CavLab All articles
Energy & Retrofit

Curves Against Cast Iron: What Weather Compensation Cannot Know About a Retrofit

CavLab
Curves Against Cast Iron: What Weather Compensation Cannot Know About a Retrofit

Photo: retrofit radiator with heat pump controller and thermometer in British bungalow, via cdn.britannica.com

Weather compensation is one of the more seductive ideas in domestic heating. The controller reads the outdoor temperature, consults a curve, and chooses a flow temperature that should match the building's heat loss. On paper it replaces a fixed high flow temperature with a lower, steadier one. In a retrofit bungalow, however, the curve is only as good as the assumptions behind it. Over two weeks this autumn I ran three air-source heat-pump controllers on the bench and then on the same 1968 bungalow, first at 2°C and then at 12°C, to see where the mathematics met the radiators.

The bench is not the building

The three controllers are anonymised as A, B and C. All were set to their manufacturers' default weather-compensation curves, with no room influence and no schedule. On the bench, each supplied a controlled water loop through a calibrated flow meter and a 6 kW load bank. I logged flow and return temperatures at ten-second intervals and recorded compressor modulation where the controller exposed it.

Controller Flow at 2°C Flow at 12°C Minimum modulation
A 45°C 32°C 25%
B 42°C 30% 30%
C 48°C 35°C 20%

At 2°C the requested flow temperatures were 45°C, 42°C and 48°C; at 12°C they fell to 32°C, 30°C and 35°C. On the bench, each held its requested temperature within 1.2 K once steady. The differences appeared at low load. Controller B, with a minimum compressor modulation of 30%, overshot its 30°C target by 2.4 K before pausing. Controller C cycled most often because its minimum output was lowest, but its pump continued at full speed for a fixed 90 seconds after the compressor stopped, moving heat that the curve had not asked for.

The bungalow and its cast-iron assumptions

The test house is a 1968 detached bungalow of 84 m² with cavity wall insulation, 180 mm of loft insulation, double glazing and its original radiators. Those radiators were sized for a 70°C flow temperature, not the 45°C or 35°C that a heat pump would prefer. The measured heat loss is roughly 3.1 kW at -2°C and 1.3 kW at 12°C.

At 2°C, controller A held the living room at 20.8°C, B at 20.2°C and C at 21.1°C against a 21°C setpoint. The differences were modest and mostly explained by sensor placement and hysteresis. At 12°C the picture changed. All three curves asked for flow temperatures between 30°C and 35°C. In a radiator system with a low emitter exponent, that is close to the temperature at which useful output begins to disappear. The rooms still warmed, but the controllers were no longer steering the result; internal gains were.

The 12°C day was damp and overcast, with cooking and a tumble dryer adding heat in the afternoon. The curve cannot see those gains. Controller A's 32°C flow left the radiators tepid and the compressor running long, yet the living room rose 0.8 K above setpoint by mid-afternoon. Controller B's 30°C flow produced four cycles per hour. Controller C produced nine. None of them was wrong according to its own mathematics. All of them were guessing about the building.

Cycling, comfort and the mild-weather problem

At 2°C, cycling was modest: A averaged two cycles per hour, B 1.5 and C three. At 12°C, A rose to seven, B to four and C to nine. The mild-weather problem is not only efficiency. It is comfort. A radiator at 30°C feels cool to the hand and gives up heat slowly. When the controller pauses, the room can drift down before the next cycle. Occupants notice that as a cool spell even when the average temperature is correct.

The curve also has no way to distinguish a dry 12°C from a damp 12°C. Latent losses and air movement change the building's response, while solar gain and internal gains change it in the other direction. Weather compensation is an open-loop estimate. It becomes a control strategy only when it is paired with room feedback, load compensation or a commissioning process that measures the actual emitters. Without those, it is a formula with a thermometer attached.

What the numbers say

The controllers did what their curves promised. The building did not always do what the curves assumed. The most useful intervention was not a steeper or shallower curve but a room sensor with modest influence, set to adjust flow by no more than 3 K. The second was lowering the minimum modulation or accepting longer cycles at mild temperatures. The third was measuring the radiator exponent rather than copying a default curve from a well-insulated new build.

Weather compensation is not an illusion. It is a useful first approximation. In a British retrofit, however, it is only the beginning of control. The curve stops being a strategy when it cannot see the room, the radiators or the damp afternoon outside. At that point it becomes a guess, and the occupier pays for it in comfort rather than in kilowatt-hours alone.

All Articles

Related Articles

Condensation and the Clock: What Three Heat-Pump Dryers Cost in Time and Kilowatt-Hours

Condensation and the Clock: What Three Heat-Pump Dryers Cost in Time and Kilowatt-Hours

The Distance Premium: A Fortnight of Power Quality at a Rural Supply Extremity

The Distance Premium: A Fortnight of Power Quality at a Rural Supply Extremity

The Shape of the Current: Twenty Domestic Loads and the Waveform Behind the Nameplate

The Shape of the Current: Twenty Domestic Loads and the Waveform Behind the Nameplate