Abstract
The creep and shrinkage of concrete play important roles for many nuclear power plant (NPP) and engineering structures. This paper benchmarks the standard and micromechanical models using a revamped and appended Northwestern University database of laboratory creep and shrinkage data with 4663 data sets. The benchmarking takes into account relevant concretes and conditions for NPPs using 781 plausible data sets and 1417 problematic data sets, which cover together 47% of the experimental data sets in the database. The B3, B4, and EC2 models were compared using the coefficient of variation of error (CoV) adjusted for the same significance for short-term and long-term measurements. The B4 model shows the lowest variations for autogenous shrinkage and basic and total creep, while the EC2 model performs slightly better for drying and total shrinkage. In addition, confidence levels at 5, 10, 90, and 95% are quantified in every decade. Two micromechanical models, Vi(CA)2T and SCK CEN, use continuum micromechanics for the mean field homogenization and thermodynamics of the water-pore structure interaction. Validations are carried out for the 28-day Young's modulus of concrete, basic creep compliance, and drying shrinkage of paste and concrete. The Vi(CA)2T model is the second best model for the 28-day Young's modulus and the basic creep problematic data sets. The SCK CEN micromechanical model provides good prediction for drying shrinkage.