Abstract
Rebar corrosion is the most commonly observed deterioration mechanism in reinforced concrete (RC) structures, severely affecting their performance and potentially resulting in premature in-service failures. Corrosion causes internal damage to RC elements, owning to the loss of steel area and the formation of associated expansive corrosion products. Corrosion damage, which may ultimately lead to inadequate performance, includes loss of steel area, loss of bond between steel and concrete, reduced concrete strength due to cracking, loss of concrete section due to spalling and reduced mechanical properties of the affected rebars, i. e. strength and ductility, mostly due to pitting formation. The severity of this damage, however, depends on the nature (i. e. uniform or/and pitting) and extent of corrosion as well as the location of its occurrence, i. e. tension, compression and/or shear reinforcements in RC beams. Accurate performance evaluation of corroding structures could allow extension of service life, where appropriate, and may contribute to a more consistent safety level across a network of structures. This would improve the efficient use of scarce resources, and minimize the impact of indirect costs through optimised inspection, maintenance and repair works. In this study, performance of corroding under-reinforced beams is examined under serviceability (i. e. deflections, cracking patterns) and ultimate (i. e. load capacity, ductility) conditions through detailed parametric and reliability analysis studies. The performance of corrosion-damaged under-reinforced beams is assessed using non-linear finite element analysis (NLFEA). Empirical models from the literature are used to consider the different effects of corrosion damage in the analysis. The NLFE models are used in parametric studies, where a number of corrosion-damage scenarios are examined. For the mid-scale beams studied within this thesis, loss of tensile steel area and concrete damage/section-loss due to corrosion in the compressive region of the beam, are found to be the main causes for loss of load capacity and bending stiffness. The numerical models predicted an increase of beam’s ductility for increasing amounts of corrosion, up to approx. Qcor = 12% loss. This behaviour was found to be due to the beneficial effect of reducing tension rebar area and yield strength on beam’s ductility. At higher amount of corrosion, beam’s ductility started to deteriorate due to a transition in beam’s behaviour at ultimate deflection from concrete crushing towards premature rupture of the corroded tension rebars. Reliability analysis, which is a widely proposed assessment tool of corroding RC structures, is used to examine the influence of uncertainties associated with different variables involved in the deterioration process. A FORM based reliability analysis methodology is developed, which uses NLFEA together with adaptive response surface method (RSM) for the approximation of the performance function. The procedure is used for the analysis of two corrosion damage scenarios on (mid-scale) under-reinforced beams. The response surface of the yield-load capacity limit state is shown to be a stationary surface, since the regression coefficients are insensitive to the amount of corrosion. The reliability index, B, was found to gradually reduce for increasing corrosion damage, Qcor. The sensitivity factors obtained for the load capacity limit state revealed that the relative importance of the random variables remained almost constant as corrosion loss increased. The approximation of the ductility limit state was found to be strongly influenced by the selection of variable ranges due to significant curvature in the response surface. The accuracy of the fitted response surface was improved by splitting the variable space into smaller ranges and evaluating the response surface for each corrosion interval. An initial increase of the reliability index, β, was observed for the two corrosion damage scenarios examined. The sensitivity factors for the ductility limit state revealed that as corrosion damage increases, a transition of the relative importance of random variables concrete compressive strength, fc and ultimate (tension) rebar strain, eu, is likely to occur.