Application of Artificial Neural Networks to Predict Insulation Properties of Lightweight Concrete

Abstract

Predicting the properties of concrete before its design and application process allows for refining and optimizing its composition. However, the properties of lightweight concrete are much harder to predict than those of normal weight concrete, especially if the forecast concerns the insulating properties of concrete with artificial lightweight aggregate (LWA). It is possible to use porous aggregates and precisely modify the composition of lightweight concrete (LWC) with specific insulating properties. In this case, it is advisable to determine the parameters of the components and perform preliminary laboratory tests, and then use theoretical methods (e.g., artificial neural networks (ANNs) to predict not only the mechanical properties of lightweight concrete, but also its thermal insulation properties. Fifteen types of lightweight concrete, differing in light filler, were tested. Lightweight aggregates with different grain diameters and lightweight aggregate grains with different porosity were used. For the tests, expanded glass was applied as a filler with very good thermal insulation properties and granulated sintered fly ash, characterized by a relatively low density and high crushing strength in the group of LWAs. The aim of the work is to demonstrate the usefulness of an ANN for the determination of the relationship between the selection of the type and quantity of LWA and porosity, density, compressive strength, and thermal conductivity (TC) of the LWC.