The first objective of this thesis was to assess the effect of casting orientation on bond strength in pretensioned prestressed concrete members. The "top strand effect" was evaluated through transfer and development length tests of prestressed concrete beams. Eight beams were cast with normal orientation, while four beams were cast with inverted orientation so that a significant depth of fresh concrete was placed below prestressing strands. Discrete transfer lengths were determined at the ends of each beam by measuring concrete surface strains. Inverted casting orientation caused an average 70 percent increase in transfer length. Some transfer lengths in beams with inverted casting orientation exceed current ACI and AASHTO code provisions. All measured transfer lengths were less than 90 strand diameters (45 in. for 0.5 in. diameter strands). Ranges of development length were determined through iterative load testing. The top strand effect on development length was more qualitative than quantitative. Ranges of development length in normal beams were conservatively less than code provisions. Ranges of development length in beams with inverted casting orientation were much closer to and sometimes exceeded code provisions. It is recommended that ACI and AASHTO code provisions for the development length of prestressing strand be modified to include the same magnification factors that are specified for the development length of deformed bars with twelve or more inches of fresh concrete placed below.

The second objective of this thesis was to compare experimentally measured prestress losses to theoretical calculations. Theoretical prestress losses were calculated according to PCI and AASHTO Refined methods. These methods produced similar results. Prestress losses were experimentally measured by vibrating wire gages and flexural load testing. Vibrating wire gages were used to monitor internal concrete strains. Two methods were used to reduce vibrating wire gage data: an upper/lower bound method and a basic method. The upper/lower bound method produced distorted data that was unreasonable in some cases. The basic method was more reasonable, but resulted in some prestress loss measurements that were greater than theoretical predictions. Flexural load testing was used to back calculate prestress losses from crack initiation and crack reopening loads. Prestress losses measured by crack initiation loads were generally greater than theoretical values. Losses measured by crack reopening loads were distorted. The distortion was attributed to difficulty in isolation of the correct crack reopening load. Large measurements of prestress losses by the basic vibrating wire gage and crack initiation methods suggested that losses occurred between the time when concrete was poured and prestress transfer occurred. Such losses are not accounted for in current code provisions. More research is recommended to determine the magnitude of these additional losses and their effect on design.