Resumen
Different models reported on specialized literature have been proposed for cuntifying concrete drying shrinkage at the design stage of concrete structures. However, questions have been raised regarding their applicability to conditions different to those used in their derivation, particularly with respect to differences in cement and aggregate characteristics from country to country. Therefore, this paper evaluates and highlights the need for updating proposed concrete shrinkage prediction models to local conditions. The experimental program was carried out on 8 concrete mixtures of similar strength, manufactured with Portland and Blended Portland Cement which represent the most commonly used local concretes. Tests were conducted on 72 specimens and drying shrinkage strains were measured up to 448 days of drying to investigate the effect of cement type, slump, aggregate maximum size and specimen dimension on concrete shrinkage. It was concluded that the type of cement had a significant effect on drying shrinkage timedevelopment and magnitude. Shrinkage exhibited by concrete mixtures made with Portland Pozzolan cement was approximately 25 % higher than those of similar strength mixtures made with Portland cement. Furthermore, shrinkage of Portland Pozzolan cement concretes increased initially at a much faster rate and then slowed down earlier than Portland cement concretes probably due to the increased in impermeability of Portland Pozzolan cement concretes. Test results were compared with strain predictions from ACI, CEB, B3, GZ, GL and Sakata models. It was concluded that all analyzed shrinkage models highly underestimated shrinkage strains observed in this study, particularly for Portland Pozzolan cement concrete and do not adequately represent the observed shrinkage behavior of concretes made with locally available materials. Therefore, prediction models were calibrated for their use at the design stage and correction factors are suggested for Chilean conditions and for concretes with characteristic compressive cube strengths in the range of 30 to 40 MPa. For model calibration to concrete of a specific project, a minimum testing time of 100 days is recommended to estimate shrinkage within normally accepted accuracy limits.