When buildings plan around their energy demands

As the global transition to renewable energy accelerates, the role of buildings in shaping sustainable energy systems has become increasingly significant. While photovoltaic systems have long since proven their potential on a small scale - for a single-family home, for example - certain questions still arise when it comes to the security of supply of our overall energy system. Is the potential of renewable energies really sufficient to ensure we have enough energy throughout the year? Or are we suddenly facing another energy shortage like in the winter of 2023? However, the primary problem here is not the production of renewable energy, but the logistics associated with our distribution system. Up until now, this has been geared towards power plants that permanently feed a certain amount of energy into the electricity grid. In order to cover our energy demand through renewable sources in the future, two factors have to be addressed: first, we need to further expand our production facilities and second, we need to make use of smart technologies that ensure grid stability at all times. This is because, unlike traditional energy sources such as coal or uranium, a solar plant does not produce the same amount of electricity at all times: it is subject to weather conditions and, above all, the day-night rhythm. Thus, energy demand must be minimized when production is low: at night, for example, while so-called production peaks must be distributed evenly in order to avoid overloading the power grid.

Automated systems seem to be ideal for managing these complex logistics. Based on local production, existing storage and availability in the grid, these systems can optimize electricity consumption in such a way that both grid stability and consumer flexibility are guaranteed at all times. In practical terms, this means that thanks to predictive energy planning, the building system ensures that I can take a hot shower or cook even when too little electricity is being produced to cover actual demand at any given time. Simultaneously, overproduced energy is not necessarily stored locally, but fed into the power grid whenever possible so that demand can be met at all times.

First of all, the researchers were able to show that their system was able to reduce the building's carbon emissions by more than 10 percent. At least as important was the realization that the building was able to communicate in advance when and how much electricity it could draw from or feed into the grid. This is particularly relevant when too much electricity is produced or demanded at peak times. The experiment has therefore shown that the flexible availability of renewable energies is not a problem a priori. However, reliable information on demand and predictive planning are essential in these circumstances: two tasks that a self-learning algorithm can handle better and more consistently than a human ever could.
In order to make the results scalable in corresponding applications, buildings must therefore be consistently digitized in the future. To ensure that the IT infrastructure required for this does not itself generate large amounts of carbon emissions, Empa researcher Hanmin Cai has already looked into salvaging reused hardware, namely old smartphones, for building automation in another study.
 

Cai and his colleague, Federica Bellizio, are working on bringing their technology to the market as part of their start-up “Kuafu”. Bellizio was recently awarded the “Empa Entrepreneur Fellowship”, a scholarship for researchers who want to set up their own company. With their data-driven system, they want to act as a bridge between grid operators and energy providers and thus make a solid contribution to energy optimization and decarbonization in the building and electric mobility sector.