Saturday, October 26, 2013

Density and the transit puzzle


How would you choose a route for a transit line? It's a difficult task, particularly in a region as big as the Twin Cities. Developed land stretches out for miles and miles before finally tapering off into farmland or patches of wilderness. Some parts of the region have natural geographic constraints due to rivers and lakes, while other places get blocked off due to human activity: buildings, highways, and railroad lines to name a few. Opportunities for perfectly straight routes often aren't as common as we wish. Even though the core cities and some older suburbs have substantial chunks of street grids, there are inevitably points where the block size shifts or entire pieces get rotated. In particular, areas near the downtowns of Minneapolis and Saint Paul were set up to follow the riverbanks rather than east-west/north-south lines. Those areas still tend to have small blocks and many opportunities for routing buses or rail vehicles—perhaps too many in some cases.

The street grid itself can be an impediment sometimes, increasing travel distance by nearly 30 percent in the worst case (if a straight line between your origin and destination is angled at 45 degrees against the grid). But still, a dense grid offers lots opportunities for getting between places, and allows for good walkability—it's often not a big deal if a bus or rail line misses pockets of density by a couple of blocks, because they can still be easily accessed on foot. Depending on their shape and size, it's not hard to walk 4-8 blocks on either end of a transit trip—or even double that. But when you go into an area where the grid fades away, it becomes much more critical to precisely locate transit stops and stations at cross streets or other locations (such as near bike/ped pathways) that maximize the number of different ways for people to access stations.


Of course, maps merely showing population density don't tell the whole story. Large chunks of downtown Minneapolis don't have any residents, for instance, but there are skyscrapers rising nearly 60 stories high. Clearly there's something there: office buildings. Some are dedicated to major companies, but the buildings are generally renting out space to many different tenants. Residential buildings are mostly excluded from the core of downtown, which is why some areas can feel dead on weekends and evenings. There are high-rises not too far away, but the daytime population of downtown remains dominated by employees who live somewhere else.

It can also be important to show where the boundaries of each region of population are. In sprawling suburbs, the data you try to use in order to make decisions becomes much more sparse. Individual blocks are larger, sometimes huge. Public streets are few and far between, with individual properties connected and yet separated by private drives and parking lots, and bounded by landscaping features such as ditches, creeks, ponds, bushes, and trees.


So, overlaying employment data can help. Unfortunately, it usually isn't available at very good resolution. For my first and third map in this post, I used Transportation Analysis Zone employment information from the MetroGIS DataFinder service overlaid on top of block-level population data from the U.S. Census Bureau. The second and fourth map only show population data (population count per block as a number, density for the color). All of the maps were created using the open-source tool, QGIS.  The first and third map also include bus lines and proposed transit corridors (LRT for Blue and Green Lines, and BRT for the Red and Orange Lines).

Frustratingly, a number of corridors were not included because of a lack of easily-accessible data. The Gateway Corridor is still being nailed down, and was not represented in the DataFinder shapefiles. Even the Bottineau corridor wasn't included, so I added that. "Arterial BRT" and potential streetcar corridors weren't included either, though I did add station objects along Route 54 which is the highest-frequency limited-stop bus route in the Metro Transit system. Other "HiFrequency" lines are shown solid, because they have stops every 1-2 blocks in most cases. The precise routing and stop locations for planned routes are also in flux, so these maps will need to be revisited.

Each TAZ is usually composed of multiple census blocks, so overlaying the two can have some strange effects (I believe the appearance of heavy density northeast of University and Snelling is largely due to employment at the Griggs-Midway building by Fairview melding into an area that's mostly residential, for instance). There also seems to be missing data in a few cases, particularly on the employment side (the University of Minnesota has a lot of employees, though the East Bank campus is seemingly empty). The maps are not the One True Way to see the world.


Still, it's pretty remarkable that we're currently planning to build LRT out to an area with population dispersed like we see above, and with such a sparse street network, when projects to connect denser areas get discarded or downgraded. The number of people per census block is actually relatively high in Eden Prairie, often measured in the hundreds, but it's difficult to move around in the area because of such a stingy layout of roadways, often lacking sidewalks (this map is at the same scale as the second one in this post which focused on south Minneapolis). Looking at aerial photos, the region seems to be bustling with activity, but a lot of that is due to low-slung office and retail buildings with massive parking lots. All the asphalt makes it look more built-up than it is.

The city also has a big employment base—more people work in Eden Prairie than live there. But despite that large number of workers, the employment density isn't all that high. There may be some missing data from employers who declined to submit per-site numbers, but it appears we'd have a much better return if we looked at improving the situation along the I-494 strip and the France Ave/York Ave corridor in Bloomington, Edina, and Richfield, for instance.

But looking at our existing Hiawatha Line seems to show that rules that you think are important can be broken and still result in something that gets broad appeal: There's very little residential density along the east side of Hiawatha Avenue, because it has been and continues to be used as a freight rail corridor for industry (mostly grain traffic). The airport and Mall of America are also critical to the line's success, but they're in areas which are otherwise pretty empty. The Central Corridor also neighbors large areas without any residential use—though that line tends to skip over the "empty" space while other routes seem to skip the populated areas.

The region is certainly running an interesting set of experiments with these routes, as well as some of the others that I couldn't get to. Which ones will turn out the best?

Friday, October 18, 2013

Japanese Shinkansen documentary

Richard Hammond hosts an episode of Engineering Connections which looks at some of the engineering challenges faced by developers of the Japanese Shinkansen high-speed rail system.

Friday, October 11, 2013

Amtrak versus the shutdown

A rail crew replaces a switch on October 4th in preparation for reconnecting the Saint Paul Union Depot to allow Amtrak service.

Amtrak, the National Passenger Railroad Corporation, could cease to be a "national" network if the federal government shutdown drags on. The company has said that it can ride out a short-term shutdown, on the order of "several weeks", but it's unclear what might happen beyond that point. The good news is that the company is in better financial health than they have been in a decade or more, somewhat extending the window of time they can operate without federal assistance. The bad news is that it would still be a extremely difficult task to find the $100 million or so per month they would need to keep running the company as it is today.

The federal shutdown has also come at an impeccably bad time for boosters of the Saint Paul Union Depot project. As you read this, work is progressing on finally connecting the station to mainline tracks in order to allow the Empire Builder to return to a downtown stop it hasn't visited since Amtrak was formed in 1971—the return may only be shortlived.

Of course, political figures have proposed zeroing out Amtrak funding numerous times over the years, or slicing it up into regional businesses, so some thought has been put into how the company might try to cling to life. Until now, at least, the railroad has always received enough support to keep limping along. There have been occasional bursts of greater investment, but never the sustained influx of cash that the company has needed to get a truly stable footing.

Probably the most popular idea has been for Amtrak to condense service down to the Northeast Corridor, the 453-mile D.C.–Philadelphia–New York–Boston line which accounts for more than a third of passenger traffic on the system, and about half of the revenue. That's because there are dozens of trains per day on the NEC, a frequency of service far higher than anywhere else (most of Amtrak's overall route mileage only has one or two trains per direction per day). The NEC is where Amtrak most consistently turns a profit against operating expenses, though that doesn't necessarily cover the costs needed to maintain and upgrade the rails and power systems (for instance, much of the line uses an antiquated 25Hz power supply, unlike the 60Hz power used throughout the rest of the country, and it will cost billions to fix).

Shrinking down to just the NEC, or the NEC plus a few branch routes, probably wouldn't be necessary. Under the Passenger Rail Investment and Improvement Act of 2008, states must take over as primary funders of routes under 750 miles in length. States had until October 1st of this year to negotiate contracts, with October 16th (next Wednesday) set as the service cutoff date. Three states were still in talks at last count. So those routes (roughly 30 across the country) would probably still be able to operate going forward—but those are regional routes which would still become isolated from each other in many cases.

Most at risk are the long-distance trains, such as the Empire Builder which runs through Minnesota. These trains are still mandated under federal law, but don't see any significant financial support from the states that they run through. Would the mandate to keep them running still be valid if the flow of federal money dries up? The long-distance routes are big, meaty targets simply because they are so long. The Empire Builder runs 2,200 miles, not counting the extra split it does near the west coast to serve both Portland and Seattle. In order to maintain daily service, four or five trains need to be in motion on that route at all times, as opposed to a short corridor route of the same frequency which would only need one or two sets of equipment. On the whole, the ticket revenue covers a similar percentage of operating costs on long-distance trains as it does on shorter routes, but shorter corridor trains have a history of needing some amount of state-level funding (which Amtrak counts as "revenue"—something to be wary of if you ever look at their monthly reports).

One possible solution, though it has its limits, is simply raising ticket prices on these routes. Passengers on the Empire Builder paid an average of 18 cents per mile in fiscal 2011, yet the train costs about 35 cents per passenger-mile to operate. Fare increases would be steep, but might barely be able to cover the gap. On the busy Saint Paul to Chicago segment, it's hard to find a regular adult fare any lower than about 32 cents per mile—the fares get much cheaper in the less-populated areas west of the Twin Cities.

Going by that, Amtrak may already run at break-even along the Saint Paul to Chicago corridor, so one alternative could be to truncate the line here. The ridership picture would change drastically, since many people who get on or off within the MSP–CHI corridor are traveling to or from points farther west, but shortening the route would massively improve the on-time reliability of the train, particularly eastbound. The train usually picks up some delay on the way as it plies its way 1,800 miles from the Pacific Northwest to Saint Paul (earning the nickname "Late #8"), and these past couple of years have been particularly bad.

But the Empire Builder is not the most expensive of Amtrak's routes. It has one of the best ratios of ticket revenue to operating cost of the long-distance services, so it may not be the first one cut if things came to that—the Southwest Chief and California Zephyr are both worse in terms of absolute-dollar subsidy. But the Builder isn't far behind on that measure—despite a good operating ratio, the nature of the train and it's long route both contribute to making it one of the most expensive routes to keep running, to the tune of about $57 million per year (the $975 million to be spent on the new Vikings stadium could keep it running for 17 years, or far longer if costs were divided up among states along the route).

Collectively, the long-distance trains need about $600 million per year to keep up with operating expenses. It's not outside the realm of possibility that Amtrak would just start taking out loans to run the services for a while. The company now has its lowest debt load in more than a decade, so that could work for a short- to medium-term period. Of course, that's dependent on credit markets still working properly. With the threat of federal default looming nearby, they could seize up again like in 2008.

What will actually happen? It's hard to say. It probably shouldn't be a top priority as compared to other effects of the shutdown, but it's another thing to keep an eye on as the days turn into weeks. While hoping for the best, transportation officials in states, cities, and counties along long-distance routes should prepare by identifying funds that could be shifted to Amtrak if the need arises.