Bond-slip constitutive model for the strand-concrete interface based on pull-out tests

Abstract

Bond transfer at the strand–concrete interface governs the anchorage behavior of pretensioned concrete structures. Monotonic central pull-out tests were conducted on 39 specimens containing 15.2 mm seven-wire strands to investigate the effects of effective anchorage length, transverse confinement, and steel fibers. Three specimens exhibiting loaded-end strand unwinding were excluded; the remaining 36 specimens exhibited interface-controlled strand pull-out accompanied by localized surface cracking. Within the effective anchorage-length range of 200–500 mm, stirrup spacings of 100 and 50 mm increased the mean peak load by 25.1–29.4% and 39.4–59.9%, respectively, relative to specimens without stirrups. For specimens with an effective anchorage length of 500 mm, the addition of 1.0% steel fibers increased the mean peak load, residual load, and loaded-end slip at peak load by 7.4%, 9.7%, and 56.4%, respectively, when averaged across the three confinement levels. A discontinuous tri-linear five-parameter local bond–slip law was adopted, and its parameters were identified through inverse analysis. For the 36 valid specimens, the median R 2 , median NRMSE, and peak-load MAPE were 0.9637, 4.29%, and 2.87%, respectively. A parameterized model incorporating effective anchorage length and equivalent confined-concrete strength was established. Independent validation produced peak-load errors of 7.4–15.9%. The identified characteristic stresses and slips differed markedly from code-based values for ribbed rebar–concrete interfaces, indicating that those empirical parameters should not be directly applied to seven-wire strands.