Skip to main content

Heat Optimization with AI

What is Heat Optimization with AI?

Energy optimization generally involves using available resources – heat, cooling, ventilation, lighting, charging – in a way that minimizes costs and environmental impact without compromising comfort or functionality. When artificial intelligence is used in this work, the conditions change fundamentally: instead of fixed schedules and manual adjustments, the system can continuously learn, predict, and adapt.

From Static Control Systems to Learning​

Traditional heating control is based on control systems programmed by a human: "start the pump when the outdoor temperature is below 17 °C," "raise the temperature on weekdays, lower it on weekends," "compensate the heat to the radiators." This works in stable environments – but reality is more complex. The building behaves differently depending on weather, occupancy, and season. This leads to unnecessary temperature variations, excessive air turnover, resulting in increased energy use and a larger carbon footprint.

AI-based heat optimization replaces or complements these fixed rules with models that are continuously updated based on real measurements. The system learns how the building reacts to cold, sun, and wind, and adjusts the control accordingly – automatically and without manual intervention. A good analogy is the weather services' "temperature" and "feels like temperature."

Climate Control of Buildings and the Weather Services' "Feels Like Temperature" – a Comparison

When you check the weather on your phone before a walk, you usually see two numbers: the actual temperature and the feels like temperature. One day, the thermometer may show +4 °C, but the feels like temperature is −2 °C due to the wind. Another day, +8 °C in calm sunshine may feel like ten degrees warmer. It is well known that the measured temperature rarely tells the whole truth.

The exact same principle applies to buildings – and it has profound implications for how buildings should be controlled.

What is Measured, and What is Experienced?​

A traditional outdoor sensor measures air temperature: it is a precise and reproducible measurement. But it is not the same as what a building is affected by.

Buildings are influenced by at least three factors in addition to air temperature:

  • Space and Solar Radiation The building's roof and exterior walls cool as heat radiates out into the atmosphere – regardless of air temperature. For example, a car roof can have frost even if the outdoor temperature is above 0 °C. Conversely, solar radiation on roofs, walls, and windows provides an energy boost that the outdoor sensor cannot measure.
  • Wind. The effect of wind chill outdoors can make the air movement indoors – from ventilation, leaky windows, or convection currents – cause the room to feel colder than what the sensor shows.
  • Humidity. A damp stone facade is affected when water evaporates through so-called sorptive cooling. You can compare it to dipping your hand in warm water. In the water, your hand is warm. When you take your hand out, the water starts to evaporate, energy is released from the skin, and you experience the sorptive cooling effect.

The Parallel to the Weather Services' Model​

Meteorologists' "feels like temperature" is about moving from a simple measurement point to a contextualized picture of reality. A building controlled solely based on outdoor temperature is in the same situation as a weather app that only shows the thermometer value. It has a piece of the truth – but not the whole.

What It Means for Climate Control​

A heating system that only uses outdoor temperature as a control parameter implicitly assumes that all other factors are constant. This is rarely the case. A sunny winter day with a clear sky and calm winds can provide significant solar heat through windows to the building – energy that a system without climate data does not "see." The result is overheating and unnecessary energy use. In the evening, when the sun has set and the temperature drops quickly, the same system may react too late.

A windy day with high wind chill and/or sorptive cooling cools the facade and infiltrates cold air through leaks – but this is not reflected at the outdoor temperature sensor. The system may underestimate the need, and the temperature drops indoors.

Just as a wind chill correction improves the relevance of the weather forecast, a composite picture of climate impact enhances the accuracy of property management.

The Misplaced Outdoor Sensor​

A heating system's heat curve is directly dependent on the outdoor temperature that the system measures. If that sensor measures incorrectly – not due to a technical fault, but because of where it is located – the system compensates for a climate that does not match reality. A common example is outdoor sensors placed in courtyards in urban environments. A courtyard is often sheltered from the wind and surrounded by facades that absorb and re-radiate heat. On a cold winter day, the temperature in the courtyard can be several degrees higher than on the exposed street side. The sensor reports −3 °C while it is actually −7 °C against the facade where the heat losses actually occur. The heating system thinks it is warmer than it is – and heats too little. The result is that operational technicians or managers over time compensate for this error by manually raising the heat curve. This solves the symptom but not the cause: the curve is now set against an incorrect measurement point, and the system is permanently misconfigured. In milder weather – when the courtyard's microclimate is more representative – the building is overheated. The energy consumed unnecessarily never appears as a single error, only as a high energy bill. The same problem arises with sensors placed in direct sunlight, near ventilation outlets, under eaves, or on facades with unusually high or low exposure. Common to all is that the measurement point is local and situation-dependent – it measures its own microclimate, not the climate that the building is actually exposed to.

From Measurement Point to Experienced Comfort – the Task of Smart Control​

The modern answer to this is to combine multiple data sources and model the building's actual thermal behavior – not just react to a single measurement point. This means incorporating:

  • Outdoor temperature, but also wind speed and solar radiation
  • Actual indoor temperature from multiple points in the building
  • Weather forecasts that allow for proactive rather than reactive action
  • Knowledge of the building's thermal inertia – how long it takes to heat up or cool down

The goal is not to measure more for the sake of measuring – it is to control towards what is actually experienced as comfortable and energy-efficient, just as "feels like temperature" provides a more useful figure than raw thermometer data.

One Last Parallel: Baseline and Deviation​

Weather services also report deviations from normal: "three degrees warmer than the seasonal average." This provides context. A measurement without a reference point tells little. The same applies to energy management. A building that consumes 180 kWh/m² per year – is that good or bad? It depends on climate zone, age, usage, and a range of other factors. It is when comparing against a well-defined baseline and tracking deviations over time that one can see if the management is actually making a difference. AI-based systems do just this: they build a model of what "normal" is for the specific building and continuously measure how outcomes relate to forecasts. Deviations drive improvement.

Optimate – Control that Understands More than the Outdoor Sensor​

Enkey Optimate is built around the insight that an outdoor sensor is not enough. The system combines weather forecasts with actual indoor temperature from wireless room sensors and the building's own history to calculate optimal control in real time. Just as "feels like temperature" provides a more relevant picture than the value you read on a thermometer, Optimate provides control based on real comfort and actual energy needs – not just a number from a sensor on the north side of the facade. The result is a more stable indoor climate, lower energy consumption, and a heating system that reacts to what affects the building's heating and cooling, not what the outdoor sensor measures.

What is Meant by "AI" in This Context?​

In energy optimization of buildings, AI primarily involves three techniques:

  • Machine Learning means that algorithms identify patterns in historical data – for example, how long it takes for a building to reach the desired temperature after a cold night – and use these patterns to make better decisions going forward.

  • Predictive Modeling combines internal measurements with external data sources such as weather forecasts to anticipate future needs. If a cold snap is expected within 12 hours, the system can act proactively rather than reactively.

  • Optimization Algorithms weigh multiple goals against each other – the lowest possible energy use, maintained comfort, reduced peak loads – and find the control that best balances all these demands under given conditions.

What Can Be Optimized?​

AI-based energy optimization can be applied to a range of systems in a property:

  • Heating and Cooling is the most common area. By dynamically controlling the supplied heat – based on actual indoor temperature and weather forecasts rather than a fixed schedule – energy use for heating can be significantly reduced, by 10-20%, without compromising comfort.

Why is it Relevant Now?

For a long time, energy was a predictable cost in the property budget. The price varied little from year to year, and there were rarely strong reasons to actively optimize consumption. That picture has fundamentally changed.

District Heating Prices Have Increased Significantly​

District heating is the dominant heating method in Swedish multi-family houses and many commercial properties. In recent years, prices have risen sharply among a large number of network owners. Behind the increases is a combination of rising fuel and production costs, increased network investments, and an energy market that has become more volatile and difficult to predict. Since district heating is delivered as a monopoly within each network area, property owners have limited opportunities to switch suppliers. This puts pressure to instead reduce demand – to use every kilowatt-hour more efficiently and avoid unnecessary consumption. Energy optimization in this context is not just about comfort, but about direct cost control in an environment where prices are determined by others.

Electricity Prices are More Volatile and Harder to Plan For​

Electricity has shifted from a stable background cost to one of the more unpredictable items in the property's operating budget. Extreme price variations – between hours, days, and seasons – have become the norm rather than the exception. The price difference between a night with high wind power and a cold winter morning with high demand can be multiple.

For properties with heat pumps, electric heaters, or electricity-dependent ventilation systems, these variations directly affect operating costs. The capacity charge – a separate cost based on the highest power draw during a measurement period – has also gained increased weight in network tariffs and can constitute a large portion of the total electricity cost. Just a few unfortunate peaks per month can make the charge noticeable.

Starting in the fall of 2025, electricity will be traded in the EU's day-ahead market in 15-minute intervals, making price variations even more granular. This increases both the need for and the possibility of precise control: a property that can adjust its consumption in 15-minute intervals can take advantage of price dips and avoid peaks in a way that was not previously possible. Enkey addresses this with the service demand flexibility.

The Requirements for Monitoring and Reporting are Increasing​

Alongside price developments, regulatory requirements are tightening. The EU's CSRD directive means that a growing number of property owners and managers must report their energy consumption and climate impact in a traceable and audit-friendly manner. Vague estimates are no longer sufficient – actual measurements, methodology, and the possibility of external review are required.

This creates dual pressure: to reduce consumption and to be able to prove that it is being done. By integrating existing meters or supplementing with new meters, Enkey Building Insight® can help you comply with the CSRD directive. Here you can read more about reporting requirements

In Summary​

Rising district heating prices, volatile electricity, and capacity charges that hit hard at the wrong time – along with increased demands for measurability and reporting – make energy optimization a strategic issue rather than a technical detail. What was previously a nice-to-have has become a direct economic and regulatory necessity for those managing properties with the ambition to keep costs under control.

Optimate – Intelligent Heating Control for Your Property​

Enkey Optimate is a concrete example of how AI-based energy optimization works in practice. The system combines wireless room sensors, weather forecasts, and the building's own temperature data, with real-time and historical energy and power data, to continuously adjust heating control – without needing to replace any existing system.

The results are measurable: more stable indoor temperature, lower energy consumption, and reduced capacity charges. And since all control is logged and visualized in real time, you get the basis required for monitoring, reporting, and future optimization.