Experimental study on the post-tensioned unbonded prestressing UHPC prefabricated retaining blocks for highway bridges

Traditional sacrificial concrete shear keys are generally employed in highway bridges to restrain the excessive transverse displacements of the bridge superstructures and protect the substructures from being seriously damaged under the earthquakes. However, the post-earthquake rehabilitation or reconstruction of the conventional concrete shear keys are difficult and inconvenient. To this end, this paper proposed and designed a novel modified post-tensioned unbonded prestressing ultra-high-performance-concrete (UHPC) prefabricated retaining block (UHPC-PRB) structure with the straight joint connection. The novel UHPC-PRB structure consisted of a prefabricated UHPC retaining block, a prestressing anchorage system, and a cast-in-situ cap beam. The prefabricated UHPC retaining block was installed to the cast-in-situ cap beam using the unbonded prestressing tendons. The proposed UHPC-PRB structure is supposed to be effective in not only providing the same functionality as that of the traditional concrete shear keys, but also is expected of simplifying the retrofit procedure and reducing the rehabilitation expense simultaneously. To investigate the seismic behavior of the proposed UHPC-PRB structure, four test specimens were designed and the pseudo-static tests were carried out. The damage process, displacements, strains, and stress states of the specimens were studied. In addition, the influences of the loading height, thickness of the UHPC retaining block, and the initial tension of prestressing tendons on the seismic performance of the proposed UHPC-PRB structures were also examined. Finally, the experimental results indicated that, (i) the proposed UHPC-PRB structures had excellent lateral deformation capacity and self-resetting ability; (ii) the initial tension of prestressing tendons had significant effects on the lateral displacement, critical rotational load, and horizontal load-carrying capacity of the UHPC retaining blocks; (iii) the prestress loss of prestressing tendons had certain effect on the development of the bilinear analytical model of the UHPC retaining blocks, but this effect could be ignored in practical highway bridges; and (iv) increase of the horizontal loading height could reduce the critical rotational load and load-carrying capacity of the UHPC retaining blocks.


Introduction
Traditional concrete retaining blocks (shear keys) have been popularly used as the seismic fuses to protect the bridge superstructures from severe damage under the potential earthquakes.However, according to the previous earthquake damage investigations, the failure of concrete retaining blocks is known as one of the most frequent damages in highway bridges after the seismic events.For example, in 2008 Wenchuan earthquake, the observed damage rate of the concrete shear keys is up to 16.8%, which is significantly larger than 2.4% of the damage rate of the bridge piers (Chen 2012;Han et al. 2009;Li et al. 2008).In addition, numerous concrete shear key damages were also observed after the 2007 Pisco Peru earthquake (O'Connor et al. 2006) and the 2010 Chile earthquake (Kawashima et al. 2011).
To better understand the seismic behavior of the sacrificial concrete shear keys or the retaining blocks, many researchers have conducted both the experimental and theoretical studies.For example, Bozorgzadeh et al. (2006), Megally and Silva (2003), Silva et al. (2009) conducted a series of quasi-static experiments to investigate the possible failure modes of the concrete shear keys, and they also studied the effects of various design parameters, such as the volume ratio of the longitudinal reinforcements, strength of the horizontal tie reinforcements, and concrete strength on the seismic behavior of the concrete shear keys.These studies illustrated that the diagonal shear failure and sliding shear failure are the two most common failure modes of the concrete shear keys after the earthquakes.In particular, they proposed an idealized analytical model for the numerical analysis of the exterior concrete shear keys by separating the contributions made by the concrete and steel components.Based on the analytical model suggested in the aforementioned studies (Bozorgzadeh et al. 2006;Megally and Silva 2003;Silva et al. 2009), Goel and Chopra (2008) illustrated the role of concrete shear keys play in improving the seismic performance of highway bridges crossing the fault-rupture zones, by simplifying and improving the analytical model.In addition, Kaviani et al. (2012), Wilches Están et al. (2017), Xiang and Li (2018), Wu et al. (2018) investigated the influences of concrete shear keys on the seismic behavior of different types of highway bridges based on their suggested analytical models.Moreover, Kottari et al. (2020) recently proposed a design method for a more predictable failure mechanism governed by the horizontal sliding of the concrete shear keys.After the 2008 Wenchuan earthquake, many scholars in China also conducted many experimental studies on the concrete shear keys according to the practical situation of the structural forms and the reinforcement characteristics of many highway bridges (Han et al. 2020(Han et al. , 2017;;Xu and Li 2016;Zheng and Tang 2013).These experiments suggested that the influences of those factors, such as the spacing and volume ratio of the longitudinal reinforcements, vertical reinforcements, and horizontal tie reinforcements, as well as the loading position on the seismic design of the concrete shear keys should be carefully taken into account (Han et al. 2020(Han et al. , 2017;;Xu and Li 2016;Zheng and Tang 2013).In summary, the above-mentioned studies lay a good foundation for the bridge owners and engineers to better understand the seismic performance of the concrete shear keys.Meanwhile, these studies also highlight the importance of the reasonable seismic design of the concrete shear keys that are widely applied in bridge structures.
For the bridge structures in practice, traditional concrete shear keys were generally casted together with the cast-in-situ concrete components (i.e., the abutment stem wall or cap beam).Therefore, with this regard, these concrete shear keys can be also termed as the monolithic shear keys.Damages of the monolithic shear keys would completely destroy their mechanical connection to the abutment stem wall or pier cap beam of the bridge.Generally, these damages could be repaired or strengthened by excision, steel replanting, and reconstruction (Xu et al. 2020).However, the internal damage degree of the cast-insitu concrete components is not clear and the reconstruction operation space for the newbuilt concrete shear keys is very limited.Therefore, once the traditional concrete shear keys are seriously damaged during the seismic events, it might be complicated, time-consuming, and costly to repair and retrofit them.In recent years, the seismic resilient bridge structures (Dong et al. 2022), which could restore the structural function immediately after a strong seismic event without significant rehabilitation or strengthening, have attracted the attention of many scholars and engineers worldwide (Kilanitis and Sextos 2018;Pang et al. 2021).As a consequence, some replaceable retaining blocks with better seismic performance are proposed accordingly to replace or reinforce the traditional monolithic concrete shear keys.For example, the steel retaining blocks installed with the embedded bolts (Deng et al. 2014;Li et al. 2016), the attached steel retaining block with the negative Poisson's ratio (Zhang et al. 2023), the movable shear keys with suitable slide ability (Kappos 2019;Mikes and Kappos 2023), and the prefabricated retaining blocks installed with the posttensioned prestressing tendons (Kottari et al. 2016;Wu et al. 2019;Xiang and Li 2016) were suggested and their seismic behavior was comprehensively investigated.In particular, the unbonded post-tensioned prestressing tendons have been proved to be beneficial for improving and retrofitting of the bridge structures or components under the earthquakes (El-Hawat et al. 2022;Markogiannaki and Tegos 2015;Ou et al. 2010).
Moreover, to deal with the issues of monolithic concrete shear keys (i.e., their postearthquake rehabilitation or reconstruction are difficult and inconvenient), Kottari et al. (2016) proposed the concept of concrete shear key (retaining block) with the post-tensioned steel rods for the bridge abutment.The feasibility and beneficial features of the proposed post-tensioned concrete shear key were verified by conducting the experimental studies.However, the normal concrete (NC) cannot provide sufficient bearing capacity to the retaining block.In other words, the outstanding deformation capability and self-resetting ability of the NC retaining block with steel rods is at the expense of its lateral restraint strength.Thus, to address this, Wu et al. (2019) suggested a replaceable prefabricated ultra-high-performance-concrete (UHPC) retaining block (UHPC-PRB) with the stronger post-tensioned prestressing tendons to better employ the superior mechanical properties of UHPC (Su et al. 2017;Li et al. 2022;Ye et al. 2022Ye et al. , 2023;;Li et al. 2023a, b).Additionally, to study the seismic behavior of the suggested UHPC-PRB structures with the mortisetenon joint, Wu et al. (2022) conducted a series of quasi-static loading tests on different retaining block specimens.Their experimental results revealed that the suggested UHPC-PRB structures exhibited much better deformation capacity and self-resetting ability.The final damage modes of the UHPC retaining blocks were slighter, and the corresponding load-carrying capacity was much higher than that of the NC ones.However, for both the NC retaining blocks with the platform (i.e., the horizontal plates used as pin connections in Kottari et al. (2016) or the UHPC retaining blocks with the mortise-tenon joint connection in Wu et al. (2022), there are multiple contact points in the rotation process of the retaining block, under the action of large horizontal loads (i.e., the horizontal seismic actions) (as shown in Fig. 1a).Due to the influence of normal resistance and friction effect, it is difficult to quantify the self-resetting function and establish the suitable analytical model for the seismic design of the concrete shear keys.Furthermore, as shown in following Fig. 2a, the platform could provide the vertical bearing effect during the practical construction and the function of pin connections.Similarly, the mortise-tenon joint (Fig. 2b) can also provide the same function.However, the retaining block is relatively small and light, it is not necessary to design the platform and the additional mortise-tenon joint for bridges in practice.In other words, the structural forms of the platform and mortise-tenon joint connection are complicated for construction.Consequently, Wu et al. (2022) further suggested another modified UHPC-PRB structure with the straight joint connection.For instance, Fig. 1b-d shows the schematic diagrams of the modified scheme and the comparison of simulation results based on the finite element (FE) analysis.Obviously, the residual displacement of the modified UHPC-PRB structure was much smaller.In addition, it can be observed that the damage to the cap beam was negligible (Wu et al. 2022).
Theoretically, both the replaceable retaining block and cap beam could be treated as the elastic components in the seismic design, which is conducive to the establishment of a reasonable and simple analytical model for the concrete shear keys.Thus, in this study, four modified UHPC-PRB structures with the straight joint connections were designed and experimentally studied through the pseudo-static loading tests.The functional feasibility and seismic performance of the proposed UHPC-PRB structures were experimentally investigated and verified.Moreover, the influences of the loading height of the horizontal load, thickness of the PRB component, and the initial tensile stress of the prestressing tendons on the seismic behavior of the proposed UHPC-PRB structures were also analyzed and discussed.Finally, based on the experimental observations and discussions, several main conclusions and possible future studies for the proposed novel UHPC-PRB structures were presented.

Description of the specimens
Schematic diagrams of the proposed modified UHPC-PRB structure for highway bridges and the designed specimen are shown in Fig. 3.As seen from Fig. 3, two precast UHPC-PRB components were installed to the cast-in-situ cap beam through the transverse prestressing anchorage system (i.e., the unbonded prestressing tendons, anchor, and anchor plates, etc.).Both the PRB component and the end section of the cap beam were made from UHPC.The corresponding splicing surface was vertical.Generally, a suitable gap was designed to allow the bridge superstructure to move in the transverse direction under the temperature loads or earthquakes.This paper would conduct the experimental studies on the mechanical properties of the modified UHPC-PRB structures under the action of the horizontal cyclic load using the pseudo-static loading tests.
Specimens of the cap beams and four UHPC-PRB structures were designed and constructed in the laboratory.The designed size of the cap beam was 2200 mm × 300 mm × 800 mm (length × width × height).Two thin-walled rectangular hollow steel tubes were embedded in the center of the cap beam as the post-tensioned prestressing ducts.The thickness of the rectangular steel tube was 2.5 mm and its cross-sectional size was 40 mm × 300 mm. Figure 4 presents the dimensions of the test specimens.As shown in Fig. 4, UHPC was used to cast the end section of the cap beam within a distance of 160 mm, but the remaining of the cap beam  and the foundation were cast by using normal concrete (NC).Several corrugated pipes with the diameter of 60 mm were embedded on both sides of the foundation base to work as the anchors to fix the foundation.Four UHPC-PRB specimens were identified as the specimens S1, S2, S3, and S4, respectively.The height of the UHPC-PRB specimens was 1300 mm.The component thickness (B) of specimens S1 and S2 was 300 mm, whereas that of specimens S3 and S4 was 400 mm.Similarly, two thin-walled rectangular hollow steel tubes were embedded in the lower center of the PRB component as the post-tensioned prestressing ducts.The corresponding cross-sectional size was 40 mm × 240 mm.The net distance between two ducts was 50 mm.To avoid the possible contact friction between the prestressing tendons and the internal wall of the steel duct during the rotation process of the retaining block, the height of steel tube in cap beam is designed larger than the height of steel tube in the retaining block.
In addition, the steel reinforcements had three different diameters of 8, 10, and 12 mm, respectively.The adopted prestressing tendon is the standard Φ S 15.2 tendon, which is widely used in bridge engineering in China (GB/T-5224-20142014, JTG-3362-20182018).Grade  and 5, respectively.Moreover, Table 1 lists the detailed design parameters of the four UHPC-PRB specimens and the cap beam (CB).
Table 2 lists the elasticity modulus, yield strength, ultimate strength of the steel reinforcements and prestressing tendons of the specimens.Since the UHPC component was cast in laboratory, the compressive strength of the adopted Grade 150 UHPC was obtained from 100 mm × 100 mm × 100 mm concrete cubic tests (i.e., the coupon tests).Table 3 lists the measured results of the tested coupon samples.

Production, test setup, and loading protocol
Figure 6 shows the production process of the test specimens.As shown in Fig. 6, two kinds of concrete materials (i.e., NC and UHPC) were used, and their curing conditions were different.Thus, during the production procedure of the specimens, the NC parts were cured naturally at room temperature for 28 days.After the NC parts were cast and cured, the UHPC parts were cast and steam cured in a high temperature of 95 ~ 100 °C for 48 h, as required by the material instructions of Grade UHPC-150 premix compound.In addition, it should be mentioned herein that, there was a vertical construction joint between the UHPC specimens and cap beam.As shown in Fig. 6c, the precast PRB specimens were installed and assembled to the fixed cap beam by using the unbonded prestressing system (i.e., the Φ S 15.2 prestressed steel strand composed of seven high strength steel wires, which has the standard tensile strength of 1860 MPa).After all the horizontal prestressing tendons were tensioned gradually to their designed initial tension forces, the formal quasi-static loading tests (lateral loading) in simulating the horizontal seismic behavior of the specimens were performed.Figure 7 shows the schematic diagram of the test setup.As seen from Fig. 7, to simulate the horizontal seismic loads transferred from the bridge superstructure to the UHPC-PRB structure, the servocontrolled hydraulic loading system, consisted of the hydraulic actuator mounted on the reaction wall and an actuator-balancing device, was used to apply the monotonic lateral loading.The capacity of the actuator is 1000 kN and ± 300 mm.The precast cap beam and foundation were directly mounted vertically to a strong ground anchor beams using the post-tensioned bolts with a diameter of 50 mm.Some instruments including the displacement sensors, load cells, embedded strain gauges were used during the quasi-static loading tests to monitor the deformation of the retaining block, tensile forces of the prestressing tendons and strains of the reinforcements.As shown in Fig. 7, four displacement sensors, named SP1, SP2, SP3, and SP4, respectively, were installed externally to the UHPC-PRB specimens to monitor the horizontal displacements and rotations of the specimens during the loading process.In addition, the sensor SP1 was also used to verify the horizontal displacement of the loading point.In the non-tensioned end of the cap beam (i.e., the left side of the cap beam in Fig. 7), several pressure sensors were used to measure the tensile forces of the prestressing tendons, including the designed initial tensile forces and the later tensile force variations during the loading and unloading process.Moreover, strains of the prestressing tendons were monitored with the laterally arranged strain gauges.Figure 8 shows the positions and numbering of these installed strain gauges on the prestressing tendons, which were numbered from Y01 to Y03 for each prestressing tendon (tendons A to D), respectively.Furthermore, strains of some important vertical reinforcing bars were also monitored with the strain gauges.Figure 9 shows the detailed positions and numbering of these laterally arranged strain gauges on the vertical reinforcements, including eight strain gauges installed in specimens S1 and S2, as well as ten strain gauges installed in specimens S3 and S4, respectively.
A reversed horizontal loading protocol consisted of a series of the force-controlled cycles (as shown in Fig. 10) was used to applied the lateral loading.As seen from Fig. 10, before the formal loading process, all the test specimens were pre-loaded initially by four different loading levels at 5, 10, 15, and 20 kN, respectively.Then, during the formal loading procedure, only one loading cycle was performed on the specimens for each loading stage.The amplitude of the horizontal loading was increased by 10 kN for each increment interval between 20 and 70 kN.After that, the incremental load interval of each loading stage changed to 20 kN.Finally, the testing of specimens S1 and S2 would be stopped when the maximum stress of the upper prestressing tendon reached 80% of the yield strength.However, due to the specimens S3 and S4 were much stronger, the testing would not be stopped until the fracture of upper prestressing tendons.

Test observations
After the quasi-static tests, Fig. 11 shows the final status of the UHPC component and the prestressing tendons.For specimen S1, the UHPC retaining block had no visible damage and it tended to have strong self-resetting ability when the horizontal load increased from 0 to 90 kN.The first micro-crack appeared in the block body as the applied horizontal load exceeded 90 kN, but the visible micro-crack was closed completely after unloading.Afterwards, the first micro-crack did not expand significantly and in the meantime, no new cracks were observed.However, as the applied load increased from 130 to 170 kN, the initial micro-crack began to further expand and extended to the main crack, which could not be closed after unloading.Also, in this stage, many new developed micro-cracks were observed around the main crack, and their widths increased rapidly.In addition, in the subsequent loading stage, as the applied load increased to 190 kN, many visible cracks appeared on the surface of the block body and a few vertical cracks began to appear in the compression zone.Finally, the loading process was stopped as the tensile force of the upper prestressing tendon reached 190.5 kN, which was close to 80% of the yield strength.Moreover, as seen from the final failure mode of specimen S1 given in Fig. 11a, specimen S1 could basically return to the initial position, but the block body and cap beam could not fully fit together completely.It should be noted that the appeared cracks in specimen S1 were concentrated around the prestressing ducts of the UHPC retaining block.This is mainly because the prestressing ducts of the UHPC-PRB structure were large, leading to the width of the UHPC-PRB structure was very narrow.
Similarly, the damage mode of specimen S2 in the initial loading stages was similar to that of specimen S1, but the first micro-crack appeared as the applied horizontal load exceeded 110 kN.When the applied load increased from 170 to 210 kN, the main crack formed and extended gradually to the compression zone and its width increased rapidly.This developed main crack could not be closed completely after unloading.Subsequently, as the load increased to 240 kN, many visible cracks appeared on the surface of the block body and a few vertical cracks were also observed in the compression zone.Then, as the tensile force of the upper prestressing tendon reached 200.8 kN, which was close to 80% of the yield strength (260 kN), the loading test was stopped for safety.In addition, as seen from the failure mode of specimen S2 in Fig. 11b, specimen S2 could also basically return to the initial position, but the UHPC retaining block could not fully fit together with the cap beam.The locations and cause of the formed cracks in specimen S2 were similar to that of specimen S1.
For specimen S3, the UHPC retaining block had no damage and exhibited strong self-resetting ability before the applied horizontal load increased to 150 kN.As the load exceeded 150 kN, the first visible micro-crack appeared on the surface of the block.During the subsequent process, no other new visible micro-cracks appeared except for the further development and propagation of the first visible crack, and it did not extend to form the main crack.The first visible crack could be completely closed after unloading.Then, as the load increased to 210 kN, the tensile forces of these two upper prestressed tendons reached 213.6 kN, which exceeded 80% of the yield strength (260 kN), specimen S3 only had the relatively slight damage, so it could return to its initial position after unloading.Thus, the loading test continued to the next stage of load, but the corresponding load increment interval was changed to 10 kN.As the applied horizontal load increased to 218.2 kN, the loading was stopped for safety.Because the fracture of two upper prestressing tendons were observed due to the distinctive sound produced during the fracture of tendons accompanied by a drop to the measured load.In this case, specimen S3 could not return to the initial position after unloading, but fell downward 10 cm vertically, as shown in Fig. 11c.
Likewise, as seen from Fig. 11d, for specimen S4, the damage in the earlier loading stages was basically similar to that of specimen S3.As the applied load increased to 230 kN, the tensile forces of two upper prestressed tendons were 217.6 kN, which exceeded 80% of the yield strength.However, specimen S4 only exhibited slightly damage mode, it could return its original position after unloading.Then, the loading test for this specimen continued to proceed the next loading stage, but the corresponding load increment interval was changed to 6 kN.Finally, as the load reached 236.2 kN, the loading test was stopped for safety.Because the upper prestressing tendons fractured and the applied load dropped significantly to 221.2 kN.However, compared with that of specimen S3, due to the larger initial tensile forces of the prestressing tendons in specimen S4, the two lower prestressing tendons with the elastic tension force could still provide enough resistance to overcome the self-weight of the UHPC retaining block.Therefore, specimen S4 did not fall downward after unloading, as shown in Fig. 11d.Furthermore, Fig. 11e shows the failure mode and fracture location of the upper prestressing tendons (PTs).As seen from Fig. 11e, several steel wires of the upper prestressing tendons of specimens S3 and S4 were observed to be pulled off at the end sections.Figure 11f shows the final state of the connection contact surfaces between the vertical UHPC retaining block and the cap beam.It can be observed that both the connection contact surfaces of the specimens were undamaged and the cap beam were being protected well.Overall, the new proposed UHPC-PRB structures were demonstrated to have small residual displacements and excellent self-centering ability.Additionally, the new structural form with the straight joint connection would not cause serious damage to the cap beam during the process of rotation, which is beneficial and convenient to the seismic design and retrofit of the UHPC retaining blocks.

Strains of the vertical reinforcements
Figure 12 shows the obtained maximum strains of the vertical reinforcements in different specimens.As seen from Fig. 12, the maximum strain of all the vertical reinforcements monitored in the test did not exceed the yield strain.It should be noted that, since the reinforcements Fig. 12 The maximum strains for the vertical reinforcements in different specimens with the greatest tensile stress were the first row of vertical bars, the other ones should be elastic, and the measured results were not given in this figure.On the other hand, as shown in Fig. 12, the maximum strain of the rebar appeared in specimens S1, and its strain was about 1500 με, which is very close to the yield strain (1675 με).In addition, the maximum tensile strain of the reinforcements in other specimens were much smaller than the yield strain, especially these two thicker specimens (S3 and S4) with bigger inertia moment in direction of anti-bending.Moreover, the strain situation of the vertical reinforcements of different specimens were consistent relatively well with their corresponding damage modes.For example, as shown in Fig. 11, there were several cracks in specimens S1 and S2, but the tiny cracks in the UHPC blocks of specimens S3 and S4 were negligible during the loading tests.It should be noted that, the maximum tensile strains of specimen S1 would be lower than that of specimen S2 due to the relative smaller horizontal load for specimen S1.However, Fig. 12 shows the inverse result, because the strain gages in specimen S1 were very close to the horizontal crack where the reinforcement strain was localized.

Stress of prestressed tendons
As illustrated in Fig. 9, four pressure sensors were installed between the anchorage plate and the side of the cap beam to monitor the stress states of different prestressing tendons during the loading tests.Since the tensile force of the lower prestressing tendon was smaller than that of the upper one, Fig. 13 shows the relationship between the applied horizontal load and tensile force of the upper prestressing tendon monitored by the pressure sensors.It can be seen that the tensile forces of prestressing tendons increased significantly with the increase of the applied horizontal load.In this study, the critical load that force the UHPC retaining block begin to rotate is defined as the critical rotational load (F 0 ).Obviously, when the horizontal load was less than F 0 , the tensile forces of the prestressing tendons tended to increase at a relatively small rate, and the corresponding residual tensile forces remain unchanged after unloading.However, as the applied horizontal load exceeded the critical rotational load of the retaining block, the increase rate of the tensile forces of the tendons increased significantly.Meanwhile, the corresponding residual tensile forces of the tendons gradually decreased after unloading.When the tensile forces of the tendons were greater than the initial tension force, the prestress loss caused by the anchorage retraction would occur.As seen from Fig. 13, the greater the applied horizontal load, the greater prestress loss in prestressing tendons would occur in specimens.As a result, the final tensile forces of the upper prestressing tendons for specimens S3 and S4 were zero after the loading test.This could be mainly attributed to the prestress loss and fractured of steel wires in the upper prestressing tendons of specimens S3 Fig. 13 Comparison of the development of tensile forces of prestressing tendons for different specimens and S4.In addition, as seen from Fig. 13a, the increase rates of the tensile forces of the upper prestressed tendons decreased with the decrease of the loading height.Moreover, as shown in Fig. 13b, when other conditions were the same, variations of the tensile forces of the tendons were independent from the block thickness.In other words, thickness of the retaining block did not affect variations of the tensile forces of the prestressing tendons.
Furthermore, as shown in Fig. 9, some strain gauges were attached on the tendons to monitor their strain states during the tests.Figure 14 shows the measured strain time-histories of prestressing tendons for different specimens.As seen from Fig. 14, for all the specimens, strains of the upper two prestressing tendons were greater than that of the lower ones, and the strain variations of the tendons in the same row under the applied horizontal Fig. 14 The measured strain time-histories of prestressing tendons for different specimens Fig. 15 Comparison of the load-displacement hysteresis curves of different specimens load were almost the same.According to the obtained strain time-histories of prestressing tendons as shown in Fig. 14a and b, the maximum strains of the tendons in specimens S1 and S2 were smaller than the yield strain (around 9000 με).However, the maximum strains of the tendons in specimens S3 and S4 were larger than the yield strain, as observed from Fig. 14c and d.Meanwhile, strains of all prestressing tendons would increase with the extension of the loading time due to the increase of the applied horizontal load.However, the residual strains of the tendons after unloading would decrease gradually with the increase of the applied load, which was consistent with the monitored tensile force variations.

Influences of the design parameters
Figure 15 shows the comparisons of the load-displacement hysteresis curves of different specimens.It should be noted that the displacement was measured at the loading point.In specific, the load-displacement hysteresis curves of two different specimens were plotted in the same figure to study the influences of various design parameters.For instance, Fig. 15a presents the comparison of the load-displacement hysteresis curves of specimens S1 and S2.It can be seen that the critical rotational loads (F 0-S1 and F 0-S2 in Fig. 15a) of specimens S1 and S2 were 72 and 95 kN, respectively.Under the same applied horizontal load, the lateral displacement of specimen S1 was larger than that of specimen S2.The rotational stiffness of specimen S1 began to degrade as the horizontal load reached 170 kN, whereas the stiffness degradation effect of specimen S2 were not observed until the horizontal load reached 240 kN.In summary, the critical rotational loads, the rotational stiffness, and the load-carrying capacity of the proposed UHPC-PRB structures would decrease with the increase of the loading height (h F ), where h F is defined as the distance between the loading point to the rotation center of the retaining block, as shown in Fig. 15a.In addition, the horizontal deformation capacity of the UHPC-PRB structures would increase with the increase of h F .
Figure 15b shows comparison of the load-displacement hysteresis curves of specimens S1 and S3.As seen from Fig. 15b, the critical rotational loads (F 0-S1 and F 0-S3 ) of specimens S1 and S3 were 72 and 74 kN, respectively.The difference in the critical rotational loads of specimens S1 and S3 was really small.Actually, it can be seen from Fig. 15b that the load-displacement hysteresis curves of specimens S1 and S3 were basically coincident in the early loading stage.Thickness of the UHPC retaining block (e.g., defined as B in Fig. 15b) tended to have little influence on the critical rotational load and rotational stiffness of the UHPC-PRB structures.As shown in Fig. 11, specimen S1 exhibited obvious cracking damage mode at the side of UHPC retaining block after the test.The final failure of specimen S3 was due to the fracture of the upper prestressing tendons, but the UHPC retaining block only exhibited the slight micro-cracking.In other words, the peak horizontal load of the hysteresis curve did not represent the ultimate strength of specimen S3.It could be predicted that the ultimate strength of specimen S3 could be further improved by increasing the number of the installed prestressing tendons.Moreover, increasing the thickness of the UHPC retaining block could improve its ultimate strength, thus enhancing the load-carrying capacity of the UHPC-PRB structures.However, the influences of thickness variation of the UHPC retaining block on the critical rotational load and rotational stiffness of the proposed UHPC-PRB structures were negligible.
Figure 15c shows the comparison of the load-displacement hysteresis curves of specimens S3 and S4.It can be seen that the critical rotational loads (F 0-S3 and F 0-S4 ) of specimen S3 and S4 were 74 and 110 kN, respectively.The critical rotational load of specimen S4 was much greater than that of the other three specimens.When several steel wires in the upper prestressing tendons were fractured, the lateral displacement of specimen S4 was smaller than that of specimen S3.This could be attributed to the following two reasons.On the one hand, the increase of the initial tension (T 0 in Fig. 15c) would lead to the increase of the required horizontal load to break the rotational equilibrium of the retaining block system.On the other hand, the lateral deformation capacity of the UHPC-PRB structure is mainly determined by the maximum elongation of prestressing tendons.An increase in the initial tensions of prestressing tendons would result in the increase of the initial elongation.However, the total elongation of a given prestressing tendon in the limit state is constant.Thus, the increase of the initial prestress would reduce the deformation capacity of the UHPC-PRB structures.In addition, it can be seen from Fig. 15c that the bearing capacity (i.e., the maximum horizontal load determined by the fracture of the upper prestressing tendons) of specimen S4 was slightly greater than that of specimen S3, due to the difference in the initial tension forces for specimens S3 and S4.The initial tensions of prestressing tendons could not only affect the critical rotational load and rotational deformation capacity of the proposed UHPC-PRB structure, but also affect their ultimate load-carrying capacity.

Discussions
In this study, the proposed UHPC-PRB structure is a combined system consists of the cap beam, retaining block, prestressing tendons, and anchorage device, etc. Theoretically, all other members of the composite system except the cap beam should be replaceable during the service life of the bridge, especially those cases when the bridge is subjected the unpredictable earthquakes or other impact loads.Meanwhile, to simplify the seismic design process, the failure of the prestressing tendons should be considered as a control factor, which is different from the design philosophy of the traditional sacrificial concrete shear keys.Based on the aforementioned design philosophy and the test results (e.g., failure modes of the specimens as shown in Fig. 11), the UHPC-PRB structure has sufficient strength and stiffness under the horizontal seismic load and it can be rotated as a rigid body around one center point.Figure 16 shows the basic rotating mechanism and 16 Schematic of the rotation mechanism and theoretical model for UHPC-PRB structure the idealized analytical model of the proposed UHPC-PRB structure under the cyclical horizontal loads.
As shown in Fig. 16, the UHPC body starts to rotate when the seismic inertia force or horizontal load (F) exceeds the critical rotational load (F 0 ).Due to the strong restoring force contributed by the prestressing tendons, the UHPC body could return to the original position after unloading.Obviously, the rotation mechanism of the UHPC-PRB structure is very simple.When the horizontal resisting force (F) is relatively small (F ≤ F 0 ), the PRB component could be assumed as a short cantilever beam which was fixed at the prestressed anchorage point.According to the differential equation of the deflection line, stiffness of the UHPC-PRB structure (K d1 in Fig. 16b) could be calculated by where E r is the elastic modulus of UHPC; I r , h r , b r , and t r are the moment of inertia, effective height, width, and thickness of the PRB component, respectively.Then, as the horizontal resisting force (F) increases to a certain extent (i.e., F > F 0 ), the equilibrium state of the rigid body block could maintain under the joint action of the horizontal and vertical resisting forces.Based on the equilibrium conditions, the following balance equations can be established as where M o is the total moment of the force around the rotation point O; h F , h T , and L w are the distances from the loading point of collision, prestressing anchorage point, center of gravity of the PRB component, respectively.G is the self-weight of the PRB component.T is the resultant tensile force of the prestressing tendons, which can be determined by where T 0 , E s , A , L, and ΔL are the initial tension, elastic modulus, cross-sectional area, stress-free length and elongation of prestressing tendons during the rotation process, respectively.As shown in Fig. 16b, ΔL and X 2 (rigid body rotation deformation) satisfy the geometric relationship as Finally, by combining the above equations, the horizontal resisting force (F) can be expressed as where F 0 is the initial critical rotational load and K d2 is the rotational stiffness of the UHPC-PRB structure, respectively.Figure 16b shows the idealized analytical model of the (1) UHPC-PRB structure.Obviously, some important design parameters (e.g., K d1 , K d2 , and F 0 ) can be calculated based on above derived formulas.With the increase of the horizontal resisting force, the rotational equilibrium state of the UHPC-PRB structure could always be maintained until the maximum tensile force of prestressing tendons reached the limit value (i.e., yielding strength).However, according to the test results, it is difficult to calculate the ultimate rotational load due to the fracture of several steel wires in prestressing tendons during the rotation of the proposed UHPC-PRB structure.Furthermore, it should be noted that, this study demonstrated that the prestress loss of prestressing tendons was considerable when the horizontal load increased to a very large value.This is because the anchor devices used to secure the free end of prestressing tendons is a tapered anchor.The deformation and retraction of the anchorage system would occur during the rotation of the proposed UHPC-PRB structures under the horizontal load.According to GB-50010-2010GB-50010- (2010)), the prestress loss ( Δ ) obtained by the anchorage device deformation and prestressing retracement could be calculated as where δ is the total shrinkage of the anchorage system.For the steel strand tapered anchorage system, the value of δ could take 5 mm and 6-8 mm, respectively, when there is or no pressure applied on the tapered anchor wedge (GB-50010-2010 2010).E s and L are the elastic modulus and effective length of prestressing tendons, respectively.According to Eq. ( 6), the prestress loss is related to the total shrinkage of the anchorage system and the effective length of the tendons.For different initial tension of prestressing tendons (i.e., T 0 = 65, 85, and 208 kN, respectively), the relationship between the effective length of the tendons and the prestress loss ratio can be plotted in Fig. 17a.Obviously, the shorter the effective length, the greater the prestress loss.For example, when the initial tension was 85 kN, the prestress loss ratio was up to 72.6% for the prestressing tendon with the 2.2 m length, but it would decrease to 8% when the effective length increased to 20 m.On the one hand, the actual width of the bridge deck and cap beam in highway bridges is much larger than 2.2 m.On the other hand, as shown in Fig. 17a, the prestress loss could be also reduced by increasing the initial tension force.In fact, another more effective anchorage device, such as the double-tensioned prestressing anchorage system (Zhang et al. 2021), (6) Δ = ⋅ E s L Fig. 17 Prestress loss effect and its influence on the retaining blocks could be applied as an alternative device to greatly reduce the influence of prestress loss.Consequently, the influence of prestress loss on the structural functionality of the proposed UHPC-PRB structures could be ignored in practical highway bridges.Additionally, Fig. 17b shows the idealized bilinear model of the proposed UHPC-PRB structures without considering the prestress loss effect.Apparently, this theoretical model could provide a simple mechanical model to predict the seismic performance of the proposed UHPC-PRB structures in the seismic design of highway bridges.

Conclusions and future work
This study experimentally investigated the mechanical behavior, working mechanisms, and structural properties of the proposed UHPC-PRB structures with the straight joint connection through the pseudo-static loading tests.The failure process of the designed large-scale test specimens, the stress states of the steel reinforcements, and that of the prestressing tendons were monitored and analyzed.Influences of the loading height, thickness of the UHPC retaining block, and the initial tension forces of the prestressing tendons on the seismic behavior of the proposed UHPC-PRB structures were also analyzed and discussed.Based on the experimental investigations, several conclusions are summarized as following, (1).The experimental results indicated that the proposed novel UHPC-PRB structures had good lateral deformation capability and self-resetting ability, and they could provide effective transverse restraints to the bridge superstructures.Therefore, as a supplement to the existing traditional concrete shear keys, the proposed novel UHPC-PRB structures could be used as the effective and reasonable structural components of the transverse anti-seismic system for highway bridges, to help them to withstand the seismic events.In particular, these three components (e.g., the prefabricated UHPC retaining block, prestressing anchorage system, and cast-in-situ cap beam) in the novel UHPC-PRB structure could be produced independently, but they could be assembled together to withstand to the horizontal seismic loads.Also, the proposed UHPC-PRB structures could be replaced easily and conveniently, which could provide fast, convenient, and efficient seismic rehabilitation to highway bridges after the earthquakes.(2).The initial tension forces of the prestressing tendons could simultaneously affect the lateral deformation capacity, critical rotational load, and horizontal load-carrying capacity of the UHPC retaining blocks.The critical rotational load and bearing capacity of the UHPC retaining block increased with the increase of the loading height of the applied horizontal load.(3).During the seismic design of the proposed UHPC-PRB structures, increasing the thickness of the UHPC retaining blocks could improve their strength and reduce their damages.Meanwhile, the critical rotational load and displacement demand of the UHPC retaining blocks could be determined by reasonably adjusting the initial tension forces of prestressing tendons and the loading height of the applied horizontal load.
The main objective of this study is to investigate the anti-seismic feasibility and mechanical properties of the proposed novel UHPC-PRB structures with the straight joint connections.It should be noted that, unlike the traditional sacrificial concrete shear keys with weak strength, the proposed replaceable UHPC-PRB structure is actually a member with the superior stiffness and strength.Moreover, the proposed novel UHPC-PRB structures could protect the bridge abutments or bridge piers from severe damages by losing part of the prestress, and they could also provide a fusing function as the traditional sacrificial concrete shear keys under the actions of unpredictable earthquakes.However, more experimental tests and theoretical studies regarding the work mechanisms of this fuse function of the proposed novel UHPC-PRB structures should be further carefully investigated in future.

Fig. 3 Fig. 4 Fig. 5
Fig.3The modified UHPC-PRB structure used in highway bridges and the test specimens

Fig. 6 Fig. 7
Fig. 6 Production of the test specimens

Fig. 8
Fig.8Locations of the strain gauges arranged on the prestressing tendons (unit: mm)

Fig. 9
Fig. 9 Locations of the strain gauges arranged on the vertical reinforcements

Fig. 11
Fig. 11 Final status of four UHPC PRB structures and other components

Table 1
Main design parameters of the retaining block specimens and cap beam Note: CB = cap beam; Φ S 15.2 = the type of prestressed tendons that through the prestressed ducts

Table 2
Material properties of the reinforcement and prestress tendons