For four years, a team of TU/e researchers, which includes Lisanne Havinga and Roel Loonen, in collaboration with TNO and Smart Twin have been working on a tool to change how everyone in the Netherlands can make decisions about renovation options for their homes. Now, the Renovation Explorer project is ready to launch this tool. The big launch will take place at the Renovation Explorer congress on Thursday June 4 on the TU/e campus.
"It was only when I started working on this project that I started to appreciate the intricate interplay between radiator types, heat pumps, insulation, and ventilation in our homes, and how this affects our home energy use," says Lisanne Havinga .
Havinga - who is an Assistant Professor in the Building Performance group in the Department of Built Environment, a role that she combines with a managing consultancy role at Common Futures - is the driving force behind the Renovation Explorer tool.
"It's a tool for homeowners to gain better and easier insight into sustainable renovation options for their homes that could save them money and lead to a more comfortable and sustainable lifestyle in the future," says Havinga.
A timely development
Since 2022, Havinga and her collaborators inside and outside TU/e have been working on the Renovation Explorer tool, which is timely given the vulnerability of our individual energy needs and costs to global energy crises caused by geopolitical upheavals.
"Homeowners can lower their long-term costs by renovating their homes, but with so many renovation options, it can be an overwhelming process for many people," says Havinga. "We want to provide homeowners with a tool to navigate these decisions."
Roel Loonen , Associate Professor in the Building Performance group in the Department of Built Environment, also worked on the project. Over the course of the project, Loonen stepped into Havinga's role during her maternity leave.
"Current tools for analyzing renovation options for a home require a lot of time to be set up for an owner's home. Added to that, existing tools focus on the assessment of single renovation options. But there are so many possibilities and the influence everything else. The Renovation Explorer tool changes all this," says Loonen.
The explorer trigger
Havinga has vivid memories about the trigger for the Renovation Project.
"As part of the 2018 national climate agreement negotiations, I was exploring ways to upscale the sustainable renovation of Dutch homes. During the discussions, I was asked: 'What do we need to do to ensure that 100,000 proposed home renovations are cost effective and of high quality?' Long discussions followed, and it was from these that the idea for the Renovation Explorer project was born."
From idea conception, Havinga then put a project together, which was eventually approved by the Netherlands Enterprise Agency (Rijksdienst voor Ondernemend Nederland) to tune of €3 million. Before approval, Havinga had to convince the government of the viability of the project that it was a good idea.
"In 2019, we were given €30,000 and three months to build a prototype, which worked. Then in 2020, we built a second successful prototype over eight months with funding of €100,000. Then we launched the tender in the summer of 2021 with partners TNO and Smart Twin. After that, we started the project in April 2022."
The unique selling points
Now, Havinga, Loonen, and their collaborators unveil the Renovation Explorer tool at the Renovation Explorer Congress on June 4 at TU/e. But what's different about this tool in comparison to other tools out there for homeowners?
"Most tools use static monthly or yearly averages. They're good at calculating your heating bill per month, but things like air quality, grid load, and user behavior are not static. They are dynamic," says Havinga. "So, the Renovation Explorer tool is a dynamic tool."
Current tools search a database based on postcode and house number, and the tool identifies the build year and house type, a row house for example, and uses the average row house from that year to provide renovation recommendations. "In reality, no two row houses are the same, but our tool allows the user to input their exact heating system and characteristics of their house," says Havinga.
Finally, the tool also accounts for household behavior. "Some people are hardly at home and energy-use conscious, while others are home all day and run their heating 24/7. Users can input their behavior such as the rooms they heat or whether they leave the windows open in their bedroom at night," notes Havinga.
What drives the tool?
So, what drives the Renovation Explorer tool?
"Real physics with help from machine learning," Loonen says. "It's based on an advanced energy balance simulation that includes key building physics, such as the principles of heat transfer. We use the physics model to assess the performance of a wide variety of scenarios that are used to train the machine learning model. After that, the tool can be run on a normal laptop!"
During one demo of the tool for this story, the tool searched through over one trillion renovation packages for a row house!
The most surprising aspect of training the model is the small number of scenarios needed to train a model that is able to accurately predict the performance for more than one trillion scenarios for all these different houses and users.
"We only need about 350,000 scenarios out of trillions to train the model. I must admit I struggle a bit with the small size of this training number," says Havinga. " Alexis Cvetkov-Iliev - a postdoctoral researcher and machine learning expert from the Department of Mathematics and Computer Science - is part of the team. I've challenged him several times by asking 'Is 350,000 simulations really enough?'"
In fact, training the model with many more scenarios does not significantly improve model accuracy. "The 350,000 scenarios provide an accurate predication of the trillions of scenarios," says Havinga. "Alexis convinced me of this!"
For Loonen, the tool addresses an access imbalance. "Machine learning models, such as the one behind the Renovation Explorer tool, are normally only available to experts. Our goal is to make this tool available to everyone, and you don't need to know anything about how machine learning models work."
Postdocs have been key to explorer success
Both Loonen and Havinga are quick to recognize that the Renovation Explorer project would have been lost without the critical contribution of a team of postdoctoral researchers over the course of the project.
"The postdocs have a great dynamic and it's so rewarding to see them develop and grow in the project. They definitely deserve more credit than I deserve in this project," says Loonen.
Amongst the postdocs who contributed to the project are Günsu Merin Abbas , Vasilis Soulios , Luyi Xu , Alexis Cvetkov-Iliev , and Evangelos Kyrou.
Havinga also recalls a fun exchange with one of the postdocs when they were one year into the project. "When one of the first postdocs started on the project he was a little stunned by the ambition and vision of the project. A year later when we had the first machine learning model running, he told me that he was impressed that the vision for the project had come to fruition and that we, as a team, had made it happen."
The exploration continues
The Dutch government announced in April this year that they will continue to fund project developments. "They want to rollout this out on a larger scale, develop a neighbourhood-level model for municipalities, and give the model the capacity to explore cooling demand, energy storage technologies, and smart control of systems," says Havinga.
With the additional government support, Havinga, Loonen, and the rest of the team will embark on the next steps in the project, and they are planning to start on July 1st.
"By the end of the year or early the next year, the government would like a model that can be implemented on a government website. And there will be TV and media campaigns to encourage people in the Netherlands to visit the website to use our tool. I have to admit this will be such an awesome outcome!"
A call for more global explorers
For significant global impact, Havinga would like the model to go international.
"We first need EU-wide adoption of this tool. I have explored this in the past, but lately I've been too busy with development the national model. My call is for everyone to be part of the efforts to bring this to the EU and further afield."
In the end, Havinga wants the Renovation Explorer tool to help as many people as possible.
"Energy crisis shocks are not desirable, but I hope it helps to raise awareness among households that they can alleviate the negatives of these shocks by properly renovating their homes with help from our tool. In that way, I hope dependencies on fossil fuels decrease, helping people to reach their energy needs in a sustainable way. And that's where the Renovation Explorer tool can have a huge impact," says Havinga.
