An experimental, statistical, and machine learning approach to evaluate the fresh and mechanical properties of graphene oxide concrete

Graphene oxide (GO) has demonstrated efficacy in enhancing the performance of cementitious composites by refining the pore structure, preventing crack propagation, and serving as a nucleation site for hydration. Nonetheless, systematic investigations into its microstructural influence on both destructive and non-destructive testing, supported by predictive modelling, are to be explored in different contexts. Addressing these gaps, this study investigates the effect of GO on the fresh and mechanical properties of concrete through experimental and microstructural analyses, statistical methods, and machine learning techniques. GO was added to concrete at dosages of 0.02%, 0.04%, 0.06%, and 0.08% by weight of cement. The mix with 0.06% GO showed the highest compressive and flexural strength, with an increase of 49% and 25% compared to the control mix at 28 days, respectively. Microstructural analysis revealed tortuous crack paths and the formation of needle and flower-like crystals in the GO concrete, indicating enhanced hydration. Regression analysis, multivariate analysis (MVA) and Artificial neural network (ANN) models were developed to predict compressive strength and slump. . The ANN model demonstrates superior performance over the MVA, which optimises GO usage and reduces the experimental trials and associated costs. Leveraging these models enables engineers to achieve targeted performance outcomes with greater confidence.