*3.2. Uniaxial Tensile Tests on Individual Fibre Strands*

For test preparation, the fibre strands were extracted from the grid approximately at the centre between two strands (10–15 mm from the stand axis). The textiles have comparatively large yarn cross-sections. At the same time, carbon filaments have high strengths, so high failure loads had to be assumed, which required high lateral compression in the clamps at the ends of the yarns. To prevent them from rupturing in the clamps, the ends were glued into aluminium foil with the aid of epoxy resin mortar. These straps were fixed between the jaws of the clamps after curing.

Only results from tests were considered where the yarn actually ruptured within the free length. Loading was applied displacement-controlled at 1 mm/min. Meanwhile, the applied tensile force and the strain were recorded. Figure 6 shows the results for tests on T-1-38 (left) and T-2-21 (right) as mean value of 5–10 samples.

**Figure 6.** Results of tensile tests on yarns for T-1-38 (**a**) and T-2-21 (**b**).

The results differ with regard to the test length (free length between the clamps). Since the probability of imperfections increases with increasing length of investigation, its influence should be checked. For T-1-38, no negative influence due to increasing test length is evident. Only in weft direction, slightly smaller stresses do occur at the largest test length compared to the medium test length. Instead, the tests on the weft yarns of T-2-21 show increasing ultimate stresses with longer test length. Only in warp direction of T-2-21, a decreasing trend is visible, whereby the tests with medium lengths showed comparatively low tensile strengths. Overall, the ultimate stresses of T-1-38 are somewhat lower than those of T-2-21 and the fibre strands in the weft direction have higher tensile strengths than those in the warp direction, which is due to the negative influence of the knitting thread and is commonly known [67,68].

### *3.3. Small-Scale Tests on the Composite*

As mentioned before, the composite system was tested in small-body tests under bending and uniaxial tension (Figure 4, step II) to determine the essential properties of the strengthening layer under tension, such as cracking behaviour and stress–strain relationship. The pure tensile stress approximately corresponds to the eventual stress of the TRC layer on the bridge. As considerably more tests and thus more parameter combinations could be tested in a shorter time with bending tests, this method was preferred at the beginning of the project. The initial setup for uniaxial tensile tests ((Figure 4, step II), left test setup for tensile tests) required manual fixing of the specimens between the clamping jaws using several threaded rods, all of which had to be uniformly tightened. For the first materials tested within the scope of the project (M1, M2, T1 and T2 according to Table 2), both methods were therefore used. The dimensions of the specimens were determined according to the geometric properties of the textiles and the layered structure of SMART-DECK. Therefore, one layer of textile was bi-symmetrically positioned. The width of the composite strip was the multiple of the textile's mesh width and the depth was 20 mm to represent a segment taken from a wide strengthening layer.


<sup>#</sup> Cyclic tests; ## concrete strength of existing structure. \* represents a repetition of S1-2.

The textiles were placed in concrete over a total length of 1000 mm. For the flexural test, the TRC strips were supported over a span of 900 mm, with support overhangs of 50 mm at both ends. The test set-up was a 4-point bending test with centre distances of 300 mm in relation to the position of the supports and load application points. Large deflections occurred (100 to 130 mm in the middle of the span at ultimate load), which correspond to five to six times the depth of the test specimen. Failure always occurred via a propagating crack by spalling of the concrete compression zone without rupture of the textile. As the reinforcement did not fail and due to second-order effects (large deflection of the specimens in relation to the horizontal axis of support), the results of the bending tests were not taken into account in the evaluations and only uniaxial tensile tests were used in the further investigations of the composite load-bearing behaviour.

Further development of the uniaxial tensile test setup [69] minimised execution time. As shown in Figure 4, step II, on the right, the clamping jaws were replaced by a hydraulic device that simplified the installation of the specimens considerably. This setup was used from series 2 onwards. It can be concluded that flexural tests are unfit to determine the mechanical properties of composite specimens that are very thin and simultaneously feature such high flexural slenderness, *l*/*d* (here: *l*/*d* = 90). In this case, the tensile test setup does provide convenient results.

A total of five test series was investigated with their main parameters depending on the progress and open questions regarding the efforts of the project partners to develop advanced materials that meet all, and sometimes contradictory demands resulting from the different targeted functions of SMART-DECK (monitoring, pCCP and strengthening). While all tensile tests aimed at receiving essential insights on mortar to reinforcement combination, each series also aimed at defining another parameter, like, for instance, the following:


**Figure 7.** Stress–strain relations of seven exemplary groups of uniaxial tensile tests on composite strips with seven different material combinations (**a**–**g**) and indication of the scatter of ultimate stresses, σ, and strains, ε, by means of the coefficient of variation (CoV).

Figure 7 shows the results of some of the uniaxial tensile tests on composite strips by means of their stress–strain relations. While the tensile force is related to both the composite and the textile area (left and right y-axis, respectively), the grey curves show the individual result of one test and the black line represents the mean curve calculated from the total of the individual results correspondent to an established procure [69]. Thus, it was determined by averaging stress values of all test curves over predefined equal strain intervals until the first curves end due to failure of the corresponding specimen. To define the mean increase in length at failure, an additional data point was defined which corresponds to the average strength of the individual specimens.

Those and the other tensile tests that are not shown in detail here, were conducted to characterise the properties in the warp direction, which is decisive for the strengthening effect in the present case. The specimen length was 1000 mm, and the anchor length at each end was 250 mm. Figure 8 shows a uniaxial tensile test until the specimen failed. The cumulative crack openings were measured over a length of 450 mm using displacement transducers attached to the front and back to derive the mean strain of the reinforcement. This is possible because of the large number of cracks in the measurement area (significantly more than five cracks, as indicated in [69]). The tests were carried out displacement-controlled at 2 mm/min.

**Figure 8.** Stages of a tensile test: installed test specimen with slight cracking (**a**), specimen featuring complete crack formation (**b**) and specimen after failure (**c**), photographs by IMB, RWTH Aachen University).

Due to the specified layered structure of SMART-DECK with a concrete cover of 10 mm, the concrete cover of the tensile strips was also limited and not optimised regarding maximisation bond or maximum tensile stress development. Therefore, it was expected that a full exploitation of the material's potential might not succeed, which was confirmed by the tests. After completion of the crack formation transverse to the load direction, longitudinal cracks often formed in the textile plane. Failure of the fibre strands occurred abruptly and was accompanied by concrete spalling. In some cases, there was no or only partial spalling. In these cases, however, the matrix clods were no longer bonded to the textile. The fibre strands in tensile direction either ruptured or remained partially intact. Since the specimens that were concreted by means of pouring and compacting (Figure 7a) featured higher ultimate loads and less scattering than comparable specimens that were laminated (Figure 7b), the former manufacturing method was used for the following tests. It was also the preferred procedure in terms of ease-of-use since it is easier and faster to apply to large surfaces like bridge decks. In terms of the specimens' depth, no significant difference was observed (Figure 7c,d). Therefore, specimens that only had one layer of textile reinforcement allow for a representative investigation of the tensile resistance of SMART-DECK. Material combinations 31 and 44 (Figure 7e,f) were used as strengthening layers for the large-scale tests described in Section 4.

In the first two mortars tested, the fibre strands generally failed smoothly. In the more advanced mortars, they were mostly frayed. An improvement in the bond properties was achieved by further development of the textiles (T-4-38). The highest failure stresses, and at the same time the smallest dispersion, were obtained in tests with sanded textiles of T-4-38 in combination with M-4-04 (Figure 7g). However, since sanding results in considerably poorer electrical properties, it was not shortlisted for the project. T4 featured the overall best properties regarding all three target functions of SMART-DECK and was therefore chosen for the large-scale tests.

### **4. Large-Scale Tests**
