Compressive strength of tungsten mine waste- and metakaolin-based geopolymers

Neuro-fuzzy approach has been successfully applied to a wide range of civil engineering problems so far. However, this is limited for geopolymeric specimens. In the present study, compressive strength of different types of geopolymers has been modeled by adaptive neuro-fuzzy interfacial systems (ANFIS). The model was constructed by 395 experimental data collected from the literature and divided into 80% and 20% for training and testing phases, respectively. Curing time, Ca(OH)2 content, NaOH concentration, mold type, aluminosilicate source and H2O/Na2O molar ratio were independent input parameters in the proposed model. Absolute fraction of variance, absolute percentage error and root mean square error of 0.94, 11.52 and 14.48, respectively in training phase and 0.92, 15.89 and 23.69, respectively in testing phase of the model were achieved showing the relatively high accuracy of the proposed ANFIS model. By the obtained results, a comparative study was performed to show the interaction of some selected factors on the compressive strength of the considered geopolymers. The discussions findings were in accordance to the experimental studies and those results presented in the literature.