Prediction of Properties of Concrete Cured Under Hot Weather Using Multivariate Regression and ANN Models

Abstract

Concreting in hot weather poses special problems, and the properties of concrete are significantly influenced by mix parameters, construction practices and environmental conditions. In this study, mix design parameters, casting temperature and curing conditions were varied to study the mechanical properties and durability characteristics with the ultimate aim of developing models relating the concrete properties and the experimental parameters. The effect of w/c ratio (0.3 to 0.45), in situ concrete temperature (25 to 45 °C) and curing method (ponding, covering with wet burlap or applying curing compound followed by curing under laboratory or field conditions) was assessed by measuring compressive strength (f′c), split tensile strength (ft), pulse velocity (PV) and depth of water penetration (WP) up to 180 days. Prediction models were developed utilizing quadratic regression models and artificial neural networks (ANNs). The regression models indicated that moist curing, in situ concrete temperature and curing period have a positive impact on the strength parameters and PV, while the former two variables had a negative effect on the WP. The w/c ratio had a significant positive impact on the WP. The concrete properties predicted by ANN models were more accurate than those predicted by regression models. A combined ANN model, developed for predicting strength and PV simultaneously, had a better correlation coefficient compared to ANN models developed for each variable separately. The developed models are expected to assist concrete technologists in the selection of appropriate concrete mixture parameters to obtain concrete with desired properties under hot weather conditions.