Accuracy Evaluation of Prestressed Concrete Girder Camber in Missouri Bridges

The objective of this study is to evaluate the accuracy of prestressed girder camber calculations compared with field-measured camber from two precast plants. Field data on 189 Missouri girders with initial camber measurements and 33 girders with later camber measurements were compared with girder camber calculations using the original Missouri DOT method. The measured initial camber was on average 23% greater than the calculated camber, with an RMSE of 20.6 mm (0.81 in.) and an average error of 35%. The camber at a later point in time was only 11% greater than the calculated camber, with a RMSE of 19.6 mm (0.77 in.) and an average error of 19%. A major source of the difference in the measured and calculated camber was the measurement error introduced by sag of the measurement string line used at the precast plant. Other errors related to the estimation procedure were, in order of importance, support conditions, daily temperature gradients, temporary temperature during curing, concrete strength/age, concrete modulus, and creep parameters. All have a significant impact on camber calculations, with changes in camber from 23% to 4%. Other factors, such as prestressing force, section properties, concrete density, strand eccentricity, shrinkage, humidity, and long-term time-dependent analysis method, were systematically investigated and found to have minor impacts (less than 5% change in camber). Incorporating modifications to account for the measurement line sag, overhang length, concrete strength, and curing temperature reduced the underprediction of camber to 1.3% with a RMSE of 14.9 mm (0.59 in.) and an average error of 21%.