A DOE approach to optimize the strength properties of concrete incorporated with different ratios of PVA fibre and nano-Fe2O3

The purpose of this research is to optimize the fresh and strength properties of concrete with various incorporation ratios of polyvinyl alcohol (PVA) fibre and nano-Fe2O3 by design of experiment approach. Central composite design of response surface methodology was used to acquire the optimal composition of process variables, and the influence of process variable on the fresh and strength properties of concrete was also verified. A total of five responses (Y Slump, f ck,7, f ck,28, f t,7 and f t,28) were considered in this study, and the factors and the level of each response was 2 and 3, respectively. Regression models for Y Slump, f ck,7, f ck,28, f t,7 and f t,28 were tested; the statistical significance of each process variable was assessed through analysis of variance and Pareto chart, and the obtained models were expressed in the form of second-order polynomial equation. The test results revealed that incorporation of PVA and high-reactive nano-Fe2O3 improved the strength of the concrete; however, the higher level incorporation exhibited unpronounced effect on the strength enhancement. The incorporation of nano-Fe2O3 enhanced the compressive strength to the maximum of 7.44%. The high-reactive nano-Fe2O3 accelerated the hydration process of concrete and enhanced the particle packing density of the concrete by filling the micropores. However, the higher incorporation of nano-Fe2O3 leached out the surplus silica from the concrete and reduced the strength enhancement of the cement composite matrix. The regression analysis results revealed that the most significant and contribution factor for f ck,7 and f ck,28 is the linear term nano-Fe2O3 (X 2), 99.81% contribution, and for f t,7 and f t,28 is the linear term PVA (X 1), 75.75% contribution. Pareto chart and surface plot analysis also revealed the same as above. The optimal proportion of design variables against f ck,7, f ck,28, f t,7 and f t,28 were obtained and the percentage of error of the confirmation tests for Y Slump, f ck,7, f ck,28, f t,7 and f t,28 were less than 3%.