Flexural behavior of three span continuous unbonded post-tensioned members with variable bonded reinforcement

•Available prediction methods underestimate predicted stress increase values in unbonded post-tensioned members at ultimate.•Bonded mild reinforcement has significant effect on the behavior of unbonded post-tensioned members.•The optimized collapse mechanism model provides the most accurate and consistent stress increase prediction values.•Tendon stress increase prediction method has marginal effect on the flexural strength. Several methods have been proposed for calculating unbonded tendon stress increase at nominal bending resistance. Some of these methods are empirical approaches that are based on experimental observations or statistical analyses. However, more recently, some researchers developed methods that are based on collapse mechanisms that consider different parameters such as span-to-depth ratio, material properties, continuity, and inelastic hinge formation. All of these methods relied on simple span test data in model calibration with less consideration for continuous members because there are very few experiments available in the literature. Furthermore, there is a great need for additional experimental data in this area as the largest assembled database in the literature contains only 80 continuous unbonded tendon reinforced members, mostly two-span, driven by the expense and difficulty of such tests. In this study, the research team tested four three-span, 59.5 ft (18.1 m) long specimens with variable prestressing and mild reinforcement ratios. The accuracy of seven different prediction methods was evaluated for predicting stress increase in unbonded post-tensioned members. The comparison between measured and predicted stress increase values showed that all prediction methods conservatively predict tendon stress increase at nominal flexural resistance. The measured-to-predicted tendon stress increase ratio ranged from 1.8 to 3.5, at flexural failure, and measured stress increase ranged from 38.4 to 68.5 ksi (265–472 MPa). An optimized collapse mechanism model and the Japan Prestressed Concrete Engineering Association methods provided the most accurate prediction for tendon stress increase when compared to other prediction methods. Specimens reinforced with bonded mild reinforcement less than the minimum required by ACI 318-14 code showed less ductility than those reinforced with the minimum required reinforcement or more and obtained lower strand stress, indicating the specification is adequate.