District Energy 101
The physics and the vocabulary, before the data: what it is, how it's classified, what's in the pipe, and what feeds the plant.
Instead of every building burning its own fuel in its own basement, a district energy system makes heat (and often chilled water, and sometimes electricity) at one or more central plants and distributes it through buried pipes to many buildings. Each building gets a heat exchanger instead of a boiler.
CEEDC's working definition, the one that decides what counts in the numbers on this page: a system that supplies thermal energy for space conditioning and water heating to more than one building. Campus systems — universities, hospitals, military bases — count. Industrial heat used inside the plant gate does not, because that energy is doing something other than conditioning space and heating water.
Every district energy system has these three parts plus an operational structure for control and, where customers are billed, an administrative one. Source: CEEDC, District Energy in Canada (2023).
Why anyone builds one
Fuel switching at scale. Changing one plant from gas to biomass, sewage heat or a big industrial heat pump decarbonises every connected building at once. Doing that building by building means hundreds of separate retrofits.
Things that only work big. Deep-lake cooling, sewage heat recovery and municipal waste heat are not viable for a single house. Spread across 180 buildings they are.
Nothing in the basement. No boiler, no flue, no fuel storage, no combustion inspection — floor space and maintenance handed back to the building.
Why it isn't everywhere
Pipe is expensive. Thermal network piping runs on the order of $1,000 per foot installed. You need enough heat demand per metre of trench to pay for it, which is why district energy clusters in dense downtowns and campuses.
You have to dig. Retrofitting a network into an existing street is disruptive and slow, so most growth happens in new development or on land already under one owner.
It is a long commitment. Systems are financed over decades — CEEDC found the average commissioning year skews old, and Canada's oldest system has been running since 1880.
Those are the physical and financial constraints. The structural ones — how gas infrastructure gets paid for versus how a thermal network does, and what nobody is required to plan for — turn out to matter more, and they're in section 03.
The generations, and why "5th Gen" is contested
The industry counts district heating in generations — shorthand for how hot the water in the pipe is, and that one number drives almost everything else. Hotter networks lose more heat through the pipe wall, need more robust pipe, and can only be fed by sources hot enough to reach that temperature. Cooler networks lose less, can be built from cheaper plastic pipe, and open up sources that were previously useless: waste heat at 25 °C, a lake, a sewer.
In February 2024 the IEA's District Heating and Cooling technology programme published a deliberately simplified ladder, defined purely on network characteristics, so that the terminology means the same thing in everyone's slides — and recommended retiring one of the labels. Tap a row to expand it.
Source: IEA DHC, District heating network generation definitions, February 2024, drawing on Lund et al. (2014, 2021), Sulzer et al. (2021) and Werner (2022). The committee notes plainly that a later generation is not necessarily better for climate mitigation — that depends on emissions, exergy and life-cycle cost for the specific case.
Heat carrier: what's actually in the pipe
Steam. The original carrier. Enormous heat capacity per kilogram, moves without pumping, and reaches any building temperature you like — but runs hot, leaks readily, needs licensed high-pressure operators, and loses a lot of heat to the ground. CEEDC's data shows steam systems need noticeably more staff than hot water ones, because provincial boiler and pressure-vessel rules demand closer supervision.
Hot water. The modern default. Lower temperature, lower losses, cheaper pipe, and — crucially — it can be fed by heat pumps, solar thermal and waste heat, which cannot make steam economically.
Chilled water. The cooling side. Supply is typically 4–8 °C; Canadian systems reporting it cluster tightly around 6 °C.
One pipe, two pipes, or more
Single-pipe. Flow goes out and is consumed, not returned — classic steam district heating, where the condensate is sometimes dumped rather than piped back. Cheap to build, wasteful to run: you throw away both the water and the heat still in it. Only 9 Canadian systems reported their condensate return rate at all, and those ranged from 45% to 90%.
Two-pipe. Supply and return. Every modern hot-water and chilled-water network. The difference between the two temperatures — the ΔT — is what actually determines how much energy each litre delivers, which is why a low return temperature is something operators chase.
Four-pipe and ambient loops. Heating and cooling delivered simultaneously on separate pairs, or — in a thermal source network — a single uninsulated ambient loop that each building's heat pump draws from or rejects into, so one building's cooling becomes another's heating.
Types of energy sources and equipment
Boilers — gas, oil, biomass. Still the backbone: natural gas appears in most Canadian systems. Oil and diesel usually appear alongside gas rather than alone, because they serve as backup and peaking capacity.
Cogeneration (CHP). An engine or turbine makes electricity, and the heat that would otherwise go up the stack is captured for the network. CEEDC counts these separately; at least 82 district energy systems in the harmonised database include cogeneration.
Heat pumps. Increasingly the whole point. An industrial heat pump lifts low-grade heat up to network temperature, and only works if the network temperature is low enough — the mechanism connecting the generations above to the sources below.
Geothermal (geoexchange). Ground-source heat pumps drawing on the near-constant temperature a few metres down, or, in a handful of Canadian cases, deeper geothermal wells. Steady and weather-independent, but slow and expensive to drill.
Waste heat and heat recovery. Sewage, industrial process heat, data-centre servers, a subway tunnel — heat that already exists as a by-product and would otherwise be vented or dumped. Only usable by a network cool enough to accept it, which is why this is a low-temperature-generation story as much as a fuels one.
Solar thermal. Collectors heating water or glycol directly into the network, rather than solar PV making electricity for it. Rare in Canada so far but growing internationally as a summer-load source that leaves fuel-fired capacity for winter peaks.
Electric boilers. Simple, cheap to install, expensive to run; useful for peaking and for soaking up cheap or surplus electricity.
Thermal storage. A big insulated tank that decouples when heat is made from when it is used. Rare in Canada so far: of 31 systems asked, only 5 use it.
Free cooling. Deep lake or sea water cold enough to cool buildings directly, with no chiller — Toronto's Deep Lake Water Cooling is the flagship.
What Canada has built, and what's coming
The national picture from CEEDC's inventory — every system on the map, how honest the data is, and what has been announced since the inventory closed in 2023.
CEEDC's public download does not reproduce the totals in CEEDC's published report: respondent-confidential values were withheld from the public file, so file totals come out lower on a smaller n. Both are correct — they answer different questions. This page never mixes them, and never presents a file-derived total as the report's figure.
How much of this do we actually know?
The most important chart on this page, and the one nobody puts on a slide.
Every system in the inventory
All 254 district energy systems CEEDC records — operating, planned and decommissioned. Click a point or a row for the full record.
What's being built now
CEEDC's inventory closed in 2023. Two things have happened since: the systems it already listed as planned, and the ones that opened after it went to press.
Policy, and why it isn't scaling
Who actually holds the levers — municipal, provincial, federal — what they've used so far, and the barriers the industry's own analysts say are holding thermal networks back.
Almost everything in section 02 describes a system built because an owner chose to build it. Whether that stays the exception is a policy question, and the answer is split across three levels of government that don't line up neatly. The clearest recent survey of this is the Building Decarbonization Alliance and Dunsky Energy + Climate's 2025 white paper, which is the source for most of what follows; where it's their finding rather than a primary document, it's flagged.
Their framing of the split: municipalities exert the most influence over where thermal networks get built, but their authority is delegated by provinces — so provincial action is usually what unlocks municipal ambition. The federal government has little direct jurisdiction, and mostly acts through money, model codes and its own buildings.
Municipal: the level with the actual tools
Mandatory connection is real, and Vancouver is the clearest Canadian example. Buildings inside the False Creek Neighbourhood Energy Utility service area must connect for space heating and domestic hot water under the city's Energy Utility System By-law No. 9552. A developer applies to the city-owned utility, and the NEU's engineer reviews the building's HVAC design against the network's operating parameters before a permit issues — a two-to-three-month step built into how buildings get approved there, not an optional add-on.
Land-use planning is the quieter lever. Secondary and subdivision plans can require new buildings to connect, or at minimum to be built network-ready, as a condition of approval; community energy plans map thermal load before anyone commits capital. These are statutory instruments most municipalities already run — they just aren't usually pointed at heat.
Performance standards for existing buildings barely exist yet. Vancouver has the only active building emissions performance standard in Canada; Montréal (with Québec expected to follow province-wide) and Toronto currently require disclosure and are considering thresholds. This matters because green development standards only reach new construction, and the buildings a downtown network would most want to connect are already standing.
Provincial: the gap
One regulator, nationwide. As of May 2025 the BC Utilities Commission is the only provincial energy regulator in Canada with a regulatory framework for thermal utilities; Québec, Ontario and Manitoba have early work only. For contrast, the same report counts 15 US states that have passed or are developing thermal-network legislation — 8 passed, 11 with bills introduced — mostly authorising or requiring gas and electric utilities to own and operate networks. Regulation isn't only red tape here: it's what lets a utility recover its costs through approved rates, which is what makes an investor comfortable with a 40-year asset.
Codes decide whether a network can compete. Without a carbon or energy requirement stringent enough to rule out a cheap gas boiler, a thermal network has to win on price against the incumbent — which, as the barriers below explain, it often can't. BC's original Energy Step Code was criticised for exactly this, which is why a separate Carbon Step Code followed.
And the municipal authority underneath it can be withdrawn. Alberta rolled back municipalities' ability to set standards beyond the provincial code in 2023, and Ontario's Bill 17 introduces changes that may constrain it similarly. Carbon pricing moved the same direction: the federal consumer carbon price was repealed in April 2025 and BC followed for the consumer portion, removing a price signal that had been making fossil heating more expensive relative to a low-carbon network. Québec's cap-and-trade continues.
Federal: money, model codes, and its own buildings
Strategy. The Canada Green Buildings Strategy is Ottawa's overarching net-zero-buildings plan; the BDA/Dunsky reading is that it focuses on electrification and demand-side management and "has little to say on thermal energy networks."
Money. The Canada Infrastructure Bank has made major investments in low-carbon networks — including Enwave's Toronto system — and Ottawa funds the Federation of Canadian Municipalities' Green Municipal Fund, which finances community energy systems.
A model-code quirk worth knowing. The national model code treats thermal energy delivered by a network as purchased energy and applies an assumed coefficient of performance of 1.0. A heat pump installed in a building basement gets credit for its real COP; the same heat pump sitting at the network's central plant, serving the same building, is counted as if it were a resistance heater. That is a technical accounting choice, buried in a code, that penalises exactly the configuration a modern low-temperature network is built around.
Leading by example, deliberately or not. The federal government is one of the largest owners and operators of district energy in the country — the NCR system in section 02 is its own — and the Greening Government Strategy sets emissions targets for new construction and major retrofits. The Centre Block rehabilitation connects to the Cliff district energy plant and adds a field of geoexchange boreholes for seasonal heat storage. Federal research money is also why this page has data at all: CanmetENERGY-Ottawa funds CEEDC's survey.
Barriers: why it isn't scaling
The BDA/Dunsky paper's own chapter title is "Why TENs Aren't Scaling (Yet)", and its answer is mostly not technical:
1 · The playing field is uneven by design. A gas utility recovers infrastructure cost through rates, amortized over decades and spread across a very large customer base. A new thermal network typically has no such structure and must recover its capital directly from the limited set of customers it serves, through long-term contracts. Same pipe in the ground, completely different financing. Gas also remains the default in new development, often for marketability — buyers want gas cooking.
2 · Nobody is clearly in charge. Connecting a building to gas or electricity follows a defined process with few variations. Thermal network projects frequently have no designated entity responsible for championing and coordinating them, so a single actor ends up being both developer and advocate, and everyone else participates according to whatever capacity they happen to have.
3 · Capacity and awareness. Municipalities carry much of the work with limited budgets, competing priorities and no in-house thermal expertise — and there is no systematic waste-heat or infrastructure mapping at the local or provincial level, so opportunities go unidentified rather than rejected.
4 · The consumer-side evidence is genuinely mixed, and worth taking seriously. A 2022 UK national survey of 2,244 heat-network customers against 1,733 matched non-users found the average annual bill lower on a heat network — £600 versus £960 — but with far more spread: 1 in 10 paid £2,000 or more. Users also reported more outages (50% vs 29%) and, among those who complained, only 39% were satisfied with how it was handled. The report's own explanation for the cost spread ties straight back to barrier 1: gas and electricity network charges are averaged across enormous ratepayer pools, while a heat network is local and its bill reflects that specific system's actual costs. Averages hide who got the bad end of it — which is the argument for regulating thermal utilities, not against building them.
What the evidence says about cost and emissions
The honest answer: it depends, and no single Canadian study settles it for every case. The people arguing hardest for thermal networks say so themselves — the BDA/Dunsky paper states plainly that networks are "not inherently low carbon", and that they "may not always reduce energy costs compared to the current fossil-fired status quo" while often providing cost advantages relative to other low-carbon pathways. That is the comparison that actually matters, and it is a narrower claim than the one usually made on a conference slide.
The IEA committee quoted in section 01 makes the same point about generations: a later, lower-temperature network is not automatically better, because the benefit comes from what feeds the plant, not from the pipe. And CEEDC says it directly about this inventory — the survey data are insufficient for a rigorous savings analysis, and some systems report energy use per unit floor area above the provincial average for the buildings they serve.
The strongest measured Canadian case is a fuel-switching project. The federal modernization of the National Capital Region District Energy System — Ottawa's downtown steam loop converted to low-temperature hot water, steam-driven chillers replaced with electric ones cooled by the Ottawa River — is projected to cut emissions 63% against a 2005 baseline. One project's engineering estimate, not a national average, and it says more about changing what burns in the plant than about pipes being inherently efficient.
What the potential figures do and don't say. Thermal networks serve roughly 3% of Canada's heating demand today. Research from McMaster University, cited in the same paper, estimates 70% of Canadians live in areas where home heating needs could be met by them, and one study estimates Québec alone could meet ~40% of its residential heating demand (64–81 PJ) by recovering waste heat from large-scale sources. These are technical-potential estimates — what the geography and the physics would allow — not forecasts, and the distance between 3% and 70% is made of exactly the financing, regulation and coordination problems listed above.
World leaders
Canada's newest systems are borrowing ideas other countries adopted first — some of them decades ago. None of these transplants directly; different climates, ownership models and electricity prices see to that. Each shows a different lever.
How this page was built
Two versions — the short one, and the one with the field names in it.
Everything on this page comes from one national inventory, kept by the Canadian Energy and Emissions Data Centre at Simon Fraser University and paid for by Natural Resources Canada. CEEDC emails a questionnaire to district energy operators across Canada, checks the answers, and publishes both a report and a downloadable database. We used the database they publish, and their 2023 report for the figures the database doesn't contain.
Three things are worth knowing before you trust a number here.
One: most systems didn't answer. CEEDC knows about 238 operating systems. In the 2023 round, 38 of them filled in the survey — about one in five. Roughly 160 have detailed information from any year. So a chart saying "45 systems do X" is telling you about the ones that replied, not about Canada.
Two: the data isn't all from the same year. Each system's entry records the year its numbers refer to, and those run from 2014 to 2023. A third of the entries are from 2017. "The 2023 inventory" is really a decade of snapshots stacked together, which is CEEDC's own stated caveat, not a criticism.
Three: the public file is smaller than the report. Some operators' figures are confidential, so they're in CEEDC's analysis but not in the file anyone can download. That means our totals are lower than the published ones. We show both, side by side, rather than picking whichever looks better.
Where a number could not be verified, it is left blank rather than estimated. Where a system reported "no", that is shown differently from a system that simply never answered.
Source. CEEDC Innovative Energy Facilities database, district energy subset — CEEDC_IEF_district energy.xlsx (254 rows, de == True) and CEEDC_IEF_public.xlsx (3,644 rows, the full IEF superset used here only as a map context layer). Both are public downloads exposed inside CEEDC's own Tableau dashboard. Sheet layout: row 1 units, row 2 field names, row 3+ data. Dashboard last updated 2024-01-24; report published December 2023.
Status split. status ∈ {Operating (238), Planned (13), Decommissioned (3)}. Every aggregate on this page is computed over Operating only unless the chart says otherwise; the 238 matches the report's headline count exactly.
Tri-state flags. de_hs / de_hw / de_cw and every *_used fuel column are Yes / No / empty. The empties are preserved as null, not coerced to false — 99 operating systems have de_cw empty and 98 have it explicitly "No", and collapsing those would invent 99 confirmed no-cooling systems. Charts show "unreported" as its own category.
Fuel sets, not a primary fuel. The workbook has no primary-fuel column. CEEDC's dashboard derives one inside its Tableau .hyper extract, which is not published, so rather than guess at their priority order this page keeps the complete fuel set per system and reports single-fuel and multi-fuel counts separately, mirroring the report's own Table 8 structure.
IEA generation assignment. Applied to de_hw_supply only. Thresholds from IEA DHC (Feb 2024): >100 °C → 2nd Gen (2G), >70–100 °C → 3rd Gen (3G), ≤70 °C → 4th Gen (4G) — this page displays the ordinal names, but the underlying data and this paragraph use IEA's own 1G–4G shorthand. Exactly 70 °C is assigned to 4G, since the 4G rule ("maximum forward flow temperature of 70 °C") is more specific than the 3G range ("between 100 °C and 70 °C") and names 70 °C as the physical DHW-disinfection threshold. 1G is never inferred from a temperature — IEA defines it by carrier (steam, independent of temperature), so the steam-network count is reported separately with its own n rather than folded into this ladder. Systems with de_hw_supply ≥ 150 °C are flagged as suspect (likely a steam temperature entered in the hot-water field) and drawn distinctly rather than dropped. Systems with supply ≤ 30 °C are flagged as thermal source networks.
Implied capacity factor = de_*_production ÷ (de_*_capacity × 8760), computed only where both are present and non-zero. Values above 100% are flagged in meta.json, not silently clipped. CEEDC's own averages run 13–20%; their stated reasons are short seasons, redundancy, planned expansion and retired-but-retained capital.
Known source quirks, all handled in the pipeline and listed in meta.json: the Metadata sheet's "Detail" column is offset from its "Field" column by 4 rows from row 60 onward (so geo_used reads as "Solid Biomass" and waste_used as "Spent Pulping Liquor" if taken at face value — the fuel labels here are the corrected mapping, cross-checked against the legend CEEDC's dashboard renders); the Metadata sheet documents an npri field the data sheet lacks and omits the operator field it has; municipality_type ships in mixed case; facility and city names carry stray double and trailing spaces.
Geometry. Province outlines are unioned from the repo's existing per-province FSA polygons (geo_json/*.json) and simplified to 0.03°. Yukon has no geo_json file, so its outline is reprojected from the national FSA file with the same inverse-Lambert used by Python/canada_boundary.py. All 254 facilities carry coordinates in the source and all fall inside Canada; none are geocoded here.
Pipeline: Python/district_energy_etl.py · Docs: docs/DISTRICT_ENERGY.md · QA counts and the file-vs-report delta: districtenergy_json/meta.json.
Sources & limitations
Inventory: Canadian Energy and Emissions Data Centre (CEEDC), Simon Fraser University — Innovative Energy Facilities database, district energy subset, via the District Energy Inventory dashboard. CEEDC is funded through government and industry sponsorship; this survey is funded primarily by CanmetENERGY-Ottawa, Natural Resources Canada.
Report: Griffin, B. (2023). District Energy in Canada. CEEDC, Simon Fraser University, prepared for CanmetENERGY-Ottawa, NRCan, December 2023.
Generation definitions: IEA DHC (2024). District heating network generation definitions, February 2024.
Building-stock context: NRCan Comprehensive Energy Use Database, 2023 — also browsable in this suite's CEUD Explorer.
What this data cannot tell you. Whether district energy saves energy or emissions relative to the alternative — CEEDC says so directly: the available survey data are insufficient for that analysis, provincial aggregates for heating and cooling supplied are hard to estimate, and many systems show energy per unit floor area above provincial averages. What the costs and benefits are. Whether there is a demand-density threshold below which it stops making sense.
What we assumed. That the 238 systems flagged Operating are currently operating, at whatever date each row's year_reported refers to. That coordinates in the source are correct — they were range-checked against Canada's bounding box but not verified individually. That the IEA's temperature thresholds can be applied to a single reported supply temperature, when a real network's forward temperature varies with outdoor conditions.
What would change the answer. A higher response rate — CEEDC is explicit that as more systems are added, the averages come down, because newly identified systems are generally smaller. Temperature data for more than 27 systems, which would turn the generation section from illustrative into representative. A refreshed survey: this inventory predates the largest district energy conversion in Canadian history, which is in section 08.