Abstract:
Geopolymer is a promising alternative binder to Portland cement. However, the importance of mix design parameters affecting the mechanical properties of geopolymer has yet to be quantitatively assessed. This work evaluates the significance of the four common mix design parameters, namely Si/Al (molar ratio), water/solids (mass ratio), Al/Na (molar ratio) and H2O/Na2O (molar ratio), in determining compressive strength of metakaolin-based geopolymers through experiments and statistical analyses. In addition, machine learning-based classifiers were engaged for strength predictions. Results showed that Si/Al ratio is the most significant parameter followed by Al/Na ratio. Unlike ordinary Portland cement system, water/solids ratio is not the chief factor governing strength of metakaolin-based geopolymers. Machine learning-based classifiers were able to predict the compressive strength with high precision. The strength predictions can potentially guide preliminary mix proportioning of metakaolin-based geopolymers to achieve required strength grade without going through tedious (trial and error) mix formulation.