Lower energy consumption means reduced costs and lower emissions. Discover how the power of AI makes it possible to develop solutions for minimizing a building’s energy waste.
Our team has built an AI solution that monitors and forecasts consumption for several types of energy at Akademiska Hus, the most interesting of which are district heating and electricity. To create reliable forecasts, the AI is trained on a long time series of high-quality hourly data for more than a thousand buildings, corresponding to nearly 4 million square meters across Sweden.
The savings come from two directions. The first is based on the AI monitoring and detecting whether the actual consumption of a building in the last few hours exceeds the estimated consumption by the AI. This indicates a probable fault in the heating systems that leads to excessive consumption. The building is then flagged and placed on a warning list and displayed in a dashboard for the operating organization so that an investigation can be initiated and remedial action can be taken. The difference from more traditional monitoring of energy consumption is that this usually takes place on a monthly basis. Larger, but difficult-to-detect, faults in buildings can in some cases lead to increased costs of up to SEK 100,000 before they are caught using traditional methods. Faster follow-up and thus reduced unnecessary consumption leads to savings both financially and climate-wise.
The second is based on the fact that the AI also makes forecasts of consumption one week ahead of time and thereby can identify days with expected high consumption. As a consumer of district heating, you pay not only for the consumption but also for the maximum consumption during one day per year (called daily power). By forecasting daily power peaks, measures can be taken to reduce these. For example, buildings can be preheated, where the concrete in the buildings is used as a kind of thermal battery, and during the day when consumption would otherwise have been very high, less energy needs to be added, thereby keeping the daily power down.
By reducing power peaks, it is estimated that power costs can be reduced by between 5% and 10%, which corresponds to several million SEK per year in total. In addition, reduced daily power leads to climate benefits. On the days when the daily power is likely to be at its highest, it is probably high for many consumers, as these often occur when it is at its coldest, and the need for heating is at its highest. When the energy demand is high throughout society, the energy mix needs to be supplemented with the most polluting and climate-damaging energy sources. Every kWh saved these days is one less kWh of burned coal or oil, which means that the positive marginal effect for the climate is very large.
During the course of the work, a robust infrastructure has been built up and the first version of monitoring and forecasting has been implemented. The next step is to expand and fine-tune the system to realize its full potential, and to implement automated actions based on forecasts.
The goal is to develop a seamless process where the AI not only warns of potential problems but also acts proactively to optimize energy systems, through automatic feedback and control. If the AI system detects that the power is at risk of being exceeded within a certain period of time, it should not only warn but it will also step in and, for example, preheat buildings and then reduce the heating during the cold days with forecasted high power need.
Don’t hesitate to contact us if you want to know more about how we can help optimize your business with AI or are curious about how you could get started with AI!