Sensitivity Analysis and Strength Prediction of Fly Ash — Based Geopolymer Concrete with Polyethylene Terephtalate using Artificial Neural Network
Dario Landa-Silva, Kevin Lawrence M. de Jesus, Bernard S. Villaverde, Crisialine Joy C. Cahilig, Jan Piola L. Dela Cruz, Jessa Maye S. Sario
Abstract
Sustainability is considered as one of the most essential developmental models which is integrated in almost all industries and fields of specialization worldwide. In civil engineering, the use of resources that were used as substitution to the traditional materials could address the concerns of scarcity of construction material resources. This study performed a partial substitution of waste polyethylene terephthalate (PET) bottles to sand and utilization of fly ash as partial to full substitution to Ordinary Portland Cement. The effects of these parameters were tested with respect to its compressive and flexural strength, 30 samples in each strength parameter, respectively. An Artificial Neural Network (ANN) model was developed by using these data sets to predict the compressive and flexural strength of concrete and a sensitivity analysis using Connection Weights Algorithm was performed to determine the parameter with the most significant importance to the strength of concrete. Prediction model results presents a very satisfactory performance based on its difference to the actual values. Moreover, the sensitivity analysis through Connection Weights algorithm demonstrated to be an efficient instrument to assess the parameter contribution of the target output of the model.