Reliability of the laboratory cob specimens manufacturing: Variability of hygrothermal properties at the wall scale

This article investigates how cob specimens manufactured in a laboratory are representative of a real-scale cob wall. Emphasis is placed on hygrothermal properties, which are a significant asset for reducing energy costs of buildings. It is a continuation of a previous paper in which the natural variability of the manufactured cob samples was evaluated at the material scale (laboratory samples) (Tchiotsop et al., 2022). Indeed, numerous works propose various laboratory fabrication protocol parameters and approaches for the cob specimens. Hence, a 3.2 m${}^2$ cob wall with 3% hemp shives content (by mass) was built, dried, and cored following a regular meshing plan. The variability of the wall specimens’ hygrothermal properties (density, thermal conductivity, volumetric heat capacity, moisture adsorption, moisture permeability, and intrinsic permeability) was measured and computed using the same protocols as in the previous article. As main results (details on the graphical abstract), the mean statistics of the dry bulk density and thermal properties showed good agreement between material and wall scales, while the variabilities were not always well represented at the material scale. For moisture transfer and storage properties, a significant gap was found in the statistics (up to 216%). Probability density functions (pdfs) of properties were assessed for both scales, and the same functions were found to be reliable. This study reveals the complexity at the real scale of cob buildings, making it difficult to obtain a representative specimen for characterization purposes.