Continued Monitoring of the Varina-Enon Bridge: Estimation of Effective Prestress

About the report:

Prestress loss due to creep, shrinkage, and relaxation can cause serviceability issues, and in the case of structures post-tensioned with unbonded tendons, can reduce the flexural capacity. The accurate estimation of prestress losses is vital for making good decisions about the remaining life of a structure. The Varina-Enon Bridge is a post-tensioned concrete box-girder bridge near Richmond, Virginia. Flexural cracks in the bridge prompted an investigation into the magnitude of prestress loss experienced by the structure.

Long-term prestress losses were estimated using two methods. First, a finite element model was created, and multiple code expressions for creep and shrinkage were applied to a time-step analysis of the structure. The code expressions investigated in this research were from the CEB-FIP 1978, CEB-FIP 1990, CEB-FIP 2010, and AASHTO (2017) codes. The second method utilized data from sensors installed on the bridge to back-calculate the effective prestressing force based on recorded openings of the flexural cracks.

For the four spans monitored in this research, the field-determined effective prestress varied between 161 ksi and166 ksi.  Using the commercially available bridge design software, LARSA 4D, along with the creep and shrinkage model used in the original design, CEB-FIP 1978, the calculated effective prestress varied between 169 ksi and 171 ksi. This indicates that prestress losses were higher than anticipated in the original design, but the measured effective prestress was still, on average, about 96% of the design effective prestress. The more modern creep and shrinkage models of CEB-FIP 1990 and CEB-FIP2010 also predicted higher than measured effective prestress, with both being very similar to CEB-FIP 1978.  The effective prestress predicted by the AASHTO (2017) model was slightly higher. Calculation of flexural capacity using the effective prestress estimated by the field measurement system resulted in estimates of strength 1 to 4% smaller than using the effective prestress estimated by the original creep and shrinkage model used for design.

Measured thermal gradients over the period studied in this project were smaller than the AASHTO LRFD design gradients; however, the restraint moment calculated for the worst case measured gradient was very similar to the restraint moment calculated using the design gradient.

Authors