Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Civil Engineering

Committee Member

Dr. Brandon E. Ross, Committee Chair

Committee Member

Dr. Thomas E. Cousins

Committee Member

Dr. Amin Khademi

Committee Member

Dr. Bryant G. Nielson


Bond-loss failures have been widely observed in load tests of precast-pretensioned concrete I-girders. This type of failure is associated with shear cracking near the support that interrupts anchorage of the strands, leading to loss of bond and slipping of the strands relative to the concrete. It has been experimentally demonstrated that this failure type can occur at load levels that are lower than the nominal shear and flexural capacities. Because bond loss can be the controlling factor for structural capacity, it is critical that strand anchorage be considered when detailing and calculating the capacity of I-girder end regions. This dissertation makes four contributions. First, a consistent terminology and characterization scheme for bond-loss behavior is presented. A review of 22 different test programs revealed that fifteen different terminologies were used to describe failures associated with bond loss. In response to these wide ranging terminologies, the different types of failure involving strand-concrete bond loss are characterized and a consistent labeling scheme is proposed. The fifteen different labels are condensed into four primary bond-loss behaviors. A flowchart is presented for assisting future researchers in characterizing and labeling bond-loss failures. Second, a bond-loss database is presented. The database was constructed in two phases. During the first phase of data gathering, 84 specimens were added from ten different test programs. During the second phase of data collecting, 36 more specimens from eleven different test programs were added. The database forms a basis for developing and testing quantitative models of bond-loss behavior. Third, models for calculating bond-loss resistance of pretensioned concrete I-girders were proposed. The initial model was developed and compared to the phase-one database of 84 experimental specimens. A refined model was also proposed. The accuracy of the refined model is examined by comparing the refined model to the expanded database with 120 specimens. The refined model improves the initial model by by using statistical linear regression analysis and the least squares method to identify best-fit equations with the experimental data. The refined model also has the advantage of being developed using a larger database. For the fourth contribution, the proposed bond loss model is used to evaluate the conservativeness of the strand debonding limitations in the current AASHTO LRFD Bridge Design Specifications. Debonding of select strands is an effective means of controlling stresses and cracking at the ends of pretensioned concrete girders, but can also have adverse effect on capacity due to loss of strand bond. The debonding limitations are evaluated by calculating the bond-loss capacity of six in-service bridge girders from different states. Capacities associated with varying levels of strand debonding are compared to the factored shear loads on each bridge. Calculations of bond-loss capacity are based on the initial model which was created as part of the third contribution.