Applied research study building

For a heat-pump owner on a dynamic electricity contract, the bill comes down to a question of timing. Take Saturday 25 April 2026 in the Netherlands: day-ahead prices went negative for eight straight hours through the middle of the day, bottoming at −€190/MWh, then climbed past €156/MWh in the evening. With the addition of renewables, day-ahead prices have grown more volatile year over year. For a household with a heat pump, that volatility is the biggest energy cost lever a homeowner has.

Three years ago, Eleflex.AI, the company behind HeatAdapt, partnered with Tallinn University of Technology on an applied research study. Over the full 2023/24 heating season, we tracked more than ten Estonian homes with a mix of air-to-water and ground-source units, all on dynamic electricity contracts. A related peer-reviewed paper followed in the CLIMA 2025 proceedings (Võsa & Kurnitski, Residential Heat Pump Electricity Price Based Control Savings Potential).

Inside the study: dynamic-price control of heat pumps

The study began with a digital twin of selected participating buildings, used to model how different price-based control patterns would affect savings and indoor comfort. Live metering was then added to the actual homes, all in daily use, to record consumption, heat pump COP and indoor temperatures. Through the winter of 2023/24, the team ran the heat pumps under dynamic-price control and analysed the results in detail. The season turned out cold, with an average outdoor temperature of −3,4 °C and the coldest nights below −15 °C. All heat pumps in the study buildings performed reliably under those conditions.

One of the study's goals was to assess buildings as grid flexibility assets. Each building acts as a large thermal battery, with heat stored in its walls and floors. The team tested several hourly price-based control algorithms across the residential buildings to find which captured the most savings without affecting comfort. The HeatAdapt prototype controller was used to preheat during cheap periods and coast through expensive ones without changing how warm the house feels. In practice that translated into 12 to 20% off the monthly bill on average, climbing past 28% in months with the largest day-ahead price swings.

Applied research study: buildings on the left, digital twin on the right
View of a single-family house from the study (left) and the simulation model geometry (right).

The study also quantified the value of small behavioural changes. Lowering the indoor setpoint from 23 to 21 °C in a 167 m² detached home delivered a 22% saving on heating costs (over 120 euros a year) purely from the temperature change. Adding dynamic price control to a moderate setpoint adjustment produced the largest combined savings, several hundred euros per year.

A third comparison examined what a deeper investment can add. A digital twin of a 1970s detached home was renovated in simulation to a B-energy-class envelope and paired with an air-to-water heat pump. Heating costs fell by 69%, and domestic hot-water energy use by around 50%. Renovation delivers the biggest absolute reduction; dynamic-price control delivers the best return on the smallest investment.

Three years on: the savings are even bigger

Three things have shifted since the study. Day-ahead volatility has widened sharply, dynamic-price contracts have spread from Scandinavia across the EU on the back of smart-metering, and our algorithms based on the study have spent multiple winters learning from real homes and household feedback.

Across the more volatile dynamic-price regions, around 90% of HeatAdapt users now save between 14% and 24% on their annual heat-pump electricity bill. That worked out to more than 200 euros for a typical household over the past 12 months, about a quarter more than the year before. Larger buildings saw bigger absolute results: for example, an apartment building with 69 kW of heating capacity saved over 1 100 euros. In the most volatile months, savings climbed to 31%.

Hybrid setups, such as an air-to-water heat pump paired with a gas boiler, can also yield significant savings. For these, the HeatAdapt controller automatically chooses the cheaper option, based on the current electricity and gas prices and the heat pump's COP drop in cold weather. Through the colder-than-average 2025/26 winter with low gas prices, the heat pump ran 60,7% of the heating season on Estonian day-ahead prices. The balance for the cheapest source can shift drastically with conditions: in December 2025 the heat pump handled 81,6% of the heating, while in the extremely cold January 2026 it was the gas boiler that took over most of it, at 79,6%.

For a homeowner with a heat pump on a dynamic contract, the takeaway is straightforward: shifting heat-pump consumption to the cheapest periods of the day is the biggest single cost lever a household has. The 2023/24 study put a measured range on it. Three years on, the lever is growing, and it works on the heat pump and contract you already have.

Reference

Võsa, K.V. & Kurnitski, J. (2025). Residential Heat Pump Electricity Price Based Control Savings Potential. Proceedings of the 15th REHVA HVAC World Congress, CLIMA 2025. Springer. link.springer.com/chapter/10.1007/978-3-032-10546-2_26

Note: The cited paper covers a single building modelled with a digital twin, one part of the broader study described in this article.