Improving Predictions for Camber in Precast Prestressed Concrete Bridge Girders

John F. Stanton

Professor, Washington State Transportation Center (TRAC), University of Washington, Seattle, Washington

Marc O. Eberhard

Professor, Washington State Transportation Center (TRAC), University of Washington, Seattle, Washington

Michael A. Rosa

Bridge Engineer, Washington State Transportation Center (TRAC), University of Washington, Seattle, Washington

This research was conducted to develop improved methods of predicting camber in prestressed concrete girders. A computer program was written to calculate camber as a function of time. It takes into account instantaneous and time-dependent behavior of the concrete and steel and performs the calculations in a series of time steps. It was calibrated by comparing its predictions with the camber from 146 girders, measured in the fabricator’s yard both after release and at a later time. The program’s long-term predictions were then compared with the responses of 91 girders that were monitored during construction at the Keys Road Bridge site. The measured deflections due to temporary strand release and deck casting were compared to calculated values by using variations in pier continuity. Long-term creep deflections were also monitored after deck placement.

The results showed that the response was sensitive to the predicted prestress losses and that the 2006 AASHTO values for prestress loss provided much better estimates than did the 2004 provisions. In addition, the camber was found to depend on the elastic modulus of the concrete, its creep coefficient, and the use of the prestress losses in the calculation of creep camber. Predicted cambers were compared to the measured cambers to calculate a predicted error. To achieve the best match with the measured cambers, the AASHTO-recommended values for the elastic modulus and the creep coefficient had to be multiplied by adjustment factors. The adjustment factor for the elastic modulus was found by minimizing the predicted error on the camber immediately after release, resulting in a factor of 1.15. The adjustment factor for the creep coefficient was found by minimizing the predicted error on the second camber measurement, resulting in an adjustment factor of 1.4. The prestress losses had to be taken into account when computing the creep component of camber.