Abstract
Increasing population and urbanization call for smarter cities where the cycles of matter and energy are optimized, notably in buildings which are actually a source of pollution consuming a lot of energy. The efficiency of building energy has been improved by modelling earth-air heat exchangers, yet selecting the suitable models is challenging. Here we review data-driven earth-air heat exchanger models used for buildings. We discuss issues brought about by assumptions, unmeasured disruptions, and uncertainties in numerical and experimental works. We found that high accuracy can be reached if sufficient data is available. Models are appropriate for real-time activity due to their structure simplicity, yet they display a poor generalization capacity. Model development is limited by the constrained parameters and the complex boundary conditions of the heat exchangers.