Hygrothermal and Economic Analysis of an Earth-Based Building Using In Situ Investigations and Artificial Neural Network Modeling for Normandy’s Climate Conditions

Abstract

This paper investigates the in situ hygrothermal behavior of a cob prototype building equipped with multiple sensors for measuring temperature, relative humidity inside the building, and water content within its walls. The experimental results show that the earth-based prototype building presents interesting thermal insulation performance. Without any heating system, the indoor temperature was found to remain stable, near 20 °C, despite large fluctuations in the outdoor temperature. This study also illustrated the ability of cob to absorb and regulate indoor relative humidity. The use of a neural network model for predicting the hygrothermal behavior of the cob prototype building was an additional objective of this work. This latter was centered on investigating the indoor ambience and moisture content within the walls. In this sense, a long short-term memory model (LSTM) was developed and trained. The validation results revealed an excellent agreement between the model predictions and experimental data, with R2 values of 0.994 for the indoor air temperature, 0.960 for the relative humidity, and 0.973, 0.925, and 0.938 for the moisture content at three different depths in the building’s walls. These results indicate that the LSTM model is a promising approach for predicting the indoor ambience of an earth-based building, with potential applications in building automation and energy management. Finally, an economic discussion of the CobBauge system is presented.