Artificial Neural Network with Sensitivity Analysis: Predicting the Flexural Strength of Concrete Pavement using Locally Sourced Dilapidated Concrete as Partial Replacement
Cristy Manahan Malasan, Bernard S. Villaverde, Dario Landa-Silva, Kevin Lawrence M. de Jesus
Abstract
Concrete is a non-recyclable material, and aggregate is one of the significant components. As the construction industry is getting better, the necessity for aggregates also gets wider. Previous study suggests that recycled concrete aggregate (RCA) can be used as a sustainable alternative to regular aggregates in concrete paving mixtures. The focus of this research is to generate a mathematical model to forecast concrete flexure strength using certain percentages of RCA substituted natural aggregates and the age of concrete. To develop the mathematical model of the flexure strength of concrete pavement containing RCA, the researcher used two-input parameters to meet one output parameter. RCA used in this study is dilapidated concrete from the Barangay roads of Nueva Ecija Province, Philippines. Before the concrete mix, the researcher also did the L.A. abrasion to test the abrasiveness of the RCA and investigate the development of flexure strength of the specimen at the age of the 7th, 21st, and 28th days. The results show that the concrete mixture with 50% RCA gained the highest flexure strength among the four design mixtures on its 28th day. Furthermore, the derived model from ANN for predicting flexure strength is compared to the performance of the model using multiple linear regression and on the actual performance of the design mixture.