The floor that drinks the rain

Rainwater collects in a shaded underground tank, where the surrounding earth holds it near ground temperature. A pump pushes it through a single continuous PEX coil laid below the red-oxide floor of a 16×16 ft room — cold in at one corner, warmed water back out to the tank. No joints under the concrete, one inlet, one outlet.

The research verdict, after adversarial verification against peer-reviewed sources: hydraulically, this is trivial — thermally, it is modest — and, as designed, it is accidentally condensation-safe. The 1 HP motor is the one clear mistake: it is wildly oversized. The honest expectation is a floor that always feels pleasantly cool underfoot (which matters in a barefoot household), taking a real but small bite out of the room's heat — not a replacement for a fan or an AC.

Hydronic loops are sized from heat, not horsepower. The standard relation is GPM = BTU/h ÷ 501 ÷ ΔT (verified against the MrPEX design manual and standard references). Even granting the room a generous 1 kW (~3,400 BTU/h) of floor cooling at a 5 °F water-temperature rise, the loop needs just ~1.4 GPM (≈5.2 L/min).

At that flow, 260–300 ft of ½″ PEX loses on the order of 8–12 ft (~2.5–3.5 m) of head — and the manufacturer method adds no penalty at all for the 16 sweeping serpentine U-turns; gentle PEX bends count as ordinary pipe length. Moving 1.4 GPM against ~3 m of head is roughly 2–3 watts of hydraulic work.

A 1 HP motor draws ~750 W to do that 3 W job — and worse, an oversized centrifugal pump run near shut-off head erodes, cavitates and short-cycles. The right tool is a fractional-horsepower wet-rotor circulator (25–60 W) — the class of pump sold for solar water heaters and hydronic systems (Grundfos UPS/Alpha and Wilo Star series are the leads to price locally; specific India-market models were the one hydraulics item the research could not verify). Bonus: a 40 W circulator running 10 hours costs about ₹3/day at TN tariffs; the 1 HP motor would cost ~₹50/day to do the same job.

Physics sets a hard ceiling on any cooled floor. Comfort standards (ISO 7730 / ISO 11855 / ASHRAE 55) require the floor surface to stay ≥19 °C, and the floor-to-room heat exchange coefficient is ~7 W/m² per °C of temperature difference (verified in Olesen's ASHRAE paper and current EN/REHVA standards). With a 26 °C room, that caps radiant floor cooling at ~50 W/m² even with ideal chilled water — rising above 100 W/m² only where direct sun lands on the floor (another reason the coil earns its keep near a west-facing opening).

Your water is not ideal chilled water. A shaded underground tank in Tamil Nadu settles near local ground temperature — ~27–29 °C (Chennai's mean annual is ~27.9 °C). With water only a few degrees below the floor, the realistic sustained output is 10–25 W/m², i.e. ~250–600 W for the 24 m² room. A peer-reviewed Thai field study using 24 °C water — cooler than your tank will manage — measured under 40 W/m². Another ground-fed radiant floor with 16–18 °C water delivered only ~300 W through the floor of a comparable room.

So: this is a cool-floor system, in the same comfort class as good shading plus a fan. In a barefoot household on red oxide, that direct conductive coolness underfoot is genuinely felt — just don't budget it as air conditioning.

The classic failure of radiant floor cooling in humid climates is sweat: a floor colder than the air's dew point condenses moisture and becomes a slip hazard, and radiant floors remove zero latent (humidity) load — a field-tested ground-fed floor changed indoor humidity by essentially nothing. Coastal Tamil Nadu dew points run 23–27 °C for much of the year, which is why chilled-floor systems there normally demand dew-point sensors and separate dehumidification. In one real-room test, condensation appeared as soon as supply water dropped below 16 °C.

Here is the elegant accident in your design: 27–29 °C tank water sits above the dew point almost all the time, so the as-proposed system essentially cannot sweat. The thermal weakness and the safety are the same fact.

The rule that binds if you ever get ambitious: never supply water more than a degree or two below the room's dew point (supply ≥ dew point, or hold the coldest spot of the floor — between pipe passes, not the average — 1–2 °C above dew point, with cutoff). Chilling this loop with any active source without a dew-point controller and dehumidification would turn the floor into a condensation plate.

The pipe is the cheap, easy part — verified against live listings (July 2026):

The ~16-pass serpentine needs ~85–92 m, so bare pipe is ≈₹8,000–16,000, and a Chennai supplier sells 100 m single-piece coils — meaning the whole room really can be one unbroken run, exactly as you intended. Prices are indicative IndiaMART quotes; negotiate, and confirm the material is genuine PEX (Viega's official India line is PE-RT/aluminium multilayer — functionally fine for this duty, but ask).

Two things the research could not verify and that need answers from your contractor or vendor before the pour: PEX-A vs PEX-B differentiation in the Indian market (listings rarely say), and the oxygen-barrier question — which conveniently matters little here, since an all-plastic rainwater loop with no ferrous components has nothing to rust.

Serpentine is a reasonable choice. A 2025 CFD study found serpentine transferred 10–15% more total heat than a spiral in a concrete slab — treat that as directional only (single study, heating temperatures) — with surface-temperature banding as the trade-off.

Spacing is worth tightening. Your 1-ft (300 mm) pitch produces markedly less uniform surface temperature than 150 mm — in the cited simulation, roughly an order of magnitude worse (>1 °C vs ~0.1 °C spot differences). With warm tank water this is merely cosmetic; if the loop is ever chilled, cold stripes become localized condensation lines. At ₹130/m, halving the pitch to 150–200 mm costs about ₹6,000 more and doubles the exchange area — the single cheapest upgrade in the whole idea.

Placement: bring the cold inlet in along the room's hottest edge (west wall / sun-patch side) so the coldest water meets the biggest load, and let the return exit the coolest corner. Even with one loop and no manifold, include: an air-purge point at the high spot, isolation valves at both ends, a strainer before the circulator, and a leaf-screen plus first-flush diverter on the rainwater feed so grit never enters the closed loop.

The tank is the system's battery, and the research flagged its endurance as the biggest open question — no verified study measured a shaded underground rainwater tank in Tamil Nadu across seasons.

A rough energy balance: at ~500 W of heat rejection, tank water rises ≈1 °C per day per 10,000 litres if the soil absorbed nothing. The surrounding earth does pull heat away — soil is a vast, slow sink and overnight it regenerates some capacity — but at 2–3 m depth TN soil is itself ~27–29 °C, so the gradient is small. Practical guidance: size the tank ≥10 kL if you can, keep the permanent shade you planned (it matters — an unshaded tank lid in TN sun becomes a solar collector), and consider running the loop at night, when the room load is lower and the tank has all day to recover.

Before trusting any of this, spend ₹500 on a data-logging thermometer, drop it in the tank for one full summer before the house is finished, and let the measured curve — not the estimate — size your expectations.

1. Pressure-test the coil while concreting. Standard hydronic practice is to hold the loop pressurised (typically ~1.5× working pressure) for the entire pour so any damage shows immediately — the exact Indian spec to hold was one of the items the research couldn't verify; agree pressure and duration with the contractor in writing.
2. Cover depth: keep a consistent screed cover over the 16 mm pipe and photograph the laid coil with a tape measure in frame before it disappears forever.
3. No joints under concrete — you already planned this; the 100 m single coil makes it real.
4. Sleeve the two slab penetrations where inlet and outlet emerge, so the pipe can move without abrading.
5. Buy the small pump, not the big one. A 25–60 W circulator, a strainer, two isolation valves, an air vent — the whole pump station should cost less than the 1 HP motor alone.
6. Log the tank temperature for a season before deciding whether to add night-sky radiator panels or simply enjoy the cool floor for what it is.