The useful output transferred to the water comes from a single relation: P (kW) = 1,163 × flow rate (m³/h) × ΔT (K), where 1,163 is the specific heat capacity of water expressed in Wh per kilogram per kelvin. The whole difficulty on site is making that measurement usable: at a ΔT of 3 K with sensors at ±0,2 K, the uncertainty on output already reaches 9,4 %, against 2,8 % at 10 K. Where the metering is used for billing, legal metrology takes over, with type examination, initial verification and a metrology logbook, and compliance presumed through EN 1434.
What a heat meter measures, and the relation that matters
The relation to know by heart
A heat meter fitted on the circuit does not read the boiler. It calculates the heat transferred to the water, from the flow rate and the temperature difference between flow and return. For water, instantaneous output is P (kW) = 1,163 × Q (m³/h) × ΔT (K). The table below gives the order of magnitude directly, handy for checking a data plate in thirty seconds.
| Flow rate | ΔT 5 K | ΔT 10 K | ΔT 15 K | ΔT 20 K |
|---|---|---|---|---|
| 0,5 m³/h | 2,9 kW | 5,8 kW | 8,7 kW | 11,6 kW |
| 1,0 m³/h | 5,8 kW | 11,6 kW | 17,4 kW | 23,3 kW |
| 1,5 m³/h | 8,7 kW | 17,4 kW | 26,2 kW | 34,9 kW |
| 2,0 m³/h | 11,6 kW | 23,3 kW | 34,9 kW | 46,5 kW |
| 3,0 m³/h | 17,4 kW | 34,9 kW | 52,3 kW | 69,8 kW |
The 1,163 coefficient only holds for water. As soon as glycol is present, density and heat capacity change, and you have to take the fluid's values from the manual, or accept a systematic error of several per cent in the wrong direction.
Why ΔT decides whether the measurement is credible
The relative uncertainty on output combines the uncertainty on flow rate and the one on ΔT. But the uncertainty on ΔT is absolute, in kelvin, while ΔT itself varies: the smaller it is, the more the relative error explodes. With two sensors at ±0,2 K each, the combined uncertainty on ΔT is about 0,28 K, and this is what it gives, before any flow-rate uncertainty.
| ΔT measured | Uncertainty on output |
|---|---|
| 2 K | 14,1 % |
| 3 K | 9,4 % |
| 5 K | 5,7 % |
| 7 K | 4,0 % |
| 10 K | 2,8 % |
| 15 K | 1,9 % |
| 20 K | 1,4 % |
That is the real basis of the field rule that says to avoid measurements below 3 to 5 K. It is not a convention, it is arithmetic, and it is what you put to anyone claiming to conclude on a ΔT of 2 K.
What you get, and what it does not prove
You get an instantaneous output in kW and a cumulative energy in kWh over the period. Cross-referencing that useful energy with the fuel consumption read off the meter gives you an on-site efficiency. It stays apparent, it does not replace a combustion measurement and it settles nothing on regulatory efficiency. For continuous consumption metering, see what a thermal energy meter involves.
Choosing the instrument and the measurement point
Ultrasonic, electromagnetic or mechanical
The choice depends mainly on the circuit water and on maintenance. Ultrasonic is often the easiest to live with, no moving parts, good stability if the system is well vented. Electromagnetic targets conductive fluids and needs careful fitting. Mechanical turbine types stay robust and affordable, but they are more sensitive to fouling and wear, so to drift over time.
Temperature sensors: immersion, pairing, class
Output depends as much on ΔT as on flow rate. Favour immersion sensors in thermowells, of the same mounting type on flow and return, paired in accordance with EN 1434. Contact sensors get you out of trouble in a diagnostic, but they add an error that is not symmetrical between flow and return, so directly an error on ΔT. Note the accuracy class stated by the manufacturer and keep the sheet in the file.
Flow rate, diameter and fluid compatibility
Check the nominal flow rate and above all the minimum measurable flow rate: below that threshold the device displays values without guaranteeing them. Check the DN and the recommended straight lengths upstream and downstream. Where glycol is present, check compatibility, correction factors and temperature range. The measurement point must sit away from any bypass loop and away from the immediate mixing zone of a 3-way valve.
Field method: measuring output step by step
Preparing the visit
Start with the boiler in steady state. Set a stable load by opening enough emitters, then let it run for 10 to 15 minutes. Check the balancing of the radiators or the underfloor circuit. Vent if you hear air or if zones stay cold. Note the outdoor temperature, the setpoint and the operating mode, those three values govern the interpretation.
Fitting the sensors
Fit the sensors on flow and return as close as possible to the generator, on a clean metal section, and respect the direction of circulation. Check the flow measurement point separately. Insulate the sensors with foam or tape to limit the influence of ambient air, otherwise the flow sensor reads low and you underestimate ΔT, and therefore output.
Calculate, plot, compare
Measure the flow rate and ΔT, apply the relation, and compare with the nominal figure on the data plate. Take readings every 30 to 60 seconds and plot the output curve to spot cycling, modulation and drift. To set those measurements against a generation efficiency, then compare the measured useful output with the energy consumed over the same period.
Interpreting the measurement to diagnose
Output too low: where to look
If measured output is well below expectation, start by checking the actual flow rate and the flow/return ΔT. A fouled exchanger, a clogged filter, a sludged system or a stuck 3-way valve cut flow rate or heat transfer. Also check pump speed, mounting direction and loop balancing. An abnormally high ΔT with low output almost always signals a low flow rate.
Output too high or unstable
Output that is too high, or that swings, usually points to a heating curve, an over-ambitious hot water setpoint or badly configured controls. On an oversized generator, short cycling appears quickly, with frequent start-ups and degrading efficiency. If ΔT is inconsistent with the flow rate, cross-check against the burner's modulation range before blaming the appliance. What remains is to set the measured output against the real demand of the home, which Argile recalculates to NF EN 12831-1 and compares with the unit selected to flag oversizing.
Cross-check with the other tests
The measurement does not stand alone. Cross-check it with a combustion analysis (CO, O2, air and gas setting), the flue gas temperature and the stability of the draught. Walk the system too, pipe insulation, plant room temperatures, returns running too hot. That confirms whether the loss comes from the generator or from the distribution, and that distinction is what shapes the quotation.
Metrology framework and traceability
What legal metrology requires
As soon as metering serves billing or contractual monitoring, the instrument falls under legal metrology. In France, the arrêté of 3 September 2010, made under the decree on the control of measuring instruments, requires type examination, initial verification before commissioning and a verification of the installation, with listed exclusions. Instruments are presumed compliant when they satisfy the tests in parts 1, 2 and 4 of EN 1434 (2007), and readings must be expressed in joules, watt-hours or their decimal multiples.
What the holder has to keep
The holder of the instrument must keep a metrology logbook, preserve the integrity of the seals and maintain the compliance of the installation. It is rarely anticipated on the quotation, even though it decides the legal weight of everything you measure afterwards. An in-line fitting requires shutting down and partially draining the circuit, and that downtime is priced before the visit, not after.
The file you leave behind
Keep time-stamped readings from before, during and after the works, sharp photos of the fitting, the thermowells, the labels and the data plates. File the sizing note and the product data sheets alongside the retrofit assessment where one exists. Finish with a readable report setting out flow rate, ΔT, calculated output, the uncertainty assumed and the gap to nominal. That document is what wins an arbitration, not the raw value on the display.
