A review on physical and data-driven modeling of buildings hygrothermal behavior: Models, approaches and simulation tools

The hygrothermal simulation of hygroscopic building materials is a real challenge, in terms of regulations and labelling, but also in decision-making. Today, we lack reference models for the hygrothermal behavior of a whole building. A scoping literature review is conducted to provide an overview of current state-of-the-art methods in order to address of these simulation methods. The most comprehensive studies are selected and examined in detail. These include physical models (White-box), which focus on solving equations that simulate the hygrothermal behavior of buildings, and data driven models, which involve implementing a prediction model using machine learning techniques (Black-box). On one hand, the white-box models are reviewed according to a two-category classification: CFD and Nodal approaches. On the other hand, the principal model used for black-box models is neural network models. The study highlights the need for a recognized method for hygrothermal simulations of hygroscopic envelopes. It provides a better understanding of the hygrothermal simulation, which helps to choose the most suitable tool or model. In addition, this review points out that there is limited application of data-driven methods to simulate the hygrothermal behavior of hygroscopic envelopes. This analysis study highlights future research gaps to overcome in order to stimulate data-driven building performance design.