Neural Network-Based Estimation of Flexural Performance for Polymer Permeable Concrete

Abstract

Pervious concrete is increasingly used to reduce runoff water and improve water quality near pavements and parking lots, but highway pavement structures cannot use it due to its high porosity and reduced strength. To address the issue of lower flexural strength in permeable concrete, this study designs and conducts 11 different tests with varying mix ratios. The objective is to ensure that the resulting concrete satisfies both permeability and compression resistance requirements. The uniform test method is employed to measure the flexural strength of the concrete after a period of 28 days. This study employs neural networks to analyze the flexural performance of polymer permeable concrete by considering various input factors such as cement consumption, water consumption, STA (4.75 to 9.5 mm stones), STB (9.5 to 16 mm stones), VAE (vinyl acetate-ethylene) polymer content, and SAP polymer content. The objective is to optimize the mix proportion of polymer permeable concrete and identify a suitable ratio that satisfies the requirements of pavement structural flexural performance.