Ask what flow temperature underfloor heating needs and the internet answers 35 °C. It is often right. But it answers a different question from the one an installer faces on site, because flow temperature is not a setting you pick - it is the result of three things worked out together.
This article shows what happens when you work them out separately: a measured table of output per square metre, the point where tightening the spacing stops helping, and the ceiling no setting gets past.
Heat is driven by a temperature difference, not by the flow temperature
A floor does not emit heat "at 35 °C". It emits in proportion to the difference between the mean water temperature in the loop and the room temperature. Mean, not flow: water gives up heat along the way and comes back cooler, so at 35/30 °C the pipe runs at an average of 32.5 °C, and in a 20 °C room the driving difference is 12.5 K, not 15 K.
Hence the first practical consequence: the temperature drop is a design parameter too. The same flow temperature with a bigger drop gives a lower mean, and therefore less power. Quoting a flow temperature without its return describes half an installation.
The covering is inside the equation, not beside it
Between the pipe and the air sit the screed and whatever was laid on top. All of the power has to cross that layer, so its thermal resistance is not a finishing detail - it is a term in the calculation. HeatAlgo's underfloor module derives deliverable output from four quantities: a coefficient that depends on pipe spacing, the thermal resistance of the covering, the mean water temperature, and a cap set by the maximum floor surface temperature. The resistances it uses:
| Covering | Thermal resistance |
|---|---|
| Tile | 0.05 m²K/W |
| Vinyl | 0.08 m²K/W |
| Wood | 0.12 m²K/W |
| Carpet | 0.15 m²K/W |
This is HeatAlgo's own model, calibrated for design decisions - it is not your system's datasheet. If a manufacturer publishes a thermal characteristic for a specific panel and screed, that takes precedence. What the figures below show is the thing one manufacturer's datasheet cannot: how much the answer moves between coverings with everything else held equal.
The measured table
Room at 20 °C, 150 mm pipe spacing, 29 °C surface temperature limit. In watts per square metre.
| Flow/return | Tile | Vinyl | Wood | Carpet |
|---|---|---|---|---|
| 30/25 °C | 30 | 27 | 23 | 21 |
| 35/30 °C | 50 | 45 | 39 | 36 |
| 40/35 °C | 70 | 62 | 55 | 50 |
| 45/40 °C | 90 | 80 | 70 | 64 |
Take the 35/30 °C row, the familiar answer. A room needing 50 W/m² is served exactly, to the watt, if it has tiles. The same installation under wood delivers 39 W/m², 22% short, and under carpet 36 W/m², 29% short. Nothing changed but the top layer - the one decided latest and least often referred back to the designer.
Spacing has a ceiling
The reflex when power is short is to tighten the spacing. It works, but less and for less long than expected. The same floor, with wood:
| Flow/return | 100 mm | 150 mm | 200 mm |
|---|---|---|---|
| 30/25 °C | 27 | 23 | 20 |
| 35/30 °C | 46 | 39 | 34 |
| 40/35 °C | 64 | 55 | 47 |
| 45/40 °C | 82 | 70 | 61 |
At 35/30 °C, going from 200 to 150 mm buys 5 W/m² and 150 to 100 mm another 7. That is 12 W/m² for twice the pipe - and it still lands at 46 W/m², below the 50 required. At that flow temperature no spacing exists that serves this room under wood. The temperature has to move: a mean of 36 °C, roughly 38.5/33.5 °C, and then 150 mm is enough.
That is the point of this article. Required watts per square metre do not determine the spacing. They determine it only together with flow temperature and covering - and sometimes they do not determine it at all, because at the assumed temperature no spacing is sufficient.
Calculate your roomThe ceiling you cannot beat
Above all of this sits a limit harder than spacing and temperature combined: the floor surface temperature. Occupied rooms are held to 29 °C, bathrooms to 33 °C - and those are the two cases HeatAlgo models. That is a comfort and physiology limit, not a controller setting.
Turning that limit into watts per square metre is a question of model rather than of the limit itself, so both numbers are worth having. HeatAlgo's model gives about 90 W/m² for a 29 °C limit in a 20 °C room; the limiting characteristic in EN 1264-2, at the same 9 K difference, gives about 100 W/m². HeatAlgo therefore leaves roughly 10% in hand, which works in your favour when sizing, but it is not the same as the figure from the standard - worth knowing if you are comparing against a system datasheet.
The ceiling is visible in the table: at 45/40 °C tile reads 90 W/m², and would still read 90 at a tighter spacing, because from that point on it is the permissible surface temperature doing the limiting, not the pipe.
Bathrooms are more interesting than they look. Held at 24 °C, a 29 °C limit would allow only 50 W/m² in this model. Permitting 33 °C lifts that ceiling to 90 W/m² - and it is that allowance, not tighter spacing, that makes a bathroom heatable by the floor alone.
If a room's heat load exceeds the ceiling, the answer is not more pipe. It is another emitter, or a lower heat loss.
The order that works
- Calculate the room's heat load - room by room, from a proper heat load calculation, not from a watts-per-square-metre rule of thumb. Without that number the rest is guesswork.
- Divide it by the heated area to get required watts per square metre. That is the number the floor has to deliver.
- Settle the covering before choosing a temperature. If the client has not decided, design for the worst case you are willing to allow - raising the temperature after the screed is poured is cheap, relaying pipe is not.
- Find the lowest temperature that works at a realistic spacing. Lowest, because every degree down is a lower bill for the whole season.
- Check the surface ceiling. If the required output sits above it, go back to the room's heat loss, not to the spacing.
- Only now work out loop length, because only now do you know the spacing. How to do that, and why the popular formula is wrong in two directions at once, is covered in a separate article on spacing and loop length.
What this means for a heat pump
A heat pump does not create heat, it moves it, and the cost of moving rises with the gap between the source and the water temperature it has to reach. Every degree the flow temperature comes down improves the unit's seasonal efficiency - for the whole winter, not just on design days.
So the spacing decision is a running-cost decision. Tighter spacing costs once, at installation. The higher flow temperature forced by wider spacing or a wooden floor costs every month for the life of the system. It is the same relationship behind sizing a heat pump - a matched installation and a matched unit are one job, not two.
The check at the end
Flow temperature is not a result in itself. The result is whether every room receives as much power as it loses. So the last step is comparing deliverable output against the calculated heat load, room by room - because a single room with a wooden floor can set the temperature for the entire circuit.
In the underfloor heating module that comparison happens on its own: every section carries its own spacing and covering, and HeatAlgo flags a deficit wherever the floor will not cover the loss.
Create a free account