PSI - Issue 84

Giulia Rossini et al. / Procedia Structural Integrity 84 (2026) 1183–1190

1188

5. Experimental results and discussion Table 2 summarizes results of tension release tests in terms of ratio between experimentally measured stress ( , and , ) and the stress obtained from the analytical model . The , is calculated using Hooke’s law (equation 2) assuming that the measured strain variation µ and the corrisponding stress variation , due to the saw cuts is equal in magnitude but opposite in sign to the stress that existed in the section prior to cutting. , = − = − µ ∙ (2) The results are reported in two different forms: (i) the direct outcome of the measurements , is provided and (ii) an adjusted value , is also calculated by applying a correction factor of 90%, according to the following relation: , = , 0.90 (3) The adoption of this correction is justified by previous experimental campaigns aimed at validating the technique. These studies revealed that the average stress release obtained from the tests was approximately 90%. In particular, tests performed on full-scale beams (Mantelli, 2023) showed an average release of 91% with a CoV of 24%, while tests on simplified reinforced concrete specimens exhibited an average release of 89% with a CoV of 11%. For this reason, both ratios , and , are reported. The , leads to more conservative outcomes, resulting in a safer estimation during structural assessment, while , provides results more representative of the actual stress state. In Table 2 x and y specify the test location coordinates referring to the horizontal distance from the closest support from and to the vertical distance from the girder intrados, respectively. Figure 4 shows the results of Bridge 2: the same plots could be done for the other bridges. Each square and circle represents the result of a single test, calculated as the average value between strain gauges SG_A and SG_B. In particular, squares represent the ratio , , while circles represent the ratio , . In Figure 4 and in Table 2 the test name follows a structured rule formed by three main components: the span number [Sn], the girder number [Gn], and the progressive test number. Only the average value between SG_A and SG_B is reported. A direct comparison between tests performed on different bridges is not meaningful. The reliability of in-situ testing depends not only on the intrinsic technique reliability but also on bridge specific uncertainties. The main factor is the number of prestressing tendons and their initial stress, which are often undocumented; even when design documents exist, they cannot always be considered fully reliable. Additional uncertainties arise from the actual boundary conditions of the structure and from the load distribution among the various girders.

Figure 4: , and , for bridge 2.