Mining the City for Warmth: Waste-Heat Recovery for District Heating

How cities capture low-grade waste heat from metro tunnels, data centres, and industry, and route it through heat pumps and fifth-generation district heating networks to displace fossil fuel for heating.

▶ Open the simulation

The city as an accidental heat source

Every dense city continuously rejects low-grade heat: metro tunnels warmed by braking trains and passenger body heat, data centres that dump nearly all the electricity they consume as heat from server racks, supermarket refrigeration compressors, wastewater leaving buildings at 15–25°C, and industrial process cooling. None of that heat is hot enough to use directly in most heating systems, but it is warm enough that a heat pump can economically upgrade it to useful temperatures, which is the basic premise of urban waste-heat recovery: treat waste heat as free lift for a heat pump's evaporator side rather than as a temperature source in its own right.

Cataloguing and sizing the sources

A typical mid-size city waste-heat inventory might find something like 18 MW recoverable from metro ventilation and tunnel air, 26 MW from data centre cooling loops, and 34 MW from industrial process cooling — a combined potential of 78 MW. Set against a district heating demand of, say, 120 MW, that 78 MW of raw potential represents about 65% theoretical coverage before accounting for network losses, seasonal mismatch, and the fact that not all recoverable heat can be economically piped to where it's needed. In practice, planners apply a realistic capture-and-delivery factor — commonly in the range of 60–70% of raw potential once transport losses and heat-pump coefficient of performance (COP) are factored in — so 78 MW of source potential might realistically deliver around 50–55 MW to the network, covering closer to 45% of total demand rather than the full 65% headline figure.

Storage and the economics of intermittency

Waste-heat availability doesn't track heating demand hour by hour — data centre load is roughly constant while metro heat peaks with ridership and heating demand peaks on cold mornings and evenings — so thermal storage is what makes recovered heat dispatchable. A 420 MWh storage facility operating through pumps at 86% efficiency delivers a usable daily cycling capacity of about 420 × 0.86 ≈ 361 MWh, letting the network bank midday industrial waste heat for the evening demand peak. Distribution losses are the other major economic variable: district heating pipe networks typically lose on the order of 1.5–2% of transported heat per kilometre of pipe run, so a network with 1.8%/km losses over a 12 km distribution radius loses roughly 21–22% of transported heat in transit — a reminder that waste-heat sources close to demand centres are worth substantially more than distant ones, even if the distant source has a larger raw potential.

Network temperature: why 5th-generation heat networks change the economics

Traditional district heating runs at 60–70°C supply temperature, hot enough to serve radiators directly but too hot to be filled economically by most waste-heat sources, which typically sit in the 15–35°C range. Fifth-generation heat networks flip this: they distribute water at near-ambient temperature (15–40°C) and rely on a heat pump at or near each building to do the final temperature lift, using electricity only for the lift rather than for the full heating job. This is the single biggest reason waste-heat recovery has become economically viable at district scale in the last decade — it turns waste heat from a curiosity into a genuine substitute for a large fraction of a heat pump's electricity input, since the heat pump's COP improves dramatically when its source is 20°C metro tunnel air rather than -5°C winter outdoor air.

The economics: fuel savings and emissions

Displacing purchased heating fuel with recovered waste heat produces savings that scale directly with the delivered MW, the avoided fuel price, and operating hours. At a fuel cost of $62/MWh and a delivered supply of roughly 52 MW running continuously across a full year (8,760 hours), the annual value comes to 52 × 62 × 8,760 ÷ 1,000,000 ≈ $28 million in avoided fuel purchases — the actual figure in real deployments is usually somewhat lower once realistic capacity factors (not every MW runs 8,760 hours) are applied. Emissions savings follow the same logic: at a grid or boiler carbon-intensity factor of roughly 198 kg CO₂ per MWh, that same delivered heat avoids on the order of 52 × 198 × 8,760 ÷ 1,000 ≈ 90,000 tonnes of CO₂ per year relative to a fossil-fuelled baseline, making waste-heat recovery one of the more cost-effective decarbonisation levers available to a city, since it requires no new primary energy generation, only better plumbing and heat pumps.

What makes or breaks a project

The projects that succeed tend to share three traits: proximity between source and demand (to keep pipe-transport losses manageable), a willing anchor source with predictable year-round output (data centres and metro systems are favoured because their heat output doesn't disappear in summer the way some industrial processes do), and a heat-purchase agreement structure that lets the waste-heat producer (who may see this as a byproduct, not a business) share in the value without taking on district-heating operational risk. Metering is not optional — without heat meters at both the source and delivery points, neither fair billing nor performance verification (checking that the modelled 65% coverage or 90,000 tonnes of CO₂ savings actually materialised) is possible.

Frequently Asked Questions

Why can't waste heat from a data centre just be piped directly into radiators?

Data centre waste heat typically sits around 25-35°C, well below the 60-70°C supply temperature conventional radiator-based district heating needs. A heat pump is used to lift that temperature economically, or the network itself is redesigned as a low-temperature 5th-generation system where each building does its own final lift.

What is a 5th-generation district heating network?

It's a heat network that distributes water at near-ambient temperature (roughly 15-40°C) instead of the 60-70°C used in traditional systems, with a heat pump at or near each connected building doing the final temperature lift. This lets the network economically absorb low-grade waste heat sources that a traditional high-temperature network couldn't use directly.

How much of a city's total waste heat can realistically be recovered?

Raw theoretical potential (metro, data centres, industry combined) commonly covers 60-70% of a district's heating demand on paper, but after accounting for pipe transport losses, mismatched timing between supply and demand, and realistic capture rates, delivered coverage is usually closer to 40-50% of demand without additional storage or backup sources.

Why does pipe distance matter so much for waste-heat projects?

District heating pipes typically lose on the order of 1.5-2% of transported heat per kilometre. A source 12 km from demand can lose over 20% of the heat it exports in transit, which is why waste-heat recovery projects strongly favour sources located close to the heating network rather than the single largest potential source regardless of distance.

How does thermal storage improve the economics of waste-heat recovery?

Waste-heat output (like a data centre's roughly constant heat load) rarely matches heating demand, which peaks on cold mornings and evenings. A thermal store lets a network bank heat produced at off-peak times and release it during demand peaks, improving how much of the recovered heat is actually usable rather than wasted when there's no immediate demand for it.