4.2. Specimen Preparation and Prestressing
The beams were cast in a prefabrication plant. In a first step the CFRP bars were prepared for prestressing. Previous investigations showed that a mechanical sleeve wedge anchorage is suitable for introducing high prestressing forces into the CFRP bar. The anchorage method used for the investigations consisted of the CFRP bars, an aluminium sleeve with a length of 120 mm, an outer diameter of 12 mm and a thickness of 1 mm, epoxy mortar and a conventional steel wedge anchorage. The CFRP bars were cut to the desired length before the aluminium sleeves were placed at the ends and the cavities between CFRP bar and sleeve were filled with epoxy mortar, see
Figure 3a. Further information on prestressing of FRP bars and the challenges involved can be found in [
42,
43,
44].
Before the CFRP bars were mounted in the U-shaped FRP mesh, a twisted single roving, as shown in
Figure 3b, serving as splitting tensile reinforcement was placed around them. Subsequently, the plain FRP mesh was fixed to the top of the U-shaped mesh using cable ties made of plastic. After applying the custom-made spacers [
38], the reinforcement cage was placed in the previously carpentered formwork (
Figure 3b) made of coated plywood.
Reinforced concrete beams, steel profiles, and steel plates were used as abutments for pre-tensioning of the CFRP bars. The formwork was placed between the concrete beams and the steel sections were situated perpendicular at the end of the beams. Steel sheets with holes were used to fix the anchorage sleeves against the abutment system. At one end of the prestressing system, two steel sheets with four threaded bars were used as spacers (
Figure 3c). To apply the prestressing force a hydraulic cylinder was placed between the steel sheets and the distance between the sheets was increased (
Figure 4, step 1). The new position of the steel sheets was secured using screw nuts before the hydraulic cylinder was retracted and the prestressing force was induced into the abutment, see
Figure 4, step 2. With this method, only simultaneous pre-tensioning of both CFRP bars was possible, see
Figure 3c, with an equal distribution of the forces depending on the position accuracy of the hydraulic cylinder. To counteract prestress losses in the anchorage system all bars were overstressed with additional 2 kN. About 45 h after casting of the beams (
Figure 3d and
Figure 4, step 3) the pre-tensioning force was released onto the beams by extracting the hydraulic cylinder. First the cylinder was extracted and the CFRP bars were pre-tensioned to a point where the screw nuts could be loosened (
Figure 4, step 4). Then the pressure of the cylinders was released within 20 to 30 s resulting in a full prestressing of the beams (
Figure 4, step 5).
4.4. Results
The experimental results of the four-point bending tests will be discussed by comparing bending moment-deflection curves, the cracking behaviour and the failure modes. It must be mentioned that the own weight of the beams was not taken into account, since it was almost equal for all beams and neglectable causing a bending moment of only 0.40 kNm.
In
Figure 6, bending-moment deflection curves are depicted, each presenting the influence of one or two investigated parameters. The beams with a web thickness of 50 mm, made of UHPC and reinforced with the carbon mesh reinforcement Q95/95-CCE 38 can be seen on the top left side. By applying a prestressing force on the beams, a clear improvement of the deflection behaviour is visible when comparing the bending moments at a deflection of 12 mm. Beam B01, for example, could only bear a bending moment of 5 kNm without prestressing, whereas the prestressed beams B02 and B04 reached bending moments of about 10 kNm and 16.5 kNm, respectively, at the same deflection. Due to the changed failure behaviour the ultimate bending moments of the prestressed beams were significantly higher than of those without prestressing. While beam B01 failed due to rupture of the bottom roving, the prestressed beams failed when the CFRP bar reached its ultimate tensile strength.
The diagram on the top right shows the influence of the different mesh reinforcements used as secondary and shear reinforcement in both non-prestressed and prestressed UHPC beams with a web thickness of 50 mm. As within the previous comparison a clear improvement of the deflection behaviour can be noticed for the prestressed beams, showing bending moments of 13 to 15 kNm compared to the 5 kNm of the non-prestressed variants at a deflection of 12 mm. The generally higher stiffness of beam B07 can be explained by a probably higher pre-tensioning force. Regarding the non-prestressed beams, a slightly lower stiffness of the beam reinforced with GFRP mesh reinforcement (B09) can be observed, whereas the ultimate bending moment shows no difference to that of the beam reinforced with the CFRP mesh Q95/95-CCE38. The highest achieved ultimate bending moment (beam B05) can be traced back to the higher tensile strength of the used CFRP mesh Q85/85-CCE21. It must be mentioned, that the test of beam B01 was stopped during the loading phase, due to overheating of the hydraulic aggregate and had to be restarted. An influence of the de- and reloading on the maximum bending moment cannot be excluded. For the prestressed beams no significant influence of the used mesh reinforcement is observed.
The bottom left diagram in
Figure 6 presents the influence of the web thickness on the bearing behaviour of UHPC beams with a Q85/85-CCE21 mesh reinforcement. While the deflection behaviour of the non-prestressed beams does not show any differences, the beam with the higher web thickness (beam B05 compared to beam B11) reached a higher ultimate bending moment. The prestressed beam B12 shows a less stiff behaviour compared to B07, which can again be explained by the above-mentioned higher pre-tensioning force. The highest ultimate bending moment of all tests was reached by beam B12 (40 mm web thickness, UHPC, Q85/85-CCE21, 50% pre-tensioning force) with 28.9 kNm.
The impact of the different concrete compression strengths is depicted in the bottom right diagram of
Figure 6. Even though the UHPC beams reached a higher ultimate bending load, no difference in stiffness can be observed for the non-prestressed beams. The prestressed HPC beam (B15) shows a flatter curve than the UHPC beam (B07) after a deflection of about 20 mm while having the same ultimate bending moment of around 26 kNm.
The cracking behaviour of the beams is discussed based on the occurrence of the first bending crack, the beginning of the first splitting cracks, the end of the cracking phase, and the bending crack width at a bending moment of 20 kNm. Using the DIC records all necessary information was obtained within postprocessing. The end of the cracking phase was determined by looking for a load step where no further bending cracks occurred in the detected area. Virtual extensometers were placed 10 mm above the bottom of the beam across each bending crack and the length change of each extensometer was measured for each load step. The cracking moment was defined to be the moment when the first crack reached a width of 0.01 mm. In addition to the bending cracks, the appearance of splitting cracks was also investigated, due to the high significance in terms of bond failure. In the presented investigations the splitting cracks were determined visually during the DIC postprocessing, and the corresponding bending moments, when they first occurred were listed. Cracking patterns of a non-prestressed (B06) and prestressed beam (B08) at different load steps can be seen in
Figure 7. The positive influence of prestressing is clearly visible in the top pictures. While pronounced bending and splitting cracks are found on the non-prestressed beam, the prestressed beam shows only small initial cracks at a bending moment of 15 kNm. The subsequent increase of the bending moment leads to more splitting cracks within the non-prestressed beam. This is not the case for the prestressed variant, which is only characterised by an increase of the number of bending cracks. The crack patterns before failure show a strongly damaged non-prestressed beam and first splitting cracks in the prestressed beam. During failure the concrete cover at the bottom of both prestressed and non-prestressed beams was blasted off due to rupture of the bottom mesh roving or the CFRP bar.
An overview of the experimental tests and the results is given in
Table 5. The ultimate bending moment M
u lies in a range between 21.3 to 28.9 kNm, with the prestressed beams generally showing higher M
u-values due to the different failure mode, namely the rupture of the CFRP bar instead of the roving. The corresponding deflection at the ultimate bending moment D
M,u decreases with increasing prestressing. While the non-prestressed beams failed at a maximum deflection of 98.6 mm in the centre of the beam, the prestressed beam B08 showed a deflection of 48 mm at failure. As previously mentioned, the cracking behaviour was analysed by looking at and comparing the bending moments of the first bending crack occurrence M
be,cr and splitting crack occurrence M
sp,cr as well as the bending moment defining the end of crack formation M
cr,fin and the mean crack width at a bending moment of 20 kNm. For all configurations M
be,cr increased stepwise with the applied prestressing force. When looking at the beams with a prestressing level of 70%, for example, the tests showed a comparable M
be,cr of 9.8 kNm and 10.4 kNm. For the non-prestressed HPC beams and the beams with a web thickness of 40 mm, the splitting crack occurred earlier compared to the other configurations, however prestressing resulted in a significant increase of the M
sp,cr for these beams. In general, the beams reinforced with the carbon mesh with the larger roving axis distance, showed earlier splitting cracks compared to the other beams. The crack initiation phase ended at lower bending moments M
cr,fin for the non-prestressed beams. The highest M
cr,fin was observed in the specimens with the mesh reinforcement G121 made of alkali-resistant glass (AR-glass). The assessment of the crack widths at a bending moment of 20 kNm (w20
kNm,
Table 5 second-last column) shows the positive influence of prestressing, reducing the crack widths significantly. For the non-prestressed beams, the configuration with the small-meshed carbon textile showed the smallest crack widths, while the beams with HPC, GFRP mesh and the large-meshed CFRP textile showed clear inferior cracking behaviour. The influence of the prestressing diminished the differences within the cracking behaviour of the specimens with the various investigated parameters, with all specimens prestressed to a level of 50% showing crack widths between 0.37 to 0.27 mm, with the exception of beam B15, made of HPC with a mean crack width of 0.44 mm. The two different failure modes which occurred were either rupture of the bottom roving (non-prestressed beams) or rupture of the CFRP bar (prestressed beams). In the case of the beams reinforced with the Q121/121-AAE38 mesh, the non-prestressed beam also failed due to rupture of the CFRP bar. The failure mode CFRP rupture is depicted for beam B3 in
Figure 8.