A close examination of the test results listed in the database led to a number of important findings regarding the influence of important parameters on the behavior of FRP-confined concrete. In the last part of the paper, a new design-oriented model developed using a database for predicting the final state of FRP-sealed concrete is presented. The database consists of the following information for each sample: encapsulation technique (wrapped concrete or pipe-encased concrete); geometric properties of the sample (diameter D and height H); unconfined concrete strength (f'co) and deformation (εco); FRP shell material properties (modulus of elasticity Efrp, tensile strength ffrp, total thickness tfrp); material properties of the fibers used in the FRP sheath (modulus of elasticity Ef, tensile strength ff, total thickness tf); compressive strength (f'cc) and ultimate axial strain (εcu) of sealed concrete and average strain of the FRP ring at break (εh,rup); and the reduction factor of the rupture strain of the ring based on the fiber properties (kε,f) and the FRP material properties (kε,frp).
In the studies reviewed, the properties of the FRP containment systems were reported in several different ways. In this study, the test database was sorted into two categories, and the results of the FRP-wrapped and FRP-tube-encased specimens are presented in separate tables. Several previous studies have focused on the influence of the types of FRP materials on the behavior of FRP-bonded concrete (e.g.
This understanding is supported by the trends in the test results reported in the database for this study [Figure 4 (a) and (b)]. The average values of the strain reduction factors determined from the database reported in the present study (Table 8) point to the influence of the fiber type on the strain reduction factor (kε,f) and thus on the hoop failure stresses.
Instrumentation details of specimens reported in the database
It is now understood that for a given confinement ratio (flu,a/f'co), the ultimate axial strain of the FRP-confined concrete increases with the increased ultimate tensile stress (εf or εfrp) of the materials used to confine the. . It is clear from these figures that the trend lines of deformation enhancement ratio are sensitive to the type of FRP, whereas the strength enhancement ratio is not strongly affected by changes in the type of FRP. Considering its direct influence on the actual confinement ratio (flu,a/f'co) and therefore the ultimate condition of FRP-bonded concrete, it is obvious that the accurate determination of hoop failure stresses plays a crucial role in predicting the ultimate condition of FRP-bonded concrete.
Similarly, different measurement methods have been used to measure the hoop inhibitions, including methods (i) and (iii) mentioned above, with strain gauges or measuring devices oriented in the hoop direction. Information regarding the specific methods in measuring both of these strains is reported in the last column of Table 2 for each study included in the database. For the specimens where multiple hoop strain gauges were used, such as the specimens tested by Lam and Teng [18], Smith et al.
36], and Wu and Jiang [24], the average values of the strain gauge measurements were recorded in the database. During the calculations of the average values, due care was taken to exclude inconsistent strain gauge readings, such as those coming from the overlapping areas of FRP plates.
Test database size and scatter
In particular, the two most important properties of the ultimate state, the ultimate axial stress (εcu) and the ring breaking strain (εh,rup), are very sensitive to the instrumentation used in testing samples. Only CFRP-wrapped specimens, which formed the largest subgroup in the database, were included in Figure 5 to eliminate the additional influences caused by differences in FRP type and confinement method. Differences in the trend lines shown in Figure 5 suggest that the recorded ultimate axial stresses may be influenced by the measurement method used in their determination.
Similarly, it is to be expected that the average recorded hoop rupture stresses (εh,rup) will be affected by the number and placement of the strain gauges used in measuring these stresses. As originally reported in Lam and Teng [3], hoop strains measured within the overlap regions of FRP jackets are known to be lower than those measured elsewhere around the circumference of the same FRP jacket. Therefore, it follows that variations in the hoop strain gauge settings of the specimens included in the database are one of the main reasons for the inherent scatter in the hoop rupture strain data reported in the database.
A NEW DESIGN-ORIENTED MODEL
- Hoop rupture strain of FRP-confined concrete
- Compressive strength of FRP-confined concrete
- Ultimate axial strain of FRP-confined concrete
- Comparison with test data
The observed dependence of the strain reduction factor on the type of confining fibers was previously noted in Ozbakkaloglu and Akin [39] and Dai et al. The statistical quantification of the influences of these two parameters resulted in the expression given in Eq. The effect of increasing force on the first peak stress (f'c1) of the stress-strain response is captured using Eq.
The prediction of the effect of the increase in strength after the first peak stress (f'c1) is based on the net confining pressure, that is, the reduction of the actual confining pressure (flu, a) after subtracting the threshold limit pressure (flo). 1] almost all of the best performing final strain growth expressions proposed in the literature have nonlinear forms in their predictions of the strain growth ratio (εcu / εco) as a function of the insulation ratios (grip,a / f 'co) eg [42, 43]). This is due to the dependence of the ratio of the increase in stress (εcu / εco) to the ultimate tensile stress of the material (εf or εfrp), in addition to the isolation ratio (flu,a / f'co), as was pointed out in a number of previous studies.
As previously discussed in Section 3.5, the magnitude of the recorded ultimate axial strains can be influenced by the methods used in measuring the strains. Figures 7(a) and 7(b) , respectively, show the average absolute errors (AAE) of the strength and strain-gain ratio predictions of these models.
CONCLUSIONS AND RECOMMENDATIONS
Two key ultimate state properties, namely ultimate axial strain (εcu) and hoop failure strain (εh,rup), are both very sensitive to the instrumentation arrangement used for sample testing. Therefore, the variability in the instrumentation arrangements used in different studies contributes to dispersion in the database. There are differences between the strength and strain increment ratio of FRP-encased and FRP-tube-encased specimens included in the database for this study.
However, due to differences in the number and parametric ranges of the FRP-wrapped and FRP-tube-wrapped specimens, no definitive conclusion can be drawn from these observations. As previously mentioned, it was not possible to include all the test results published in the literature in the database presented in this paper due to the lack of information regarding the material properties, geometric properties or end conditions of these samples. Therefore, future studies should strive to ensure that the test results are presented with a complete set of information that provides as much relevant information as possible about the material and geometric properties of the specimens, the test setups and instruments, the recorded performance of the specimens and their failure modes.
Furthermore, in future experimental studies, due attention should be paid to the instrumentation of the samples for the accurate measurement of ultimate axial stresses and hoop rupture stresses.
ACKNOWLEDGMENTS
NOTATION
