2026/09/29
A radiant floor can be designed correctly, installed correctly, and still underperform — because of a decision made at a flooring showroom months later.
The heat leaves the pipe. Whether it reaches the room depends on what is lying on top.
This is the one part of a radiant project chosen for how it looks, by someone standing in front of samples. It is also, thermally, a component of the heating system.
One Number Decides Most of This
Thermal conductivity, measured in W/m·K, describes how readily a material passes heat through itself. High numbers conduct. Low numbers resist — which is what you want in wall insulation and precisely what you do not want between a warm pipe and your feet.
The spread across common floor coverings is larger than most people expect.
Ceramic and stone: roughly 1–2 W/m·K. Solid wood: around 0.12–0.15. That is better than a tenfold difference. Same boiler, same tubing, same water temperature — two completely different rooms.
One qualifier before the rest of this article leans on that number. Conductivity governs how fast heat crosses the covering; it does not by itself decide whether a material survives being heated. Wood fails on a second axis entirely, and carpet is judged on a different scale again. Conductivity is the main variable, not the only one.
Tile and Stone Are What the System Is For
Ceramic, porcelain and natural stone are the default pairing, and there is no real argument about it.
High conductivity carries heat from the tubing to the surface quickly. High thermal mass means the floor keeps radiating after the boiler cycles off. Tile essentially gets out of the way.
The only practical maintenance note is sealing: slate and the more porous stones want resealing every two to three years. Otherwise there are no temperature limits worth worrying about and no compatibility questions to ask.
Korean housing arrived at the same answer without ever quoting a conductivity figure. The standard finish over a heated floor for decades was 장판 (jangpan) — a thin PVC sheet, cheap, and thermally almost transparent. Before that, traditional 온돌 (ondol) rooms were finished with clay plaster and 장판지 (jangpanji), layered paper sealed with perilla or soybean oil. Thin, dense, conductive. The principle held for centuries before anyone had a unit for it.
The shift since is worth describing accurately, because it is often overstated. As incomes rose, Korean households moved from jangpan to 강마루 and similar engineered wood products — not to thick solid planks. The thermal penalty is real but modest, because Korean flooring stayed thin even as it became wood.
Wood: Two Different Problems
Wood is where what people want their home to look like meets what the floor can physically do, and it fails in two separate ways that are worth keeping apart.
Problem one is conductivity, and it applies to all wood. At 0.12–0.15 W/m·K, wood sits closer to insulation than to a heat-transfer surface. A wood floor over radiant can cut delivered output by up to around 30 percent against a compatible covering. The system has to run warmer water to get the same surface temperature, and that costs money every day it operates.
Problem two is dimensional stability, and this is where solid and engineered part company.
Wood expands as it warms and contracts as it cools, and a radiant floor puts it through that cycle daily. Solid planks — each one a single piece of timber — move aggressively and eventually show it as cupping, crowning and gaps that open every winter.
Engineered wood's cross-laminated core resists exactly this. Which produces a point worth stating plainly, because the numbers do not say it on their own: engineered wood conducts heat no better than solid wood. Its advantage is entirely stability. If someone tells you engineered is more efficient over radiant, they have the right recommendation for the wrong reason.
Both share a hard limit: a maximum floor surface temperature of 27°C (80°F), with some manufacturers specifying lower. Above it, wood dries from beneath — gaps first, then cupping, then cracking, then damage nothing repairs. It is slow and cumulative, which is why it is usually noticed too late.
One thing radiant does better than the alternative, in wood's favour: a floor holding a steady 27–29°C is gentler on timber than forced air, which swings both temperature and indoor humidity far more violently. Engineered wood over radiant is a sound choice. Solid wood over radiant is a bet with a known outcome.
LVP and LVT — The Workable Middle
Luxury vinyl plank and tile land at roughly 0.25–0.30 W/m·K — about double wood, well short of tile. Combined with moisture resistance and easy installation, that makes them a reasonable compromise rather than a concession.
Two specifications matter.
Temperature: keep the surface at or below 29°C (85°F), per typical manufacturer specification. Beyond it vinyl can warp, shrink or discolour.
Thickness: most manufacturers cap radiant-approved planks around 5 mm. Thicker boards have identical conductivity and worse real-world performance, because heat has further to travel. Thin and dense beats thick and identical — the rule that applies to every material on this page.
Carpet, and the Tog Ceiling
Carpet is not disqualified. It is simply the easiest way to throttle a system you have already paid for.
Carpet is measured by thermal resistance rather than conductivity, in tog. Research by the Underfloor Heating Manufacturers Association puts the working ceiling at a combined carpet-plus-underlay value of 2.5 tog.
Two things make that number more useful than it first appears.
The unit converts cleanly. One tog equals 0.1 m²K/W, so 2.5 tog is 0.25 m²K/W — which in the US R-value scale is roughly R-1.4. That matters because American carpet and padding are specified in R-values, and a padded carpet commonly rated R-1.5 to R-2.0 works out to somewhere around 2.6–3.5 tog. Comfortably over the line, using numbers most buyers never think to compare.
And 2.5 is not generosity. The UHMA testing found that carpet and underlay combinations behave in practice as though their resistance were about one tog lower than their published combined figure. The 2.5 ceiling exists because published values overstate real-world resistance by roughly that much — not because 2.5 tog of genuine insulation is acceptable.
If a room needs carpet, the specification narrows: low pile, a hessian or woven backing rather than rubber or thick felt, and a combined value in the 1.0–1.5 tog range.
With a heat pump driving the system, hold it at 1.5 tog or less. A heat pump earns its efficiency by running the lowest supply temperature it can; a resistive floor covering forces that temperature back up and spends the advantage you bought the heat pump for.
Compatibility at a Glance
| Covering | Conductivity (W/m·K) | Verdict | Condition |
|---|---|---|---|
| Ceramic / porcelain tile | 1.0–2.0 | Optimal | No meaningful limits |
| Natural stone | 1.3–2.2 | Optimal | Reseal slate and porous stone every 2–3 years |
| Exposed concrete | 0.8–2.25 | Optimal | Excellent mass; slow response |
| Engineered wood | 0.12–0.15 | Workable | Max 27°C surface — chosen for stability, not conductivity |
| LVP / LVT, ≤5 mm | 0.25–0.30 | Compatible | Max 29°C surface |
| Solid hardwood | 0.12–0.15 | Not recommended | Moves too much; up to 30% output loss |
| Low-pile carpet | — | Conditional | Combined ≤2.5 tog; ≤1.5 with a heat pump |
| Thick carpet + dense pad | — | Not recommended | Functions as insulation |
Note the two wood rows carry the same conductivity and different verdicts. That is the point: conductivity explains the efficiency cost, stability explains the failure risk, and they are not the same question.
Why Wood Feels Warmer Than It Is
Wood underfoot feels warmer than tile. The sensation is real. The usual explanation for it is not quite right.
What your skin registers is not the material's temperature but how fast the material draws heat out of you — a property called thermal effusivity, combining conductivity, density and specific heat. Tile is highly effusive, so contact pulls warmth from your foot quickly and reads as cold. Wood is far less so, and reads as warm. Both are sitting at exactly the same temperature.
Over a heated floor, that same property inverts from a feature into a cost.
The tile that felt cold is the one carrying the system's heat efficiently into the room. The wood that felt warm is the one resisting it — which means a larger temperature drop across the covering, which means the water underneath must run hotter to reach the same surface temperature.
That is the whole mechanism. Not that wood "keeps" the heat — wood has little thermal mass to keep it in. It resists the passage of heat, so you pay for the resistance in supply temperature, every hour the system runs.
The old systems understood the requirement even without the vocabulary. The Roman hypocaust put stone and concrete between the fire and the room. Traditional ondol put 구들장 (gudeuljang, capping stones) over the 고래 (gorae, flue channels), then clay, then oiled paper. Every layer chosen to conduct.
The Sensor That Protects All of This
Everything above depends on one piece of hardware that is easy to leave out of a quote.
A thermostat measures air temperature. It does not know what your floor is doing. The surface directly above tubing or cable can run considerably hotter than the air reading on the wall, which means a system that looks obedient on the display can be sitting well past 27°C where it matters.
A floor sensor thermostat — a probe set into the floor assembly, wired to a controller that limits on floor temperature rather than air — is what actually enforces the limits in this article.
For tile it is a convenience. For wood, vinyl and carpet it is the component that keeps the warranty valid, and it is not optional on any temperature-limited covering.
One Rule That Survives Every Material
Do not exceed the manufacturer's maximum surface temperature, and install only products explicitly approved for radiant heat.
That reads like warranty boilerplate. It is the clause that decides who pays when a floor cups. Serious cupping or delamination across a large area is a full replacement, which can exceed what the heating system cost to install in the first place — and a floor run past its rated temperature is not a warranty claim anyone wins.
So the specification sheet outranks the showroom photograph. Conductivity tells you what the room will cost to heat. The temperature limit tells you whether the floor survives. The tog or R-value tells you whether carpet is a finish or a lid.
Choose a covering that works with the system and it returns even, cheap, quiet warmth for decades. Choose one that resists it and the same pipes, the same boiler and a larger bill produce a room that never quite arrives.