Abstract:
Although the majority of shrinkage and creep models are relative recent and comprehensive, there is a lack of consensus in their utilisation due to substantial scatter in their predictions, even when comparisons are made under relatively well controlled conditions. On one hand, creep and shrinkage are complex phenomena that depend on several factors such as concrete composition and mechanical properties, shape and geometric parameters, curing and environmental conditions, etc. On the other hand, models are typically assessed on a deterministic basis without incorporating information related to input variability.
In addition, the scarcity of long-term measurements has been a major impediment in validating creep and shrinkage models over substantial periods of typical design lives, i.e. over decades rather than years. It is worth noting that although most large prestressed concrete bridges are designed for a lifetime of at least 100 years, only a small fraction of the publicly available data sets cover more than a few years. This type of information is even rarer for measurements from in-situ specimens exposed to realistic environmental conditions.
In this paper, a selection of creep and shrinkage models is assessed by considering measurements from concrete specimens located on actual bridges and, in view of the above remarks, a probabilistic approach is implemented. The suitability of the selected models in producing deformation-time profiles based on in-situ measurements and the influence of case-specific input variability is investigated via sensitivity analysis and Monte Carlo simulation. The work is supported by two well-documented testbeds offering extensive field data - the Lezíria Bridge and São João Bridge, for which a set of creep and shrinkage measurements at specimen level are available with a comprehensive characterization of the employed concrete and good understanding of the prevailing environment.
