Data-enabled comparison of six prediction models for concrete shrinkage and creep

Abstract

Modeling the shrinkage and creep of concrete is a demanding task due to the large number and high complexity of the parameters that contribute to these two mechanisms. A range of models have been developed to date to predict shrinkage and creep over time. Among them, this study focused on some of the most widely used models, including those developed by the American Association of State Highway and Transportation Officials, the American Concrete Institute, the European Committee for Standardization, the Fédération Internationale du Béton, and the Comité Européen du Béton. This holistic investigation aimed to provide in-depth insights into the input requirements and prediction capabilities of the identified models. For this purpose, using various data sets selected from the NU-ITI database, the performance of each shrinkage and creep model was first assessed, and a calibration approach was then employed to further refine their outputs. The calibration was performed with the objective of adjusting the short- and long-term prediction accuracy, including the rate of shrinkage and creep development over time. The models were evaluated side by side through comparing the outputs of each calibrated model to data from shrinkage and creep experiments. The calibration steps explored in this study were found to improve the performance of the shrinkage and creep prediction models by up to 20%. This helped reduce the possibility to deviate from the expected strains and stresses. The outcome of this detailed study paved the way to properly select and utilize shrinkage and creep models, taking into consideration the key contributing factors for the highest accuracy.

Keywords

• Concrete
• Shrinkage
• Creep
• Predictive models
• Input and output parameters
• Data-assisted calibration