Prediction Thermal Conductivity of The Novel Developed Light Weight Concrete with Artificial Neural Networks

Abstract

The dependence on energy is increasing in a growing population and a rapidly developing global world. Around 40% of the energy consumed is consumed in buildings. Building heating and cooling have boosted energy consumption and costs dramatically. As a consequence, in order to boost energy efficiency in buildings, it becomes inevitable to develop new construction materials with thermal insulation properties. Vermiculite, waste basalt powder, molten tragacanth, and cement-reinforced samples were produced for this purpose. Mechanical and thermal conductivity tests were performed on 48 samples produced at various rates. The findings of the experimentally measured thermal conductivity were modelled and compared with the outputs of the created artificial neural network. The Matlab software was used for modelling. The mechanical properties acquired experimentally using the Artificial Neural Networks (ANN) approach were used as an input, and the correlation of the samples with thermal conductivity was investigated. The findings obtained were consistent with one another, and the thermal conductivity values were predicted with an error ranging between 7.6701% and 0.0091%, and the ANN yielded successful results at a rate of 99%.