STUDY ON PREDICTION METHOD FOR INSULATION OF DOUBLE-LAYER STRUCTURE USING “INCLUSIVE THERMAL CONDUCTIVITY” IN FIRE-RESISTANCE TEST

In this study, a method for predicting the unexposed surface temperature of a double-layer wall is proposed based on "inclusive thermal conductivity," which considers heat and moisture transfer and influence of cracks in materials containing water. The proposed method will enable the estimation of thermal conductivities of materials that are difficult to assess using conventional technologies. Fire-resistance tests were conducted on specimens with fiber-reinforced cement siding and gypsum boards overlaid. Fire-resistance tests were conducted based on ISO834. Inclusive thermal conductivities are estimated based on the results of temperature measurements from actual fire-resistance tests. The inclusive thermal conductivities are calculated by using the finite differential method. On the estimation of the inclusive thermal conductivity, the calculated internal temperatures of the materials, temperature between materials, and unexposed surface temperature were compared with the experimental results. Additionally, on estimation of inclusive thermal conductivities the influence of the moisture behavior and cracks in the materials are considered. Since the inclusive thermal conductivities have some peaks and/or dips at various temperatures, they are approximated to polynomial approximation for each temperature region. In order to estimate that unexposed surface temperature of similar double-layer specimen whose thickness of gypsum board layer is decreasing, the number of inclusive thermal conductivities are reduced. In the case of estimating the specimen thickness whose thickness is changed, the inclusive thermal conductivities of predictive calculations were performed by removing layers of the same thickness as the divided layer. Calculations using this inclusive thermal conductivity sufficiently captured the temperature history. The good accuracy on unexposed surface temperature prediction result is obtained as the difference the thickness of estimated test specimen and predicted specimen is smaller. The prediction accuracy of numerical analysis using the inclusive thermal conductivity was found to be practically sufficient.