1. Introduction
In freeway reconstruction and extension projects, widening is typically performed on both sides or on one side only [
1]. For long-span bridges crossing large rivers, the extension approach of building a new bridge on one side can be adopted to reduce the land occupation of the new bridge [
2]. The new bridge carries traffic in one direction, and the old bridge carries traffic in the other direction, as shown in
Figure 1.
However, for the existing bridge constructed with separate spans, the traffic lanes must remain separated on each span due to the disconnected ends of adjacent flanges. The problem of separated driving also occurs at the bridgehead position, which is prone to traffic accidents [
3]. It is necessary to connect the two separate spans of the existing bridge to ensure the safety of passing vehicles. Therefore, the concept of integral splicing for the existing bridges is proposed, as shown in
Figure 1. It can be defined as implementing connections between the adjacent flange plates of the existing box girder bridge (constructed with separate spans), which can realize direct vehicle crossing.
At the splicing position in an actual project, several layers are involved: the asphalt pavement layer, the integral layer, and the flange plate structure layer [
4]. Existing research results have primarily focused on these layers, and virous bridge connection schemes have been proposed.
In existing research, studies have focused on disconnecting these layers, typically by adopting expansion joints. This approach results in an unconnected bridge deck and an unspliced structure. Krishna Mochtar et al. [
5] conducted a literature review on the expansion capacity of different types of expansion joints and compared their unit price and installation time consumption. However, the expansion joints currently in use are primarily of the transverse type, which may not meet the requirements of the longitudinal expansion joints needed in this research. Li et al. [
6] applied longitudinal expansion joints in a short-span wide bridge to reduce deck cracking caused by the temperature effects and shrinkage. Additionally, Li et al. [
7] studied longitudinal and transverse cross-expansion joint structures in a short-span bridge extension project. This structure can also meet the requirements regarding the temperature effect and shrinkage, to some extent. Although longitudinal expansion joint structures can be selected for bridge expansion projects, they remain difficult to popularize. Because longitudinal expansion joints are parallel to the direction of the traffic and are mostly made of steel, problems, such as wheels becoming stuck or vehicle staggering and slipping, may occur when vehicles travel across them.
Conversely, connecting these layers should be adopted to ensure smoother traffic flows in actual projects. Depending on the width of the splicing joints, the connection schemes of the pavement layer (asphalt) and the integral layer or the flange plate structure layer (concrete) can be implemented. Wang et al. [
8] applied high-elasticity asphalt concrete in the longitudinal joints of bridge widening. Swanson et al. [
9] developed a material of silicone foam sealant for small-movement bridge expansion joints. Lu et al. [
10] proposed a type of polyurethane-modified asphalt to optimize bridge expansion joint interfaces. These new materials can be reasonably applied in widening bridges by virtue of their excellent deformation resistance. Therefore, the pavement layer connection schemes are limited to applications using small joint widths.
However, significant deformation differences may occur at the splicing position in long-span bridges. From a material perspective, whether these materials can accommodate large vertical deformation differences requires further investigation. Therefore, more studies have focused on structural connections. Tan et al. [
11] and Chen et al. [
12] studied different widening and splicing methods for medium-span and short-span bridges. These splicing methods offer valuable insights for splicing long-span bridges. Wu et al. [
13,
14] proposed different types of transverse splicing connection devices in the widening projects of a long-span continuous concrete box-girder bridge. These new solutions have been numerically analyzed, but not verified in actual projects. In addition, it is worth noting that studies have been conducted [
15,
16] to ensure structural splicing, specifically targeting uninterrupted traffic during splicing new and old bridges. Therefore, research on the splicing technology used for the separate spans of long-span existing bridges still lacks sufficient support, and needs further study.
In this research, the existing Xinfengjiang bridge (henceforth called ‘the old bridge’) served as the case study. The integral splicing design method and corresponding construction technology were studied.
Firstly, the splicing requirements of the old bridge were investigated by analyzing the relative deformation difference between its two separate spans from different dimensions. Based on the results comparison of the theoretical calculation, static-load test, on-site long-term monitoring, and random traffic flow simulation, an appropriate design basis of integral splicing the old bridge was also proposed.
Secondly, an integral splicing structure for the old bridge with separate spans was proposed, and its mechanical properties were tested. A pair of semi-loop reinforcement bars were incorporated into the joint connection segment of the proposed ISC-Structure. Ultra-high-performance concrete (UHPC) was specifically developed for this application. Three full-scale model specimens were subjected to four-point flexural loading tests. Additionally, a refined simulation model was developed for size optimization of the ISC-Structure.
Finally, the key technologies of integral splicing construction were studied based on actual site conditions. A layered and segmented construction scheme tailored to the ISC-Structure was formulated. Radar equipment was applied to determine the actual layout of the prestressing wire bundles, and a local concrete chiseling scheme for the flange plate ends was formulated. In addition, a movable construction platform was designed, and reinforcement planting methods were optimized. The reasonable concrete curing duration was also determined using on-site temperature monitoring.
This research of integral splicing design and construction technology presents the following novelties. Firstly, the combined application of a direct measurement method for deformation difference between the separate spans and a simulation method utilizing traffic flow data from the freeway at the bridge site can effectively improve the rationality of splicing requirement analysis. This combined approach has not been mentioned in the current research on bridge splicing, but they are particularly important. Secondly, a novel integral splicing composite structure (ISC-Structure) was proposed. This structure realizes integral splicing between the separate spans with limited intervals, and enhances the longitudinal mechanical characteristics of the old bridge. Thirdly, construction technologies and equipment are optimized to fulfill the demands of integral splicing. In the process of implementation, the end of flange plate containing transverse prestressed wire bundles was partially chiseled.
3. Splicing Requirements Analysis
In this section, various methods are adopted to clarify the actual needs of the integral splicing of the old box girder. The relative displacement differences between two separate spans of the old bridge and between the ends of the adjacent flange plates are compared and analyzed.
3.1. Theoretical Calculation and Static Load Test
The mid-span section of the second span was selected as the key section for analysis, which represents the maximum positive flexural moment condition of the middle span. A 3D simulation model of the old bridge was established, as shown in
Figure 6a. The theoretical calculation results obtained according to different versions of general specifications are presented in
Figure 6b.
As mentioned before, the old bridge was designed based on the old version of general specification JTJ 021-1989 [
17]. According to specification JTG/T J21-01-2015 [
19], the maximum test moment of the second span can be calculated based on JTJ 021-1989. Therefore, an arrangement of six standard test vehicles (36 t) is equivalently converted. The load efficiency ratio is equal to 1.03, which can meet the range of 0.95–1.05 specified in the specification JTG/T J21-01-2015.
In the process of the static load test, the method of step loading is adopted, which is divided into three levels to complete the loading of all vehicles. The bridge deformation is measured, and the deformation curve is drawn in
Figure 6b. The theoretical calculation results are compared with the static load test results.
By comparing the load test results with the theoretical calculation result (JTJ 021-1989), the structural verification coefficient is in the range of 0.48–0.54, which can meet the requirements of specification JTG/T J21-01-2015. This means that the old bridge is still working in an elastic state and has sufficient performance redundancy.
In addition, the theoretical value obtained by the new general specification JTG D60-2015 [
18] is compared in
Figure 6b. This result is slightly larger than that of the former version. This is mainly due to the fact that the vehicle load standard is slightly improved in the new general specification.
Actually, the static load test is carried out according to the most unfavorable stress principle of bridge structure. However, the vehicle load on the bridge has obvious randomness [
20]. Therefore, the static test results may not reflect the actual splicing needs of the old bridge.
3.2. On-Site Monitoring and Data Analysis
The deformation of the old bridge can be monitored by directly arranging the measuring points on site. If the measuring points are arranged at the end of the flange plate of the old bridge, the differential deformation at the splicing position can be directly measured. The mid-span section with the largest differential deformation is typically selected for the test, and the arrangement of the measuring points is shown in
Figure 7a. It should be noted that the new bridge has been built, but has not yet been opened to traffic during the period of on-site monitoring. Therefore, there are no other additional loads applied to the new bridge, except for the temperature effect, in a short period of time.
Differential deformations between the old and new bridges and between two separate spans of the old bridge are monitored in this test. The arrangement scheme between the old and new bridges is presented in
Figure 7b. At this position, a rigid cantilever frame with displacement meters is arranged at the edge of the flange plate of the new bridge. The deformation obtained can represent the deformation of a single half of the old bridge. Furthermore, the arrangement scheme between the halves of the old bridge is presented in
Figure 7c. At this position, a set of rigid support frames with displacement meters is arranged between adjacent guardrails of the old bridge. The deformation obtained can represent the deformation difference between the two separate spans of the old bridge.
From 26 April 2024 to 6 May 2024, a 12-day on-site monitoring was carried out with a sampling frequency of 2 Hz. Among them, from 1 May to 3 May is the toll-free period of the freeway, and the number of vehicles passing through the old bridge was significantly larger than that in other time periods, which can be used to characterize the dense traffic flow.
The effective monitoring results of each time point are plotted as time history curves, shown in
Figure 8.
In this curve, the deformation of the bridge caused by temperature effects is eliminated. It can be learnt from the statistical analysis that the frequency distribution of deformation difference generally satisfies the Pearson VII distribution. The deformation difference between the old and new bridges is mostly in the range of −2 mm–3.6 mm and concentrated in the range of −1 mm–0 mm, while the deformation difference between two separate spans of the old bridges is in the range of ±4.2 mm and concentrated in the range of ±1 mm. Thus, the peak value of the deformation difference between the new and old bridges and that between two separate spans of the old bridges are both about 10 mm.
Although the method of obtaining the deformation difference using on-site monitoring is more direct, the length of the monitoring cycle is still limited, the monitoring cost is high, and many uncontrollable factors may have an influence. Therefore, it is necessary to introduce relevant simulation methods to further expand the test data acquisition cycle and further improve the rationality of the deformation difference value.
3.3. Random Traffic Flow Simulation
The vehicle information recorded in the freeway toll system where the bridge site is located is directly adopted. Individual vehicle characteristics information (including vehicle weight, axle number, axle load, wheelbase, etc.), overall characteristics information of the traffic flow (such as traffic volume, vehicle type composition, vehicle speed, lane selection probability, etc.), and the influence line of the key components of the target bridge can be extracted using Python 3.11.
A random traffic flow simulation method is applied, which was verified in a previous study [
21]. In this method, the vehicle information obtained is converted into a simulated load sequence with spatial location, and it directly acts on the target bridge to calculate the corresponding load effect according to the specified influence line. The deformation response at the mid-span section of the old bridge is analyzed using this method. The simulation period is 400 days, and the simulation frequency is 2 Hz.
The peak deformation caused by vehicles in different lanes is extracted hourly. The probability of different deformation differences in each lane is counted in
Figure 9. It can be seen from the statistical results that the random traffic flow on the emergency lane will not produce obvious displacement effects in most cases. In the case of the existing splicing deformation difference, the mean and peak values are about 4.1 mm and 9.6 mm, respectively. The mean values caused by random traffic flow on lanes 1 and 2 are about 9.7 mm and 10.9 mm, respectively. The peak values on lanes 1 and 2 are about 14.5 mm and 15.8 mm.
3.4. Determination of Splicing Requirements
The relative deformation difference results of the old bridge obtained using the different methods mentioned above are compared in
Table 1.
By comparing the results in
Table 1, the following rules can be found:
Firstly, the theoretical calculation results of different specifications are significantly higher than the values obtained by other methods. This means that, in the analysis of splicing requirements, if the theoretical calculation results are directly adopted as the reference value of the integral splicing deformation, the strength of the designed splicing structure may be significantly higher than the actual demand.
Secondly, the static load test result of the old bridge is close to the peak value of the monitoring results between the new and old bridges. This indicates that the actual traffic flow on the old bridge has been close to the most unfavorable effect of the original design. However, the on-site monitoring results obtained between two separate spans of the old bridge are significantly higher than those obtained between the old and new bridges. This indicates that the tested deformation of a single old bridge cannot adequately reflect the actual deformation difference between two separate spans of the old bridge. Therefore, it is more reasonable to directly adopt the deformation difference between two separate spans of the old bridge as the design value of the integral splicing requirement.
Thirdly, the simulated value of random traffic flow is close to the on-site monitoring results between two separate spans of old bridge. The peak value of the former is only about 6% lower than that of the latter. Both can reflect the maximum deformation difference demand of integral splicing. In addition, the peak value of random traffic flow simulation is slightly higher than the mean value of the on-site monitoring results between the two separate spans of the old bridge by about 30%, which can ensure the safety redundancy of the demand for the integral splicing under normal loading conditions to a certain extent.
Therefore, for the target splicing bridge in this study, 18 mm should be considered as the design control value of the maximum deformation of the integral splicing composite structure (ISC-Structure).
4. Structural Design and Mechanical Characteristics Analysis
4.1. Structural Design
After fully investigating the relevant research results regarding the existing box girder bridge structure connection, a type of integral splicing composite structure for the long-span box girder bridge is innovatively proposed as illustrated in
Figure 10.
The proposed integral splicing composite structure (ISC-Structure) can be divided into three parts. From top to bottom are the composite segment serving as the integral layer (segment CI), the joint connection segment (segment JC), and the composite segment below the flange plate (segment CF), respectively. The design scheme is specified as follows:
The lower surface of the segment CI should be connected to the upper edge of the flange plate of the old bridge.
The width is 2 × 425 cm, which is the distance between the inner chamfers of the adjacent webs of the old bridge.
The thickness is 12 cm, which is equal to the design thickness of the integral layer of the old bridge.
The width is 15 + 40 + 15 = 70 cm, of which 15 cm is the width of the concrete chiseling region at the end of the flange plate. It is determined according to the design documents and field test results.
The thickness is 20.2 cm, which is equal to the mean thickness of the flange plate at 15 cm from the end of the flange plate.
For the concrete chiseling region at the flange plate end, the concrete must be chiseled out before integral splicing, while retaining the existing steel bars in the flange plate of the old bridge.
The upper surface of the segment CF is connected to the lower edge of the flange plate of the old bridge.
The standard width is 2 × 200 cm, which is determined according to the outermost position of the longitudinal steel wire bundle arranged in the flange plate.
The thickness of this segment gradually changes from the end to the center of the splicing structure. The standard thickness at the end is 15 cm, and the maximum thickness at the center is 40 cm.
In addition, a triangular decorative structure is set at the end of the segment CF to prevent the occurrence of local stress concentration at the lower edge of the flange plate.
To effectively transfer the force between the flange plate and the ISC-Structure, the following reinforcement connection measures are taken:
In segment CI, the reinforcement net is formed by horizontal reinforcement and longitudinal reinforcement.
In segment CF, the reinforcement frame is composed of transverse reinforcement frame, longitudinal reinforcement and vertical reinforcement.
L-shaped anchor-bars should be planted into the flange plate surface of the old bridge to effectively connect the interface between the ISC-Structure and the flange plate surface. In particular, the position of the steel bar planted on the flange plate should not conflict with that of the existing transverse and longitudinal prestressing wire bundles.
A complete loop bar is formed in the segment JC using a pair of semi-loop reinforcement bars. It is connected to the longitudinal steel bar to form a reinforcement frame in this segment. It is also connected with the steel bar retained at the end of the flange plate after concrete chiseling so that the reinforcement of the flange plate of the old bridge and that of Segment JC can form a whole frame.
In addition, vertical reinforcement should be set among the segments CI, JC, and CF, so that the three parts of the ISC-Structure can effectively transfer forces to each other.
4.2. Material Development
In the previous study, the application of loop bars in the joint can effectively transfer the force [
22]. But the lap length of loop bars needs to meet the following requirements:
where
Hmin is the minimum lap length of loop bars;
s denotes the layout spacing of the loop bar; and
φ denotes the plane diffusion angle of concrete anchorage failure cone, which is closely related to the strength of the concrete.
Since the width of segment JC in the ISC-Structure proposed in this study is small, the effective lap length of loop bars is only 10 mm, which is difficult to meet the minimum requirements for lap length with normal concrete. Therefore, a new type of ultra-high-performance concrete (UHPC) is developed. The mix proportion of UHPC mixture is summarized as follows: P II 52.5 cement is 6.82 kg, S105 slag powder is 2.43 kg, silica fume is 0.49 kg, quartz sand is 14.60 kg, sulfate aluminum cement is 0.43 kg, superplasticizer is 0.23 kg, water is 2.25 kg, and steel fiber is 1.50 kg.
A material performance test of the UHPC is also carried out. The tested compressive strength is 111.3 MPa and the flexural strength is 28.0 MPa after 28 days of curing. For the old bridge after splicing, the ISC-Structure is mainly subject to the tension force, so that the ability of UHPC to resist cracking needs to be explored. A dumbbell-shaped specimen is applied to test the axial tensile performance of UHPC. The stress–strain curve of the specimen in the axial tensile performance test is shown in
Figure 11.
The axial tensile strength of UHPC is 10.22 MPa, and the ultimate tensile strain is about 2000 με. The stress–strain curve can be divided into three stages:
Stage I: When the strain is less than 250 με, the relationship between stress and strain is close to linear.
Stage II: The tensile stress increases slowly, while the strain increases quickly. When the tensile strain reaches its ultimate value, the tensile stress has also reached the maximum value.
Stage III: As the strain of the material increases, the tensile stress begins to decrease slowly. The steel fiber inside the material begin to fracture quickly.
4.3. Mechanical Characteristics Analysis
4.3.1. Overall Force Characteristics Analysis
The finite element analysis model of the existing Xinfengjiang Bridge is established, as shown in
Figure 12a. The mechanical properties of the old bridge before and after the implementation of the proposed ISC-Structure are calculated and compared.
Firstly, the force characteristics are analyzed from the longitudinal direction. The maximum deformation of the old bridge is compared based on different load combination regulations. According to the results listed in
Table 2, the longitudinal deformation of the old bridge can be reduced by 3–6% after the implementation of the ISC-Structure. Among all kinds of loads, vehicle load has the most obvious influence, and its effect after the implementation can be directly reduced by about 42%. It can be considered, from the analysis results, that the implementation of the ISC-Structure has a certain beneficial effect on the longitudinal mechanical performance of the old bridge.
Secondly, the force characteristics are analyzed from the transverse direction. For the upper and lower edges of the flange plate after splicing, typical sections are selected, and corresponding transverse stresses are analyzed.
Taking the mid-span section with unfavorable stress as an example, the transverse stress distribution is presented in
Figure 12b.
By considering the frequent combination of various loads, including the transverse prestress in the flange plate, it can be learnt that the upper edge of the flange plate is in a state of compression. However, the maximum tensile stress at the lower edge of the flange plate is only 1.66 MPa, which can still meet the tensile strength limit of the old bridge concrete (Grade C50). It can be considered that the flange plate will not crack, theoretically, after the implementation of the ISC-Structure.
4.3.2. Local Force Characteristics Analysis
A local part of the splicing region was selected for specimen design, and three full-scale model specimens were made. The size of the specimens is 5.6 m × 3.0 m × 1.2 m. Each specimen contains two precast segments (pre-segment) and one cast-in-place segment (cast-segment). The pre-segment and the cast-segment are designed, respectively, according to the actual size of the flange plate and the proposed ISC-Structure. Each pre-segment contains three prestressing wire bundles with an adjacent spacing of 1.0 m. The specimens are fabricated according to the actual construction steps of the splicing scheme, as follows: Two pre-segments were first poured, and the prestressing wire bundles were tensioned. Then, the concrete at the end of the pre-segment was partially chiseled. Finally, steel bars in the pre-segments were planted or installed, and splicing concrete was poured.
The four-point flexural loading test was conducted on the local full-scale model specimens to test the flexural resistance of the ISC-Structure, as shown in
Figure 13a.
The test results can be summarized as follows:
The application of UHPC can ensure the effective lap length of the loop bar in the segment JC, but the normal concrete cannot meet the requirements.
Increasing the width of segment CF is beneficial to delay the interfacial debonding failure, and increasing its thickness can effectively delay the cracking of the flange plate.
The planted L-shaped anchor bars on the upper and lower edges of the flange plate can effectively delay the interfacial debonding failure between the flange plate and the ISC-Structure.
The failure mode of the ISC-Structure is as follows: the interface debonding occurs firstly, followed by cracking of the lower edge of the flange plate.
In addition, a refined simulation model for size optimization of the ISC-Structure is proposed, as presented in
Figure 13b. The simulation model is built in Abaqus 6.14. The solid element of C3D8R is adopted for the concrete and supports, and the beam element of T3D2 is used for the reinforcement and the prestressed wire bundles. The element size is primarily 20 mm. The interaction between the C50 concrete and UHPC is simulated by the surface-based cohesive behavior; the reinforcements and the prestressed wire bundles are embedded in the concrete. The prestress of prestressed wire bundles is applied using the temperature method. After running the simulation model, it can be found that the proposed simulation model can match the test results well and can meet the size optimization requirement of the ISC-Structure.
More detailed test schemes and the test and simulation results are available for reference in the published research article, ‘Local Full-Scale Model Test on Mechanical Performance of the Integral Splicing Composite Structure of Adjacent Existing Box Girder Bridges’ [
23].