District heating · Heat pump
Heat from a treatment plant outlet
How a heat network turns wastewater of 10 °C into an 80 °C flow – and why heat density, not network length, decides the losses.
Read the article →Technical article
A heat network is persistently losing make-up water. Four detection methods, a walk-through of every connected building, one street dug open – and for more than eighteen months, no find. A case study in what measurement can do and where it stops.
A heat network circulates treated water in a closed loop. Closed means nothing should get in and nothing out. The reason is less the water itself than what comes in with it.
Fresh water carries dissolved oxygen. That reacts with the steel pipes, and from there a chain of consequences follows – all of which occurred in this case:
A leak does not cost you the water you lose. It costs you the plant. And because the damage builds slowly, it is often first noticed at the pump that fails for the third time.
The documented loss runs at around 650 litres a day. That sounds manageable – a garden hose, a few minutes. Scaled up it looks different:
| Quantity | Value |
|---|---|
| Loss per day | around 650 litres |
| Loss per year | around 237 m³ |
| Treated water fed in over four years | around 1,000,000 litres |
| Additional flushing water needed by the treatment plant | around 1,500,000 litres |
| Additional heat needed to warm it up | around 80,000 kWh |
Two of these figures deserve a second look.
The flushing water. For every litre of treated water, one and a half litres of flushing water are needed – the treatment plant uses water of its own to regenerate its filters and resins. Actual water consumption is therefore two and a half times the loss in the network.
The heat. Make-up water arrives from the mains at around 10 °C and has to be raised to network temperature. The arithmetic is simple: 1,000 m³ times roughly 60 kelvin times 1.163 kWh per cubic metre and kelvin gives just under 70,000 kWh – the stated 80,000 kWh fits if slightly higher network temperatures are assumed. That is the annual heat demand of several family homes, spent purely on warming up water that then leaks away.
And one more order of magnitude: with 400 m³ of network volume and 237 m³ lost a year, the entire network content is replaced roughly every one and a half to two years. In a closed loop, that is not a marginal issue.
Four physically quite different approaches were used one after another. That is typical: every method has blind spots, and one hopes they do not overlap.
Modern district heating pipes have monitoring wires embedded in the insulation. If moisture enters the insulation, the electrical resistance between those wires and the service pipe changes. A travel-time measurement – sending a pulse down the wire and timing the echo from the fault – locates the point to within a metre or so.
It is the best method available, and it has a hard limit: it only sees what happens inside the monitored pipe. Connecting pipework inside buildings, service connections and valves lie outside it.
Forming gas, a mixture of nitrogen and hydrogen, is added to the water. Hydrogen is the smallest and lightest molecule there is; where water escapes, the gas escapes with it, rises through the soil and can be detected at the surface with a hydrogen probe.
Elegant, but dependent on the ground. Under sealed surfaces, in cohesive soils or with a high water table, the gas never finds its way up.
Escaping heating water is warm and heats the surrounding soil. A thermal camera can pick that up as a trace at the surface.
This works well for larger leaks at shallow depth – and poorly for everything else. The deeper the pipe, the more the signal smears out. And there are many reasons why one strip of asphalt is warmer than the next.
Water escaping under pressure through a crack makes noise. A ground microphone listens for that flow noise, particularly at welded joints and defects.
The same caveat applies: a leak with high pressure and a sharp outlet is loud. A diffuse escape at a corroded spot can be almost silent – and that is exactly what was being looked for here.
When the methods found nothing in the street, every connected building was inspected. Two things were being looked for:
For detection this is the most awkward case of all: no wet patch, no noise, no heat trace, nothing in the ground. The water merely changes circuit. With 400 connected transfer stations that is a needle in a haystack – and it explains how the search in the street could remain fruitless.
At one junction, repeated measurements showed a pressure loss. The section fell within the warranty period of an earlier construction stage, the street was opened – and there was nothing there.
That is everyday leak detection, and the reason it is so expensive. Every excavation is a bet. A pressure loss in a section says something is wrong somewhere in there, but not where exactly – and certainly not whether it is in the pipe at all.
The documentation also records that a dowsing rod was tried at one suspected spot – explicitly noting that the hit rate "does not go beyond chance". One can smile at that. What I find more remarkable is that it was recorded openly. Anyone who, after eighteen months of fruitless searching, tries every idea once and then writes it down is handling their data more honestly than plenty of people who document only their successes.
When no method finds the spot, you invert the question. Instead of searching for where the water escapes, you make the water visible – and wait to see where it turns up.
Uranine is a fluorescent dye that turns a bright yellow-green at high dilution, remains detectable in tiny quantities under UV light, breaks down within days, and is neither hazardous to water bodies nor toxicologically concerning. It has long been used in hydrogeology to trace underground flow paths.
The plan: dye the whole network at 20 grams per cubic metre – about 8 kilograms per filling at 400 m³ – and then watch buildings, the sewer system and the inlet of the treatment plant. Wherever dyed water appears, that is the flow path. Residents and drainage staff are asked to report any yellow or green discolouration.
| Item | Cost |
|---|---|
| around 30 kg of uranine, for repeated use | around €7,500 |
| preparing and introducing it into the water | around €4,000 |
| building and street inspections | around €6,000 |
| Total | around €17,500 |
The 30 kilograms are enough for just under four fillings – so one attempt is not expected to be sufficient.
The obvious move would be to blame the network's losses on the leak. The investigation states explicitly that this is not the case: the roughly 22 per cent heat loss is not attributable to the leaks but to heat density per metre of route – that is, how much heat is actually drawn off per metre of pipe.
That distinction matters. Water loss and heat loss are two different problems with two different remedies: one is found with measurement, the other solved with more customers per metre of pipe.
Few people have a heat network. Almost everyone has a heating system with a closed water circuit, and there the same rules apply on a smaller scale:
That is exactly where recording pays off: not to produce pretty curves, but to see a slow trend that you would never notice in passing.
Because every litre of make-up water brings dissolved oxygen with it. That causes corrosion of the steel pipes and forms magnetite sludge, which burdens pumps, valves and heat exchangers. The damage is not done by the water lost but by the water added.
Four common ones: electrical measurement via monitoring wires in the pipe casing, gas detection using forming gas and a hydrogen probe, thermography to find heat traces at the surface, and a ground microphone listening for flow noise. In addition, buildings are inspected and heat exchangers checked.
Because every method has blind spots. Electrical monitoring covers only the pipe, not service connections. Gas detection needs a path to the surface, thermography a sufficient heat trace, the ground microphone an audible noise. A faulty heat exchanger lets water cross into the building circuit rather than escape – then there is nothing to find from outside.
A fluorescent dye that turns yellow-green at high dilution, is detectable under UV light and breaks down within days. It is added to the water; wherever dyed water appears in buildings, in the sewer system or at the treatment plant, the flow path can be traced back.
Occasional topping up after maintenance is normal. Regular topping up is not – that indicates a leak or a failed expansion vessel. Either should be found, because every top-up brings oxygen into the system.
Figures on loss volumes, damage patterns, the methods used and the planned dye tracing come from an investigation by the consulting engineers, presented to the municipal council in October 2023 and published. The annual figures and heating energy are own calculations on that basis. The source documents are in German.
Personal names and company names from the investigation are deliberately not reproduced here. This article describes methods and orders of magnitude; it does not assess construction work or warranty questions.
Want to know whether your heating system is slowly losing pressure – recorded, rather than noticed at the next breakdown? Send me a message – the first conversation is free.