Understanding How A Heating Curve Actually Works

The Heating Curve For Water is a line on a graph that tells a boiler what supply water temperature to aim for based on the outdoor air temperature. It is the backbone of outdoor reset control in hydronic heating systems. You set a couple of points, the controller interpolates between them, and the system adjusts. That is the theory anyway. The reality is messier, and most installers get it wrong on the first run. I have spent years tuning these curves across residential and light commercial buildings. What works on paper rarely matches what happens in practice without adjustments. The curve itself is simple. It runs from a minimum outdoor temperature to a maximum, and at each point it specifies a required supply temperature. When it is freezing outside, the curve demands hotter water. When it is mild out, the curve backs the temperature down. The goal is to keep the indoor space stable while wasting as little energy as possible.

The Practical Heating Curve For Water

Let me skip the textbook definition and talk about how you actually dial one in. Start by picking two anchor points. The standard convention is the design outdoor temperature and the design indoor temperature, but the numbers matter more than the labels. For a well-insulated house in a cold climate, you might set the low outdoor point at -18°C (0°F) and the high outdoor point at 15°C (60°F). The corresponding supply temperatures would be around 75°C (167°F) and 40°C (104°F). These are starting positions. They are not answers. Once you have those two points entered, watch the system run for several days across different weather conditions. Pay attention to whether the spaces are too hot or too cold and how quickly the system recovers after setbacks. If the rooms are consistently 2-3 degrees too warm on moderate days, the curve is too steep. You need to lower the supply temperature range. If the house never gets comfortable during a sustained cold snap, the curve is too flat or the minimum supply temperature is set too low. I encountered a specific problem with a retrofit a few years back. A two-story building had been heated by a cast-iron boiler with a fixed heating curve that the previous installer had left untouched for eight years. The upper floor was always uncomfortably warm while the ground floor barely met setpoint during peak loads. The curve itself was fine on paper. The issue was that the upstairs had smaller radiancy surface area relative to its heat loss due to being on the top floor with significant ceiling losses, while the downstairs had a large perimeter and older single-pane windows. My workaround was splitting the system into two independent zones, each with its own pump and its own heating curve. The upper zone ended up running around 10 degrees cooler on the curve than the lower zone. This cost an extra circulator and a few control boards, but it eliminated the constant complaint cycle and dropped the gas bill by about 18 percent over the following heating season.

Setting Up The Curve In Practice

Most modern modulating boilers have the curve built into the controller. You enter the two outdoor supply endpoints, sometimes a third parallel shift point, and the boiler handles the interpolation. The controller also applies a parallel shift if you need to nudge the whole curve up or down without recalculating every point. A parallel shift of +3°C raises the supply temperature by 3 degrees at every outdoor condition. It is a quick fix when the building has a consistent offset from the curve prediction. For systems using external mixing valves or third-way valves, the curve translates directly into valve position commands. The controller outputs a percentage opening based on where the current outdoor temperature falls on the curve. A linear curve means the valve opens at a constant rate as outdoor temperature drops. A quadratic curve starts more conservative and ramps harder as it gets colder. Linear is easier to understand and tune. Quadratic can reduce overshoot on mild days but makes debugging more confusing when something goes wrong. You need an outdoor sensor. It should be mounted on the north side of the building, shaded from direct sun, and positioned at least 1.5 meters above ground. If you mount it on a south-facing wall that absorbs solar radiation, the curve will think it is warmer outside than it actually is, and the boiler will run too cool. I have seen this happen repeatedly. The building complains about cold, the installer blames the curve, and the real problem is a sensor baking in afternoon sun. Move it to a shaded location and the whole system usually resolves itself within a day.

There is a common misconception that you need a separate curve for each radiator type. Radiator size and output capacity are handled by the initial sizing, not by changing the curve shape. The curve assumes the emitters are correctly sized for the building heat loss. If they are not, no amount of curve tweaking will fix it. You will just end up chasing symptoms instead of the cause.

Where The Curve Breaks Down

The heating curve assumes steady-state conditions. It does not account for thermal mass in the building, solar gains, internal gains from occupants or equipment, or wind effects on heat loss. During a clear winter day with strong sun, the curve will keep pushing hot water through the system because the outdoor sensor reads cold, but the building is gaining heat from the sun. The result is overshoot and short cycling. Some controllers have a solar gain compensation feature that lifts the curve slightly on sunny days. If yours does not, you can manually apply a negative parallel shift during the daytime and revert it at night. It is a manual intervention, but it keeps things comfortable without adding expensive hardware. Another failure mode is when the curve is applied to a system with a mixing loop that has a large buffer tank. The tank smooths out the temperature swings, which sounds good until the boiler cycles too frequently because the water in the tank never reaches the temperature the curve demands. A 50-liter buffer on a 20-kilowatt boiler is undersized for this purpose. You end up with short firing cycles that reduce efficiency and increase wear on the burner. The workaround is either adding a larger buffer or running the boiler in a dedicated low-temperature circuit with a separate mixing loop that is governed independently from the curve. This adds complexity but prevents the cycling problem entirely. Flued combustion appliances have a minimum firing rate constraint. If the curve calls for a very low supply temperature on a mild day, the boiler may not be able to modulate down far enough to match the load. It fires at its minimum output, the water gets too hot, and the system oscillates. The solution is a weather-compensated outdoor air damper on the flue or a thermostatic bypass valve that maintains a minimum flow through the boiler even when the load is low. The bypass valve is simpler and cheaper. It does add a small parasitic loss, but the difference is usually under 2 percent of annual fuel consumption.

The most honest thing I can say about heating curves is that they are a starting point, not a finished solution. A properly tuned curve on a well-designed system can cut heating fuel use by 15 to 25 percent compared to a fixed-temperature thermostat. But if the building envelope is leaky, the emitter sizing is off, or the controller is misconfigured, the curve will either overheat the space or fail to maintain comfort, and you will waste more energy trying to compensate. There is no substitute for measuring the actual indoor temperatures over a full heating cycle and adjusting from there. The graph on the controller screen is only as good as the data you feed into it.