Towards comfortable, interactive and zero-carbon buildings

Published

In the future, buildings will provide enhanced, more personalized comfort to occupants while also achieving zero carbon emissions. This will be possible thanks to advances in sensors and AI, enabling new ways of interacting with our environments while leaving ultimate control in occupants’ hands. Yet, as users, we’ll also need to make lasting changes to our habits.

The construction industry – the second-biggest greenhouse gas emitter and the biggest end user of energy – is shaping a new future, driven by technological innovation and the response to climate change. Buildings have long been viewed as passive structures, but they’re in the process of becoming smart, interactive and sustainable environments. Will it be possible to effect this transition and create more comfortable homes and offices while cutting buildings’ energy use? The answer is yes, but only if we also change our habits. Engineers are studying methods for reconciling the switch to clean energy with aspirations for more comfortable buildings that also improve occupants’ well-being and adapt to their behaviors. One project in this vein is SWICE, led by EPFL and carried out by a consortium of numerous Swiss universities together with partners from the private and public sectors. The project is being funded by the Swiss Federal Office of Energy as part of the country’s strategy to reach net zero by 2050.

Control: a key element of comfort

Indoor spaces are a prominent feature of our daily lives. According to an independent study, we spend 90% of our time indoors, whether at work, at home or during leisure activities – this figure also factors in commuting time. The well-being of building occupants is, therefore, an important issue, but it’s also a subjective one. “When we talk about occupants’ comfort, we need to make a distinction between their needs and their expectations,” says Marilyne Andersen, coordinator of the SWICE project and head of EPFL’s Laboratory of Integrated Performance in Design (LIPID). Whereas needs refer to the minimum conditions required for us to stay in good health, expectations are shaped by the society we live in, our personal preferences, outdoor climate and cultural factors.

Marilyne Andersen. © 2024 EPFL – Illustration by Jeanne Guerard

“It’s clear that lifestyles oriented towards needs rather than expectations would be better for the environment,” says Andersen. “But ironically, there are some cases where some of the occupants’ core needs aren’t met – for instance, sufficient access to daylight – while expectations linked to technological progress, such as the ability to control indoor lighting with voice commands, would be. Assuming the primary purpose of buildings remains to satisfy their occupants, architects should ultimately strive to meet both needs and expectations, while making sure resources are used economically. Comfort is a subjective, relatively personal concept, although we’ve identified certain trends that apply to most people.”

For example, a study by Andersen’s research group found that occupants’ thermal comfort can be influenced by the color of the lighting in the room, where redder colors gave occupants the perception of a slightly higher ambient temperature compared to bluer ones. “Studies have also shown that occupants report a greater sense of well-being when they have agency over their environment, like if they know they’ll be able to open a window,” says Andersen. The impact of factors such as indoor air quality, natural light and noise levels can be very important, with increasingly documented effects on occupants’ health. “Not getting enough daylight exposure, for example – which often occurs when people spend a lot of time indoors, where there’s typically around 100 times less light than outdoors – can have negative effects on concentration, productivity, mood, the immune system and even quality of sleep,” says Andersen.

Dolaana Khovalyg. © 2024 EPFL – Illustration by Jeanne Guerard
“Strong, almost empathetic ties with their building”

Modern buildings are equipped with an array of sensors for measuring temperature, smoke, air quality, occupancy (to adjust the lighting) and more. Some buildings also have devices that automatically control the lighting and the heating, ventilating and air-conditioning (HVAC) systems for maximum efficiency. What’s more, these devices could soon be equipped with artificial intelligence (AI) to “learn” occupants’ habits and adjust ambient conditions accordingly in real time. For example, an AI-driven system could automatically lower a room’s temperature in the winter after confirming that it’s empty, leading to significant energy savings. Engineers at EPFL’s Laboratory of Integrated Comfort Engineering, headed by Dolaana Khovalyg, have developed these kinds of smart controllers and trained them with reinforcement learning. “Our devices can continuously and autonomously adjust the control settings for an indoor environment while aiming for several targets at once, such as lowering energy use and maximizing occupant comfort and safety,” says Khovalyg. Her team is now taking this one step further by developing models to predict the metabolic rate of individuals as they perform everyday indoor activities. These data will be used to develop control policies that automatically adjust heating or cooling in the immediate surroundings of a person.

Denis Lalanne. © 2024 EPFL – Illustration by Jeanne Guerard

Recent advances in AI have made it a powerful tool for enabling personalized, efficient building control systems. But experts agree that humans must be the ones with the ultimate say on their indoor environments. If nothing else, the feeling of control psychologically improves the sense of comfort. Denis Lalanne, an expert in human-machine interaction at the University of Fribourg, explains: “With smart buildings, it’s almost as if occupants enter into a computer. They live and work inside an intelligent structure, and there needs to be strong, almost empathetic ties with it so that it can adjust ambient settings in exactly the right way for maximum comfort and minimal carbon emissions.” Human-Building Interaction (HBI) is a fast-growing field that puts the occupant back in the building as a player, not just a passive user.

At EPFL, Andrew Sonta, the head of the Civil Engineering and Technology for Human-Oriented Sustainability Laboratory, is conducting important research in this area. “Indoor sensors not only collect data on ambient conditions, but also provide insight into how occupants interact with a building and how they affect the building’s performance,” he says. His research group has studied human-building interaction by mea-suring indicators such as power consumption and CO2 concentrations in different rooms. “People breathe in more oxygen as they speak, which could lead to an increase in the CO2 concentration of the air,” he says. “By mapping out these data, we can get an idea of the occupants’ social use of a building.”

Andrew Sonta.© 2024 EPFL – Illustration by Jeanne Guerard

Innovative technology will go a long way towards reaching the twin goal of greater occupant comfort and less energy use. But as a society, we also need to adapt our expectations. Scientists in the SWICE project are thus trying out new approaches in a number of “living labs” – real-world experimentation spaces where they work directly with the occupants of residential and office buildings to see how new approaches fit in with occupants’ daily activities. “Interventions in living labs are co-created through an active collaboration with their residents and thus future users, thereby heightening awareness about these interventions,” says Andersen. “Their effects are then tested, measured and studied under real-world conditions to give us concrete data. This serves as a pilot project to design future interventions that would be relevant to more open contexts and could be conducted on a larger scale.” When it comes to human-building interaction, engineers can use different kinds of sensors to better understand occupants’ needs and improve buildings’ ability to meet them. Sensors can also enhance the way information about a building is transmitted to its occupants. “Input is required from both sides,” says Lalanne. “Machines can’t predict everything on their own, and work still needs to be done on how they can communicate their operating conditions to occupants.”

© 2024 EPFL – Illustration by Jeanne Guerard
Tomorrow’s housing will be collective and integrated into urban planning

Is the clichéd house with a white picket fence on its way to extinction? According to 2022 figures from the Federal Statistical Office, single-family homes currently account for over half of Switzerland’s residences (although the size of these homes has shrunk considerably over the past 50 years owing to the high cost of land). Yet in view of the housing shortage in large Swiss cities, the push to stop land take and, especially, the goal of preserving biodiversity and cutting the country’s greenhouse gas emissions, the trend will probably be to shift towards collective housing designed according to a new, more ecological form of urban architecture that responds to the challenges of climate change.

“As part of SWICE, we’re examining how the notion of energy savings can be applied to Switzerland’s urban centers,” says Andersen. “This involves reconciling the complex factors making up an individual’s well-being with the degree of behavior change that people are willing to accept. Our approach incorporates the concept of quality of life through designs for public spaces, vegetation – including the capacity for vegetation to lower urban temperatures – the spacing of buildings, and the potential dynamics of building use. We’re also looking at the energy footprint of urban transportation systems, notably pertaining to commuting.”

Authors: Cécilia Carron

Source: EPFL

Share

You might be also interested in

Can we convince AI to answer harmful requests?

New research from EPFL demonstrates that even the most recent Large Language Models (LLMs), despite undergoing safety training, remain vulnerable to simple input manipulations that can cause them to behave in unintended or harmful ways.

(more…)

Can AI influence election outcomes?

Artificial intelligence (AI) may be a weapon of mass disinformation, but a recent report has demonstrated that its impact thus far has been limited.

(more…)

IC PhD graduate recognized with the ELLIS PhD Award 2024

EPFL’s Anastasiia Koloskova has won this year’s ELLIS PhD award, which recognizes European dissertations in the area of artificial intelligence and machine learning-related fields.

(more…)