Effective thermal conductivity model of straw bales based on microstructure and hygrothermal characterization.

• The interconnections distribution in fibers determines the heat transfer model. • The development of a thermal conductivity equation considers the radiative heat transfer. • The bale's thermal conductivity varies according to the fiber orientation. • Thermal conductivity depends on the chemical composition of the straw. Straw fibers are natural fibers that have a strong insulating property and a minimal environmental effect, making them suitable construction materials. In the construction sector, to predict the building energy consumption, the thermal conductivity of the materials composing the envelopes must be precisely known, which is not the case for the straw material. The properties of a straw bale are variable and mainly depend on its density, fibers orientation, chemical composition, temperature and relative humidity. Consequently, this study consists of determining a mathematical equation to predict the effective thermal conductivity of straw bales that can be applied to all straw types, in different circumstances. For this purpose, the size, distribution, and morphology of straw fibers are first determined from microscopic images. Second, a mathematical model for estimating the thermal conductivity is suggested using the heat transfer models of porous and fibrous materials. The numerical model is validated by an experimental study that measures the thermal conductivity of straw bales by altering their density between 80 and 120 kg/m3, their relative humidity between 15 % and 95 %, and their temperature between 15 ˚C and 55 ˚C. The experimental results show that the thermal conductivity increases from a minimum of 0.047 W/(m∙K) to a maximum of 0.09 W/(m∙K) when increasing the three altered factors. For a straw fiber having 40% cellulose content, the thermal conductivity increased from a minimum of 0.05 W/(m∙K) to a maximum of 0.0832 W/(m∙K). This behavior is found the same for the numerical findings. The comparison of numerical and experimental values shows a good agreement, with a root mean square error of 0.005 W/(m∙K) and a scatter index of 5.5 %. [ABSTRACT FROM AUTHOR]