Underfloor heating warms the stone, then the room
Underfloor heating does not warm a stone floor so much as charge it. Where the comfort of a heated floor is decided, and what to send with the order.
16 November 2026 · 19 min read · GranTerra Stone, Hui'an
An underfloor heating system is specified in metres of pipe, litres per minute and a manifold in a cupboard. The stone is not in that description at all, although it is the layer the client will actually touch with a bare foot. That gap is where most of the disappointment in a heated stone floor begins: the heating is designed as a heating system, the stone is chosen as a finish, and nobody designs the thing the two of them make together.
The useful way to hold both in one hand is this. The stone is not the heating. It is the part that stores what the heating gives it. Almost everything a client notices afterwards — how quickly a room comes up, how long it stays warm once the valve has closed, why one room feels generous and the one next door feels mean — comes out of that single property, and out of the two or three decisions that sit under the stone rather than in it.
The stone is at the end of a queue
Heat produced in a pipe does not arrive at a floor surface. It arrives at the screed, which has to be warmed before it will pass anything upward; then at the bed or the adhesive, which is thin but not instant; then at the stone. Each layer has a mass, and each mass has to be raised before the next one gets its share. The granite is not slow because granite is difficult. It is slow because it is last in a line of four.
The same queue runs in reverse when the heating stops. The pipe cools in minutes, the screed takes much longer, and the stone keeps releasing what it has already taken. A floor that took an evening to warm will still be giving heat back the following morning. Nothing in that is a fault, and no finish, thickness or supplier changes its direction.
It does explain two complaints that sound like opposites. The first is that the floor never seems to warm up on the first cold week of the year. The second is that the room is still warm hours after the heating has been switched off, sometimes uncomfortably so. Both are the same behaviour seen from either end of the queue, and both are answered by the same thing: the floor should be brought up before it is needed, not when it is needed. A heated stone floor responds to a schedule. It does not respond to a switch.
This is the first place where a buyer’s expectation and a stone yard’s scope separate. We cut the surface. The schedule belongs to whoever is running the building. If the commission is a hotel with a fixed opening date, the useful question is not how fast the stone warms but when the system is first switched on, because the answer decides whether the floor is ready for the first guests or for the second week of them.
A wall thermostat reads the air, and the air is the last to know
The device on the wall measures the temperature of air. Air is light, and it is warmed by the floor rather than by the pipe. So the thermostat is not reading the thing that is producing the heat; it is reading a consequence of it, with a delay, from a position that is often a metre and a half above the source and in the corner of the room least representative of it.
What follows is a cycle that has nothing to do with the stone and everything to do with the sensor. Sun crosses the floor in the afternoon, the air warms quickly, the thermostat decides the room is satisfied and closes the valve. The floor, which was already warm, goes on releasing into a room that no longer needs it. By evening the air is too warm. Then the floor finally cools, the air drops, the valve opens, and the room runs cold while the floor spends another few hours catching up. The client experiences a floor that is either stuffy or cold, and concludes that stone behaves badly.
The remedy is a floor sensor rather than a wall one, or a control strategy that anticipates the lag instead of reacting to it, and in most buildings it is cheap at the design stage and awkward afterwards. It is also outside our scope: we do not design heating. The contrast worth holding on to is that heat arriving from below is a completely different problem from heat arriving from above, which is what a fire beside the same material is about: one is uniform, slow and patient, the other is local, sudden and unforgiving. But the control logic changes what a client should be told at the point of choosing stone, because a room that will be run from a wall thermostat is a room where a heavy floor will feel unpredictable, and a lighter one will feel merely slow.
There is a version of this that catches out refurbishments. A stone floor is laid over an existing system that was set up for carpet or timber, with the same manifold, the same valve and the same sensor. The system now has several times the mass it was designed around. It will still work. It will simply answer the thermostat later than the thermostat expects, and the further the room is from the manifold, the more pronounced the effect.
A hollow piece is a blanket laid over the pipe
Every stone floor is tested by ear in its first week, whether anybody intends it or not. A piece that sounds hollow when tapped has a void under it, and over a heated slab a void is not a cosmetic issue. Air is the best insulator in the whole assembly. Where there is a gap, the heat stays in the screed and the surface above runs cooler than its neighbours, so a single dull-sounding tile can be the coldest square metre in a bathroom and nobody can find a fault in the boiler.
The cause is nearly always the bedding. A full bed of adhesive or mortar leaves no space to trap air. Spot bonding, or a bed laid thick in the middle and thin at the edges, leaves plenty. This matters more on a heated floor than anywhere else, and it matters even more where the stone is large, because a large piece cannot be pressed down evenly onto a bed that is not already flat.
We do not lay floors, and we are not in a position to inspect a bed we did not build. What we can say is that a hollow note in the first week is worth acting on immediately, while the piece can still be lifted and re-bedded without damage. Left until the floor has been in use for a season, the same void has usually collected moisture and stained the stone from beneath, and the repair becomes a replacement. That progression is covered in more detail in what sits below a stone floor and decides its fate, and it is the same mechanism whether or not there is a pipe in the slab.
Thickness buys delay, not warmth
The assumption that a thicker stone is a warmer floor is common and it is wrong in a specific way. A thicker piece holds more heat, so it releases for longer after the heating stops. It also takes longer to reach temperature in the first place, and it takes longer to change when the room’s demand changes. What thickness buys is stability. What it costs is response.
For a floor in continuous use — a lobby, a restaurant, a bathroom in a hotel that is occupied every night — that trade is a good one. The floor never has to catch up because it is never allowed to fall behind, and the mass smooths out the small variations of a day. For a floor used twice a day, in a house where the bathroom is cold from nine in the morning until seven in the evening, the same mass is a liability: the room is asked to warm a floor that has spent ten hours cooling, and it will be asked again tomorrow.
So the question is not which thickness heats fastest. It is how often the room is used, for how long, and whether the heating will be left running. A client who wants a floor that is warm within twenty minutes of asking is not describing a stone floor at all. They are describing a different system, and no change of material will deliver it.
There is a second consequence, less discussed. A thicker piece is more forgiving of a floor that is not perfectly flat, and less forgiving of a joint that has to close a gap. On a heated floor the two effects pull against each other: the mass you want for stability is the mass that makes a low spot harder to disguise. That is why the thickness and the bedding are decided together, and why changing one of them late is more expensive than it looks.
One thermostat, two floors
Stone in a hallway and timber in the room beside it, on one zone and one valve, is one of the most common arrangements in residential work and one of the least comfortable. The two materials conduct heat at very different rates and they have very different limits. The valve is set to satisfy one of them, and the other lives with the result.
Set the zone for the timber and the stone is under-heated, so the stone feels cold underfoot and the client blames the stone. Set it for the stone and the timber runs at a surface temperature it was not meant to see, which shows up over a season as gapping, cupping or a finish that dulls in the traffic lanes. The failure is not in either material. It is in the assumption that a zone is a room.
Three honest options exist. Split the zones, so each material has its own control and its own schedule. Accept the stone as the reference and keep the timber out of the heated area, or keep the heated area small enough that the mismatch is limited to a doorway. Or change the specification so the two floors are the same material, which clients rarely want and which is at least a decision rather than an accident.
The version of this that reaches a stone yard is a drawing with a single floor finish note and a schedule showing stone in four rooms and timber in three. What the drawing cannot show is whether the pipe layout follows the room boundaries or runs straight through them. When it runs straight through, the timber is being heated by a loop designed for stone, and the gap at the doorway becomes the place where the two behaviours are forced to meet. That junction is the subject of where stone meets a different material and who ends first, and a heated floor makes it worse by adding a temperature difference to a movement difference.
The first warm-up belongs to somebody else
Before a stone floor is laid over a heated slab, the slab has to be dry enough, and the usual way to get it there is a controlled warm-up: bring the system up in stages, hold it, then let it come down, so that moisture leaves through the surface rather than being trapped under a finish. Where that sequence is done well, the stone is laid on a slab that has already been through its first cycle. Where it is skipped, or compressed to protect a handover date, the moisture leaves later, through the joints and the bed, with the stone already on top of it.
What fails in that case is not the granite. Granite does not crack because a floor reached its working temperature. What fails is the bed, the joint, or a piece that was fixed hard at both ends with nowhere to go. The stone gets the blame because the stone is what the client can see, and the report that follows names the material rather than the sequence.
We are not on site for this, and we would not pretend otherwise. What we do ask, before an order for a heated floor is released, is whether the warm-up has happened and whether it was recorded. A date, a maximum flow temperature and a signature are enough. There is a second question that gets asked less often and matters as much: whether the floor will be finished early in the programme and then spend months under other trades. A heated floor that is laid first and protected last is a floor being asked to survive what a working site does to new stone while the joints are still taking up their first movement. Where the answer is that the programme did not allow for it, the honest response is to expect trouble at the joints rather than at the surface, and to keep a spare piece from the same block, which is the argument in why a spare belongs in the first shipment rather than the second.
Underfoot, the finish changes contact, not temperature
A polished floor and a flamed floor in the same room, over the same pipes, are at the same temperature. They do not feel the same. What differs is how much of the sole of a foot is actually in contact with the stone, and contact is what moves heat. A polished surface meets skin over a large area and takes heat from it efficiently, so it reads as cool in a cold room and as distinctly warm on a heated one. A flamed or bush-hammered surface touches skin at many small points and leaves air between them, so it reads as less cold and less warm, and it feels the change more slowly.
This is the part of a heated floor that surprises clients in both directions. Somebody who has chosen polished granite for a bathroom because it looks right, and then finds it almost too warm to stand on at the end of a heating cycle, has not been mis-sold anything. They have chosen a surface with maximum contact and then heated it. The same person in an unheated room would describe the same floor as cold.
The consequence for a specification is that the finish should be chosen for the room’s use and then the comfort accepted, rather than chosen for comfort and then hoped to look right. The seven surfaces we work, and what each does to a floor, are set out in the finishes we cut and where each one belongs. On a heated floor the useful rule is simpler than the list: the rougher the surface, the more the floor behaves like a floor and the less like a hot plate.
When a piece has to come up
A repair on a heated floor is a different job from a repair on a cold one, and the difference is depth. The pipe sits below the stone, at a depth that varies with the system, and every tool that goes down has to stop before it reaches it. A chisel becomes a decision rather than a habit. A grinder that would be routine on an unheated floor is a risk to the loop, and a cut pipe in a screed is a repair that costs more than the whole floor.
Once the piece is out, the bed has to be rebuilt rather than patched, because a patch will leave the same void that caused the problem and will run cooler than its neighbours. The joint has to be remade on four sides. And the replacement has to match a floor that has been down long enough to have changed: a granite that has been walked on, cleaned and exposed to light for a year is not the same colour as a piece from the crate. This is exactly the case what a repair can and cannot restore sets out, with the heated floor adding one more limit — the floor has to cool before the work and warm again afterwards, so a two-hour job becomes a two-day one.
The cost of the whole episode is almost never the stone. It is access, the bed, the joint, and the disruption. That is the reason the spare piece is worth carrying from the first order, and the reason a hollow note in the first week is worth a phone call rather than a note for the defects list.
The cold floor with no heating under it at all
Most complaints about cold stone have nothing to do with heating. A stone floor in an unheated room feels cold because it takes heat out of a foot and spreads it into a mass that is large and, in most buildings, connected to the ground. The stone is not cold. The floor is doing what a dense material does, and the client is feeling the direction of the heat rather than its amount.
That distinction matters because it changes the remedy. If the complaint is comfort in an unheated room, the answer is the assembly beneath: insulation, a thermal break, or a different substrate. Changing the granite from one dense stone to another dense stone will not change the sensation by any amount a person can feel. The one material decision that does change it is the surface, for the reason in the previous section, and even that is a smaller effect than a layer of insulation nobody can see.
Where this becomes a live issue is in a building that mixes both cases: a heated ground floor and an unheated basement, or a heated bathroom and a landing with no loop under it. The same stone in both, laid the same way, behaves differently, and the client reads the difference as a defect in the stone. It is not. It is the same piece of granite in two different assemblies, and the note that explains it is worth writing at handover rather than after the first winter, which is part of what the first year of a stone floor asks of whoever is caring for it.
What a heated floor asks of the order
None of the above changes what we cut. It changes what we need to know before we cut, and it changes what a buyer should expect the stone to do afterwards. Four things cover most of it: whether the heat comes from water or from an electric element, whether the floor is one zone or several, what else shares the floor, and how each room is actually used through the day. A fifth, where it applies, is whether any of it is a wet area, because a heated wet floor combines two sets of decisions that are usually made by different people.
From those answers, three things follow that are ours to control. The thickness stays consistent across a single floor, so that no piece responds faster than its neighbours. The pieces are numbered and packed in the order they will be laid, which matters more on a heated floor because a piece that waits in a corridor is a piece that has to be lifted twice, and a crate that is opened in the wrong order is a version of the number that decides where a piece ends up going wrong at the worst moment. And the joints are cut to the width the design calls for rather than closed tight, because a floor that is warm for half the year has to be allowed to move at the edges. What we will not do is design the system, warrant the assembly, or promise a surface temperature. Those belong to the heating engineer and the flooring contractor, and saying so early is cheaper than being asked later.
Where our part ends is set out plainly in the half of a stone job that never appears in a quote, and a heated floor is a good example of why the line is worth drawing before the order rather than after. We can hold a spare from the same block. We can keep a floor consistent in thickness and in colour. We cannot make a heavy floor answer a thermostat quickly, and no supplier can.
The two documents worth sending before the stone is chosen
The most useful thing a buyer can send with an enquiry for a heated floor is not a specification of the stone. It is a section through the floor and a note of how the building will be run. The section shows the layers, which is where the response time lives. The note shows the schedule, which is where the comfort lives. With those two, the stone choice becomes a decision with a reason attached; without them, it becomes a preference, and preferences are what get changed at the last minute.
If the project already has a flooring contractor, the questions worth putting to them are the ones that decide the outcome: whether there is a floor sensor, whether the zones follow the rooms, and what the warm-up record looks like. Those are not stone questions and we cannot answer them. But they decide whether the stone the client chose will be described in a year as generous or as cold, and the difference between those two descriptions is almost never the granite.
A heated floor is a slow floor, and the slowness is the product rather than the problem. The stone stores what it is given and gives it back later, on a curve that no finish and no thickness can straighten. Once that is understood, the whole specification simplifies: keep the water going, keep the mass even, keep a spare, and give the far end of the queue enough time to arrive. Everything a client will say about the floor afterwards is decided in those four sentences, and all of them are settled before the first cut. A short note to the sales desk with the section and the schedule is enough to start.
Every figure on this page comes from our own order sheets. Where something is a trade norm rather than a number we can hold, we say so — send a spec and we will confirm it against a real block before you commit.
More from the yard
Send the drawing. We do the rest.
Reply within one working day, in English or Chinese. Send a spec, a drawing or just a photo of the sample you are matching — we will tell you honestly whether we can hold it.
MOQ 1 × 20FT · FOB XIAMEN · SAMPLES IN 7 DAYS