firstly placed precast deck portion averaged 42.3 MPa, and that of the second placement in the connection reached 40.1 MPa.

2.2. Determination of the Fatigue Load

The fatigue load for the specimens was determined with reference to the design live load moment. The design live load moment acting on the PFC deck reaches 47.7 kN× m/m, which corresponds to the loading of two vehicles. Analysis was performed to obtain the load generating such live load moment on the specimen. In the analysis, all the diverse members used for the installation of the specimen were also modeled to achieve the actual boundary conditions of the specimen. In the pulsating fatigue test, the H-beam representing the cross-beam is directly fixed on the support blocks, while an additional H-beam is adopted in the rolling fatigue test (Figure 6).

For the DCFP specimen (pulsating fatigue specimen), a moment of 21.6 kN×m/m develops under application of the live load of 96 kN. Converting this load with respect to the design live load moment, this load becomes 212 kN. Considering a safety factor of 20%, the fatigue load is computed to as 254.4 kN. Besides, the moment developed in the DCFR specimen (rolling fatigue specimen) under application of the live load of 96 kN is 22.6 kN×m/m. The corresponding fatigue load obtained through conversion with respect to the design live load moment is 202 kN, which becomes 242.4 kN considering an additional safety factor of 20%.

2.3. Testing Procedure and Measurement

The specimens were fixed on the support blocks to enable measurement of the negative moment behavior above the cross-beam. For the static loading test and pulsating fatigue test, the load was applied on a loading surface of 230 mm (longitudinal direction) × 580 mm (transverse direction) at the center of the deck. In the case of the rolling fatigue test, loading plates of 230 mm × 580 mm were continuously installed and a steel plate was disposed over the plates to enable loading by the moving vertical load. Figure 7 shows the fatigue loading tests.

After each specimen experienced the specified number of fatigue cycles, static loading test was conducted to evaluate the degree of damage with respect to the number of cycles. Since the degree of damage can be assessed by means of the deflection or the cross-sectional strain, displacement sensors were disposed at the bottom of the center of the specimens, and strain gauges were installed at locations distant longitudinally from the center by 0, 250, 750, 1250, and 1550 mm as shown in Figure 2. Moreover, 2 strain gauges were also installed at different heights in the FRP module at each location together with 2 strain gauges embedded inside concrete to observe the distribution pattern of the strain inside the section as well as to measure relative slip that might occur at the interface.

3. Experimental Results and Discussion

3.1. Static Failure Pattern

In order to examine the position of the cracks occurring in the negative moment zone of the static loading specimen (DCS), preliminary loading test was carried out up to cracking before the tests. Cracking initiated under loading of about 230 kN above the location corresponding to the end of the haunch. The occurrence of cracks at the end of the haunch indicates that the effective length of the deck may be shortened due to the presence of the haunch. After removal of the preliminary load, crack gauges were installed and loading was applied through displacement control at speed of 0.5 mm/min.

Nearly circular crack with its center at the loaded portion was observed at load of about 250 kN in a partial zone at the top of the deck. Furthermore, the cracks initiated at the end of the haunch propagated transversally with the increase of the load. Finally, punching failure occurred at 1098 kN as shown in Figure 8.

3.2. Fatigue Damage Pattern

The values of the vertical load applied on the fatigue specimens were 254.4 kN for the pulsating fatigue and 242.4 kN for the rolling fatigue, corresponding respectively to 23% and 22% of the static failure strength. The DCFP and DCFR specimens subjected to the repeated action of this vertical load did not fail even after 2 million cycles. Residual performance test was conducted on the specimens that did not experience fatigue failure. The results are discussed in Subsection 3.3. Negative moment cracks developed in the first fatigue cycles at the top of concrete at the end of the haunch. Thereafter, microcracks initiated in partial zones under continuing cyclic loads but did not propagate significantly.

Figure 9 plots the variation of the deflection and strain at the center of the deck (DCFP specimen) according to the number of fatigue cycles with reference to the applied load of 254.4 kN. In fact, strain gauges were also installed at the top surface of the FRP panel but the gauges were damaged since the start of the test. The de-

flection was 3.8 mm at the beginning and increased by 17% to reach 4.4 mm at 2 million cycles. The strain at the bottom of FRP also experienced change from 487 × 10^–6 to 569 × 10^–6 showing an increasing rate of about 17%. Besides, the strain of concrete changed to exhibit reduction of the curvature of the concrete section and downward shift of the neutral axis. This phenomenon appears when crack occurs in the surrounding concrete. The behavioral pattern in the graph verifies indirectly the damage pattern of the surrounding cracked concrete. The variation of the curvature with respect to the initial curvature reaches approximately 11% showing that the degree of damage is relatively small.

Figure 10 shows the strain distribution pattern in the longitudinal direction from the center of the deck (DCFP specimen). The comparison of the state of the deck before application of the fatigue load and after 2 million cycles indicates that larger damage occurred at the center

of the deck than at the supports. This phenomenon can be explained by the larger moment acting on the loading surface. The main cause of such damage can be found in the cracking of concrete surrounding the interface and the resulting damage of the interface. Generally, the interface tends to experience brittle failure in which damage propagates suddenly. However, such failure could practically not be observed in specimen DCFP, where damage occurred proportionally to the size of the applied moment. In addition, the extent of damage being relatively small, it can be asserted that the coarse sand coating and FRP connector chosen to compose FRP and concrete for the PFC deck bears sufficient applicability.

3.3. Residual Performance

Static loading test was conducted to examine the residual performance of the two specimens that did not experience fatigue failure. Loading was applied until failure at speed of 0.5 mm/min. Similarly to the static loading specimen, circular cracking with the center located at the loaded portion was observed according to the increase of the load in a partial zone at the top of the deck. Finally, punching failure occurred as shown in Figure 11.

Figure 12 plots the load-deflection and load-crack width curves measured during the static loading tests of the fatigue and static test specimens. Compared to specimen DCS, specimen DCFP exhibits a failure strength smaller by about 5.6% while specimen DCFR shows nearly identical behavioral pattern with a failure strength larger by approximately 10.9%. This verifies that the PFC deck is practically immune to the occurrence of fatigue damage.

Table 3 compares the design criteria and experimental results. The design criteria under service load state involve the deflection and crack width. The measured deflection ranging between 2.8 mm and 3.5 mm satisfies the deflection criterion of 4.625 mm (= L/800, L: span length). Moreover, the crack width occurring at the negative moment zone also satisfies sufficiently the allowable crack width of 0.3 mm.

The shear strength and flexural strength under ultimate state should also be checked. Since all the specimens experienced failure due to punching shear, the shear strength can be known but the flexural strength cannot be verified exactly. However, the flexural strength of these specimens is at least larger than the shear strength. Therefore, the value of the shear strength of the specimens can be assumed to be the minimum flexural strength, and this value can be set as a reference for comparison with the design criteria.

The PFC deck can be divided into two parts: the FRPconcrete composite structure and the reinforced concrete structure around the support. Accordingly, adequate strength reduction factors should be applied with respect to each of the structural type. For the reinforced concrete structure, strength reduction factors of 0.85 for flexure and 0.8 for shear were applied according to the Highway Bridge Design Code of Korea [10]. Besides, the following strength reduction factors were decided with reference to the ACI Committee 440 [11] for the FRP-concrete composite structure. For structures reinforced by FRP bar, it was decided to apply the strength reduction factor for the shear in reinforced concrete structure as it is. Accordingly, the strength reduction factor of 0.8 specified for reinforced concrete structures in Korea was applied. For the flexure, decision was made to apply a shear reduction factor ranging between 0.55 and 0.65 according to the failure pattern. Here, a value of 0.55 was applied for conservative purpose.

Table 3 arranges the values of the reduced strengths of the specimens. The reduced shear strengths were computed by multiplying the shear failure strength of specimen DCS (1,098 kN), DCFP (1,037 kN) and DCFR (1,218 kN) by a strength reduction factor of 0.8. The comparison of these reduced strengths with the design strength relative to shear of 249.6 kN shows that the specimens satisfy sufficiently the shear design criterion. In addition, the minimum flexural strength determined with reference to the shear strength is also seen to satisfy the design criteria even when the flexural reduction factor is applied.

These results show that the PFC deck satisfies the design criteria even after 2 million pulsating or rolling fatigue cycles. This indicates that the PFC deck offers sufficient endurance to fatigue.

4. Conclusions

The durable performance of the precast FRP-concrete composite (PFC) deck was investigated through pulsating fatigue and rolling fatigue tests. Three types of fullscale specimens were fabricated: a reference specimen for the evaluation of the static performance, a pulsating fatigue test specimen and a rolling fatigue test specimen. The results of the fatigue tests conducted using a vertical load corresponding to the service load weighted with a safety factor of 20% showed that the pulsating and rolling fatigue specimens did not fail even after 2 million cycles. Thereafter, static loading tests were performed in order to evaluate the residual performance of the two types of specimens that did not experience fatigue failure. The results revealed that the fatigue specimens exhibited failure strength similar to that of the static test specimen. In addition, the comparison with the design criteria indicated that the PFC deck satisfied the serviceability criteria involving the deflection and crack width as well as the strength criteria that are the shear strength and flexural strength. The coarse sand coating and FRP connector applied to achieve the composition of FRP and concrete were seen not only to provide sufficient fatigue performance but also to offer features preventing sudden propagation of failure. Consequently, it can be concluded that the PFC deck exhibits sufficient endurance to fatigue.

5. Acknowledgements

The “Development of deck systems for hybrid cablestayed bridge” project, an in-house project of the Korea Institute of Construction Technology has supported this research. The authors express their gratitude for the supports.

6. REFERENCES