Industrial change created a peaking problem for Chicago transit

[An entire month without blogging -- let's fix that. This post started with a Twitter conversation about the unusually low peaks in how Montreal schedules its Metro trains, perhaps because it's not as 9-5 as other cities. A note about the charts: it turns out that I can't embed Datawrapper charts on WordPress.com, so the ones below are screen caps. Just click on the chart to go the original chart and see the source data.]

Along the lines of “the best transportation plan is a land use plan,” sometimes land use changes can impose huge costs upon the transportation system. As an example, let’s examine how industrial change in central Chicago triggered vast, and costly, shifts in how the CTA arranges its services.

Chicago skyline in 1970

Chicago

Chicago skyline in 2010 (slightly narrower view)

Popular perception understandably saw downtown Chicago as a boomtown: Enough skyscrapers were built to house all of downtown Philadelphia’s offices, plus all of Glenview or Moline’s residents. Within the high-rises, private-sector office jobs (in business services and finance) grew by 53%. Yet the total number of jobs in Chicago’s Central Area (source) grew surprisingly little in the 1980s and 1990s — by just 10.4%.

The growing skyline masked a sharp decline in nearby industrial jobs. Together, the manufacturing, transportation/utilities, and wholesale sectors lost 42% of their center-city workforce. This bifurcating job market, common to many deindustrializing American cities but occurring on an leviathan scale in Chicago, exacerbated the city’s social divides, plunging some neighborhoods into despair and richly rewarding areas just blocks away.

This tremendous economic shift remade the paths of Chicagoans’ daily travel, and to a large extent demanded a reconstruction of the city’s transit system. Despite the Loop’s triumphant skyline, everyday Chicago was for many years a collection of factory towns stitched together along streetcar seams. The factories lined up along the various rail or river routes leading into the city, and the high-level services they required were provided downtown, but their workers came from all over. Terry Clark writes in the essay “The New Chicago School”: “immigrants naturally lived in neighborhoods where they could talk, eat, relax, and worship with persons of similar national background. They would commute even to distant factory jobs to preserve this neighborhood-cultural-ethnic heritage.”

Since so much employment was at three-shift, all-day factories, service levels were remarkably consistent throughout the city and the day; that combined with the city’s grid to create the gridded bus network we all know well. The comprehensive transit system even worked overnight: The 1957 route map lists 69 surface routes and nine elevated lines running all night. Yes, the “L” system did its work of shoveling people into the congested Loop, but even there it only carried 25% of all transit passengers — even to downtown, 75% arrived via the surface lines.

Just like manufacturing, transit is also a capital-intensive enterprise, and having steady ridership all day/all night makes sure that the equipment (and labor) is optimally used. There’s no need to buy streetcars and pay drivers just to shuttle one giant crowd in at 8 AM — and then keep the fleet parked until they leave at 5 PM. Also, it’s all-day transit, not peak service, that enables urban life: as Jarrett Walker writes, “Low-car or no-car lifestyles, in turn, mean that transit has to be available for many of life’s purposes, not just the peak commute.”

The deep spiral of deindustrialization that I mentioned above also changed where and how Chicagoans commuted. Instead of dispersing themselves across the city at all hours of the day — a flow that became better suited to driving anyways — people began piling onto Loop-bound trains for 9-to-5.

Commuter trains always ran highly “peaked” service, with many more vehicles during rush hour, but these services’ peaks have dramatically grown. The commuter line from Hyde Park to the Loop used to run a 2:1 ratio of peak : midday trains in 1939; now that’s a 7:1 ratio.

Bus ridership, particularly crosstown, dropped off — setting off a vicious cycle of cuts (chronicled by Joshua Mason and Graham Garfield) that reduced crosstown bus service to a shadow of the former streetcar empire. Today’s route map counts a mere 17 all-night surface routes; three-fourths of the corridors that used to have nighttime transit now don’t.

Yet parking buses overnight is relatively easy to do, even though idle capacity is expensive in the long run. What’s been much more difficult, and costly, is adding new capacity to accommodate the ever-larger rush hour crowds, particularly for the growing (Loop-centric) rail system and commuter express buses. Already, CTA spent $530 million on the Brown Line Capacity Expansion Project, which increased train lengths by one-third, and more recently spent over $1 billion on a train order that increased its fleet by 17%. Many of its other planned capital projects, like rebuilding the North Side Main and untangling Clark Junction, will also sink huge sums into upgrading the system to accommodate rush hour crowds.

A small countervailing trend has more recently emerged, though. The city as an entertainment destination — as a site of 24-hour consumption, rather than production — has pushed the system to slightly extend evening hours. That said, the efforts will always pale in comparison to the overnight network that once existed, serving not the few who partied all night, but rather the many who worked all night.

Lumpiness: in cities’ property values, and in metro structure

Two only tangentially related thoughts on lumpy growth:

1. Richard Florida in The Atlantic Cities was one of the few major outlets to cover a report from the Demand Institute (a collaboration between Nielsen and The Conference Board) called “A Tale of 2000 Cities.”

The top 10% of American cities account for more housing wealth than the next 90%. The gains in the 2000s were tilted towards the already wealthiest communities.

The report includes an extensive look at a typology identifying nine types of American communities primarily by the strength of their local housing markets, post-recession. In keeping with the name, the results show a striking divergence, with a select handful of healthy markets sweeping up much of the gains — and leaving half of American cities and towns “currently facing fundamental economic pressure.” The report’s summary says: “In today’s global economy, nothing is more important than the strength and sustainability of the local labor market, regardless of whether employers are serving customers in Chicago, Chile, or China.”

If anything, today’s telecom-centric, information economy has resulted in the geography of opportunity getting lumpier, not more diffuse as earlier expectations had predicted — “reports of the ‘death of distance’ have been much exaggerated.” We telecommuters haven’t all decamped to mountaintops. The most valuable places are becoming even more so: they account for not only an outsized share of wealth but also the gains of recent years.

The underlying economic reality, that human capital is what drives most prosperity today, is why I differ from my colleagues who believe that “investment ready places” can thrive based on previous investments in capital goods like housing.

(I’ll have more thoughts in a later post about how macroeconomic changes, and in particular greater economic inequality, have left their mark on “gateway cities.” In the meantime, I highly recommend Ryan Avent’s ‘The Spectre Haunting San Francisco,’ which ties in man-of-the-moment Thomas Piketty as well.)

On another note, the report also has a good omen for suburban retrofits in “favored quarter” suburbs, in the form of an interesting but familiar disconnect between housing supply and demand in “Affluent Metroburbs.” 58% of housing stock in these communities is detached, “but fewer than half [of those seeking to move] say they are seeking a detached single-family home, compared with a national average of 60 percent.”

Among residents of “historic skyline cities,” a broad category that includes both healthy and less-healthy cities, there isn’t exactly a stampede to the exits. 54% of those who intend to move still “intend to stay in an urban area,” and “nearly one in five” wants to move for better schools (hardly the unanimity some cry about).

2. Alon Levy has a great post about how, on a macro scale, the gridded West has a suburban layout that fosters high-coverage bus networks, whereas more organically settled Eastern suburbs have a dendritic, hub-and-spoke layout that lends itself to commuter rail. (Yes, he points out that Johnny-come-lately Washington has, through Metro, grown into the latter pattern.)

This might go some way towards explaining “the Western Paradox” in Brookings’ findings regarding transit access to jobs. In short, Western cities (particularly in the desert southwest) had a strange spread: many jobs were technically accessible by transit, but low transit-to-work mode shares. The highest mode shares were found in older eastern cities, where a large fraction of suburban service jobs are inaccessible by transit.

How would a carbon tax affect DC?

Nature's fuel

The right thing in climate policy for all the big countries is a carbon tax, which is simpler and less vulnerable to fluctuations in emissions than cap-and-trade schemes.” – The Economist

A recent discussion spawned the idea of implementing a carbon tax within DC, and so I wrote up this brief.

What and whom would a carbon tax affect?
A carbon tax, technically a tax upon the carbon content of energy and fuels, would primarily affect electric generation, gasoline & diesel, and heating fuels (natural gas, fuel oil). A narrower tax could affect only fuels, or electricity. The UK’s carbon tax, for instance, taxes various energy sources at differing rates.

Who consumes energy in D.C., and how?
The EIA reports that DC’s total energy consumption is 70.5% imported electricity, 18.7% natural gas, 7.9% gasoline, and 2.7%fuel oil. 66.3% of energy is consumed by the commercial sector (i.e., offices), 19.9% by residences, 12.1% by transportation (i.e., cars & trucks), and 1.6% by industry.

Of carbon emissions within DC proper in 2010, natural gas was 54.6% and petroleum 45.2%. Because DC imports all of its electricity, it has the least carbon intense economy among the states, emitting 91.6% less CO2 per dollar of GDP than the US average. This does not, however, include fuel burned for electricity used by DC end users; 59.2% of DC electricity originated from fossil fuel generators.

Have carbon taxes been implemented elsewhere?
Yes, several jurisdictions have. Finland and Sweden were first, in 1990 and 1991. In North America, the provinces of British Columbia and Quebec have carbon taxes, as does the city of Boulder (on electricity only). Dozens of multinational corporations, including most oil majors, use an “internal carbon price” to evaluate corporate decisions: ExxonMobil’s is $60/ton.

How have these fared?
British Columbia’s carbon tax, unique in its broad reach even though the province works within the framework of a high-carbon-emitting country, “has been remarkably effective in reducing fuel use, with no apparent adverse impact on the province’s economy,” according to a University of Ottawa study. GDP growth paralleled Canada’s, income tax rates fell to the lowest nationwide, and fuel consumption fell by 17.4% per capita.

Have carbon taxes been proposed in U.S. states?
A bill has been introduced in the Massachusetts legislature, and a ballot measure is currently collecting signatures. In Washington state, Governor Jay Inslee has specifically directed a legislative commission to study a carbon tax, and an NGO has proposed draft legislation.

What level of tax would be appropriate?
An easy guideline for measuring the impact of a carbon tax is that a tax of $1 per ton of CO2 results in just less than 1¢ in tax per gallon of gasoline. DC’s current gas tax rate of 23.5¢ per gallon thus implies a tax rate of $27.98/ton of carbon dioxide. (Maryland’s gas tax is now 27.1369¢.) This rate is very similar to the C$30/ton that British Columbia charges.

Where do proceeds of carbon taxes go?
In most cases, as in British Columbia, carbon taxes are a “tax swap,” whereby other taxes — notably on income, capital, etc. — are reduced. Some bills, like that proposed by Citizens Climate Lobby, feature a “dividend,” or direct rebate back to taxpayers. Sometimes, climate actions are funded with a portion of proceeds as well; the Massachusetts bill, for instance, directs $90 million in revenue towards transportation debts and 10% to clean energy. In DC, ambitious plans have been launched, but not yet funded, for transit expansion (by WMATA and DC) and for cutting emissions, and a carbon tax would be one way of funding implementation of those plans. (Boulder’s tax was implemented to fund its climate action plan.) In addition, DC currently pays its annual operating subsidies for both WMATA ($275 million in FY2014: $58M bus, $42M rail, $22M paratransit) and DDOT transit out of general funds, and a carbon tax could be a stable, dedicated source of transit operating funds.

Who are winners and losers?
A carbon tax that includes electricity would have a much broader base and thus wider impact. It would primarily affect the office sector, and as such mostly commuters, but it might also attract Congressional attention. A carbon tax solely on fuels would mostly impact building heating/cooling; again, this would largely fall on offices, but also on DC residents’ heating bills.

Although a carbon tax typically is somewhat regressive, there are many ways to design a carbon tax to mitigate impacts on lower income consumers. In particular, a DC carbon tax could use targeted measures to offset higher home heating costs for low income residents: income tax credits, weatherization or LIHEAP assistance, and transit improvements.

Further reading
Sightline Institute: Carbon Tax Fact Sheet
Resources for the Future: Carbon Tax FAQs
Citizens Climate Lobby: DC Chapter

No subways for you, rowhouse neighborhoods

Nationals Park View
Maybe for these rowhouses, since they’re in a “Land Use Change Area.” Photo by Mr. T in DC.

Metro’s recently publicized 2040 plan for a new downtown loop line elicited a lot of consternation by many city residents, wondering why the new subway line hewed so closely to the existing downtown. The short answer is: because they won’t need new subways. It ultimately comes down to cost and benefit: new subway construction is so expensive that it can only serve the highest-density neighborhoods. Instead of a plan to connect every rowhouse in the city to a subway, it’s a plan to maximize capacity to and through the regional core at a minimal cost.

As Matt Johnson wrote in a recent GGW post, “Metro’s studies found little need for a new subways outside of downtown based on the expected travel patterns in and density of those areas in 2040.” Ben Ross clarifies in a follow-up: “Why isn’t Metro planning more rail lines inside the Beltway? One big reason is that political pressure and federal regulations require it and other transit agencies to look only at current zoning and master plans… This forecast, in practice, is prepared by cobbling together the master plans adopted by local governments, which are not anyone’s best guess of the future, but mostly reflect the desires of locally dominant political forces.”

WMATA’s staff echo this on their PlanItMetro blog: “Many of you have said that we missed or have asked why we don’t have a line to . As part of this plan, we have analyzed almost every corridor or mode that you have identified.”

A recent Cal study by Erick Guerra and Robert Cervero examined the cost-effectiveness of various rail transit extensions vs. the population and employment density of the areas served, and determined a cut-off aggregate density of 100 persons (residents and jobs) per acre as a cut-off for high-cost heavy rail:

Rail cost-effectiveness & density

In this region, per research from Terry Holzheimer at Arlington Economic Development (full PDF), only downtown D.C. and a handful of high-rise suburban centers surpass that threshold:

Gross land use intensities, 2007

Notably, though, the historic neighborhoods of Georgetown and Old Town — while surprisingly surpassing Edge Cities in intensity — fall well short of that density threshold. Laurence Aurbach calculated per-acre intensities for a few other historic neighborhoods, and of them only Dupont Circle [112.7] achieves downtown-level intensity, appropriate given that it’s chockablock with mid-rise office and hotel. Capitol Hill [44.8] and Logan Circle [57.1] also fall well short of 100 persons/acre.

The DC comprehensive plan gives precious little scope for other non-downtown areas within central DC to intensify much further. Nor do they seem to want intensification: DC neighbors will have a cow over one 6-story building one block from Metro. Contrast that to the 33-story apartment towers next to Metro in olde-timey Alexandria, 15-story office towers surrounding the Metro stations in mostly-rural Loudoun County, and heck, 40-story towers proposed along a Beverly Hills subway that’s at least ten years away. Those transit-supportive densities are impossible to build in the “neighborhood conservation areas” that comprise most of the District’s land area.

Meanwhile, the areas designated for “land use change” (in essence to be annexed to the mid-rise downtown) are NoMa and Capitol Riverfront. Connect those two neighborhoods to Union Station and Georgetown,* via bypasses of the Rosslyn and L’Enfant choke points, and add a few connecting points at the edges of downtown so that people can reach destinations like Union Station without going through Metro Center, and you’ve clearly defined WMATA’s proposed downtown loop subway.

DC comp plan policies map

Another instructive comparison to Paris might be in order: Washington, with its low density and low transit cost effectiveness, has 41 heavy rail stations and single-family rowhouses just one or two miles from the core, whereas Paris has flats stretching to the city limits, lower rail construction/operating costs, and almost 300 heavy rail stations within the city.

* We can perhaps give Georgetown a pass, since the crowds of tourists visiting its shops don’t show up in either the resident or employee numbers.

Shorts: movements

Striding

1. Susan Silberberg et al (via Angie at Streetsblog write that placemaking’s true value stems less from physical transformation than social transformation: “The act of advocating for change, questioning regulations, finding funding, and mobilizing others to contribute their voices engages communities.”

In short, it’s not about the bike, or the parklet: it’s about creating social space for a social movement to free now-privatized but publicly-controlled spaces, returning them to public use.

Years ago, this was a key (and under-appreciated) accomplishment of early Critical Mass rides. The event is just a means to an end, a safe space through which a social movement organized; to this day*, many confuse those ends and means.

* it’s arguably lost its urgency now that there are many other organizing venues.

not a maglev

2. There have been a few proposals to build maglev trains in the USA before, including this cross-Maryland proposal ten years ago. So what’s different about the latest version?

In a meeting with President Obama last winter, Mr. Abe offered to provide the maglev guideway and propulsion system free for the first portion of the line, linking Washington and Baltimore via Baltimore-Washington International Airport, a distance of about 40 miles. – Eric Pfanner, NYT

Those previous plans, however, did not feature Abenomics and its tidal wave of printed yen. As much as I’m skeptical of proprietary technologies, a fast and efficient connection between the two cities would certainly be momentous.

3. Thad Hall from the University of Utah (via Washington Monthly & Mischiefs of Faction) graphically shows how the House GOP has marched rightward, using DW-NOMINATE data:

The 50th-percentile average Republican in 1995 (104th Congress) — the red bar — was as conservative as today’s “RINO” moderate. Meanwhile, 1995’s firebrand 90th-percentile revolutionaries (the blue bar) then are *below* average now. The entire bell curve has shifted.

Shorts: getting to the office, and its implications for developers

1. Jonathan O’Connell asks, “With no office tenants and no financing, is the Southwest Waterfront redevelopment in trouble?”

The trouble they’ve had in raising capital underscores the importance of a good phasing strategy. The Wharf needs to start with a big bang for both top line and bottom line reasons:
– the retail won’t pay out without a critical mass of activity on the site; indeed, the 140K sq. ft. proposed for Phase 1 is at the low end for a viable lifestyle center
– the large amount of underground infrastructure (two levels of parking, new seawall, parks) that must be built before the first building pays out would be cost-prohibitive for a smaller project.

2. Earlier from O’Connell, a sign of how much the local market has embraced urbanism (with almost none of this space receiving local subsidies):

Of the 5.5 million square feet of office space under construction in the region, about 4.6 million of it, or 84 percent isn’t just near a Metro station but within a quarter mile of one, according to data from Jones Lang LaSalle, CoStar Group and Delta Associates.

The trouble is that each one of these buildings feeds off of Metro’s positive externality of greater access, but any one building can only make a relatively small contribution towards that transit infrastructure. Even the BIDs and special taxing authorities set up around most of these WUPs focus primarily on low-cost, high-return placemaking projects rather than the much more expensive, long-term work of buliding value with transit. Meanwhile, buildings that attempt a greater level of infrastructure investment, like those at the Wharf, can’t get the running start that they need. Areas with that transit infrastructure in place thus have an edge over those where it remains (and, given scarce funds, will likely still remain) to be built.

Some have taken this to mean that retrofitting suburbia will be necessarily be too expensive a proposition to be economically feasible. On that point, I only half-agree: there are a few cities, primarily in the Rust Belt, where intact and high-quality urbanism is cheap enough to feasibly redevelop. Yet with a growing population — America’s population will grow by about 140 million between 2000-2050, nearly matching the entire 1950 population of 150 million — that’s now concentrated in an entirely different set of cities, retrofitting may well prove to be the cheaper alternative.

3. Michael Andersen in the Green Lane Project notes that it isn’t just Metro access that moves office space:

Kathy Card, the general manager of two office buildings in DC’s fast-changing Chinatown, said last week that she doesn’t ride a bike herself. But watching traffic at the bikeshare station at 8th and H, she said, has convinced her that some of H Street’s four auto travel lanes and two parking lanes should be repurposed for protected bike lanes. Dedicated bike infrastructure is what’s needed, Card said, to help her buildings appeal to the private-sector firms she’s marketing to. “There’s plenty of ability to put bike lanes in,” Card said last week. “Obviously the demand is there.”

4. So perhaps it’s only appropriate that the latest MoveDC plans, leaked via DCBAC, floats the idea of cycletracks on just about every arterial in the city: Independence, Constitution, Connecticut, Massachusetts, Rhode Island, Florida, Bladensburg, Minnesota, Alabama, N Capitol, S Capitol, U, 9th, 14th, etc. — oh, and a bike bridge to Alexandria, too.

5. A generation ago, Joel Garreau wrote in “Edge City” about how one axiom shaped everything about suburban office parks: four car parking spaces per thousand feet of office. Yet today, both numerator and denominator have changed beyond all recognition: the cars are gone and the offices are smaller. As reported by Laura Kusisto, the conversion of Brooklyn’s vast Watchtower printing plant into loft offices proposes a bike parking ratio of 8.3 bikes per 1000′ of office. The resulting indoor garage will house 5,000 bikes — 15-16X larger than dedicated bike stations in Chicago or Santa Monica, and undoubtedly as a result requiring special accommodations like ramps. It would be interesting to know whether bike parking is a way to soak up the dark space in the middle of the floors (that would be convenient) or if they’re warehoused in a basement.

6. Besides, who ever uses that much car parking at their suburban offices? Even the car lobby (!) would rather have cute farmers markets than parking lots:

Brighter isn’t always lighter

Which of these stations is better lit?

Canada Line Station

TransLink system lighting standard for subway platforms: 4 foot-candles

IMG_2524 - Washington DC - WMATA Metro Chinatown Station - After Genesis concert
WMATA system lighting standard for subway platforms: 10 foot-candles

Believe it or not, WMATA hardly has the darkest stations in the business. I was amazed to learn recently that Vancouver’s transit agency specifies platform lighting 60% dimmer than WMATA’s: their standard is 4 foot-candles, vs. 10 foot-candles for WMATA. I usually read when I’m aboard transit, and whereas I have to seek out light on Metro subway platforms, I’ve never thought twice about the brightness on TransLink platforms (admittedly, I’ve spent much less time on the latter, partly due to the automated system’s startlingly low headways).

The difference is that TransLink also specifies high-reflectance, light-colored walls and floors, and directs light into occupied areas so that they feel much brighter. With “passive illumination,” it’s not just how much light is used, but also what the space does with that light. Seemingly minor increases in reflectance for surfaces like walls and ceilings (particularly for indirect lighting scenarios) proportionately increase the brightness one can achieve with a given amount of light.

By comparison, much about the classic Metro station design thwarts attempts at improving lighting — and intentionally so, in fact. Our standards of brightness have increased, partly because illumination has become so cheap. Yet the dark material palette chosen for the stations (unpainted concrete walls and ceilings, burgundy tiles, chocolate brown panels, even the bronze railings) absorb both what little light is added and dirt, which further darkens the stations over time.

Metro points to the efforts that it’s taken recently, including regular power cleaning of the concrete station vaults, existing efforts to add fixtures, and a system-wide re-lamping with more modern (and thus brighter and more energy-efficient) equipment. The fruits of these can be seen at stations like Judiciary Square, which does indeed seem like a beacon of light compared to others in the system.

Yet using more reflective materials can also improve station lighting. That’s the gist behind recent changes that Metro announced to the Bethesda station (previous GGW discussion here), like replacing brown metal panels and concrete walls with brushed metal and clear glass. These changes will definitely help, but a more comprehensive approach could look at other changes that can improve lighting without dramatically impacting the stations’ canonical appearance.

  • A clear polymer coat (not paint) could increase reflectance of existing concrete surfaces, reduce porosity and thus the problem of embedded dirt, and make cleaning easier. Painting the station vaults has proven controversial throughout Metro’s history: Zach Schrag’s book The Great Society Subway points to a 1968 disagreement between the designers Harry Weese and William Lam as to whether to paint the vaults, and notes Weese’s “commitment to ‘pure structure in plain concrete’ ” in criticizing a 1990s decision by WMATA to paint some vaults. Yet materials advances now mean that light reflectance surprisingly has less to do with color as one might expect: a darker color with a slight gloss reflects more than a brighter color with a dull finish.

  • The existing fluorescent tubes are recessed within wells that are out of sight, beyond the platform edge or between the tracks. Since these surfaces are so close to the light sources, small changes here will result in big changes throughout. Cleaning and brightening surfaces within these wells, adding reflectors below the tubes to “catch” light that’s currently pointing downwards, moving wire conduits so that they’re below lights instead of blocking them, and replacing bronze-colored diffusers above the between-track tubes with clear plastic diffusers, will all result in more light to reflect upwards into the station.
  • Acoustic panels in coffer recesses can be brighter. These panels cover a surprising amount of the vaults’ surface area, but because they’re literally in the concrete’s shadows, we don’t tend to notice them very much. These, too, accumulate dirt and dust over time, and as they’re replaced their reflectance could be increased. The new Rosslyn entrance has highly reflective panels embedded within its coffers, which I didn’t even notice the first few times I walked through it.
  • Similarly, the drop-ceiling tiles underneath station mezzanines can be replaced with tiles that reflect more light. Given the low ceiling heights in these spaces and the fact that they’re largely hidden from view, a more ambitious upgrade could replace these with ceiling tiles with embedded LED lamps, reducing both shadows and glare in these areas while improving efficiency over the existing can lights. LED ceiling tiles might sound gaudy, but look no different than the fluorescent panels embedded in most office drop ceilings.

Attention to these details can ensure that the maximum possible amount of light is available within Metro’s subway stations, improving energy efficiency, safety, comfort, and accessibility without altering their iconic appearance.

[A version of this is cross-posted at Greater Greater Washington]