Houses can learn to save electricity

By Kristin Straumsheim Grønli

This article was originally published on Forskerforum on December 19, 2023

With the help of "collective intelligence", clusters of houses can save electricity together say researchers in Ålesund.

The goal is buildings that automatically reduce consumption and costs and, at the same time, respond to the needs of the energy grid. All without compromising comfort, says project manager Mohammadreza Aghaei.

Crazy electricity prices scream at you from an online newspaper or other app. You turn down thermostats, turn off switches and introduce a shower, laundry and dishwashing ban for the whole family.

You monitor the price development and barely take the chance to do laundry at four o'clock one morning while you look for a wool sweater. The inside temperature has dropped to 16 degrees. You are afraid that the bathroom floor may cause frost damage.

Good decisions about electricity are easier said than done. How much and how long can the heat be turned down without the people living in the house starting to freeze? When is it most economical and environmentally friendly to run the dishwasher?


Project in six countries

Perhaps it is a consolation that researchers are working with systems that can make houses smart enough to make such decisions on our behalf.

- There is great potential in simpler, faster and not least automatic control of electricity use - for all kinds of buildings, says Aghaei, project manager and senior researcher at NTNU Ålesund.

This can save electricity and money and become an important resource for the electricity grid to balance production and consumption and prevent overloading.

Smart buildings may be asked to reduce consumption for a while. Together, many such houses can become a force to be reckoned with. The up-scaling must be done by imitating swarm behaviour in nature, such as, for example, fish estimations.

The project has partners in six European countries and includes municipal buildings in Ålesund, apartment blocks in Milan in Italy, a "living lab" in Grenoble in France and several university buildings in Nicosia in Cyprus. The funding comes from the EU program Horizon 2020 and amounts to NOK 40 million over four years.


Cheap equipment

We equip the houses with equipment that enables them to learn the patterns and needs of residents and users and to see this in connection with both the electricity prices and the needs of the energy system, explains Aghaei.

Imagine a period of extreme cold. Thousands of smart homes let the dishwasher run right after dinner and instead pause it until late in the evening. In this way, they can help prevent the network from overloading when everyone comes home from work and simultaneously turns up the heat.

The project has many industrial partners at home and abroad, and a large part of the equipment is specially developed, including sensors that measure temperature, air quality and light level. It is necessary to ensure that the comfort lasts.

Medya Temelli Fenerci is coordinator in the project and advisor at NTNU Ålesund. She emphasizes that it should not be expensive. "We invest in making it easy and cheap to upgrade the smart level of both old and newer houses", she explains. For example, the researchers are using smart plugs and smart thermostats for electrical appliances that are not smart in the first place, such as dishwashers, refrigerators and radiators.



Construction currently uses 40 percent of the energy in Norway, according to SINTEF. The number is equivalent for the EU.

The potential for energy efficiency using smart solutions is great. Almost 75 per cent of the building stock in the EU are existing buildings with a smart level of zero, without opportunities to interact digitally with the power grid and users, says Aghaei.

Our system will reduce electricity consumption in participating buildings by at least 16 per cent.

Switching from fossil fuels to electricity is key to achieving climate goals and the green transition. Therefore, the strain on the power grid will increase, and power saving will become more important.

In this country, we have good access to energy, but the power grid is not designed for many people to use a lot at the same time. Pushing part of the consumption away from peak hours can help save society from the network's expensive and often unpopular development.

The technical term here is "flexibility". "The increase in energy use from variable sources such as solar and wind power is helping to make this increasingly relevant", says Aghaei. Such consumer flexibility has already become a commodity.


Data at the edge

Aghaei refers to the market operator NODES, which is owned by Agder Energi and offers local marketplaces for flexibility. Here, network companies buy from intermediaries such as the electricity supplier Tibber.

Such an actor can gather flexibility from customers who have installed smart solutions, for example, charging the electric car. This will be cheaper in return for the power company to control the charging time and speed. The government body Enova provides subsidies to housing estates that acquire this intelligent control solution.

In the NTNU project, the researchers want the houses to adjust their consumption based on what the building has learned about the users and what the users or occupants have entered regarding preferences.

"Most similar systems use cloud storage of data. In our system, data about what happens in the house will stay there", says Aghaei. The project manager emphasizes that this strengthens privacy and reduces risk. Information about the conditions in a residence can, for example, help burglars assess whether the house is vacant for holidays.

The researchers use something called edge computing. It is about bringing calculation and data storage closer to the data sources. In the wall or ceiling, the researchers install a small and affordable single-board computer (Raspberry Pi). Everything comes here, from measurements of the indoor environment and the pattern of the users to settings that the users have entered into an app on the mobile phone, as well as information about electricity prices and any inquiries from the electricity network.

"On the single-card computer, all this is fed into simple and easy algorithms that enable the house to make sensible decisions", says Aghaei. The development of the algorithms and the machine learning itself is the foundation for the artificial intelligence in the system and constitutes a large part of the research in the project.


Build in shoals

The setup facilitates scaling up using collective intelligence. Bee hives, fish nests, anthills and flocks of birds are all examples of advanced formation in a complex system. The school of fish turns at lightning speed when the killer whale approaches, so quickly that it can look like the fish are thinking with the same brain. In the NTNU project, the demand from the power grid is the orca, while the houses in the neighbourhood or town are the fish.

"The shoal does not need each fish to have high intelligence", explains Amin Moazami. He is the project's initiator and was the project manager for the first period before switching to a research position at SINTEF in Oslo. He still has a role in the project and supervises doctoral students who participate.

For us, the energy flow becomes the dynamic phenomenon (the shoal) that can be controlled with the help of collective intelligence, says Moazami.

The houses decide how to react to signals from the power grid, including what they want to do to adjust consumption for a while.

We will test such collective intelligence in Ålesund in seven buildings during the project.


The municipality is involved.

Ålesund municipality has two children's schools, one youth school, one office building, two care centres and one sports hall in the project. Per Martin Leinan is the building manager and ENØK responsible for Ålesund municipal property.

The motivation is more than just the possibility of reduced electricity bills. As a municipality, we must be drivers of the green shift. It is interesting to be close to the development and get the chance to have a small impact, says Leinan.

The project also aims to use relatively simple steps to make buildings several notches smarter. "That is a carrot in itself", says the building manager. He adds that projects like this also create commitment - not only in and around the municipal management but also among the users of the building - who get to participate and influence the result through surveys and the like. "I clearly believed in this research", says Leinan.

The project is close to the market, and the researchers invest in commercializing the package solution with hardware and software when the project period is over.


Article published on Forskerforum (in Norwegian) on December 19, 2023