Realistic Long-Term Prediction of Prestress Forces in PSC Box Girder Bridges

The purpose of the present paper is to propose a method to give a more accurate prediction of time-dependent prestress force changes due to creep and shrinkage of concrete in prestressed concrete (PSC) structures. Updating of long-term prediction of prestress forces is achieved using Bayesian statistical inference. By Bayesian statistics, prior predictions of time-dependent effects due to creep and shrinkage of concrete are used with the information obtained from in-site measurements to develop updated predictions, or posterior predictions. The prior prediction contains the uncertainties with regard to creep and shrinkage of concrete. The present study also deals with the uncertainties in the long-term prediction of creep and shrinkage effects using a sampling method. The proposed theory is applied to long-term prediction of prestress forces of an actual PSC box girder bridge. The present study indicates that the width of mean ± two standard deviation for posterior predictions of prestress forces with nine measurement information is about half that of mean ± two standard deviation for prior predictions of prestress forces. Therefore, the present numerical results prove that a more accurate long-term prediction of prestress force changes in PSC structures due to creep and shrinkage of concrete can be achieved by employing the proposed method.