Using artificial neural networks for estimating the compressive strength of andesite-substituted cement-based composites

Abstract

Abstract In this study, the effect of waste andesite dust (WAD) substitution used in the production of cement-based composites on the strength values of the composite material was investigated using artificial neural networks (ANN). In the production of cement-based mixtures, WAD was substituted with cement at six different ratios as 5%, 10%, 15%, 20%, 25% and 30% by weight. In addition, the effect of curing times on compressive strength was investigated at two different curing times as 28-days and 90-days curing times. While the cement and WAD replacement rates constituted the main input data for the ANN, the 28 and 90-days compressive strength values constituted the output data. When the data obtained from the compressive strength estimation conducted by ANN and the experimental data obtained under laboratory conditions were compared, it was determined that the compatibility between them was good with a 99% coefficient of determination. In addition, the 5% substitution rate used in the mixtures provided higher strength values among other substitution rates.