Role of uncertainties on time dependent behavior of prestressed and cable stayed concrete bridges

The behavior of pre-stressed and of cable stayed concrete bridges may be strongly affected by the time dependent response due to the creep and shrinkage of concrete. The assessment of safety and serviceability performance of these structures during their whole service life requires a reliable estimation of the influence of these phenomena. The conventional design of pre-stress losses and, in general, of the time dependent behavior of non homogeneous structures, like cable stayed bridges, uses tools of analysis which involve many uncertain quantities, such as environmental factors, the material characteristics and the intensities of the tension in cables. Also construction imperfections may strongly influence the final structural response. After some brief recalls regarding the time dependent analysis in presence of creep and shrinkage, this paper studies, through a probabilistic approach, the effects due to a large variations of the basic variables. Two cases are studied: the first one is less sensitive to uncertain data, while the second is strongly affected by them. In the case of a cable stayed bridge, made of a concrete deck, subjected to creep effects and suspended to a set of pretensioned cables, the role of uncertainties in the pretensioning forces, in the relative humidity and in the concrete strength do not influence the time dependent behavior sensibly. The standard deviation with respect to the mean tension in the cables is relatively small and, most of all, it progressively reduces. Hence the system seems to be self-stabilizing over time. On the contrary, in the case of a prestressed cantilever beam, the effects of uncertainties in the pretensioning forces and in the concrete strength cause a significant variance of the tip deflections. Both deflections and their variance increase over time and, speaking about bridges, they may strongly modify the vertical attitude of the structure. These effects are emphasized when the prestressing is applied a few days after curing. Such results outline the limits of the traditional deterministic analyses and suggest the need of further studies on this topic.