Project Details

In recent years, NCDOT has experienced increasing construction problems related to discrepancies between the predicted and measured camber for prestressed concrete bridge girders, as well as problems with differential camber between identical girders. In addition, current prestress loss predictions used by NCDOT are based on the 2004 AASHTO LRFD Bridge Design Specification, which has been superseded by the 2010 edition. ​

This report examines the accuracy of the current NCDOT method for predicting the prestress losses and camber for prestressed concrete girders as compared to field measurements. Other methods available in the literature are also reviewed, including the PCI method and the AASTHO 2010 method. ​

The report presents the findings from the testing of a large number of concrete cylinders that was conducted to evaluate the properties of the concrete. It also presents the findings of several site visits to precasting plants that were conducted by the research team to identify factors related to girder production that could potentially affect the accuracy of the camber predictions. Specific findings related to the concrete properties and other production factors include the following:The concrete compressive strength at transfer was found to be on average 25% higher than the specified design value.The concrete compressive strength at 28 days was found to be on average 45% higher than the specified design value.The elastic modulus of the concrete was found to be on average 15% less than the value predicted by the AASHTO specifications using a unit weight of 150 pcf for the concrete.Concrete properties can potentially vary from girder to girder within the same casting bed due to the use of multiple batches of concrete along the bed as well as delays in concrete batching that occasionally occur during a casting.Deformation of the internal void forms in box beams and cored slabs caused the camber to be overestimated by a significant amount and should therefore be considered in the prediction of camber.The prestressing force was found to be significantly affected by the temperature fluctuations of the prestressing strands during fabrication.Temporary thermal gradients in the girder could cause significant scatter of the measured camber data.The debonding and transfer length of the prestressing strands were found to be significant sources of error in the camber predictions for girders with debonded lengths greater than ten feet and should therefore be considered in the prediction of camber. This report provides specific recommendations to account for several of these factors to enhance the prediction of camber. ​

The research introduces two methods for the prediction of camber for prestressed concrete bridge girders, including an ""approximate"" method based on multipliers and a ""refined"" method based on the detailed losses calculations given in the 2010 AASHTO specifications. The current NCDOT method was also modified to account for the factors related to girder production. The current NCDOT method, the modified NCDOT method, and the two proposed methods were compared with measured cambers of 382 prestressed concrete girders in the field, some of which were taken by the research team and others that were collected with the help of NCDOT inspectors and Resident Engineers. The girder types that were considered in the study include AASHTO Type III and Type IV girders, box beams, cored slabs, and modified bulb-tees. The findings from the comparison of the prediction methods are summarized briefly as follows: The current NCDOT method was found to overestimate the camber of prestressed girders by an average of 52%. The modified NCDOT method overestimated the camber by an average of 39%. The proposed approximate method overestimated the camber by an average of 16%. The proposed refined method underestimated the camber by an average of 6%.The accuracy of the predictions of camber at prestress transfer was found to vary between different girder types and curing methods. Steam cured box beams and cored slabs exhibit lower cambers at the time of prestress transfer than the moist cured members. However, the accuracy of the predictions at later stages is less significantly affected by the curing method and girder type.

Based on the findings of this research, the two proposed methods are recommended to provide the most accurate prediction of camber. The proposed approximate method is more convenient for simple hand calculations, while the proposed refined method is suited for more accurate computer calculations. ​