The EHU’s ENEDI group has created a tool to estimate the heating consumption of a neighbourhood, and to decide which retrofitting measures should be implemented to save energy and reduce emissions
One of the research lines of the group of the EHU-University of the Basque Country is seeking solutions to generate districts or neighbourhoods whose annual net energy consumption is close to zero. Using machine learning techniques, Dr Milagros Álvarez has achieved greater accuracy in energy demand estimates and in the assessment of retrofitting strategies to reduce thermal consumption on the neighbourhood level.
Urban buildings account for around 30% of total global energy consumption. In the European Union, over 64% of energy consumption goes on heating. However, according to European Commission data, 75% of European building stock is considered to be energy inefficient. So there is an increasing momentum to extend the concept of nearly zero-energy buildings to entire districts, with the aim of accelerating retrofitting and moving towards reducing greenhouse gas emissions and improving energy efficiency.
In this context, the researcher Milagros Álvarez-Sanz has developed a new model to map the heating demand of buildings on an urban scale: “We’re talking about a simple model to make initial heating demand estimates (which can be broken down by building) and to assess a range of retrofitting strategies that could be implemented to reduce that demand.” Given the urgency of addressing sustainability and the global energy crisis, “improving energy efficiency in buildings is essential, and achieving this on a neighbourhood level is a pressing challenge for cities,” said Dr Alvarez.
The study was carried out by the EHU’s ENEDI group, which works in the field of energy efficiency and savings in buildings. “Planning municipal energy strategies begins by finding out, quickly and accurately, how much energy buildings need. In this regard, more and more research and energy transition policies are moving away from analysing individual buildings towards studying entire neighbourhoods, as this allows for a more comprehensive approach to reducing emissions," added the researcher Jon Terés-Zubiaga, who is working on several projects in the field of energy vulnerability and planning on an urban and provincial level.
This practical tool is fast, accurate and easy to use by professionals working in the fields of energy planning, architecture and urban management, as it allows them to assess different alternatives for addressing this challenge: “Once the current state of a neighbourhood has been analysed, this model allows one to analyse, from an energy and economic point of view, the impact that different solutions could have on some buildings or on the entire neighbourhood, such as passive retrofitting, modifying the energy system by using solar panels or replacing boilers, for example.”
Machine learning and public data
The ENEDI group of researchers used machine learning techniques to determine an “energy quality” indicator of a building or district. “The model enables a base temperature to be determined for each building, depending on its characteristics, and converts it into an indicator of its energy performance; this parameter can be determined by using publicly available variables or data.” Álvarez and Terés stressed the importance of taking advantage of these open data sets, which are being increasingly used in urban energy research and planning, “as they provide valuable information for analysing buildings and neighbourhoods. These are easily accessible data and allow the model to be fed reliably without any need to collect complex information.”
One of the main contributions of this research has been to demonstrate that simplified models can provide sufficiently accurate estimates of large-scale heating demand without the need for complex simulations. By using machine learning techniques, "it was possible to improve the accuracy of energy demand estimates and efficiently assess energy retrofitting strategies in neighbourhoods. This model allows informed and scalable decisions to be made for energy retrofitting, even without detailed data on each building," said Álvarez. What is more, this simplification allows the tool to be integrated as a module into broader models, thus contributing to energy assessments on an urban scale.
To demonstrate the practical usefulness of the model, they reported that they had conducted a district-scale study in Otxarkoaga and Txurdinaga (Bilbao), in which they assessed “some energy retrofitting strategies and analysed energy and economic indicators, which allowed them to prioritise the least efficient buildings for specific interventions”. It should be noted that the model presented in this study "is restricted to the climatic conditions of southern Europe. So, one of the next steps identified is to make this model more general by expanding the range of scenarios assessed", they pointed out.
Additional information
This study is part of Milagros Álvarez-Sanz's PhD thesis, written up in the ENEDI group under the supervision of Dr Jon Terés-Zubiaga and Dr Álvaro Campos-Celador, both of the EHU. The thesis focused on developing simplified models to estimate the heating demand of buildings and neighbourhoods, using geographic information systems, public data, and key design and operational parameters. Cristina Villanueva and Pello Larrinaga, researcher and lecturer in the Department of Energy Engineering, respectively, also participated in the research.
Jon Terés-Zubiaga is a tenured lecturer in the Department of Energy Engineering of the Faculty of Engineering - Bilbao.
Bibliographic reference
Milagros Álvarez-Sanz, Cristina Villanueva-Díaz, Álvaro Campos-Celador, Jon Terés-Zubiaga, Pello Larrinaga
Combining heating degree days method and machine learning techniques: implementation of a new model for mapping buildings heating demand at urban scale
Energy Conversion and Management: X
DOI: 10.1016/j.ecmx.2025.101503