Optimization of foam concrete characteristics using response surface methodology and artificial neural networks

Abstract

In this study, influences of waste marble powder (WMP) and rice husk ash (RHA) partially replaced instead of fine aggregate and cement into foam concrete (FC) on compressive and flexural strength, porosity, and thermal conductivity coefficient were investigated using Response Surface Methodology (RSM) and Artificial Neural Networks (ANN) methods. The foam parameter was determined as two levels in the experimental design, and the WMP and RHA parameters were determined as three levels. With the RSM analysis, the most influential parameters for compressive and flexural strength were determined as Foam WMP and RHA, respectively. Likewise, the order of effective parameters for porosity and thermal conductivity coefficient was found as foam WMP and RHA. With the RSM method, R2 values were obtained as 0.9492 for compressive strength, 0.9312 for flexural strength, 0.9609 for porosity, and 0.9778 for thermal conductivity coefficient. Correlation coefficients with the ANN method were found as 0.98393, 0.96748, 0.9933, and 0.96946 for compressive and flexural strength, porosity, and thermal conductivity coefficient, respectively. The ANN method was found to be suitable for estimating the responses. The RSM method was found to be suitable both for estimating the responses and for determining the effective parameters. In addition, the optimum parameters were determined by the RSM method.