UFH on uninsulated solid ground floor

[JamesPa]

This is just mad right?  

 

Or is there some part of the physics I don't yet understand which means that its somehow sensible?

 

Just checking, I don't have it nor do I plan to, but I have been asked and want to check whether my immediate reaction is right or not.

[JohnMo]

The running regime a little different to most expect, on at start of heating season and off at end - do not stop , no setback etc. stopping system means you need to rebuffer the ground and wastes a bucket load of energy.

 

The heat will only travel a given distance underground based on distance and flow temperature and that becomes a heat buffer for the house. Past that travel distance the ground doesn't absorb any more heat as the temperature is at equilibrium. You need close spacing on the UFH loops so you can run the lowest possible temperature.

 

It allows a low flow temperature as long as the rest of the house insulated. So with a heat pump you can get a good SCoP.

 

Adding a little insulation in this situation cause more losses, as once the energies below the insulation it cannot come back up and is consistently lost.

[Nick Laslett]

@JamesPa, that was my reaction when I watched an Urban Plumber YouTube video, hand waving away the need for insulation under a slab. This guy is a Heat Geek certified installer, even done a few videos with head Geek Adam. 
 

Reading the YouTube comments it appeared I was wrong and that the ground actually works as a thermal store!

 

You just need to be wary of high water table and underground water courses, that will remove heat more quickly. 
 

For clarification, I would personally advocate for an insulated raft foundation. 

Edited 17 hours ago by Nick Laslett

[JohnMo]

9 minutes ago, Nick Laslett said:

I would personally advocate for an insulated raft foundation

Not easy on a retrofit - or is but expensive and would you save that money in heating costs - most likely not

[Roger440]

7 hours ago, JamesPa said:

This is just mad right?  

 

Or is there some part of the physics I don't yet understand which means that its somehow sensible?

 

Just checking, I don't have it nor do I plan to, but I have been asked and want to check whether my immediate reaction is right or not.

 

Johnmo makes some excellent points which are, essentially as I understand it.

 

On a new build it wouldn't make any sense, but on an existing house, makes more financial sense than digging up a perfectly good floor.

 

I know, I ran the numbers. I’d be long dead before break even.

 

 

 

[JamesPa]

Thanks everyone, great discussion.

 

Lets leave aside the financial for a moment (I accept its a sound argument but its another dimension which different people will view differently).   Lets just consider the physics and what the urban plumber said.  He said that (in round numbers), for a room at 20C, its DT 10 to the floor (because the soil is at say 10C), and DT20 to the room (because the  outside air is at 0C).  But for the walls and ceiling its the internal air that is in contact with the cold elements whereas in a floor with UFH its heated pipework that is in contact with the cold elements.  So it may be DT from room to soil but its DT20-25 from pipework to soil, and that's what matters because that is where the heat is lost from.  He also said he has no evidence to support his assertions and 'gets good scop'.  Well yes, you can get good scop with a house that consumes 10kW or with a house that consumes 3kW, the SCOP is not dependent on the loss, its dependent on the flow temperature.  Im not saying he is wrong, but he admits he has no evidence to prove it and some of his arguments do appear to be wrong.

 

1 hour ago, JohnMo said:

The heat will only travel a given distance underground based on distance and flow temperature and that becomes a heat buffer for the house. Past that travel distance the ground doesn't absorb any more heat as the temperature is at equilibrium. You need close spacing on the UFH loops so you can run the lowest possible temperature.

 

There has to be a thermal gradient from the pipe temperature of 30-35 to the soil temperature of 10, and heat will continue to fall down that gradient.  So I accept its a thermal buffer but it is a lossy buffer, same as a river, the river level stays constant but water still pours out of the estuary and has to be replenished at the source.

 

Not convinced, I wonder if anyone has actually modelled this?

 

 

[JohnMo]

Asked Chat GPT and this is the answer it gave

If heating UFH with no insulation below the pipes what is the thermal gradient downwards and how far will the heat travel - assume 10 degs ground temperature

 

ChatGPT said:

 

Thought for 25s

 

Short answer up front — without insulation under the pipes most of the heat goes into the ground rather than up into the room. On a daily timescale the warming is confined to the top ~0.1–0.2 m of soil (significant effect to ~0.5–0.8 m). On a seasonal/yearly timescale the heat can travel metres (seasonal penetration ~2–3 m; detectable effect even deeper over many months). That’s why insulating under UFH is usually recommended.

Below I show how those numbers were obtained and give some typical heat-flux examples so you can see the magnitude of the loss.

---

Assumptions used

• Ground (deep) temperature = 10 °C (you stated).

• Typical soil thermal diffusivity α ≈ 0.6–1.0×10⁻⁶ m²/s (reasonable range for loam/compact soil).

• Thermal conductivity k ≈ 1.0–1.8 W/(m·K) (typical for topsoil, consolidated fill, etc.).

• Typical UFH floor/near-pipe surface temperatures used for examples: 25 °C and 30 °C.

• Rough 1-D conduction / semi-infinite solutions and simple resistance approximations — these give engineering-useful intuitive answers.

If you want different soil properties or a different floor temperature, I can re-run the numbers.

---

How far heat penetrates — theory and numbers

Penetration depth for periodic temperature changes

For sinusoidal/periodic surface temperature variations (useful to understand daily or seasonal cycling) the thermal penetration depth

δ=2αω\delta=\sqrt{\frac{2\alpha}{\omega}}δ=ω2α

where ω is the angular frequency of the cycle.

Using α = 0.6–1.0×10⁻⁶ m²/s:

• Daily cycle (24 h):
  δ ≈ 0.13–0.17 m.
  The amplitude of day/night temperature swings falls to ≈1% at about 4.6·δ ≈ 0.6–0.8 m.

• Yearly/seasonal cycle (365 d):
  δ ≈ 2.45–3.17 m.
  The seasonal amplitude falls to ≈1% at ≈ 11–15 m.

Interpretation: if the floor is being heated on a daily schedule, the thermal effect is mostly in the top ~0.1–0.2 m and small by ~0.6–0.8 m depth. If you run heating continuously for months, the seasonal/long-term warming can penetrate metres into the ground.

Steady / long-term heating

If you sustain a higher temperature at the floor for many days/weeks, the warmed layer will progressively deepen (transient conduction — see error-function solutions). Over weeks–months it moves from the decimetre scale towards metre scales (matching the seasonal values above as time grows).

---

Example conductive heat-flux numbers (simple resistance estimate)

A quick way to see losses is to assume a temperature drop from floor surface to some soil depth L and compute q ≈ k·(T_floor − T_ground)/L. This is crude but shows order of magnitude.

Examples (k, Ts = floor temp, L = effective depth):

• k=1.5 W/m·K, Ts=25 °C, Tg=10 °C:

  • L = 0.2 m → q ≈ 45 W/m²

  • L = 0.5 m → q ≈ 22.5 W/m²

  • L = 1.0 m → q ≈ 11.25 W/m²

• Range across k=1.0–1.8 and Ts=25–30 °C gives roughly 15–180 W/m² for thin to very thin effective layers (see table below). Even the mid-range numbers (20–60 W/m²) are substantial compared with typical UFH design heat fluxes (~30–80 W/m² in rooms).

(So: if the heat instead goes downwards because there’s no insulation, you can easily lose tens of watts per square metre into the ground.)

---

Lateral spread

Heat diffusion is 3-D, so lateral spread from a line or strip of pipe is similar in scale to the vertical penetration for the same time scale. In other words, on the daily scale the warmed zone sideways is of the order 0.1–0.5 m; over seasons it spreads metres laterally. This is why uninsulated slabs or ground-mounted UFH can create a warm “footprint” in the soil that extends beyond the pipes if running long enough.

---

Practical conclusions / recommendations

• Short term (daily heating): most heat stays within the top ~10–20 cm of soil; significant losses to the ground are still likely unless you have under-slab insulation.

• Long term (continuous/seasonal heating): heat penetrates metres — you will steadily heat the ground if you run continuously without insulation. That is wasted heat from an energy-cost perspective.

• If your goal is to send heat into the rooms efficiently: put a continuous insulation layer under the pipes (typical UFH practice) — it dramatically reduces downward loss and reduces the heat you must supply.

• If you want, I can:

  • run the same numbers with your exact pipe/floor temperature, slab thickness, and a chosen soil k/α; or

  • estimate expected W/m² losses for a given floor construction so you can size the ASHP/controls accordingly.

Would you like me to calculate a tailored example with your floor temperature, slab thickness, or specific soil properties?

[JamesPa]

@JohnMo  I didnt think of that and its definitely interesting.  I guess the bit that purports to justify the the key statement it makes ie 'Short answer up front — without insulation under the pipes most of the heat goes into the ground rather than up into the room.'  is the section 'Example conductive heat-flux numbers (simple resistance estimate)' and the key variable here is L, the depth into the ground before you reach the 10C assumed ground temp.  The bit about seasonal variation is also interesting, suggesting it takes months to get to a depth of metres.

 

Another thought - when we bury flow/return pipes we worry a lot about insulation, surely somewhat the same logic applies to UFH!

 

 

Edited 14 hours ago by JamesPa

[JamesPa]

@JohnMo  building on your idea I asked chatgpt a more specific question.  I have quickly read through and it seems plausible, I might ask another AI tool the same question.  If its right then UFH on uninsulated slab is mad, just put radiators in!  However I dont know if its right.  It is however prima facie more convincing than the handwaving arguments made by UrbanPlumber, but it could of course be BS which I have certainly experienced with ChatGPT in the past.

 

 

 

Q: Suppose I have underfloor heating with pipes at 150mm centres buried 15mm deep into concrete that is 100mm thick, sitting directly on damp soil assumed temp at sufficient depth 10C. Suppose that on top of the concrete is say 5mm of luxury vinyl tile and the room above is at 20C. How much heat goes down and how much goes into the room? You may assume, for simplicity, that the heating continues for several months, that the room is infinite is size and that the pipes are supplied with water at 35C

 

 

 

A: Nice — I’ll do a simple heat-resistance (steady / long-run) calculation and give numbers, plus a short sensitivity check and practical notes. I’ll keep the math short and show the assumptions so you can change any number if you want.

Model & assumptions

• Pipes run at 35°C (we take the water/pipe as an isothermal heat source at 35°C).

• Room air temperature above floor 20°C. Soil at “sufficient depth” fixed 10°C (an infinite sink).

• Geometry: slab total 100 mm thick; pipe level 15 mm below top surface → 15 mm to top, 85 mm to bottom. Pipe spacing 150 mm (0.15 m) centre-to-centre (so each metre of pipe serves 0.15 m² of floor area). Vinyl tile 5 mm thick on top.

• Material properties (typical values): concrete thermal conductivity k = 1.7 W·m⁻¹·K⁻¹, vinyl k ≈ 0.25 W·m⁻¹·K⁻¹.

• Convective (and radiative) heat-transfer coefficient at the floor surface h ≈ 8 W·m⁻²·K⁻¹ (natural convection + radiation typical value).

• Slab sits directly on damp soil; I add a small contact/soil surface resistance of 0.01 m²K/W (per m²) to account for imperfect contact and soil conduction.

• Steady / long duration (months) so temperatures and fluxes are taken as time-steady.

This is intentionally simple (1-D vertical resistances from the pipe location to the top surface and to the bottom surface). The main limitation: real piping is a line source, soil is semi-infinite and will warm over time, and pipe internal resistance / pipe outer surface temp would change the numbers — see caveats below.

---

Calculation (concept)

Treat upward and downward paths as thermal resistances per metre of pipe (area associated with one metre of pipe = spacing × 1 m = 0.15 m²).

Upward path (pipe → concrete 15 mm → vinyl 5 mm → surface → convective to room):
[
R_{up} = \frac{t_{conc,top}}{k_{conc}A} + \frac{t_{vinyl}}{k_{vinyl}A} + \frac{1}{hA}
]

Downward path (pipe → concrete 85 mm → soil sink):
[
R_{down} = \frac{t_{conc,bottom}}{k_{conc}A} + R_{soil_contact_per_m^2}/A
]

Heat flows per metre of pipe:
[
Q_{up}=\frac{T_{pipe}-T_{room}}{R_{up}},\qquad
Q_{down}=\frac{T_{pipe}-T_{soil}}{R_{down}}.
]

---

Results (base assumptions above)

Per metre of pipe (spacing 0.15 m):

• Upward heat to room: (Q_{up} \approx 14.6\ \text{W per m of pipe})

• Downward heat to soil: (Q_{down} \approx 62.5\ \text{W per m of pipe})

• Total supplied to ground+room: ( \approx 77.1\ \text{W per m of pipe})

Split by percentage:

• Upwards to the room ≈ 19%

• Downwards into the soil ≈ 81%

Per square metre of floor (divide the “per metre of pipe” numbers by spacing 0.15 m):

• Upwards to room ≈ 97.5 W·m⁻²

• Downwards to soil ≈ 417 W·m⁻²

• Total ≈ 514 W·m⁻²

(Those per-m² numbers look large because the model assumes the pipe is at 35°C and the slab/soil are cold — it’s a purely conductive picture with a warm source close to the bottom. The per-m values and percentages are the most useful result for comparing up/down split.)

---

Sensitivity (short)

A few changes strongly affect the split:

• Lower convective coefficient h at the top (poor surface heat transfer) → less heat goes up (down fraction increases).

• Higher soil contact resistance or insulation under slab → more heat goes up. Example: if soil contact resistance increases to 0.05 m²K/W (larger insulation effect), downward heat falls and the upward fraction increases substantially.

• Closer pipe spacing (more pipe per m²) → increases total heat available per m² (but the % split depends on resistances).

• Higher concrete conductivity → more symmetric flow; with high k more heat reaches soil faster (but details depend on thicknesses).

I ran a few numeric variants (keeping other parameters same):

• Lower h = 5 W/m²K → upward fraction drops (less to room).

• Raise contact resistance (soil R = 0.05 m²K/W) → Q_down falls from 62.5 W/m to ~37.5 W/m and upward share increases (about 28% up instead of 19%).

• Tighter spacing (100 mm) reduces per-pipe Q but increases per-m²; percentage changes modest if material resistances unchanged.

---

Practical interpretation & caveats

1. Most of the heat is lost downward with the geometry you gave (pipe fairly close to slab bottom relative to top temperature difference). The key reason: although the temperature difference to the room is smaller (35→20 = 15 K) than to the soil (35→10 = 25 K), the conduction resistance downward (85 mm concrete) is lower per unit area than the whole upward path (thin concrete + tile + convective/radiative exchange), when evaluated per pipe area. The geometry and the fact the pipe is closer to the top or bottom can swing the split strongly.

2. A real system normally includes insulation under the slab precisely to stop the large downward losses. If you add even modest insulation under the slab, the downward path resistance rises fast and most of the heat will go up into the room. In the sensitivity run, increasing bottom contact resistance from 0.01 to 0.05 m²K/W cut downward heat by ≈40% and boosted the upward share.

3. Transient soil warming. Over months the soil immediately beneath the slab will warm and the downward heat flux will reduce compared with the initial value. I assumed an infinite 10°C sink (worst case continuous draw) so the numbers are conservative for downward loss. If the heating has been on for a long time, the steady downward flux will be smaller (because the soil warms).

4. Pipe/pipe wall & internal convective resistance were ignored (we treated pipe water as 35°C source). In reality pipe outer temperature will be slightly below water temp; that would slightly reduce flux numbers but not qualitatively change the split unless the pipe/fluid resistance is large.

5. Line source & lateral spreading. A rigorous solution would model the cylindrical pipe as line source in a slab and account for lateral spreading and the semi-infinite soil transient. The simple 1-D resistive model is fine for quick engineering estimates and to show how to reduce losses.

---

Quick takeaways / recommendations

• With the numbers you gave roughly 80% of the energy goes downward and only about 20% into the room (under the simple assumptions above).

• If that is undesired, add insulation under the slab (or a thermal break) — even modest insulation under the slab will shift heat upwards dramatically.

• Reducing spacing (more pipe per m²) raises the heat available to the room for a given supply temperature; shifting pipe closer to the top surface also helps.

• If you want, I can:

  • rerun the calculation with different soil contact resistance (insulation thickness/k), different h, or different spacing;

  • compute the required underside insulation (thickness/k) to get e.g. 70% of heat upward; or

  • do a more realistic line-source slab transient estimate (requires a bit more detail).

Which follow-up would you like? (I can re-compute numbers right away with any changed assumptions.)

.

Edited 14 hours ago by JamesPa

[JamesPa]

Duck.ai came up with ~50:50 room/soil with a slightly different methodology.  For that I has to specify a room 7m square (ie a house), but otherwise the same assumptions.  I also gave it 10m square and it came up with a similar ratio.   I am not sure which underlying model Duck.ai uses  so this may not be 'independent'.

 

Im now definitely not buying the handwaving argument that most of the heat goes up though.  I would say the jury is out, do this at your own risk!

[JohnMo]

Think the takeaway is expensive heating bill even if you get a scop of 5+

[Roger440]

I'm not convinced, though that's based on the fact, that, practically, this would be done on an existing slab, so the pipes wont be buried as has been used in the examples above. They will be lying in slots cut into the surface, and depending on the proposed floor covering, left like that or grouted in. 

 

I'll leave it others to ask ai, but the real world solution I suggest will, logically alter the ratio of what heat goes where.

 

Interesting though, as this is exactly what I will be doing. Much as having insulation under the slab would be great. I'm not ripping up six inches of perfectly good concrete to do so. £5-7k (even doing it myself) buys a lot if energy in whatever form. Never mind the disruption, and ridks if undermining shallow foundations.

 

There's also some benefits in a traditional house with no dpm of gently warming the base of the walls and driving the moisture out.

Edited 13 hours ago by Roger440

[JamesPa]

1 hour ago, Roger440 said:

I'm not convinced, though that's based on the fact, that, practically, this would be done on an existing slab, so the pipes wont be buried as has been used in the examples above. They will be lying in slots cut into the surface, and depending on the proposed floor covering, left like that or grouted in. 

 

I'll leave it others to ask ai, but the real world solution I suggest will, logically alter the ratio of what heat goes where.

 

Interesting though, as this is exactly what I will be doing. Much as having insulation under the slab would be great. I'm not ripping up six inches of perfectly good concrete to do so. £5-7k (even doing it myself) buys a lot if energy in whatever form. Never mind the disruption, and ridks if undermining shallow foundations.

Obviously there is quite a bit dependent on construction detail (something builders are notoriously sloppy about!), but if the pipes are in good contact with 'grout' ('grouted in') then that wouldn't be much different to being buried in concrete.  Only if there is a void underneath (ie insulation!) can I see it making much difference.

 

On the strength of this I wouldn't do it personally, nor would I dig up the floor because, as you say, it costs a lot.  Id fit radiators!

[Roger440]

1 minute ago, JamesPa said:

Obviously there is quite a bit dependent on construction detail (something builders are notoriously sloppy about!), but if the pipes are in good contact with 'grout' ('grouted in') then that wouldn't be much different to being buried in concrete.  Only if there is a void underneath (ie insulation!) can I see it making much difference.

 

On the strength of this I wouldn't do it personally, nor would I dig up the floor because, as you say, it costs a lot.  Id fit radiators!

I already have radiators. Which are fine. But I do like UFH. Was fantastic in our last place, which I installed, though that did have foamed glass insulation under the slab. The original slab had to come up anyway on that house though.

 

I'll take my chances. Mainly because i dont believe 85% of the heat will go down. And, of course, currently im already losing heat down through the slab as its already in my room!

 

It's certainly not an uncommon thing to do. I can easily rent the machine to cut the grooves.

[JamesPa]

48 minutes ago, Roger440 said:

And, of course, currently im already losing heat down through the slab as its already in my room!

Yes, but currently you have air at 20C in contact with the slab not pipework at 35C, makes a big difference.  Your choice of course!