Time-Dependent Deformations of Concrete for Precast/Prestressed Bridge Components

Abstract

Self-consolidating concrete (SCC) is a material that has properties capable of improving the quality and durability of structures. However, much is still unknown about the hardened properties of this material, including the creep and shrinkage behavior. This thesis presents research aimed at improving knowledge in this area by investigating the creep performance of SCC mixtures and corresponding conventionally-vibrated concrete (CVC) mixtures used for the construction of precast/prestressed concrete bridge girders in Alabama. All mixtures contained Type III portland cement and included several chemical admixtures as well as slag cement. While the girder concrete was placed, representative cylindrical test specimens were also cast. Each of these mixtures was cured using two forms of accelerated curing, which are representative of typical curing conditions used in prestressed concrete, and two loading ages. Match-cured specimens were loaded at prestress release and tarp-cured specimens were loaded at 26 hours. Time-dependent deformations of the concretes were measured using standard creep testing procedures. The accuracy of the following nine creep and shrinkage models was also investigated as part of this research: ACI 209, AASHTO 2004, AASHTO 2010, NCHRP 628, MC 90, MC 90-99, MC 90-KAV, MC 2010, and Eurocode. The test results were compared against the prediction models investigated and overall, the CVC performed no better than the SCC used in this project. For the prediction of load-induced strain, AASHTO 2010, MC 2010, and ACI 209 all provided acceptable predictions for both SCC and CVC produced in accordance with ALDOT specifications for precast/prestressed concrete girders. For the prediction of shrinkage strain, none of the methods investigated provided accurate predictions for SCC or CVC. Of the most current methods, AASHTO 2010 was the most accurate method for estimating shrinkage strains for the concretes tested in this project. When older methods were considered, the original MC 90 method was more accurate than AASHTO 2010.