Hybrid approach to representative building archetypes development for urban models-A case study in Andorra

Building archetypes development has been the focus of numerous research works over the past decades and are considered one of the biggest challenges, as well as one of the main sources of inaccuracies in Urban Buildings Energy Models (UBEM). The development of machine learning and data mining techniques, such as clustering, as well as the access to building data at disaggregated scales, has opened new horizons in this field. With the aim to reduce additional simulation errors derived from the fragmentation process of the archetype approach in UBEMs, this paper presents an alternative hybrid approach to identify representative building archetypes combining the classic deterministic and a data-driven clustering approach using building data at both building and cadastral unit scales. The methodology was tested in the Escaldes-Engordany building stock, a city of the Principality of Andorra, and the resulting archetypes have been compared with the archetypes obtained by the application of both approaches applied separately. A total of 71 archetypes were identified in the Escaldes-Engordany building stock. The results show that both approaches complement each other and allow to overcome the identified barriers when applied separately. In addition, the results also reveal that there is an important heterogeneity of certain building aspects not only between buildings, but also within buildings, which cannot be detected if the fragmentation is still carried out at the building scale.

Automated summary

The development of building archetypes has been a major focus in research for the past decades, and it is considered a challenge and a source of inaccuracies in Urban Buildings Energy Models (UBEM). This paper presents a hybrid approach to identify representative building archetypes, combining deterministic and data-driven clustering methods, and the results show that both approaches complement each other.