*4.1. Design and Materials*

For investigation of the strengthening effect of SMART-DECK, a total of 17 results from component tests are available, as the overview in Table 2 shows. Twelve of them were carried out on strengthened slab segments. From another project, a double reference test with a longitudinal reinforcement ratio of ρ<sup>l</sup> = 1.0% can be referred to [70]. The aim was to investigate the influence of SMART-DECK on the component behaviour for the two decisive failure types, bending (M) and shear (V), as well as failure in the interface (I). In Table 2, a distinction is made between planned and occurred failure types. For mixed forms, the sequence corresponds to the sequence of failure modes in the test. For example, V + I means that primary failure was due to flexural shear and the interface failed secondarily.

To control the bending moment, the load distance *a* = 0.7/1.0/1.3 m between load and support was modified and to additionally vary the exploitation of the flexural reinforcement, the longitudinal reinforcement ratio was set to either ρl,s = 1.0/0.5/0.2% in terms of the steel in the existing slab. For the high and medium longitudinal reinforcement ratio, threaded steel bars with a strength class of St900/1000 (Ø15) were used, and for the small longitudinal reinforcement ratio, a ribbed steel of B500 quality (Ø10) was used. The reinforcement layout of the reinforced concrete slab segments is shown in Figure 9. One half of specimen S8 was reinforced like S2 and S7, while the other half was equal to the reinforcement of S5 and S6, respectively. The non-strengthened reference slab segments featured a reinforcement that was identical to the corresponding specimens with TRC-strengthening.

No shear reinforcement was provided in the shear spans, as its installation in slabs is costly and therefore unusual in trunk road bridge slabs for practical reasons. A longitudinal reinforcement of Ø10/20 in transverse direction to the load transfer direction of the test bodies (longitudinal direction of the bridge) was provided in accordance with the normative minimum value for bridge slabs. The height of the RC base specimens was 28 cm, and the concrete cover was 20 mm all around. All slab segments were designed with a width of *b* = 50 cm, while the load plate's widths were 40 mm to involve the entire cross-section in load transfer. No significant impact was expected by the additional TRC layer on stress redistribution in the slab originating from the concentrated load. In the near past, the influence of the slab width was extensively investigated for reinforced concrete slabs [71–75] and it is assumed that the findings are generally applicable to TRC strengthened slabs.

The reinforced concrete base bodies were concreted indoors with a ready-mixed concrete featuring a target strength corresponding to a C30/37 and a maximum aggregate size *d*<sup>g</sup> = 16 mm. The specimens were compacted by means of an internal vibrator. Material samples (cylinder Ø = 150 mm/*h* = 300 mm and cubes with an edge length of 150 mm) were produced, which were stored next to the slab segments and tested at the time of the component test in order to be able to draw conclusions about the developed concrete strengths of the slab specimens (Table 2). Prior to the application of the strengthening layer, the upper surfaces of the reinforced concrete components were roughened by means of solid blasting. Using the sand surface method [76], a mean roughness of at least *R*<sup>t</sup> = 1.0 mm was determined (three measurements per slab segment). SMART-DECK was manufactured by the project partner Eurovia under construction site conditions.

Specimens S4, S5, S6, S7 and S8 were produced together and strengthened some weeks later at once. They had the same material combination in the TRC layer. The specimen which was used for tests S2-1 and S2-2 was manufactured at an earlier stage with the material combination then available.

### *4.2. Test Setup*

Two separate tests were carried out on each specimen (Figure 10). The load was always applied to represent a load resulting from a truck driving on the outer lane and thus loading the cantilever of the bridge slab. The load distance, *a<sup>i</sup>* (load axis to axis of the support close to the load), was varied according to the specifications in Table 2. The support with a larger distance to the load was designed to take the lifting forces. The load was applied via a hydraulic cylinder and transferred to the specimen via a square load plate of 40 × 40 (cm) corresponding to the wheel contact area for trucks according to European standard [77].

**Figure 10.** Specimen dimensions and test setup for the large-scale tests on slab segments (graphic by IMB, RWTH Aachen University).

All tests were statically loaded with stepwise increments until failure, except S8-1 and S8-2, which were loaded cyclically. The first four load stages were introduced load-controlled. From about half of the expected failure load, the load was applied deformation-controlled.

## *4.3. Results of Tests with Static Loading*

### 4.3.1. Cracking

In some tests, there were production-related imperfections in the interface between old and new concrete. Therefore, the TRC layer was partially detached from the existing concrete during the tests. However, these delaminations only occurred in the concrete interface and always originated from pre-existing imperfections. In case of intact interfaces, the load did not cause the joint to open, which could already be observed in previous tests [78].

Figure 11 shows the crack pattern the specimens exhibited after failure. During loading, a finely distributed crack pattern developed in the TRC layer at the top of the specimens (tension zone). The significantly larger number of cracks with simultaneously reduced crack widths in the strengthened components in comparison to the non-strengthened reference tests is to be regarded positively with respect to the influence of SMART-DECK on the serviceability.

Based on the crack patterns, the tests can essentially be divided into three groups, differentiated by primary cracking that introduced failure, or could be observed at the time when ultimate load was reached (Figure 12). Flexural failure was characterised by a wide vertical accumulated crack in the area of the support (test S6-1 and S6-2 in Figure 11). Relatively few other cracks occurred in the base body. In case of shear failure, an inclined flexural shear crack emerged, which exhibited a large crack opening at failure. This crack pattern was observed among the majority of tests in Figure 11, including the reference specimens. If delamination in the concrete interface occurred prior to failure, the shear crack usually propagated along the flexural reinforcement of the RC structure and the propagation of the crack did not continue into the TRC layer. In case of an intact interface between the existing structure and SMART-DECK, the crack propagated into the interface or the strengthening

layer. Then, considerably higher ultimate loads could be achieved than in the case of primary interface failure, which will be discussed in more detail later.

**Figure 11.** Crack patterns of static tests on specimens S1 to S7.

**Figure 12.** Essentially distinguished cracks related to failure (photos and graphic by IMB, RWTH Aachen University).

With a few exceptions, activation of the carbon concrete supplement could be achieved in the tests presented here, as shown in the previous section. However, the findings from manufacturing the strengthening layer for the specimens of this experimental programme provided crucial indications for the further development of the mortar and the production methods. In subsequent applications [50], unimpaired bonded interfaces were achieved so the complete functionality of the system could be realised.
