New York’s skyscrapers soar above a century-old steam network that still warms the city. While the rest of the world moved to hot water, Manhattanites still buy steam by the megapound.
Since 1882, Manhattan has delivered steam into the homes and businesses of its citizens. It is used for: pressing linens at The Waldorf Astoria; cleaning crockery and heating food in restaurants; washing clothes at dry cleaners; sterilizing medical equipment at NewYork-Presbyterian Hospital; by the Metropolitan Museum of Art to control the humidity levels and temperatures around its artwork. And it is used by Manhattanites, in both iconic buildings and regular apartment blocks, to heat their space and water.
Steam functions like any other utility: produced centrally, metered, and delivered into homes and businesses through a 105-mile-long grid of pipework. Like electricity, sewage, and water, it plays an integral part in the daily operation of the city.
Today, the Manhattan steam system is responsible for heating 1.8 billion square feet of residential, 700 million square feet of commercial, and 90 million square feet of industrial floorspace. This represents over three quarters of Manhattan’s total residential footprint.
Steam enabled the growth of Manhattan, providing efficient heating to an increasingly vertical city. Many other cities have adopted district-level heating systems, but very few use steam to distribute that energy. Why does some infrastructure survive while others become obsolete?
A short history of heating
Heating represents around half of all global end-user energy consumption. Much of this energy is spent warming homes, offices, and water – for cooking, cleaning, and washing – but roughly half goes toward various industrial uses and a smaller percentage toward agriculture. Keeping spaces warm is amongst the most foundational uses of energy there is.
Before modern central heating, keeping homes warm was inefficient, inconvenient, and sometimes deadly. Most residential buildings would be centered around a fireplace or stove, often burning wood, coal, surplus crops, or dung. Traditional fireplaces are immensely inefficient, drawing in 300 cubic feet of air per minute for combustion and expelling up to 85 percent – along with the heat – up the chimney. They are inconvenient because the fuel source needs to be regularly replenished: felling, hauling, and splitting firewood could take up to two months of human labor per person per year. And they are deadly because these sorts of solid fuels, full of impurities, are highly polluting. A recent study found that a fireplace pumps out 58 milligrams of particles under 2.5 microns in diameter (PM2.5s) per kilogram of firewood burnt. This means that every hour you spend in a room with a fireplace burning wood reduces your lifespan by about 18 minutes, equivalent to smoking 1.5 cigarettes..
Efficiency and pollution concerns motivated the rediscovery and refinement of central heating systems, moving the production of heat away from its consumption. James Watt laid the groundwork in the late 1700s, connecting a system of pipes to a central boiler in his own home.
In 1805, William Strutt heated cold air using coal and distributed it throughout homes using ducts; around the same time, some wealthier homes in France began using similar fire-tube hot air furnaces.
By 1863, the radiator had taken definitive form, perhaps the most significant breakthrough in modern household heating. Running steam through pipes that have been folded over many times allows for a large amount of surface area in a compact space. The heat is then transferred from the steam into the ambient environment via convection and by directly radiating into the room. Individual radiators can also be fitted with valves, which allows for heating control room-by-room.
While the radiator allowed for much better distribution of heat within a building, each building needed its own boiler. But on-site boilers, especially those for larger buildings, came with both economic and ergonomic limitations. Steam heating required large boilers, coal storage areas, and extensive piping systems, which take up valuable floor space. As the systems became more advanced, they often also became more complex, necessitating skilled technicians for installation and maintenance.
Coal logistics presented its own set of problems. The regular delivery, storage, and handling of large quantities of coal is labor- and space-intensive: heating the Empire State Building would need more than 45 tons of coal per day, roughly two shipping containers’ worth. Thomas Edison’s first electric power station on Pearl Street opened in 1890 and needed upwards of 20 tons of coal each day. All this coal needed to be moved across bridges or via boat onto Manhattan Island for storage in a city where space was increasingly at a premium. The price of prime parcels of land increased tenfold between the 1790s and the 1880s, spiking from $400 an acre in 1825 to over $1 million an acre at the end of the century (over $30 million in today’s dollars).
Nor is all coal created equal. Boilers and furnaces need an expensive class of coal to burn hot enough. Anthracite coal, with its high carbon content (between 92 and 98 percent) and few impurities, burns hot and clean but is difficult to light: one exasperated user declared that ‘if the world should take fire, the Lehigh coal mine would be the safest retreat, the last place to burn’. (This would have been a bad strategy: in 1962, the anthracite seam under Centralia, PA – 25 miles to the west of Lehigh County – caught on fire and is still burning today.)
Bituminous coal is much softer, with a lower carbon content (between 45 and 86 percent); it therefore burns less energetically and pollutes much more. And all coal needs to be stored, handled, and its ashes cleared out regularly.
The unpleasantness of soot and smoke led US cities to begin to legislate against it in the 1880s, although these restrictions were often ignored. When strikes drove up the price of anthracite in 1902 and some New York City plants switched to bituminous coal, the resulting smoke choked the city, violated local ordinances, and alarmed residents. Pollution encouraged invention: from the 1830s to the 1880s, the ‘patent office . . . groaned with inventions to save fuel, abolish smoke, evaporate water and utilize steam . . . still leaving a great need unsupplied’, according to a contemporary account.
The growth of Manhattan in the nineteenth century
Between 1850 and 1900, New York’s population ballooned. Manhattan alone grew from 515,000 to 1.8 million residents. This increase led to severe overcrowding, with population densities in some areas reaching 632,000 people per square mile, ten times higher than today.
Overcrowding led to both moral and public health concerns, and anti-slum legislation was introduced to try to improve living conditions. This made space in Manhattan still more scarce. Elevated train lines, such as the Ninth Avenue Line, which opened in 1878, allowed the city’s effective size to increase and the partial suburbanization of the outer city. But fares weren’t cheap enough to allow low-wage laborers to commute regularly to and from the new communities forming uptown. If the city were to continue to grow, it would need to build upward.
Two major innovations allowed the development of ever-taller buildings. Physically climbing up stairs – bringing food, water, and fuel with you – places a practical limit on how high a building can practically be: as high as land prices rose, no Ancient Roman, medieval Byzantine, or Hausmannian residential building went above six storeys. The introduction of the safety elevator, by Elisha Otis in 1852, made taller buildings practical and desirable. Otis’s design incorporated a safety brake that would engage if the hoisting cable broke, alleviating fears of catastrophic falls.
Concurrent with safety elevators was the use of steel building frames. Traditional brick and stone structures face several limitations that prevent them from exceeding around 12 storeys, the most important of which is weight distribution: as a building gets taller, the weight of the upper floors puts enormous pressure on the floors below.
Advancements in steel production, including the Bessemer process – oxidizing impurities in molten iron by blowing air through it – made it both affordable and economical to use steel frames, which allowed for a ten times higher strength-to-weight ratio. Designing the building around a skeleton of vertical steel columns and horizontal I-beams allowed buildings to redistribute their weight into this structure, freeing up tensile pressure from the walls and floors. This meant that the rest of the building could be made from lighter non-load-bearing materials, making it possible to build higher still.
So New York shot up. In 1890, the tallest building reached 18 storeys. By 1899, the Park Row Building stretched to 31 stories, and in 1908, the Singer Building soared even higher to 41 stories. Vertical expansion allowed the city to create more residential and commercial space, but that space needed to be heated, powered, and watered.
The New York Steam Company
To solve this, New York turned to district heating, an invention of Birdsill Holly (1820–94), a self-made man from Lockport, New York. Holly understood the problems of nineteenth century cities deeply. As urban populations swelled, fire risk became catastrophic: large parts of Manhattan would be destroyed in three great fires between 1776 and 1845. Early hand-pumped fire engines could deliver only 30 gallons of water per minute. Among Holly’s earliest patents were water pumps such as an ‘elliptical rotary pump’ – the precursor for his steam-powered fire engine in 1856.
His ‘Holly System of Direct Water Supply and Fire Protection for Cities, Towns and Villages’ provided a constant supply of water at pressure to both homes and fire hydrants, lowering costs and removing the need for reservoirs and standpipes. It was so successful that 23 other cities copied it, without crediting Holly, and were later forced by the U.S. Supreme Court to pay him damages. In the 1870s, he began work on a skyscraper project in Niagara Falls, which he then took to Long Island, suggesting that overcrowding could be improved by building upward. Mocked as ‘the farmer from the west’, he abandoned the project altogether and went home to Lockport.
Holly returned to plumbing. Drawing from his research in steam-powered pumps, pipework, and urban density, he began to create plans for a district-level heating system. If he could distribute steam to buildings as he had water, then multiple buildings could be heated by a single boiler, reducing fire risk and centralizing heating logistics. He began by running steam through one and a half inch pipes, later upgraded to three inch pipes supported by wooden trenches insulated with asbestos. The steam would be produced in a boiler in his basement at 31 Chestnut Street and passed across his garden into the houses of his neighbors.
By 1877, he launched the first practical district-scale test of his system. Over the course of the year, the network was extended to the surrounding area as Holly ironed out the problems and registered customers. Each stage necessitated new inventions: valves used to draw steam off the mains, regulators to stabilize pressure, steam traps to handle condensation, brackets to handle the thermal expansion of the pipes, and meters to measure the flow rate and bill customers accordingly.
By the time winter had arrived, 20 houses had been registered. By this stage, the network encompassed three miles of pipe, supplied by three boilers, at a pressure of 25 to 30 pounds per square inch. According to a contemporary account, ‘all the houses, to which steam had been supplied during the winter, had been most comfortably heated, and the discomforts of the old methods done away with . . . heat could be supplied at a much lower cost.’
The test was a success, so he incorporated the Holly Steam Combination Company, patenting his more than 50 inventions and exporting district steam heating to other cities. By 1882 at least twenty Holly steam systems were in operation, including: Denver, CO; Auburn, NY; Troy, NY; Springfield, MA; Detroit, MI; and Hartford and New Haven, CT. (The oldest, in Denver, has been running continuously since 1880.) By far the biggest system, however, was in New York City.
Manhattan had some municipal utilities already: in the 1830s, the city had begun to lay water pipes, bringing fresh water from the Croton River in Westchester County through 41 miles of enclosed aqueduct. In the 1850s, private companies such as the Manhattan Gas Light Company started to run gas mains down the avenues, replacing oil lamps across the city.
But the idea of distributing steam at city scale was a little more audacious, even in the context of Gilded Age America. It was an associate of John D. Rockefeller, an entrepreneur and ‘popular dandy with a flair for equipage and flowered vests’ called Wallace Andrews, along with his engineering partner, Charles E. Emery, who came across Holly’s designs and decided to implement them in New York City. Andrews worked to arrange the financing and convince the New York City council to permit the development of the steam system. (A task made easier by Edison, who was already digging up the streets to install electrical cabling; Emery and Edison even crossed paths underground.)
In 1882, its first year of operation, New York Steam made $200,000 (about six million dollars adjusted for inflation) in gross revenue. An early steam station was built at 172–176 Greenwich St, on the site of today’s World Trade Center memorial; it contained 64 boilers of 250 horsepower each, distributed across four floors. Coal was brought into the upper levels of the building, where it fed the boilers through a chute; ashes were dropped into the basement of the building in the same fashion. By 1884 the Company had five miles of pipes in active use – feeding steam to 250 consumers, from the Battery in the south up to Murray St, by the modern-day Civic Center – and was profitable.
Replacing individual boilers with a central system had significant town planning and public health benefits. In a city as dense as Manhattan, coal logistics was a growing concern – the New York Steam Corporation estimated that its central system displaced 1.2 million tons of coal and the smoke from more than 2,500 chimneys, requiring 700 five-ton truckloads per day for 300 days per year to handle the coal and ash – and the risk of dying in a fire was increasing (as were insurance premiums).
Andrews himself would die in a house fire in 1899. But the system he helped create outlived him and proved remarkably flexible, adapting to the city’s rapid growth both uptown and skyward. As New York expanded, numerous new buildings tapped into its growing steam network, saving them the costs and complexity of building their own heating systems. Each new building made the system that much more efficient because the costs of laying down the pipework and moving the coal could be amortized over more customers’ bills.
The first customers were mostly office buildings, such as the First National Bank at Broadway and Wall Street, but users included busy restaurants (one serving ten thousand meals per day), electrical power generators, laundries, and public baths. Network steam was even used to melt snow in the streets. Most of the Art Deco skyscrapers of the 1920s and 1930s were supplied by the New York Steam Company, including the Standard Oil Building, the Ritz Tower Building, the New York Life Insurance Company, the Paramount Building, the Chrysler Building, the Waldorf-Astoria, the Rockefeller Center, and the Empire State.
In 1931 – emerging from the financially turbulent 1910s, when the company went into receivership in 1918 – its annual profits had reached $2 million. By 1932, it was the largest district heating system in the world, with 65 miles of pipes, bringing heat to more than 2,500 buildings.
The modern New York steam system
Today, the network, now owned by utilities giant Consolidated Edison – ConEd – operates four steam production sites in Manhattan and one in Queens. They also purchase additional steam from a 322 megawatt plant in Brooklyn, which is managed by a separate company. ConEd’s largest facility sits along the East River, taking up the lion’s share of a city block on 14th Street. At this plant, a dozen massive boilers heat water to approximately 177 degrees celsius and push it into the network at 150 pounds per square inch (roughly equivalent to ten atmospheres).
The scale of all this is considerable. The six boiler sites together have a total capacity of roughly 11 and a half million pounds of steam per hour; at peak times they use over nine million. Every gallon of water produces just over eight pounds of steam, which means that the system consumes nearly two Olympic swimming pools’ worth of water per hour during the winter.
Once heated, steam is sent out from plants through large pipes called mains, which are usually between two and three feet in diameter, before spreading through a network of 105 miles of smaller pipes latticed under the city’s streets.
The older pipework is made of cast iron, often still coated in the asbestos in which Emery clad them, which can cause problems during regular maintenance and when the pipes crack. (When the last major rupture occurred in 2023, the city shut down seven blocks and washed neighboring buildings and streets out of ‘an abundance of caution’.) New pipes are made from a carbon steel alloy and wrapped in insulation, usually mineral wool, and sit within prefabricated concrete ducts, which both lessens heat loss during transit and also prevents nearby pipes and wiring from being affected.
Most of these pipes sit deeper than the other utilities, anywhere from four to 15 feet below street level. The deepest mains pipe runs below the Park Avenue Tunnel, 30 feet underneath the trains shuttling commuters to and from the Bronx and Grand Central. The pipes are accessible through manhole covers in the street, with special telescopic tooling used by workers to open and close the valves that regulate system pressure. Individual sections of pipe are welded together and bound into the concrete with strips of metal called anchors. Metal expands and contracts when heat levels change, so at various points along the mains, the pipes run through expansion joints to absorb this movement safely.
Whenever a new building wants to tap into the network, it installs a service valve and the corresponding piping, which rises up a few feet and carries the steam into the basement, where its pressure is lowered to 125-ish pounds per square inch. (This is structurally similar to electricity grids, where current is transmitted over longer distances at a higher voltage and then reduced at a local substation for local distribution into homes and offices.) Once the steam has passed through the meter and the pressure-reduction substation, most buildings then pipe it through a heat exchanger for distribution into individual apartments and rooms.
Consolidated Edison most commonly uses an ‘orifice plate’ meter to bill its customers. A thin, flat metal plate with a precisely machined hole in the center, an orifice plate is set within the pipeline, perpendicular to the direction of flow. As the steam is forced through the hole, sensors measure a drop in pressure, and work backward to calculate the flow rate.
Steam is billed by the megapound (one million pounds), calculated from a fairly baroque hierarchy of service types and tiers. All apartment building connections pay a standing service charge of $3,555.43 per month. Usage fees are then added on top, paying anywhere from $11.35 per megapound in the summer to $35.013 during peak season.
According to data from the NYC Department of Buildings, 201 East 62nd Street, a mid-century residential building in Midtown with 70 apartments, consumed approximately 5,962 megapounds of steam in 2021. Its steam service cost the building roughly $16,741.23 per month or $239.16 per apartment per month. Lincoln Towers, at 150 West End Avenue, built around the same time, has 454 apartments and pays no less than $47,370.80 per month for its 19,483 megapounds of steam, or $104.34 per apartment per month; the pricing structure strongly rewards density.
A large part of the engineering challenge in maintaining the network is managing how the pressure changes as steam moves throughout the system, being drained into buildings as it goes. Engineers can control the inbound pressure for any given customer very accurately using valves on either side of the service valve and monitor usage across the system with sensors. When the pressure gets too high, valves open and vent the steam through a grate, delighting tourists and enraging taxi drivers. If the pressure is too high for a sustained period, ConEd shuts off some of the boilers. If the pressure is too low, ConEd can increase the supply by switching on new boilers or closing valves elsewhere. Large parts of this feedback mechanism are now automated, and modern electronic sensors and high-precision valves allow the system as a whole to run at a high level of efficiency: 60 percent, end-to-end.
As steam cools, it condenses into water droplets called condensate, a substance that accumulates at low points in the piping system. Condensate poses several significant problems. It absorbs oxygen and carbon dioxide, which makes it highly corrosive. The difficulty of accessing the underground infrastructure, especially in older parts of the network, can make this corrosion costly to repair and, in some cases, highly dangerous. Condensate left in pipes can freeze during the winter, leading to blockages and bursts.
But the most dramatic failure is called a waterhammer. Sometimes the condensate can form a ‘slug’ of water that fills the entire cross-section of the pipe. When a slug comes into contact with the high-pressure steam, it gets pushed through the piping and accelerated to ridiculous velocities, often reaching over 100 miles per hour (50 meters per second).
When this fast-moving water reaches a closed valve, sharp bend, or other obstruction in the pipe, it crashes into whatever it has found, producing a spike of pressure much higher than the standard operating pressure of the system. This can cause significant damage and in some cases total system failures.
The last total system failure was in 2007, when an 82-year-old pipe at 41st and Lexington exploded, showering Midtown in debris. Heavy rainfall had cooled the pipes, producing large amounts of condensate quickly, and a clogged steam trap meant that the system was unable to expel the water. When this build-up hit a critical level, the internal pressure shot up, causing the explosion. Almost fifty people were injured, and one woman died of a heart attack while fleeing:
District heating in the twenty-first century
Steam was a logical choice of heat transfer medium at the turn of the twentieth century, since there was a ready source of it from existing power plants, and its higher temperatures meant it was useful for industrial processes as well as domestic needs. Steam rises, making it a natural choice for tall buildings in dense urban environments. Although steam loses heat as it travels – especially through older systems with poor insulation – its initial high temperature and high energy content mean it can deliver sufficient heat over longer distances.
By the 1930s, designers of new systems had switched from using steam to hot water to carry heat. Hot water is now the preferred medium for several reasons.
Water systems distribute heat via hot water at temperatures as high as 180 degrees celsius (under a high 147.9 pounds per square inch of pressure so that it doesn’t turn to steam) and as low as 40. Water produces a much higher operating efficiency, measured as the ratio of usable heat delivered to customers to the total energy used at the heat source. Lower temperatures mean a smaller heat differential between the water and the ambient environment, which means less heat loss overall. Similarly, insulation materials are generally more effective at lower temperatures since thermal conductivity tends to increase with temperature. All in all, steam systems usually lose 10-25 percent of their heat during distribution; low temperature water systems can get this as low as 3-5 percent.
Lower temperatures and pressures have other benefits: they reduce the risk of high-pressure accidents and reduce the risk of leaks, meaning easier, safer – and therefore cheaper – system installation and maintenance.
Managing condensate also adds much complexity. Condensate needs to be separated from the steam and drained away or pumped through a return pipe back to the boilers to be reused. A majority (usually 50–60 percent) of the capital costs of district heating systems are in the pipework and pumps, which can make return pipes far too expensive to use for larger systems. But draining away the condensate causes a loss of 15 percent of the total energy input, which harms operating efficiency.
Hot-water systems also allow for easier integration with renewable energy sources since water can easily be heated to a sub-boiling temperature using an electric-powered immersion heater. Nuclear power plants famously use a lot of water – between 1,514 and 2,725 liters of water per megawatt-hour, of which roughly 15 percent is turned into steam – to cool their reactors. Rand reactors in Beznau, Switzerland, and Haiyang, China already provide hundreds of gigawatt hours of heat output per year to district heating systems. Another source might be data centers, which convert almost all of their electrical power input into heat. An individual large data center can produce 20–50 megawatts of heat; in 2023, global data centers consumed 7.4 gigawatts of power in total, enough energy to heat London. Some new developments are even using residual heat from the London Underground via heat pumps to power their district heating systems.
Besides their natural efficiency and amenability to using a variety of renewable heat sources, water-based systems have another interesting benefit: water has a thermal capacity 2.08 times greater than steam at 100 degrees, which means that it stores twice as much heat per unit mass. This capacity allows water-based systems to capture energy during periods of low electricity demand or high renewable generation and discharge it as heat later, acting as a form of city-wide energy storage. Using a city’s heating system as a battery is not just useful for the electricity grid, it also enhances the overall efficiency of district heating networks since the networks are able to heat their water when electricity prices are cheaper.
There are now tens of thousands of district heating systems across the world, especially in Northern and Eastern Europe. Helsinki (870 miles), Copenhagen (over 1,000 miles), Stockholm (1,200 miles), and Vienna (over 750 miles) are all heavy users of district heating. Former Soviet states, with their long history of centralized planning and heating, are particularly reliant: Moscow’s network alone extends over 10,000 miles. These systems are used by a large percentage of cities’ citizens: Copenhagen provides more than 99 percent of the city’s heating needs via district heating, and the Moscow United Energy Company provides heat to 90 percent of Moscow residents.
In Britain, there are 2,000 heating systems that serve more than one building, more than one third of which are in London. This represents around two percent of UK heat. Perhaps the most famous UK system was a hot-water system in Pimlico, which was opened in 1950 and drew waste heat from Battersea Power Station, providing heat for over 3,000 homes. That system still functions today, although the City of Westminster is considering refurbishing or closing it due to its age and carbon intensity.
Some systems are moving away from steam, and new systems are likely to run on hot water. But engineering is the science of the art of trade-offs, and steam does have its benefits. Steam moves around by itself, so it doesn’t need to be pumped to where it’s wanted; water systems have to pump the water back. Steam systems that drain their condensate require roughly half the amount of pipework. And many industrial processes require steam directly. Thus, there may still be circumstances where steam is preferred.
Where district heating makes sense
District heating is a truly district-level technology. It needs a high density of heat demand: heat loss and infrastructure costs generally make it unviable below a linear heat density (heat sold per meter of network pipe) of two to three megawatt hours per meter per year. This corresponds to a minimum density of roughly 12 dwellings per acre (about 30 dwellings per hectare), approximately the average for new suburban developments in the UK, but a fairly high level of density for American suburban development: R1 residential zoning in Los Angeles, with its minimum lot size of 0.114 acres, wouldn’t make the cut, ruling out three quarters of all residential land in the city.
District heating systems therefore target smaller, denser, developments: housing estates, central business districts, government complexes, and campuses. Many systems are anchored around hospitals, universities, or industrial facilities, which give a reliable base load demand helping to nudge viability upward.
But the upfront capital costs are high: according to one 2014 study, British district heating networks cost on the order of £1,242 per meter of piping. The dysfunction of infrastructure development in the Anglophone world, with its high construction costs and baroque planning regimes, only compounds these costs. There are commercial incentives for residents to support these schemes: in Denmark, 94.4 percent of district energy (admittedly via a regulated tariff) was cheaper than an alternative independent source; and some UK calculations suggest annualized saving of £350 per year compared to gas boilers and £950 per year compared to air-source heat pumps. However, if district heating is to thrive, it will likely need significant policy support and financial incentives to overcome these initial barriers.
New York City’s municipal steam system is an iconic anachronism: a fascinating part of the city’s daily life and visual language, a foundational part of its history, and a system that has been exported and refined worldwide. It was an essential technology for its time as the buildings of Manhattan began to reach up and nudge the sky even higher. Future cities likely won’t be powered by hundreds of miles of steam pipes, but significant parts of them may well be heated and cooled by district systems pumping water through highly insulated pipework, stretching out like tentacles under the streets.