In this study, static loading tests were conducted on a real PC bridge to confirm the load-bearing and failure behavior of the main girder at the ultimate state. An analytical model was developed to evaluate the load-bearing capacity of the PC superstructure by partially extending the linear framework model which is commonly used in practical design. Failure tests revealed that even when the main girder placed as the superstructure reached its shear capacity, the load was redistributed to the entire superstructure through the joints of the transverse girders and infill parts between the upper side flanges, and, consequently, the deformation performance was maintained. The advanced framework model constructed based on the results of this study was found to be able to reproduce the load-bearing behavior of the entire superstructure moderately. Furthermore, the advanced framework model was found to be an effective method for evaluating the safety of the entire superstructure at ultimate state and verifying the failure process using the loads employed in the design of new structures.

 This study proposes a prediction method of inclined crack widths of reinforced concrete (RC) beams when the beam action collapses based on energy theorem. The method was developed as an applicable one to evaluation model for capacities of shear resistance mechanisms towards deteriorated RC beams. The crack widths were derived from condition to maximize surface energy accumulated by mode II opening deformation on the inclined crack surface, subjected to the equilibrium condition of internal work done by internal forces and surface energy accumulated in the opening process of an inclined crack of the free body of beam action. The applicability was examined based on experimental and analytical results for sound RC beams. The results indicated the highest applicability to slender beams in all examined cases, and that the evaluation model which applied developed method could predict strengths of beam action in similar accuracies compared with prediction equations in design codes.