PSI - Issue 64

Aeneas Paul et al. / Procedia Structural Integrity 64 (2024) 1287–1294 Aeneas Paul et al. / Structural Integrity Procedia 00 (2019) 000 – 000

1292

6

( z -coordinate) is color coded. An unweighted moving average filter (Guiñón et al. (2007)) was applied over a length of 10 cm to reduce the white noise from the data. In general, positive concrete strains occur as a tendon is cut. The previous compression caused by the prestressing is reduced. This holds true for all measured strains shown in Fig. 4. Just over the height the alteration of strains differs since it is superposed by the strains induced by self-weight.

Fig. 4. Concrete strains measured on the surface a) ε xx for y = 0.15 m on NSC b) ε xx for y = 0.15 m on HPC c) ε xx for y = -0.15 m on NSC d) ε xx for y = -0.15 m on HPC after tendon break and corresponding stresses from FE simulation e) σ xx and σ zz for NSC f) σ xx and σ zz for HPC.

Nevertheless, all curves qualitatively run similar, decreasing symmetrically from the maximum at the break point to almost zero. Around the maximum, a hollow of varying intensity is observed. By value the strain peaks differ for the two concretes. They amount to 15-30 µm/m in NSC, compared to 10-15 µm/m in HPC. The length of impact on the strain differs, too. To both sides from the peak, it is about 50 cm (NSC) or 35 cm (HPC), respectively. FE simulations of the beams were performed to highlight the mechanics of the re-anchoring process. They were subjected to longitudinal forces corresponding to the prestressing forces P ( x ) obtained in section 4.1 for the two lengths l pt . The predicted longitudinal ( σ xx ) and transverse ( σ zz ) surface stresses at z = 0 are shown in Fig. 4e-f. σ xx shows similar results for both lengths: Over the first 5 cm from the break point slightly increasing, then decreasing to a constant value of 0.95 N/mm² over the associated length l pt . σ zz increases to its maximum at about 10 cm distance towards the break point, before converging to zero. The peak value computed for l pt = 22 cm (≈0.12 N/mm², HPC) is about 50 % higher than for l pt = 39 cm (≈0.08 N/mm², NSC). This shows an increase of σ zz with declining l pt , which also reduces ε xx . A change of σ xx is only obtained over l pt . It must be noted that the FE simulation neither captures the Hoyer effect nor accounts for slip. The numerical results are further compared to the measured strains in the tendons and the concrete in section 4.4. 4.3. Pull-out tests All pull-out tests resulted in bond failure. The results, presented in Fig. 5, show an increase of the bond stress for all configurations, until a maximal bond stress τ b,max is reached. For each series, the mean maximum bond stress ̅ b,max