4.3.2. Load-Bearing Behaviour

In addition to crack formation, SMART-DECK influences the load-deflection behaviour of the slab segments (Figure 13 for static tests). Here, a differentiation was made according to the load distance, *ai* , and the longitudinal reinforcement ratio, ρl,s, referring to the steel reinforcement. Depending on the load distance and the flexural reinforcement, the ranges of abscissa and ordinate were adjusted, which should be kept in mind while comparing the results. The strengthened specimens (solid curves) show less deflection than the plain RC specimens (dashed curves) at the same load level. The deflections,

*w*, were measured continuously by means of displacement transducers beneath the specimens in the load axis.

**Figure 13.** Load deflection curves of large-scale tests on slab segments with static loading (**a**–**f**) separated by combination of flexural reinforcement and load distance.

For quantifying the strengthening effect, increases in ultimate loads were compared to the results of the corresponding non-strengthened reference tests. These quotients are indicated in Table 2. Particularly noteworthy are the shear tests S2-1, S7-1 and S7-2, which illustrate the considerable potential of SMART-DECK for increasing shear capacity. S2 and S7 only differ in the material combination of textile and mortar. The degree of strengthening of specimen S2 could even be increased with optimised materials in the reinforcing layer (specimen S7). The comparability of the results and the observed increases in shear capacity allow the conclusion that a simple top-side supplement of the flexural tensile reinforcement also has a positive influence on shear capacity.

In both partial tests on specimen S5 with a small steel reinforcement ratio, the type of failure shifted due to the strengthening with SMART-DECK. Capacity could be increased to such an extent that shear failure occurred instead of flexural failure.

While testing specimen S6, a loud popping noise occurred several times during the last quarter of the loading process, which was characterised by a slight load drop. This came along with the visible opening of a bending crack that appeared in the area of the cantilever support. Furthermore, the strain gauge attached to the flexural reinforcement in the RC base body failed after very high strains had been measured beforehand. The rebars in the support section (Figure 14) show that the steel reinforcement failed. Despite the partial opening of the interface, the stresses could clearly be redistributed that completely by the TRC layer.

were released when the steel reinforcement reached its ultimate strength. The subsequent increases in loading provide the evidence that the released stresses could be taken completely by the TRC layer. *Materials* **2020**, *13*, x FOR PEER REVIEW 19 of 27 subsequent increases in loading provide the evidence that the released stresses could be taken

**Figure 14.** Plastic deformation of flexural tensile steel reinforcement in the cantilever section of S6 (photo by IMB, RWTH Aachen University). **Figure 14.** Plastic deformation of flexural tensile steel reinforcement in the cantilever section of S6 (photo by IMB, RWTH Aachen University).

The results of the tests on S5 and S6 illustrate the enormous strengthening potential of SMART-

DECK with regard to flexural capacity of bridge deck slabs in transverse direction. The strengthening degrees determined for the tests on S5 and S6 are beyond the values which could be achieved in bending tests within the previous experimental programme [78]. Since then, advanced materials for the strengthening layer could be provided by the project partners, resulting in better mechanical properties for the mortar and the textile. Delaminations in the concrete interface only occurred in case of pre-existing imperfections. Those represented the origin of cracks along the concrete interface at high-load levels. It was also noticed that such delaminations had a negative influence on the participation of the strengthening layer in shear load transfer (tests on S4), while high degrees of flexural strengthening remained possible (tests on S6). Therefore, it could be assumed that flexural capacity of the increased cross-section is relatively independent of the quality of the concrete interface. Fortunately, imperfections could be prevented later on by enhancing mortar and application method. *4.4. Results of Tests with Cyclic Loading*  To investigate the influence of predominantly cyclic loads due to the impact of traffic, two tests were carried out under load collectives (red curves in Figure 15). At least 2 × 106 load cycles were aimed at. S8 already showed imperfections of the interface in the test area of the first partial test prior to loading. Therefore, an initial load was applied that corresponded to approximately 75% of the The results of the tests on S5 and S6 illustrate the enormous strengthening potential of SMART-DECK with regard to flexural capacity of bridge deck slabs in transverse direction. The strengthening degrees determined for the tests on S5 and S6 are beyond the values which could be achieved in bending tests within the previous experimental programme [78]. Since then, advanced materials for the strengthening layer could be provided by the project partners, resulting in better mechanical properties for the mortar and the textile. Delaminations in the concrete interface only occurred in case of pre-existing imperfections. Those represented the origin of cracks along the concrete interface at high-load levels. It was also noticed that such delaminations had a negative influence on the participation of the strengthening layer in shear load transfer (tests on S4), while high degrees of flexural strengthening remained possible (tests on S6). Therefore, it could be assumed that flexural capacity of the increased cross-section is relatively independent of the quality of the concrete interface. Fortunately, imperfections could be prevented later on by enhancing mortar and application method.

failure load of the non-reinforced test. The first partial test featured a high longitudinal reinforcement

### ratio (shear test). Just as in the static tests, the load was applied stepwise. Subsequently, about 80,000 load cycles with an amplitude of 10 kN were applied at a frequency of *f* = 5.243 Hz. The upper load *4.4. Results of Tests with Cyclic Loading*

was 120 kN (maximum peak load) and the lower load 100 kN (minimum peak load). This load range corresponds to about 12.5% of the ultimate load of the RC reference specimen. Hardly any stress changes in the reinforcement were measured and there was no significant change in the crack pattern. To investigate the influence of predominantly cyclic loads due to the impact of traffic, two tests were carried out under load collectives (red curves in Figure 15). At least 2 × 10<sup>6</sup> load cycles were aimed at. S8 already showed imperfections of the interface in the test area of the first partial test prior to loading. Therefore, an initial load was applied that corresponded to approximately 75% of the failure load of the non-reinforced test. The first partial test featured a high longitudinal reinforcement ratio (shear test). Just as in the static tests, the load was applied stepwise. Subsequently, about 80,000 load cycles with an amplitude of 10 kN were applied at a frequency of *f* = 5.243 Hz. The upper load was 120 kN (maximum peak load) and the lower load 100 kN (minimum peak load). This load range corresponds to about 12.5% of the ultimate load of the RC reference specimen. Hardly any stress changes in the reinforcement were measured and there was no significant change in the crack pattern. Therefore, the amplitude was doubled while maintaining the upper load at 120 kN and approximately 0.5 × 10<sup>6</sup> load cycles with a doubled load oscillation width were applied (lower load: 80 kN). In the meantime, the interface between the existing slab and the strengthening layer opened up, starting from the aforementioned imperfection that already existed before the start of the test. Nevertheless, no increase in the strain of the reinforcement and the concrete compression zone could be observed, which is why an increase in the average load was targeted.

**Figure 15.** Comparison of load-deflection curves of cyclic (S8) and static tests (S2, S6 und S7). *F*<sup>M</sup> = mean load, ∆*F* = peak-to-peak amplitude: shear test (**a**) and flexural test (**b**).

Therefore, a lower load of 140 kN and an upper load of 160 kN were selected which was less and more than the capacity of the non-strengthened static reference test, respectively. However, shortly after reaching the upper load for the first time, a wide shear crack formed (Figure 16). It propagated horizontally at the level of the flexural steel reinforcement. The specimen thus failed due to interface failure and secondary flexural shear failure after only ≈0.6 × 10<sup>6</sup> load cycles. Both the crack pattern and the ultimate load allow the conclusion that the interface damage prevented the strengthening layer to participate in the load transfer at high load level.

**Figure 16.** Crack patterns of cyclic tests on specimen S8.

The second cyclic partial test S8-2 was also performed as a load-collective test. In total, three load levels were applied (Figure 15b). First, the amplitude was retained at an average cylinder load of 35 kN and a load range of ∆*F* = 10 kN for approximately 4 × 10<sup>5</sup> load cycles, whereby hardly any changes in stress occurred in the strengthening material. So, the mean stress was increased by 15 kN to 50 kN, which was already significantly higher than the capacity of the non-strengthened slab. During the following approximately 4 × 10<sup>5</sup> load cycles at the second stage, no difference in the crack pattern and material stresses occurred (Figure 17), which is why the amplitude was doubled while the lower load remained the same.

The average load of the third cycle stage corresponded to approximately twice the capacity of the non-strengthened slab. At this stress level, the specimen was loaded up to a total number of more than 3 × 10<sup>6</sup> load cycles. During the third cycle stage with more than 2 × 10<sup>6</sup> load cycles, no significant increase in the material stresses could be determined, although clearly visible delaminations were found in the interface. Since the target number of load cycles was already exceeded, the load was taken off. Subsequently, the residual capacity was determined (Figure 15b).

Figure 17 shows a comparison of the strains measured during the test in the support axis (support at cantilever section) over the applied number of load cycles, whereby the three load levels are clearly visible. The strains of the steel reinforcement were determined using strain gauges (one measuring point averaged from two strain gauges), while the strains in the concrete compression zone and in the textile were determined using displacement transducers (approximation of the mean value from two measurements with a measuring length of 280 mm, >5 cracks in the strengthening layer, see above). The displacement transducers for determining the textile strain were located on the upper side of the slab segment, so the values shown are slightly higher than the actual strains in the textile plane.

**Figure 17.** Development of the strains in the support axis during cyclic loading of test S8-2 referring to the peaks of the cycles.

In the course of each load level, an increase in concrete compression can be seen. A redistribution of the stresses from steel to textile can be assumed. During the shift from the first to the second load level, the static capacity of the non-strengthened slab was exceeded. The steel strain suggests that the rebars started yielding. During the following 10,000 load cycles, however, the strain decreases again. At the same time, an increase in strain in the textile occurs. When the stress is increased again to level 3, a decrease in stress is visible in the steel reinforcement, while the strain in the strengthening layer increases. It can be concluded that the entire additional stress due to the increase in the upper load is transferred by the textile. An examination of the cross-sections of both reinforcements shows that the equivalent textile area weighted to the tensile strength is more than eight times the steel cross-section. Despite the subsequently applied 2 × 10<sup>6</sup> load cycles at high-load levels, no sign of fatigue failure of the steel reinforcement can be detected (e.g., disproportionate increase in strain). The specimen was prised open after the test to reveal the steel reinforcement in the support section of the cantilever. Slight confinement was visible which was much less distinct as in specimen S6 (Figure 14). This indicates that the carbon reinforcement indeed transferred the majority of the load during the last load stage. This suggests that it features good fatigue behaviour, which has been observed in other projects [40,79]. However, further investigations are required for verification. Furthermore, the presented tests could not provide any information on the fatigue behaviour of the interface. The separation of the strengthening layer from the RC base during loading was due to an already existing imperfection. Figure 18 shows the results of the slip measurements for the cyclic shear test S8-1 (Figure 18a) and the cyclic flexural test S8-2 (Figure 18b). The values refer to the head end of the slab segment and are the average values of two measurements with displacement transducers.

The slip during S8-1 increased continuously while the augmentation was less distinct in S8-1. Also, the total slip was larger in S8-1, which can be attributed to the higher load level of the shear test which lead to more pronounced stress in the interface. However, no conclusions can be drawn from those tests results regarding concrete-to-concrete bond with an intact interface. Considering it is an unreinforced interface, it is of particular importance to attest that an intact interface can be maintained despite fatigue loading.

**Figure 18.** Measured development of the slip in the concrete interface compared to load duration of the cyclic tests: (**a**) shear test S8-1 and (**b**) flexural test S8-2.

### **5. Summary and Conclusions**

This paper presented the results of tests on concrete bridge deck slabs with an additional layer of carbon textile-reinforced concrete. It is supposed to be applied between the existing structure and the road surface. This TRC layer (so called SMART-DECK) is intended to provide a monitoring system, preventive cathodic corrosion protection if necessary and the possibility to enhance the deck slab's flexural and shear capacity in transverse direction of T-beam or hollow-core concrete bridges.

An experimental campaign was introduced comprising small- and large-scale tests. The small tests were conducted using TRC samples to investigate the interaction between suitable high-performance mortars and potential textile reinforcement materials aiming at material refinement and selection of proper materials and characterising its properties. The large-scale tests were conducted on slab specimens strengthened using SMART-DECK. They were partly statically or cyclically loaded until failure.

The outline of the specimens was varied by means of


Therefore, load cases' bending and shear could be addressed in twelve static and cyclic tests on strengthened specimens. A comparison to test results gained from non-strengthened reference specimens showed that SMART-DECK can enable high increases in capacity:


The results demonstrate the high potential of carbon concrete strengthening for ULS load cases in bridge deck slab design. Additionally, the fine crack pattern at the tensile side of the slab also resulted in significant advantages regarding serviceability.

For verification purposes, further investigations should be carried out, with the focus on fatigue loading with alternating loads and higher amplitudes typical for bridges. Further test results provide the basis for generalised design approaches which not only quantify the flexural but also the shear and fatigue strength.

**Author Contributions:** Conceptualisation, V.A. and J.B.; investigation, V.A. and C.D.; writing—original draft preparation, V.A.; writing—review and editing, V.A., J.B. and C.D.; visualisation, V.A.; supervision, N.W. and J.H.; project administration, N.W.; funding acquisition, N.W. and J.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the German Federal Ministry of Research and Education, grant number 13N13108.

**Acknowledgments:** The presented investigations were carried out as part of a research project of the German Federal Ministry of Education and Research (BMBF). The authors would like to express their gratitude to the BMBF for the project's funding and the VDI Technologiezentrum GmbH, which was entrusted by the BMBF with the consultation and implementation of the grant guidelines, for their valuable support. The authors would also like to express their gratitude to the project partners, the Federal Highway Research Institute (BASt), Eurovia Beton GmbH NL Bauwerksinstandsetzung, solidian GmbH, Massenberg GmbH, instakorr GmbH (Darmstadt), Sto Cretec GmbH, subsidiary of Sto SE & Co. KGaA and the ibac of the RWTH Aachen University for the cooperative collaboration.

**Conflicts of Interest:** The authors declare no conflict of interest. The funding institution had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
