PREDICTION OF EFFECTIVE THERMAL CONDUCTIVITY OF MOISTENED INSULATION MATERIALS BY NEURAL NETWORK

In harsh climates, utilizing thermal insulation in the building envelope can substantially reduce the building thermal load and consequently its energy consumption. The performance of the thermal insulation material is mainly determined by its effective thermal conductivity, which is dependent on the material’s density, porosity, moisture content, and mean temperature difference. The effective thermal conductivity of insulation materials increases with increasing temperature and moisture content. Hence, thermal losses may become higher than the design values. The availability of measured data of the thermal conductivity of insulations at higher temperatures and at elevated moisture contents is poor. In this article the Artificial Neural Networks (ANN) is utilized in order to predict the effective thermal conductivity of expanded polystyrene with specific temperature and moisture content. The experimental data was used for training and testing ANN. Obtained results from the ANN method give a good agreement with experimental data.