Predicting post-fire behavior of green geopolymer mortar containing recycled concrete aggregate via GEP approach

Abstract

In the present study, 20 models were developed using gene expression programming (GEP) to predict the compressive strength and mass loss of geopolymer mortar (GPM) containing recycled concrete aggregate (RCA) exposed to elevated temperatures. To do so, the results of 160 specimens manufactured out of 32 different mixture designs in an experimental effort were used. In developing the models, 80% of the total datasets were employed in the training phase, with the remaining 20% used in the validation phase. Three input variables were taken into account, namely the applied temperature (T), recycled concrete aggregate (RCA) replacement level, and superplasticizer (SP) addition percentage. The training and validation phases with the coefficient of determination of 0.95 to 0.99 demonstrated that there was proper consistency between results predicted by the proposed models and the experimental results. Moreover, the results of statistical analyses gave another reason for the ability of GEP to predict both the compressive strength and mass loss of GPM containing recycled concrete aggregate under elevated temperatures.