On a freezing morning in Toronto, a 35-foot-wide shaft descends 165 feet through frozen ground and clay, taps into a municipal sewer, and draws warm wastewater into a row of stainless steel plate heat exchangers. Toronto Western Hospital, a hospital complex of roughly 1.35 million square feet, now takes about 90 percent of its heating and cooling from that source.
This is sewage heat recovery, and the reason it matters is simple arithmetic. Every shower, dishwasher, toilet, and industrial process connected to that sewer puts warm water into the pipe. The city then spends enormous energy and burns natural gas or electricity to warm up fresh cold water to replace it. The energy already in the pipe is, in engineering terms, a low-grade heat source sitting under a city block.
It is not a speculative technology. It is running in Toronto, Denver, and Warsaw; is being piloted in Boston, Syracuse, and Duluth; has been commissioned across several German towns; and has been installed at more than fifty sites across Europe. What is still unsolved is not whether it works. It is why a resource that is warm, constant, free, and already installed fails to get built almost everywhere else.
What sewage heat recovery actually is
The phrase sounds more exotic than it is. Sewer heat recovery, also called wastewater heat recovery or sewer thermal energy, is a heat pump setup that borrows the sewer as its cold reservoir instead of the outdoor air.
An air source heat pump, which is what most of us call a heat pump, has to work against the weather. On the coldest day of the year, when a house needs the most heat, the outdoor air is at its least useful temperature, which is when the machine runs hardest and delivers the least. Water is far more forgiving. The same logic explains why ground source and air source heat pumps perform so differently and why cities across Europe are now wiring river water into their district networks.
It is worth being precise about the unit, because it is often misused. A coefficient of performance above 1 does not mean the machine is giving away energy or breaking the laws of thermodynamics. It means the electricity is doing the work of moving heat, and that heat was already there. That is the basis of every heat pump ever built, and a sewer-fed unit is simply moving more heat per unit of electricity than an air source machine because the water is warmer.
Wastewater arriving at a treatment plant or flowing in a large interceptor main typically sits between 10 and 25 degrees Celsius, and unlike air, it barely moves. That stability is what makes sewer heat a better heat source than ground loops in cold climates, where the ground itself changes temperature with the season. Blair Wisdom, energy manager for Denver’s Metro Water Recovery, describes the practical appeal plainly: “Wastewater is a consistent and reliable source of thermal energy because communities continuously generate warm wastewater year-round through everyday activities such as showering and dishwashing.”
The recovery process has four stages, and keeping the two fluid circuits physically separate is the part that matters for safety and equipment longevity.
- Tap the sewer. A wetwell, essentially a concrete shaft, reaches down to a large interceptor pipe and diverts a portion of the flow. The wastewater is screened at the sewer level so only sieved liquid moves forward.
- Transfer the heat. In a plate-and-frame heat exchanger, the dirty wastewater runs down one side of a stack of steel plates, and a closed loop of clean water runs down the other. Heat crosses the metal; the two fluids never mix.
- Lift the temperature. Water-to-water heat pumps raise the captured heat to whatever temperature the building actually needs, which is usually in the 55 to 80 degree Celsius range for hot water and heating.
- Distribute it. In a district system, insulated pipes carry the hot water to multiple buildings, which connect through their own heat pumps. Some systems skip the network and serve a single large building directly.
Two variants exist. In an indirect system, the heat exchanger and its clean loop do the work. In a direct configuration, the wastewater itself feeds the heat pump, which requires heavier pretreatment and filtration. Almost all municipal-scale projects in North America use the indirect arrangement, which is why the Toronto system separates solids before the water ever reaches a heat exchanger.
Why sewer water beats the outside air
Efficiency in a heat pump is measured by its coefficient of performance, or COP, which is the units of heat delivered for each unit of electricity consumed. A COP of 3 means three times as much heat comes out than electricity goes in. Because water is warmer and steadier than air, a sewer-fed machine posts noticeably higher numbers than an air source unit of the same nominal capacity.
The measured results across published field studies and review literature land in a consistent band. Sewer source heat pumps typically achieve a COP between roughly 3 and 4.6, with the best field installations reaching higher.
The single most useful line in the academic literature is also the least reported. In a review of sewer heat exchanger applications, Swiss researchers found that more than 15 percent of the thermal energy supplied to buildings was lost through the sewer system itself. The pipes running from a building to the treatment works are acting as a radiator, dumping recoverable heat into the ground.
That figure is a useful corrective to the enthusiasm. Some of the “lost” energy is not recoverable in practice, but it does establish that the resource leaving a building is larger than the resource arriving at the plant.
The counterweight is capital cost and scale. A 2021 modelling study by University of Colorado Boulder researchers, published in the Journal of Sustainable Development of Energy, Water and Environment Systems, tested sewer heat recovery against US Department of Energy reference buildings and found a sharp divide. A large hotel model paid back in 4.0 to 6.6 years across four US climate zones. A mid-rise apartment model returned 73 to 114 year payback periods, and in three of four climate zones it actually increased greenhouse gas emissions.
The difference was not engineering. It was hot water volume. Hotels generate enough wastewater flow to justify the capital. Apartments do not, and cheap natural gas made the arithmetic unworkable. If you are weighing options for a home rather than a large facility, our breakdown of current heat pump efficiency standards is a better place to start. The study concluded that sewer heat recovery is appropriate for large commercial buildings and not for small residential settings.
Toronto Western Hospital, the largest system in the world
The most detailed operating data available belongs to Toronto Western Hospital, part of the University Health Network. Noventa Energy Partners developed the project with Enbridge Gas and the City of Toronto, and it was approved for construction in 2022 and completed in 2025.
The scale is what distinguishes it. The installation uses 16 Huber BG8 RoWin heat exchangers, which Noventa describes as making it eight times larger than any other installation worldwide, alongside six Rotamat pumping station screens. Total thermal capacity is 19 megawatts, split roughly 10 megawatts of heating and 9 megawatts of cooling, feeding more than 2,400 tons of cooling capacity and over 33,000 MBTU of heating capacity. The system is rated to handle 1,800 cubic metres of wastewater an hour, equivalent to 500 litres a second.
According to the University Health Network, the system is expected over thirty years to supply 90 percent of the hospital’s heating and cooling requirements and save more than 141 million kilowatt-hours of electricity, 130 million cubic metres of natural gas, and 1.3 billion litres of cooling water.
The emissions numbers need care, because the project’s own backers have published several different versions. Noventa states the figure as about 8,400 metric tonnes a year, converts that to the equivalent of removing 1,811 cars from the road, and projects roughly 250,000 tonnes over thirty years. Trane, which supplied the heat pumps, gives the same annual number but writes it as “8,400 tons.” UHN’s groundbreaking announcement instead put the annual reduction at about 10,000 metric tonnes, while still attaching the 1,811 cars comparison to it. The Canada Infrastructure Bank, which put $19.3 million into the project, describes the public impact as roughly 59 percent emissions reduction, and an earlier Huber figure from the 2021 announcement put cumulative savings at about 169,000 tonnes.
Only one of those sets is internally consistent. A typical passenger vehicle emits about 4.6 tonnes of carbon dioxide a year, which makes 1,811 cars equal to roughly 8,300 tonnes, so Noventa’s annual figure, its thirty-year projection and its car comparison all line up. UHN’s larger number is the outlier, and its own release attaches a car comparison that does not match it. All of these remain projections rather than metered results, and since the system only entered service in 2025, none has yet been independently confirmed.
One detail rarely reported is a side benefit. Because the wetwell makes wastewater physically accessible, the hospital gains the ability to take routine samples from the sewer, opening up research access on antimicrobial resistance and pandemic monitoring that most institutions simply do not have.
It is also worth noting who paid for what. The project cost $42.9 million in total, of which the Canada Infrastructure Bank committed $19.3 million, with further support from the Government of Canada’s Low Carbon Economy Fund, Vancity Community Investment Bank, Enbridge Gas and Noventa itself. That funding stack is itself a data point about why these projects are rare. It is a very different profile from the district-scale systems now being planned, and the contrast is worth keeping in mind: a single hospital with a private developer attached can attract finance, while a citywide loop serving hundreds of unrelated buildings has no obvious single owner to underwrite it. The Infrastructure Bank’s own explanation is revealing, describing its financing as the mechanism required “to bridge the project’s commercial and economic gap.”
For the hardware details, see the published project specifications. For the financing, see the Canada Infrastructure Bank’s investment record.
The specification, in brief:
- 19 MW total thermal capacity, 10 MW heating and 9 MW cooling
- 16 Huber BG8 RoWin heat exchangers and 6 Rotamat screens
- Over 2,400 tons of cooling and over 33,000 MBTU of heating
- 1,800 cubic metres of wastewater per hour, about 500 litres a second
- 1.8 billion kilowatt-hours of energy over the life of the project
- A wetwell 35 feet across and 165 feet deep, reaching the mid-Toronto interceptor sewer
- A $42.9 million cost, of which the Canada Infrastructure Bank provided $19.3 million
For context on how a building’s heating load is actually calculated in the first place, our guide to sizing a heat pump correctly covers the temperature-difference maths that determines the COP in the first place.
How much energy is actually sitting in the pipe
The temptation in a story like this is to describe an unlimited resource. The measurements say something more precise and more interesting. Water has a very high specific heat capacity, which means a small temperature drop releases a large amount of energy. But you can only take that energy without pushing the water below a temperature that the receiving ecosystem can tolerate.
A 2026 case study of a sewage treatment plant in Gorzyce, Poland, published in the peer-reviewed journal Energies, worked through this with a full year of operational data. The plant serves 11,593 population equivalents and treats an average of 4,200 cubic metres a day. Its treated effluent held steady between about 10 and 15 degrees Celsius.
Cooling that effluent by just one degree Celsius, the researchers calculated, would release about 1.16 gigawatt hours of heat per year. The water source heat pump they specified achieved a COP of 3.0 to 3.4 and a seasonal coefficient of performance of 3.21. The plant’s own heating demand was 45.1 megawatt hours a year, so the available heat exceeded demand by a factor of roughly twenty-six.
The authors’ own conclusion is more interesting than the headline figure. Because the heat available vastly exceeds the demand, the system is what they call demand-limited rather than source-limited. In a large district system serving many buildings, the sewer stops being the constraint, and building demand becomes the limit. That reframes the engineering question: this is not a resource scarcity problem, it is a capital and coordination problem.
There is a real environmental boundary, too. The same study notes that excessive cooling of wastewater is not recommended, and that discharge temperature into a receiving body should stay above about 277 kelvin. A 2024 study of a multipurpose building in Italy found the effect of a recovery system on the wastewater itself was negligible, changing it by no more than 0.056 degrees Celsius in winter.
Why it is not everywhere: the four blockers
Every one of the four obstacles below is real, documented and none of them is a reason the technology does not work.
1. The economics only clear at scale
The 2026 Energies study found annual savings of about 2,310 euros against gas heating for the Polish plant, on a capital cost of 30,308 euros. That is a simple payback of roughly 13 years, which lengthens to about 15 years if the heat exchanger degrades from 80 to 50 percent effectiveness through fouling. Pairing the heat pump with on-site photovoltaics cut the payback to 7 or 8 years. Under that combination, net present value turned positive in 10 to 11 years, helped by subsidies that covered roughly two-thirds of the capital cost.
So the honest version is that small, standalone installations often do not pay for themselves, and the study says so. Financial viability appeared only because the plant received a water infrastructure grant and a subsidised loan.
2. The best projects keep losing their funding
St. Paul, Minnesota, has the largest hot water district heating system in the United States. Ever-Green Energy has drawn up plans to tap the heat in the roughly 172 million gallons of wastewater flowing daily out of the Metropolitan Council’s treatment plant into the Mississippi River, and use it to replace the natural gas currently fuelling half the district system.
An analysis put the recoverable thermal capacity at 60 megawatts. The city, the Metropolitan Council, Xcel Energy and District Energy St. Paul pursued a roughly $150 million Climate Pollution Reduction grant application from the US Environmental Protection Agency. They were not selected, and a parallel Ever-Green project in Duluth was also turned down. As of March 2026, District Energy St. Paul was still describing Clean Heat St. Paul as a proposal that awaits financial and policy support and remains in planning and stakeholder engagement stages. Minnesota lawmakers have since been asked to appropriate $3 million for a wastewater heat recovery pilot programme. Cities starting now have somewhere to look: Massachusetts runs a funded state grant programme specifically for assessing and building these projects.
A fully designed, fully permitted, decades-scale project that removes a major share of a city’s fossil fuel load, sitting unfunded, is the clearest evidence that the obstacle is not technology.
3. Three agencies have to agree, and none of them owns the outcome
These projects sit at the intersection of a water utility, a district energy operator, a city planning department and a private developer. The Denver case makes it explicit. Metro Water Recovery’s own energy manager put it plainly: “These projects require strong partnerships and long-term planning because they sit at the intersection of water, energy, infrastructure and development.”
The economic payoff usually lands on one party’s balance sheet while the capital sits with another, and the heat source belongs to a third.
4. Fouling, and the honesty problem that comes with it
Wastewater carries solids, biological growth, fats, grit and minerals. Heat exchangers foul, and their effectiveness falls over time. The Polish study modelled this explicitly, showing recovery dropping from 1.16 to about 0.73 gigawatt hours per year as exchanger effectiveness fell from 0.8 to 0.5. Large operating systems use self-cleaning mechanisms and mechanical screening to stay ahead of it, but maintenance is not optional and it is rarely cheap.
Some of the most confident published figures come from design-point modelling rather than long field operation. The Gorzyce authors were unusually candid, noting their dataset covers only a single operational year, that their forecast models do not account for rainfall-driven inflow, and that the reported performance figures “should be interpreted as expected ranges rather than guaranteed operational metrics.”
Denver’s twist: the pollution rule is the business case
The most interesting thing about the Denver effort is that heat recovery there is not primarily a decarbonisation project. It is a permit compliance strategy.
In 2007 the Colorado Water Quality Control Commission adopted statewide temperature standards for waterways, set to protect fish and aquatic invertebrates. Metro Water Recovery has monitored the Robert W. Hite Treatment Facility since then and found that its final effluent runs about 4 degrees Celsius warmer than the chronic stream standard in winter, producing regular temperature exceedances in the South Platte River as far as seven miles downstream at 104th Avenue. Under the standard, Metro has to cut its discharge temperature by roughly 27 percent, from about 15 degrees Celsius down to 11 or 12 degrees Celsius.
It is a catch the industry did not choose. Wastewater treatment plants have to solve a thermal pollution problem, and pulling that heat out of the water before it enters the river is the cleanest method available. Metro requested additional time to meet the limits in January 2025, and its filings state that it “plans to use innovative thermal energy recovery technologies to reduce the temperature of water it discharges to the South Platte River.”
It is the same underlying logic that has cities across Europe tapping river water for district heating. A river is a free, enormous, year-round heat source, and so is a sewer. The difference is that nobody objects to a pipe running under a river, but they do object to one running past their property.
That reframes the whole category. Rather than a city building a heating network because it wants to, a regulator creates a thermal obligation, and the heating network becomes the cheapest way to discharge it. Blaire Wisdom of Metro Water Recovery made the connection directly: wastewater thermal energy represents an opportunity to “help address a regulatory challenge associated with reducing the temperature of our effluent prior to discharge into the South Platte River.”
Denver is now converting its Cherokee Boiler House into a central hub for an ambient loop network that would eventually reach up to ten or eleven city-owned buildings, potentially including around 500 geothermal wells. A 2025 feasibility study estimated full build-out at between $280 million and $320 million over a decade, and found the network could be up to 75 percent less expensive than other decarbonisation strategies for the pilot buildings. A micro-loop linking two buildings and a sidewalk snowmelt system is intended to run within two years, with as many as nine buildings connected by 2030.
The working example already running is Denver’s National Western Center, a 250-acre campus where a system developed through EAS Energy Partners, a consortium led by Enwave Energy with AECOM and Saunders Construction, supplies district heating and cooling. The campus says it will avoid an estimated 2,600 tonnes of carbon dioxide a year.
Thirteen US states are now exploring or developing thermal energy network initiatives: California, Colorado, Connecticut, Illinois, Maine, Maryland, Massachusetts, Minnesota, New Jersey, New York, Texas, Vermont and Washington.
Where the largest and newest projects are
Recovery is not a US story, and the largest and most recently funded schemes are mostly outside North America.
| Location | Scale | Status |
|---|---|---|
| Toronto Western Hospital, Canada | 19 MW, 16 heat exchangers, up to 90% of campus heating and cooling | Completed 2025 |
| Warsaw Żerań, Poland | Up to 50 MW thermal, heat for around 20,000 dwellings, adds 3% renewable heat to the district system | Funded April 2026, PLN 125.4 million from the national environmental fund |
| Denver, Colorado | Micro-loop of two buildings, expansion to nine by 2030, $280 to $320 million full build-out | Pilot in development |
| St. Paul, Minnesota | About 60 MW thermal potential, would displace gas for half a district system | Fully designed, unfunded |
| Syracuse, New York | Proposed energy centre serving the Inner Harbor | Permit request filed 2026 |
| Germany: Tuttlingen, Darmstadt, Paderborn and Frankfurt | Individual projects ranging from roughly 360 kW to 675 kW of thermal extraction | Commissioned 2026 |
| Massachusetts | State grant programme for wastewater energy recovery pilots | Open, amended August 2026 |
Warsaw’s Żerań project is the most instructive recent development, because it is funded specifically to meet tightening environmental requirements. The national environmental fund agreed to co-finance the installation in April 2026, working with ORLEN Termika to recover heat from raw sewage arriving from the Białołęka pumping station. Two heat pumps will feed up to 50 megawatts of thermal capacity into Warsaw’s district heating network.
The Syracuse proposal is the most structurally novel. If it proceeds as filed, National Grid could become one of the first investor-owned utilities in the country to run a thermal energy network with wastewater as its primary resource, which would change the usual equation, since the utility earns on the heat rather than only treating the water.
Can you capture the heat from your own home?
Almost certainly not through your municipal sewer, and the research is unusually clear on this.
Researchers at the Université Libre de Bruxelles investigated heating individual homes from a Brussels sewer using small heat exchangers of roughly 6 metres paired with 6 kilowatt heat pumps. Recovered heating power ranged from about 2.5 to 5.5 kilowatts, delivering 3.2 to 7 kilowatts of space heating, enough for a heatable floor area of about 63 square metres in an average Brussels home. Their cost analysis found a higher levelized cost of energy than conventional systems except in buildings with high heating demand, and their modelling suggested only about 35 percent of the region’s residences would be eligible even in principle, because of the distance between candidate sites and difficulty accessing the sewer network.
There is a second, more practical route that does work: capturing heat at the building rather than at the sewer. This is the same principle behind the drain water heat recovery systems now sold for homes, though those work on the small warm flows from an individual property rather than a municipal main. A Shanghai spa study found a system reaching a COP of 3.38, and a 2017 field test of a sewage source heat pump in a 15,000 square metre Chinese hotel recorded a heat pump COP of 6.0 and a whole-system COP of 3.9, with a 4.7 year payback. In that hotel, two-thirds of the electricity went to the compressor and one-third to the system’s pumps, and the researchers recommended variable speed pumps so the pumping load could track a varying demand.
That last point generalises to the whole sector. The heat is free. The compressor and the circulation pumps are what you pay for, and how well you manage them is the difference between a good project and a disappointing one.
Frequently asked questions
How hot is sewer water, realistically?
Typically between 10 and 25 degrees Celsius. A Polish treatment plant measured 10 to 15 degrees Celsius in treated effluent across a full year, Brussels recorded 8 to 16 degrees Celsius between December and April, and a review of European systems documented a general 8 to 25 degree range. That is too cool to heat a building directly, which is why the technology depends on a heat pump to lift the temperature.
Is it renewable energy?
It is very often described as a renewable or recovered energy source, and it is certainly a waste heat recovery measure rather than a new fuel source. The honest framing is that it avoids burning something, because the heat was already in the water. Whether it counts as renewable depends on the electricity source that drives the compressor, which is why the Seattle results in the modelling study were so strong, as the regional grid runs heavily on hydropower.
Does taking heat out pollute the river?
That is the constraint, and it is taken seriously. Cooling the water before discharge is the point in Denver’s case, because it brings effluent temperatures back into compliance with Colorado’s aquatic life standards. The limit is real, with researchers typically cautioning against cooling discharge water below about 277 kelvin. Where recovery is properly designed, studies including a 2024 Italian case found the temperature change to the wastewater itself was negligible.
How much does it cost?
It varies enormously with scale. The Polish treatment plant study put a small installation at 30,308 euros with a 13 year payback. Toronto Western Hospital, which supplies a whole hospital complex, cost $42.9 million. Denver’s full network is estimated at 280 to 320 million dollars. St. Paul’s district-scale proposal is around 150 million dollars. The consistent pattern is that small standalone systems struggle on payback while large systems serving many buildings clear it.
Why has this not spread faster?
Funding is the most common reason. A St. Paul project that would remove a large share of a district system’s natural gas use failed to win an EPA grant and remains unfunded years later. Beyond that, the projects require a water utility, a district energy operator, a city and a developer to align on a timeline and a financing split, and each has a different definition of success. Where a regulation forces the issue, as in Denver, projects move much faster.
The buried energy under every city
Sewage heat recovery sits in an awkward position in the energy conversation. It is technically mature, demonstrably efficient, and in one case large enough to heat and cool a major hospital almost entirely. It is also nearly invisible, because it is a set of pipes and plates buried between a sewer and a building rather than a power plant anyone can photograph.
What the evidence actually shows is that the constraint has moved. In Poland, in a single small sewage plant, recoverable heat exceeded local demand twenty-six times over. In Colorado, a state regulation on water temperature created the business case that climate arguments alone could not. In Toronto, six separate funders assembled to make a single hospital work.
The question is no longer whether the energy is there. It is whether a city is willing to sign a sewer, a pipe and a heat pump, and whether someone is willing to fund the paperwork. On the evidence so far, the cities getting sewer heat built are not the ones with the most enthusiasm. They are the ones where somebody had to.
For cities starting now, Massachusetts runs a funded pilot grant programme, Minnesota is seeking money for a similar programme, and thirteen states have active thermal energy network initiatives. For anyone hoping to do it at home, the evidence is a clear no. The heat is not under your house. It is under your street, running on its own, and it is not going anywhere.
