Stochastic model of constitutive relationship for corroded steel bars

The existing database for mechanical properties of corroded steel bars was extended by further tensile tests and collection of correlated data. Three-dimensional (3D) geometric models of corroded steel bars were established by using a 3D laser scanner, and the area of any cross sections could be obtained. The ratio of the average to the minimum cross-sectional areas of corroded steel bars, R, was adopted to quantify the longitudinal variation of the cross-sectional area. It was found that R could be characterized by the Gumbel distribution. Both the location parameter and the scale parameter for the probability distribution of R increase linearly with increase in the corrosion degree. Tensile tests of corroded steel bars showed that the nominal yield strength, the nominal ultimate strength and the ultimate strain of the steel bar decreased, and the yield plateau became shorter and even disappeared with the development of reinforcement corrosion. Based on tests results, a deterministic model of stress-strain relationship is introduced for corroded steel bars with different corrosion degrees. It is found by statistic analysis that mechanical properties of steel bars with various corrosion degrees keep normally distributed, while the coefficient of variation rises with the increase of corrosion degree. The probability distribution models of mechanical properties for uncorroded steel bars and the stochastic process models for the deterioration of mechanical properties due to corrosion are established. Consequently, the deterministic model of stress-strain relationship for corroded steel bars is developed to a stochastic one, which can be used to model time-dependent resistance and predict remaining service life of corrosion affected reinforced concrete structures more truly. ©, 2014, Tongji University. All right reserved.