In the Netherlands, district heating networks are still presented as the logical solution for the built environment. The question of when a heat network actually performs better than individual solutions — technically, economically and socially — is rarely asked out loud.
In plans for gas-free neighbourhoods the heat network is a much-used instrument. Municipalities, housing associations and energy companies regularly opt for it as the collective route: one source, one infrastructure, one supplier. Administratively that is easy to understand — organising thousands of individual projects at once is considerably more complicated than a single tender.
Ease of organisation and technical-economic optimality do not automatically coincide, however. That question deserves a more honest answer than planning practice usually gives.
What makes a heat network work
A heat network becomes attractive as soon as there is a large, reliable heat source nearby that would otherwise go to waste: industrial waste heat, waste incineration, geothermal energy or another source with low marginal heat costs.
Heat demand density matters just as much. Flats, offices, care institutions and compact urban blocks close together deliver a lot of offtake per linear metre of pipework. The higher that heat demand per metre, the more favourable the infrastructure cost per connection.
Public studies and meta-analyses point the same way: in densely built areas heat networks can work out well, provided source, temperature regime and building stock genuinely match. Those same analyses also show how wide the spread is, and how strongly the business case depends on local circumstances. [1]
What makes a heat network vulnerable
Against that stand structural drawbacks that receive little attention in enthusiastic presentations.
First, dependence on a single supplier. Anyone connected to a heat network cannot switch. That is precisely why the Dutch regulator ACM investigates the financial returns of heat suppliers and sets a benchmark return, to prevent households from paying too much. [2] [3]
Second, distribution loss. Transporting heat costs heat — every heat infrastructure has that property, permanently and unavoidably. In low-rise neighbourhoods with low heat density those losses can mount up considerably.
Third, the certainty of the source. Industrial waste heat depends on the survival and the production profile of that industry. A heat network is an infrastructure investment for forty to fifty years; both source and demand profile have to keep adding up over that whole period. That is an assumption deserving more attention than it usually gets — fifty years is roughly the span between the first moon landing and today.
How much heat is actually lost?
Time for the back of an envelope. Heat distribution through an underground network always costs energy: the water in the pipes is in permanent thermal contact with the soil. The size of that loss depends mainly on three factors — the supply temperature, the pipe diameter and the insulation quality.
For a pre-insulated district heating pipe with PUR insulation, a soil temperature of 10 °C and a burial depth of 0.80 metres, the heat loss per metre of pipe is of this order:
| Pipe size | 60 °C supply | 90 °C supply |
|---|---|---|
| DN50 (steel ø60 mm, casing ø140 mm) | 9.5 W/m | 15.2 W/m |
| DN80 (steel ø89 mm, casing ø200 mm) | 9.9 W/m | 15.8 W/m |
| DN100 (steel ø114 mm, casing ø250 mm) | 10.3 W/m | 16.5 W/m |
These values follow from a straightforward heat flow calculation using insulation resistance plus soil resistance, with λ_PUR = 0.027 W/mK and λ_soil = 1.2 W/mK. Those assumptions are in line with common values for pre-insulated pipes. [4]
The most striking figure is in the comparison between the two columns. A traditional high-temperature network at 90 °C loses roughly 60 per cent more, structurally, than a modern low-temperature variant at 60 °C. Same pipe, same soil, different thermostat.
What this means at neighbourhood level
Take a neighbourhood of 500 homes. A compact neighbourhood quickly amounts to some 5 kilometres of single route. Supply and return together make that 10 kilometres of pipe.
At an average heat loss of 10 W/m that yields:
- 10 W/m × 10,000 m = 100 kW of continuous loss
- 100 kW × 24 hours = 2,400 kWh per day
- almost 73,000 kWh per month
- 100 kW × 8,760 hours = 876,000 kWh per year, or 876 MWh
Against that stands the combined consumption. Recent figures from Statistics Netherlands (CBS) show an average consumption of 20.4 GJ per home in 2023 for homes on district heating, around 5.7 MWh per home per year. For 500 homes that comes to some 2,800 to 3,000 MWh of heat per year. [5]
Reckon on 3,000 MWh delivered, and those 876 MWh of distribution loss amount to almost 29 per cent of the heat supplied.
That loss runs all year round: in summer too, at night too, even when demand is low. The heat network never stands still. Almost three in every ten kilowatt-hours going into the system reach no home — and that energy is paid for again every day by the connected households.
The coordination problem of high supply temperatures
Many existing Dutch heat networks still run at high temperatures, often around 85 to 90 °C. Historically that is easy to explain: older buildings, high peak demand, existing radiators, little comfort margin.
Physically and economically that regime works out badly. Pipe losses rise almost linearly with the temperature difference between pipe and soil, so lowering the supply temperature towards 60 °C helps immediately — the table above shows by how much.
In practice there is no switch for it. A lower temperature regime requires a building stock designed or adapted for it. As long as a meaningful share of the customers needs high temperatures, the whole network keeps running harder than is technically desirable. One poorly insulated house at the end of the loop keeps a thousand neighbours at temperature.
Pump energy: a forgotten cost item
Alongside heat loss through the pipe, a second structural energy item appears on the bill: the electricity needed to keep circulating the water.
For that same neighbourhood of 500 homes, take the following assumptions: a peak heat demand of 2,500 kW, a temperature difference between supply and return of 20 K, a mass flow of around 30 kg/s, a pressure drop of 30 Pa per metre, a total pipe length of 10 km and a pump efficiency of 70 per cent.
From that follows a volume flow of about 0.03 m³/s, a total pressure drop of 300,000 Pa, a hydraulic power of around 9 kW and an electrical pump power of around 13 kW.
Maintained continuously, that comes to 113,880 kWh per year, so roughly 114 MWh of electricity. Against a neighbourhood heat demand of 3,000 MWh per year that is another 3.8 per cent of extra system consumption.
The real system balance
Added up: 876 MWh of pipe loss plus 114 MWh of pump electricity is 990 MWh of system overhead per year. Against a heat delivery of 3,000 MWh per year that comes to roughly 33 per cent overhead. Broadly one in every three kilowatt-hours the system draws does not reach the home as usable heat.
That is what a heat network in a low-rise neighbourhood with substantial pipe length does physically. No exceptional situation, no badly built network — simply the outcome of ten kilometres of pipe in cold ground.
Dutch practice is less rosy than the brochures
Look at Dutch practice and you do not see a sector rolling out effortlessly under its own steam. Central government supports the construction of new heat networks with the Heat Network Investment Subsidy (WIS), a scheme open in 2025–2026 precisely because many projects do not add up financially without subsidy. [6]
That is the reality behind many Dutch business cases: without subsidy, risk cover or public participation, projects do not add up. TNO and CE Delft speak explicitly in this context of the unprofitable top of heat networks. [4] [7]
On top of that comes affordability for residents, which remains a structural concern. ACM points out that households on a heat network cannot choose a supplier, and therefore actively investigates whether returns remain reasonable. [2] [3]
None of this amounts to a general death sentence. A compact network with plenty of stacked housing, a stable source, low temperatures and few metres of pipe per customer can work out excellently. The spread in performance, costs and risks is simply too wide to treat heat networks as the self-evidently best route in Dutch practice.
When individual solutions fit better
For low-rise neighbourhoods with low heat density, at a distance from large and stable heat sources, a heat network becomes a heavy infrastructure solution with high fixed costs and structural distribution losses.
In those situations decentralised or small-scale solutions are more obvious: per home, per block or per small cluster. That removes the long distribution routes, the standing losses and the dependence on a single supplier. The storage sits where the heat is used, charges when energy is available and discharges when demand appears. Transport loss over ten kilometres of pipe does not occur in that setup, for the simple reason that there is no ten kilometres of pipe.
The choice between central and decentralised is therefore a physical and economic assessment, not an ideological one. And that is exactly where things regularly go wrong in the Netherlands: the collective solution is assumed as the standard while local conditions do not always suit it.
In theory a heat network can work excellently: a short loop, plenty of stacked housing, high heat density, a stable source, low temperatures and few metres of pipe per customer. In Dutch practice that combination is more an exception than a basic recipe for the average neighbourhood.
Sources
[1] Netbeheer Nederland, Meta-analyse warmtenetten, 2025.
[2] ACM, ACM start onderzoek naar financiële rendementen van 4 warmteleveranciers, 2025.
[3] ACM, Besluit redelijk rendement warmteleveranciers 2026–2028.
[4] TNO, Financiering warmtenetten, 2022.
[5] CBS, Warmteleveringen bij woningen, 2025.
[6] RVO, Warmtenetten Investeringssubsidie (WIS), geraadpleegd maart 2026.
[7] CE Delft, Factsheet onrendabele top collectieve warmtesystemen, 2023.