Set the curve from two points: your design outdoor temperature paired with the design flow temperature from the heat loss calculation, and mild weather around 15 °C paired with a flow of 23–28 °C. Then lower the whole curve 2–3 °C at a time, allowing 48 hours per step, until a room stops reaching target — and go back one step. Each degree removed is worth 2–2.5% of COP.
What is the weather compensation curve doing?
It replaces a fixed flow temperature with a sliding one. Weather compensation reads the outdoor sensor and sets the flow temperature from a straight line: cold outside, hot water to the emitters; mild outside, barely warm water. The heat pump then runs continuously at low output instead of blasting and stopping.
That matters because the design flow temperature is only needed on the coldest day of the year. A fixed 50 °C setting delivers the January maximum for the entire season and wastes 2–2.5% of COP for every unnecessary degree — which, over an average European heating season, is most of the running cost difference between a good system and a mediocre one.
What two points define the curve?
Every controller expresses the curve differently — a slope number, a "heating curve 0.4", or two editable points — but they all describe the same line. Fix it by its ends.
Point one: the design condition. Your design outdoor temperature (typically −1 to −5 °C in western Europe, colder inland) paired with the design flow temperature from the heat load calculation. If nobody gave you a design flow temperature, that is the first thing to ask for.
Point two: the no-load condition. The line should pass through the point where outdoor temperature equals room temperature, because at that point no heat is needed. Setting the curve through 20 °C outdoor / 20 °C flow makes the whole line fall out of the design point automatically:
| Emitters | Flow at −3 °C | Flow at +7 °C | Flow at +15 °C |
|---|---|---|---|
| Underfloor sized to 35 °C | 35 °C | 28 °C | 23 °C |
| Radiators uprated for 45 °C | 45 °C | 34 °C | 25 °C |
| Original boiler-era radiators | 55 °C | 40 °C | 28 °C |
Most factory defaults sit well above these lines, which is why so many heat pumps arrive over-efficient on paper and expensive in practice.
How do I tune it without freezing the house?
Tune downward, slowly, in winter — never in April, when mild weather flatters any setting.
- Open everything first. Fully open the thermostatic valves in the reference rooms and set any room thermostat 2 °C above target so it stops interrupting. You cannot tune a throttled circuit.
- Lower the whole curve by 2–3 °C at the design end, keeping the mild end anchored.
- Wait 48 hours. A house has thermal mass; 24 hours is not enough to see the result, and a single cold night proves nothing.
- Check every room, not just the hall. The coldest room sets the limit.
- When a room stops reaching target, go back one step and stop. That is your curve.
- Then rebalance. If one room lags while the rest are warm, the answer is a lockshield valve or a bigger radiator in that room — not a higher curve for the whole house.
A season of this typically ends 5–10 °C below the installer's default, which is worth 10–25% of heating electricity.
Do I still need a room thermostat?
As a limit, yes; as the primary control, no. Weather compensation should decide the flow temperature and the heat pump should run continuously; a room thermostat is there to stop the house overheating on a sunny afternoon or when a wood stove is lit. Using a thermostat as the main control forces the machine to full flow temperature and then off again, producing exactly the cycling that damages efficiency — see why does my heat pump keep switching on and off?
Setback schedules deserve the same caution. Dropping the target 4 °C overnight saves little on a heat pump and forces a high-temperature recovery in the morning at the worst possible outdoor condition. A shallow setback of 1–2 °C, or none, usually wins.
Why does my curve seem wrong?
- One cold room is a balancing or emitter problem, and raising the curve to fix it heats every other room too much.
- The house is warm but the bill is high — the curve is too high and there is margin to take out.
- Rooms overheat in mild weather — the mild end of the curve is set too steep; pull the +15 °C point down towards 23–25 °C.
- The unit cycles in mild weather — the curve may be fine and the problem is minimum output against a small load; check system volume, as in buffer tank vs volumiser.
- Hot water seems unaffected by the curve — correct. Cylinder charging uses its own fixed setpoint, not the heating curve.
No manufacturer, Solimpeks included, can set this curve for you at the factory: it is a property of your building’s heat loss and emitters, not of the machine.
Record what you change and when. A curve tuned across one real winter is the single cheapest efficiency upgrade available to a heat pump owner, and it costs nothing but patience — see what flow temperature should I run my heat pump at? for the underlying targets.
Frequently asked questions
What should my weather compensation curve be set to?
To a line running from your design outdoor temperature at the design flow temperature — 35 °C for underfloor, 45 °C for uprated radiators, 55 °C for original ones — through the point where outdoor temperature equals room temperature. Then lower it until a room stops reaching target.
How much does weather compensation save?
Typically 10–25% of heating electricity compared with a fixed flow temperature, because each degree of flow temperature is worth about 2–2.5% of COP and most of the season is much milder than the design day.
Should I use a room thermostat with weather compensation?
Only as an upper limit. Weather compensation should set the flow temperature and let the heat pump run continuously; a thermostat used as the main control forces full flow temperature followed by shutdown, which causes cycling and higher bills.
How long should I wait between curve adjustments?
At least 48 hours. A building's thermal mass means a change of 2–3 °C in flow temperature takes a day or more to show in room temperatures, and judging it overnight leads to overcorrection in both directions.