Safety Evaluation of Concrete Structures with Nonlinear Analysis

Doctoral thesis, 2011

Concrete is the most widely used construction material in the world. To obtain effective new constructions and to use existing concrete structures in an optimal way, accurate structural models are needed. This requires that good approximations of important model parameters be available and that the nonlinear material response of concrete can be accounted for. However, uncertain model parameters can significantly influence the structural response modelled which leads to high modelling uncertainty. To estimate uncertain parameters a methodology is proposed and applied to the new Svinesund Bridge to improve the initial finite element model through finite element model updating using on-site measurements. To account for the nonlinear material response, it is also necessary to have a safety format suited to nonlinear analysis. However, the available safety formats for nonlinear analysis have been questioned and the need to quantify the modelling uncertainty of nonlinear analysis has been highlighted, Carlsson et al. (2008). Therefore, the modelling uncertainty of nonlinear analysis was quantified based on available data. It was found that the uncertainty varies significantly depending on the failure mode obtained and that this uncertainty was often the factor that governed the safety evaluation. Based on this observation, a new safety format is proposed which allows the modelling uncertainty be explicitly accounted for. To facilitate realistic modelling the mean in situ material parameters are used in the nonlinear analysis; the reliability is assured by a, so called, resistance safety factor. Apart from the modelling uncertainty, the resistance safety factor depends on the material and geometrical uncertainty. It was found that the material variability can be estimated by using a sensitivity study, which involves two to three additional nonlinear analyses with reduced material strengths. Applying the safety format to short columns loaded by a normal force and to beam sections loaded in bending, shear, and the combination of bending and shear, led to a reliability level that was in good agreement with the target reliability. Other safety formats for nonlinear analysis, according to EN 1992-2, CEN (2005), and Model Code 2010, fib (2010a), fib (2010b), were found to underestimate the modelling uncertainty of difficult-to-model failure modes, leading to a reliability level below the target reliability. To study the consequences of assuring the safety on the structural level by an inequality of forces, as proposed in Model Code 2010, four safety formats were applied to a concrete portal frame bridge. It was shown that an inequality of forces on the structural level does not necessarily lead to the intended reliability level, unless the deformation capacity used is reliably available.