems and bikeways work for the city and metro scale, but intercity lines have to be resurrected or built up from scratch to handle travel over longer distances. Trains help restore walkability to the towns and cities they serve, which will help us turn the tide on climate change.

There is a significant slice of the travel market that isn't well-served by existing options, and I believe the window is large enough for new operators to come in and make money. I've focused on what can be done for conventional-speed trains at short distances, but states with larger populations and different distributions have a wider variety of options. Who might take the plunge, though?

Finding the right pocketbook

Whoever funds the next passenger rail line in the region will need to have a lot of funding available, since the initial costs are pretty high no matter how you look at it.

It would be great if Amtrak itself was pushing for new routes, but they've usually been hamstrung with just enough funding to pay for their operations and their maintenance backlog. Not much has been available for upgrade or expansion. They're almost certainly out of the picture unless a state or other partner organization comes along with the money.

The major freight rail companies were collapsing under their own weight at Amtrak's formation in 1971, but have made remarkable turnarounds. The largest companies now make billions of dollars annually and could easily run passenger service again while barely denting their budgets. They know their lines the best and control the flow of traffic, so they would be in prime position to implement the services as inexpensively as possible.

Unfortunately, they have many employees who only see passenger rail as a drag on their freight business, and would be opposed to any expansion. There are a couple of freight companies bucking that notion, particularly one in Florida. It isn't clear if any others will change their positions, though.

Foreign passenger rail companies from Europe and Japan have also poked around at a number of markets in the U.S., so there might be a push from outside our borders at some point. But they would need to have strong partnerships with American companies and investors to pull it off.

Airlines would be interesting candidates for entering the passenger rail market. It's difficult for them to make money at short distances, so it would make sense to build ground-based feeder networks to airports that specialize in long-haul travel. Many corridors would only cost as much as one or two jumbo jets.

In the end, I don't particularly care who funds and builds any future passenger trains that will be available to me—it's just clear that our current efforts at expansion are moving way too slow to affect things like climate change or handle our aging population. Any new entrant who has the right formula and the right funding has a pretty wide open field to build something successful.

Monday, February 16, 2015

Following the tracks of the Empire Builder



Here is a map I put together late last year showing the route of Amtrak's Empire Builder in a way I hadn't seen before. I followed aerial imagery from Chicago to the Pacific Northwest, using different colors to mark out which parts of the route are single-track (red) and which parts have sidings or two or more tracks (yellow). I made it to help shed light on why the train has been plagued by long delays over the past few years, becoming especially bad in 2014. The train usually ranks as the busiest long-distance route in the Amtrak system, but saw ridership tumble nearly 16% to 450,000 in the most recent fiscal year—knocking it out of the top spot for routes of its type and placing it behind the Los Angeles–Seattle Coast Starlight. As work progressed to try and resolve congestion for both freight and passenger service on the line, Amtrak spent several months running the train on a more lax schedule than usual and operating with a partial reroute. Since January 12th, the service is back on its normal route, and on-time performance has improved greatly.

(Note that the map isn't a perfect snapshot—the aerial photography was done at different times for each area, but for the most part it looks like it ranges from being a few years old to only a few months old.)

A combination of several factors led to this trouble over the last few years. The train takes an epic journey of more than 2,200 miles each trip, wending across vast glacial plains and up into a series of mountain ranges before reaching its western endpoints in Portland and Seattle. Between Chicago and Saint Paul, the train runs on tracks owned by Canadian Pacific Railway, while BNSF Railway is the owner of tracks used from Saint Paul on to the west. Both of these are major transcontinental lines for their respective railroads. The route it follows has been hit by extreme weather multiple years in a row, ranging from the "Polar Vortex" to flooding along the Missouri and Souris River valleys, and also a permanent flood that has been slowly raising the level of Devils Lake in North Dakota. Combine that with rebounding levels of freight traffic following the Great Recession, and it became a recipe for trouble. These problems all hit at a time when the railroad began working on significant capacity expansion projects stretching the length of the corridor. Delays from each issue compounded the others and, just like with road construction, the effort to upgrade the line exacerbated delays for all types of rail traffic.

Time for an upgrade

Rail upgrades along the corridor have mostly progressed in a west-to-east pattern over the past few construction seasons, though they haven't all been concentrated in a single spot. One particularly worrying issue for Empire Builder passengers was the deterioration of the line near Devils Lake. BNSF's main line runs diagonally through North Dakota on the Surrey Cutoff (more properly knows as the KO Subdivision) between Fargo and the town of Surrey just east of Minot, but Amtrak has used the line through Devils Lake because it allows access to Grand Forks and its metro-area population of 100,000.

BNSF stopped running their own trains through the Devils Lake area in 2009 and only minimally maintained the track afterward. Amtrak stuck it out a couple years longer despite having to run at restricted speeds. They finally started detouring onto BNSF's main route in 2011 as water lapped at the edge of a bridge by the town of Churchs Ferry. However, the railroad's growing traffic congestion to the west led them to a new agreement with Amtrak and the state of North Dakota where each party would pay 1/3 of the cost to repair and upgrade the Devils Lake and raise the bridge to a level that the lake waters could never reach.

That happens to be on the longest piece of single-track along the Empire Builder route—27 miles—and it's notable how the spacing of sidings widens out on other parts of the Devils Lake line versus BNSF's main line to the south. You can click on the side menu of the map to see an extra layer showing the KO Subdivision (along with a related segment, the Prosper Subdivision).

West of the Devils Lake area, there has also been a major project to add double-track between Minot and eastern Montana, which has been carried out in phases. Initially, some long sidings have been added to the line over the course of a couple of construction seasons, and the final stage will be to connect those sidings together to create a long stretch of double-track. A similar strategy has been at work out in Washington State. You can even see a blurring between old and new imagery of single- versus double-tracking it this spot near Hatton, Washington.

Right-sizing the rail network

It probably seems a bit strange that a major rail route is mostly single-tracked, but as long as trains are scheduled well and the sidings are the right sizes and in the right locations, a single-track line can handle dozens of trains per day. A trade-off is that locomotive engineers have to operate their trains much more strictly than the free-for-all we're accustomed to from driving on streets and highways—a higher level of planning and coordination needs to take place. Trains need need to operate with strict adherence to pre-planned schedules in order to make full use of the line.

Each mile of track comes with an annual maintenance cost of tens of thousands of dollars, so it makes sense to have as little of it as necessary to get the job done—something that all rail companies have learned the hard way. The rail network grew dramatically in the first century of railroading in the U.S., and there were many cases of overbuilding. Some lines were never profitable and ended up being abandoned early on, but the addition of new track outpaced abandonments until the network peaked in 1915. Since then the network has been shrinking due to competition from the automobile and other transportation modes, but we appear to be moving through another inflection point—"Trough Rail", if you will.

Even though railroads have lost a great deal of mileage since the system's peak, the actual volume of traffic has had a decades-long trend of growth. North Dakota has had a spike in traffic due to oil production in the Bakken shale region, but the overall trend has been up for all types of freight including agricultural products and goods shipped in intermodal containers. Remaining rail lines have seen traffic become more concentrated as railroad companies have found ways to make their systems more efficient. However, there's a certain point where there is no more room for efficiency and the rising volume of traffic means that additional track is needed—either replacing old segments that were once torn out or building entirely new tracks.

Many of the major railroads have been making big expenditures to boost capacity, though BNSF has been setting the biggest spending records—$4 billion in 2013, $5.5 billion in 2014, and a planned $6 billion this year. They're expected to slow down after this year, but they're not quite done along the Empire Builder's route yet. There are only a few short stretches of single-track remaining on BNSF's main route through Minnesota, and they'll be spending spending $326 million in the state this year to close some of those gaps, which should be practically eliminated by the end of 2016. The double-tracking project west of Minot will also still be in high gear this year. Amtrak riders may still find some extra time on their hands as they deal with construction-caused slowdowns, but it lays the groundwork for better service for the future. Freight users will be the primary beneficiaries, though at least this fixes one notable gap between Big Lake and Becker that would benefit any extension of the Northstar Line to St. Cloud, or potential added Amtrak service up to Fargo or beyond. BNSF has previously said that they'd need to add some extra track between Minneapolis and Coon Rapids for that to happen, though.

I haven't said anything about the Canadian Pacific track that the Empire Builder uses between Chicago and Saint Paul, but that's mostly just because they haven't been as self-promotional about maintenance and upgrade activities as BNSF. One leg of their transcontinental route west of the Twin Cities also passes through North Dakota, crossing the Empire Builder route in Minot, so they've also dealt with the booming rail traffic in that state. One of their major projects is adding Centralized Traffic Control (CTC) signaling on that route west of the Twin Cities and up to Moose Jaw in Saskatchewan, a process that will take several years.

More on the map

I found it interesting to compare the spacing and length of sidings on the Canadian Pacific tracks to those along BNSF's corridor—they tend to be a bit longer, generally in the range of 2.5 to 3 miles rather than BNSF's sidings which seem to be more like 1.8 to 2.5 miles in length. It's hard to say if that makes a huge difference, though it can allow trains to pass each other at higher speeds, or potentially allow more than one train to fit onto a siding at a time at times of high traffic congestion (freight trains currently max out around 7,000 feet long, so two could fit fairly comfortably on a 15,000-foot siding). That likely gives CP's route a bit more flexibility and capacity than BNSF's typical single-tracked areas, but it's hard to say how much more.

It was a surprise to see that the longest stretch of double-track for the Empire Builder is in Minnesota, stretching 115 miles from the town of Philbrook (southeast of Staples) to Moorhead (just before the line crosses into Fargo). The next-longest segment is about 103 miles from Chicago to the Milwaukee suburb of Pewaukee (I show a segment in Brookfield in red, but that's just because the two tracks split to be half a mile apart for a little bit). The exact length and placement of planned double-tracking in western North Dakota isn't entirely clear, but could end up being even greater than either of those—Minot is 121 miles from Williston, for instance, and that's pretty close to the center of the Bakken action.

There are a couple of major features of BNSF's line that can significantly constrain route capacity going forward—tunnels. The line runs through the 7-mile Flathead Tunnel in western Montana and the northern leg that carries the Empire Builder to Seattle also passes through the 7.8-mile Cascade Tunnel—the longest railroad tunnel in the United States. Each time a train passes through these tunnels, powerful turbines have to fire up to clear them of exhaust generated by the locomotives, and that process can take 20 or 30 minutes. If traffic on the corridor continues to grow, there may be new tunnels in the future, which will be pretty amazing to see.

Final notes

It will be interesting to come back and look at this route again in a few years, to see how things have changed and what the effect has been. The recent decline in oil prices has led to a tapering off of shipments to and from the Bakken region, so the remaining traffic should have a smoother and faster ride than what would otherwise be expected. This should be good news for Amtrak passengers either way.

Friday, July 18, 2014

Transit budgets expose hidden costs of roads

img_8777

Transit projects make for easy political targets because of high construction costs. They have prices in the many millions of dollars, and occasionally tip into the billions, which brings critics out of the woodwork. It's difficult to know whether certain projects are justifiable or not, particularly because different modes of transportation concentrate and spread out their costs in different ways.

The $957 million Green Line can't be directly compared against a street or highway project, because the rail project's budget contains many things that are (almost) never included in budgets for roads. You might just think of the Green Line as "a train", but there are a lot of components to that. The rails in the ground are the most comparable to a roadway, but that's just one piece of the overall picture. Take a trip on the line, and you'll stand on a platform, pay a machine, and sit inside a vehicle—one that's riding on rails, taking power from overhead lines which are fed by nearby substations, and progressing according to those dastardly signals we don't quite know what to do about yet. At the end of the day, the train will exit the line and go off to a maintenance facility to be cleaned and repaired—a place that also provides room for the operator while they're not driving the train along.

Streets and highways are primarily built for cars, but their budgets don't include the "rolling stock" of the roadway. Unlike buses for on-street transit and steel-wheeled vehicles for rail, most cars and trucks on the road are owned by individual families and businesses. Only a small minority are parts of government-owned fleets. Unlike the maintenance facilities for rail and bus transit, parking lots and repair shops aren't included in the cost of a highway. We don't include the cost of gas stations or the fueling infrastructure leading to them either (comparable to the overhead lines and power substations on a route).

The rail vehicles for the Green Line cost $3.3 million apiece, which certainly looks and probably is higher than it should be, each vehicle carries a huge number of people each day. I was reminded when looking through the National Transit Database that Metro Transit had just 27 light rail vehicles in 2012 for the Hiawatha Line. The line's capacity was stretched so thin at that point that every single vehicle was being pressed into service. Because they operated from the early morning to late into the night, each individual vehicle hosted about 1,200 trips each weekday. They were usually coupled into 2-unit trains, so each set was carrying around 2,400 trips per day. How many trips does an average auto carry each day? How many cars does it take to carry that many people? How much would those cars cost? How much would it cost to provide parking for that many cars? How much space would they take up?

Since 2012, Metro Transit's fleet of light-rail vehicles has roughly tripled, but ridership is probably at least double now that the Green Line is up and running. The number of trips provided by a single LRV unit each day has gone down, but is still at least 700 daily. Three-car trains are now common, so the number of riders per train set remains fairly high.

A lot of the infrastructure for rail projects is also built to stand up better over time. Rail vehicles last longer than cars or buses—In 2012, the average age of a light-rail car was 16 years old, compared to 11 years for automobiles and 7 years for transit buses. The vehicles also have lower operating costs per passenger—they're typically more expensive per vehicle to operate than buses, but each one carries more passengers, resulting in lower costs per passenger-mile. In 2012, it cost 50 cents per passenger mile to operate the Blue Line, versus 84 cents per bus passenger-mile in the Twin Cities.

Transit has a great return on investment as the years go by, especially when coupled with proper land-use patterns. Recovering value from land that would be dedicated to parking can recover tens to hundreds of millions of dollars in value in out-of-pocket expenses for land owners and allow more development in less space. That also leads to a better tax base for the corridor.

It's often very short-sighted to shoot down transit projects because of high upfront costs. There are certainly legitimate reasons to trim costs and ensure that the right type of transit is being chosen for a particular route, but they pay back in ways that are difficult to see at first. We should be much more ambitious in this region about upgrading existing lines and implementing new ones as appropriate—we need to find better shortcuts to cut through the drawn-out, decades-long debates that have typified transit projects around here because we're leaving a lot of value on the table by skipping past good routes.

Friday, June 20, 2014

Saint Paul's oversized downtown streets

IMAG0168
5th Street in downtown Saint Paul, noon on a Thursday.

If you grew up in Minnesota, you may have heard the story of Saint Paul's narrow streets—In downtown, the streets were so tight that the city eventually decided to shave off the fronts of buildings downtown, in order to make the streets wider.

Today, however, many of downtown Saint Paul's streets are too wide, or at least they grant way more space to cars than what is actually needed. Here's a map I made of streets in the area that appear to be wider than necessary to handle the amount of traffic they see (orange means roughly one lane too wide, red means roughly two lanes too wide):


View Oversized downtown St. Paul streets in a larger map

The new Green Line is hoped to be a new big economic driver for Saint Paul, but to maximize the benefit, we have to make sure that the areas around stations—and throughout the city—are built for people and not just the movement of cars. Sidewalks need to be widened, bike lanes need to be added, and bus lanes need to be retained and improved, including upgrading bus stops that are often put in tight spaces (especially the one on 6th Street at Cedar Street).

Personally, I like the idea of converting 5th Street to a (bus) transit/pedestrian mall with bi-directional bus traffic, since it connects directly to West 7th Street and the I-94 ramps to/from Minneapolis, links Rice Park and Mears Park, and also goes straight past the Green Line's Central station and is just one block from Union Depot. Large chunks of 5th and 6th are about twice as wide as they need to be for the amount of car traffic they see, so they could be combined into a two-way street for cars on 6th. I'm not sure which street would be better off with bike facilities—in this concept, 5th would probably be more comfortable for many riders. Buses are big vehicles, but there are far fewer of them than cars (combined, the 5th and 6th street bus lanes now appear to carry a bit over 1,000 transit buses per day).

Minnesota Street should really become a two-way, as should the stretch of Cedar Street south of 5th. Jackson Street could become a two-way (probably in addition to Sibley Street, even though that's relatively appropriately sized when the amount of on-street parking is considered). Jackson and Sibley each narrow down to just two lanes as they reach Shepard/Warner Road, so it doesn't make much sense for them to be wider to the north. Jackson is ridiculously wide north of 7th Street (a segment where it is a six-lane two-way), and really needs a major overhaul to convert it to a more pleasant space rather than a river of concrete.

There's a potential to re-make downtown Saint Paul to be a much more walkable/bikeable space. There are pockets of great spaces in the area, but the streets need work to link those places together to really revive the streets.

Friday, June 6, 2014

Validating Green Line travel time estimates

green-line-sim-full-accel

Following my post last week, Bill Lindeke posted a comment which he summed up as, "What is the theoretical shortest possible run time for the Green Line?"

That's a really good question, since it sets a guidepost for what can realistically be achieved from changing signal timing along the route, and helps us to understand the assumptions that planners of the line originally made.

Of course, the answer depends on exactly how theoretical you want to get. I decided to do a simple simulation based on what speed limit signs I could see along the route. I had started taking attention of them a few months ago as they went up, but spent a few evenings this past week trying to verify the information I already had and to add more to it. Unlike roadways where you typically have a single speed limit over great distances, rail lines tend to have special speed limits for each curve, junction, and other special areas. I also made an assumption that station stops would average 20 seconds.

The main leg of the Green Line along University Avenue is mostly set at 35 mph, though there is a curve at Fairview avenue set for 25. Other curves on the line are mostly set at 10 or 15 mph. Cars and trucks tend to naturally slow at sharp turns, but they move through pretty quickly. The slower speeds that trains take through bends tend to be more pronounced, since the trains themselves are often a lot longer than the curve itself.

As seen in my graph above, it looks like a train could make the run from Target Field to Union Depot in under 38 minutes if it wasn't encumbered by signals. That includes downtown Minneapolis—my simulation had the train crossing from Target Field to Downtown East station in under 5 minutes, or about 3 minutes faster than the Blue Line is scheduled to take today. That instance also assumed that the LRV operators were extremely good at starting and stopping their trains, starting to decelerate at the precise moment needed to hit the platform dead-on. In the interest of safety, a 38-minute end-to-end time while making all stops should probably remain theoretical.

I did take one step beyond, looking at what might happen if there was an "express" version of the Green Line, which only ran end-to-end, making zero intermediate stops. Here's the result:

green-line-sim-express

28 minutes from end to end, if you had an operator with nerves of steel and didn't get held up by signals or other traffic. I've heard people ask about express trains on the Green Line in the past, and have tended to brush it off as an unwise idea. According to these two charts, it might be possible to save 10 minutes on the trip, but the complexity of switching from track to track to pass other trains would make it very vulnerable to delay, and would probably have a negative impact on regular trains. They'd also have to run pretty often in order to get benefit from that 10-minute time savings. Running express traffic only every half hour, for instance, would probably be a net loser all around.

Here's another graph where I attempted to be a bit more realistic, assuming that train operators would only accelerate and decelerate their trains at 50% of the maximum (full service acceleration/deceleration is 3.0 mph per second, though emergency braking is greater). The result was a run time of 41 minutes, which sounds pretty familiar:

green-line-sim-50pct-accel

Based on that and my first graph, it seems that planners of the Green Line expected some very strong signal coordination, priority, or preemption (whatever word you want to put on it), allowing trains to travel with a minimal amount of delay. I believe it's fairly customary to set schedules around a simulated time based on best-case travel time, plus a fudge factor of 10% or so. Of course, in this case, any fudge factor mostly gets eaten up just by delay in downtown Minneapolis (where there hasn't been any signal coordination for the Blue Line, as far as I can tell).

There are several things that may be different as built in the real world compared to how they were originally set up in computer models for making estimates. It seems pretty likely that that the three stations added to the line late in the process (Hamline, Victoria, and Western in Saint Paul) weren't included in the planners' simulations. That probably contributes to about one and a half minutes of extra time with the way I set things up. I also seem to remember that the Washington Avenue Bridge was originally going to allow trains to go relatively fast (perhaps 45 mph), but as far as I can tell, it's limited to 25 mph for trains. There may be other changes along the route that were never put into the schedule simulations.

Finally, since this all involved calculating stopping distance, I'll make a note about safety around light-rail vehicles—assuming a 2-second reaction time and normal braking, it takes 400 feet for a 35-mph train to stop, or 2/3rds of a long block along University Avenue. Even with emergency braking, it could take 280 feet or more for a train to come to a halt, so keep your eyes open when visiting the Green Line.

Friday, May 30, 2014

Green Line travel time update

Yesterday evening, I boarded a route 50 bus at Minnesota and 6th Street in downtown Saint Paul at 4:43 p.m. and rode it to the University of Minnesota campus, getting off at Oak Street and Washington Avenue. That was at 5:29 p.m., or 46 minutes after I had boarded in Saint Paul. The bus still had about three miles to go before ending its run in downtown Minneapolis. Under a schedule posted last weekend, the Green Line is supposed to make an equivalent trip (Central station to East Bank station) in 31 minutes. The bus was 48% slower than the current expectation for LRT (and the 50 is a limited-stop route—the 16 is even slower).

Metro Transit posted updated bus schedules for routes along the Green Line as my last article went up, and posted the schedule for the Green Line itself last weekend. I've updated my graphs from that last entry to show expected downtown-to-downtown travel times:

Eastbound, with the existing schedules and old Green Line projection first, and current planned Green Line and route 94 schedules second (the 50 will be replaced by the Green Line—the 16 will still exist, but won't run downtown-to-downtown, and therefore can't be compared):



Westbound, with the existing schedules and old Green Line projection first, and current planned Green Line and route 94 schedules second (again, the 16 will still exist, but won't run downtown-to-downtown, and therefore can't be compared):



I also went a bit further and tried to show how the frequency of service affects these travel times. These graphs assume that you've arrived at the bus stop halfway between two runs of the same service—if you're a punctual person (and the services are running on time), these times may be able to be shortened a bit. I didn't try to account for people switching between different services (16 to 50 for instance), and assumed that people would typically take the next route 94 rather than keeping track of whether it's a B, C, or D:

Eastbound:



Westbound:




(Data: old spreadsheet, new spreadsheet)

The good news is that the Green Line should be faster than the old routes 16 and 50 by a big margin, except it may be a bit slower late at night (11 p.m. to 5 a.m.). Morning/evening commuters using route 94 should also see some improvement—particularly in making travel times more consistent—as that route is being realigned to use 5th and 6th streets in downtown Saint Paul, won't have the split of some routes running up to the Capitol and back, and will use 7th Street for westbound runs into downtown Minneapolis rather than jogging over to 4th Street. However, the frequency of service for route 94 has been cut back in midday so that it's not much different than taking the Green Line, and evening service after 7 p.m. has been eliminated—those service hours have been folded into increasing frequency on lines that criss-cross the LRT route, such as the 87, 84, and new route 83.

Of course, we'll have to wait until the first days/weeks of service to really know how these travel times work out. With more than 60 intersections and other signaled locations where the trains can be forced to stop, just removing or adding an average of one second across all of those signals can change the Green Line's end-to-end travel time by a minute. Slight errors in setup can be magnified, so it's critical that signals are configured correctly.

The Green Line will be around for decades to come (and hopefully longer than that!), so we shouldn't too be dismayed by hiccups at the start. It will take a consistent effort to improve and maintain good travel times along the corridor, though.

We may also see changes to related services—the 16 and 94 in particular. Will we keep the 16 in its truncated, lower-frequency state (becoming every 20 minutes most of the day rather than every 10 minutes as it is today)? Will the 94 still have a reduced frequency at midday? That route might get folded into future Red Rock or Gateway Corridor bus services.