LONG-TERM DEFORMATION OF HIGH PERFORMANCE PRESTRESSED CONCRETE BRIDGES

Abstract

Prestressed concrete bridges are in increasing demand to cater for the enormous needs of developing transportation networks and infrastructure in general. Concrete undergoes volumetric changes throughout its service life. These changes coupled with environmental factors result in timedependent loss of prestress and additional deformation of post-tensioned concrete girders arising mainly from shrinkage and creep of concrete and relaxation of prestressing steel. The use of HPC in Prestressed bridge girders can allow higher prestressing forces which in turn permit longer practicable span lengths. The combination of longer span lengths and higher prestressing forces may lead to excessive long-term deformation. Hence, from designer's point of view it is not only enough to pay attention on adequate safety factor against failure, but also proper attention should be paid to take care of long-term deformation in HPPC bridges have been addressed in totality. Creep and shrinkage mainly depend on parameters related with the composition of the concrete, the effect of local materials and environmental conditions. The assessment of creep and shrinkage for the purpose of design has been carried out by empirical relationships developed for different regions. Also, the same have been incorporated in their respective codes of practices. Nevertheless, there is no proper Indian model for predicting creep and shrinkage. In this regard in the present work, creep and shrinkage strains for a HPC mix have been measured for a prolonged period of 850 days on two different test specimen sizes. Comparisons of test results with eight different conventional models, which are used extensively in practice, have been carried out. The CEB 90 model has been found to show better match with experimental results of this study. However, at higher ages some deviations have been observed in both creep and shrinkage predictions. For the sake of improvement, the ii experimental data and CEB 90 predictions have been combined to develop an appropriate training set for the ANN. The optimum architecture ofANN trained in this work is 5-11-4-2. Relaxation losses are sensitive to variations in stress levels over time. Eurocode-2 reduced relaxation formula and low relaxation formula by PCI have been used in the present work. Using the developed ANN model, the two relaxation formulae commonly used by researchers and the incremental time-step method, an application tool for the calculation of long-term deformation in HPPC bridges has been developed. For the validation of this tool, experimental long-term deformation results reported by Stallings et al. for HPPC girders have been used. It has been seen that there is a reasonably good match between the experimental and analytical results. To check the significance of long-term behavior in the overall performance, three different case studies have been carried out to check the long-term behavior of the conventionally designed HPPC bridges. Two simply supported HPPC box girder bridges and one three-span continuous HPPC box girder bridge designed as per relevant codes of practices and being built in India, have been considered for the long-term deformation studies. During case studies, it has been found that the low relaxation formula by PCI gives higher long-term losses than that by Eurocode-2 reduced relaxation formula and the difference between the two increases with the increase in span. The ratio of the final percentage loss to the initial percentage loss in prestressing tendon force, for simply supported bridges has been found to be greater than 2 at the mid span section and with the increase in the span the ratio further increases. The studies also shows that for longer-spans more time steps will be appropriate for better prediction of behavior. Also, it is recommended to use short-time steps at the beginning of sequence and a longer time step at the later stages of service. in