Performance of Pier-to-Pier Cap Connections of Integral Bridges under Thermal and Seismic Loads

In general, most highway bridges are constructed using prestressed concrete or steel girders. Mechanical joints are provided at the end of each span, to allow for the expansion of the bridge deck due to shrinkage of concrete, thermal effects, and deflections, among others. Smooth riding ability, low noise, wear resistance, and water tightness should be provided by expansion joints. In recent times, the increased traffic volume, along with heavier vehicle movements, adversely affects the performance of expansion joints in the bridge girder, causing a possible failure in one of the above-mentioned mechanisms. The deterioration of the expansion joint may result in leakage of water, concrete cracking, and potential problems in the underlying substructure. In this paper, we study the pier-pier cap connections in integral bridges subjected to thermal and seismic loads using analytical methods and experimental tests.


Introduction
Integral bridges are constructed without joints or connection frames. ese are characterized by a monolithic relation between the bridge deck and the substructure (piers and abutments). e intermediate connection between the two abutments is provided without any joints in the deck. Around the world, integral bridges have been built with different connection systems. ese types of connections offer several advantages over simply supported bridges. e expansion joints and bearings provided in the simply supported bridge may lead to poor structural performance given the separation of the superstructure into two or more parts. Furthermore, integral bridges have a lower maintenance cost and lower corrosion issues when compared to bridges with expansion joints. e riding quality on the integral bridge is smoother. Moreover, redundancy is added to the structure by virtue of its monolithic property Holombo et al. [1] and Wassef et al. [2].
In spite of the relevance of integral bridge construction, there is still a lack of research on the analysis and design of integral bridge connections. For instance, in India, there are no standardized procedures for the design of integral bridge connections. Although engineers have developed guidelines based on the experience gained by conducting field tests and laboratory experiments, there are no proper guidelines, especially for the pier-pier cap connections. In particular, little attention has been paid to the thermal and seismic response of this type of connection. erefore, in this paper, we investigate the behavior of integral bridges with pier-pier cap connections subjected to thermal and seismic loads. e main novelties of the present research work are the following: (1) Experimental investigation on the behavior of the connections of an integral bridge subjected to thermal and seismic loads (2) Computational modeling of the above thermal and seismic behavior (3) Validation of numerical predictions e present paper is organized as follows. Section 1 introduces and reviews the main research works carried out in the context of integral bridges and their connections. Section 2 describes the experimental program followed in this paper. e casting details are detailed in Section 3. Section 4 gives information on the finite element modeling of the structure. e corresponding experimental results are given in Section 5. Section 6 presents the numerical results predicted by the computer model. e validation of the model is presented in Section 7. Finally, the main conclusions of this work are summarized in Section 8.
Abendroth and Greimann [3] carried out a research program to evaluate the design of integral abutment bridges using prestressed concrete (PC) elements and to validate the assumption of thermal loading conditions in the current design procedure for such types of connection in integral bridges. Over a period of two years, two skewed or bent PC girder integral abutment bridges were instrumented in order to test experimentally the structural behavior. roughout the monitoring time, displacements, rotations, and temperature distributions along the longitudinal and transverse directions of the integral abutments were recorded for the PC girders. For different PC girders investigated under dry and saturated conditions, the concrete coefficients of thermal expansion and contraction were measured. Furthermore, the longitudinal displacements of the integral abutments were validated with experimental results obtained with changes in the bridge temperature. Such experimental data were utilized for the calibration of finite element models. e differences between the predicted and measured thermal expansions and vertical rotations of the bridge and integral abutments were not completely clarified.
Panday and Tandon [4] investigated the integral flyover at the intersection of Maa Anandamoyee Marg and Khegaon Marg on Outer Ring Road in Delhi and an intersection at Moti Nagar. A detailed study on the substructure and superstructure of the Kheagoan Marg Bridge was carried out. e superstructure was 150 m long with 5 spans, with a carriageway of 9 m and a depth of superstructure of 1.7 m. e substructure consisted of twin piers per carriageway per support location. Piers were rectangular with rounded short edges with a flute finish. e thermal load was decreased according to the stiffness and length of the structure, while the flexibility of the outer pier was gradually increased by the thickness from 1.15 m to 1.5 m. At the same time, the height of the pier was gradually increased from 0.5 m to 2 m at the base of the footing to the center of the structure. In this way, the flexibility was increased around 6 times at the end of the pier.
Laman [5] conducted a study on the superstructure stresses in prestressed girder integral abutment bridges by thermal induction. e stresses and forces were developed on the superstructure of prestressed concrete integral abutment bridges due to thermal loads. e loads applied on the superstructure showed a constant temperature change.
ermally induced superstructure forces were investigated on the bridge length, number of spans, abutment height, and pile orientation. ermally induced superstructure stresses and forces reached maximum values near the abutment. It was found that the abutment height and bridge length highly influence the thermally induced superstructure forces. e thermally induced superstructure stresses were greatly influenced by the number of spans. e results showed that the thermally induced shear forces and superstructure stresses were comparable in magnitude to those produced by the live load.
Rodriguez Leo [6] monitored the temperature response of integral abutment bridges in Utah and California for approximately one year, using arrays of thermocouples through the height of girders and depth of the bridge deck. Temperature changes in concrete were recorded in addition to temperature gradients. A finite element model (FEM) was validated with the changes of deflections and strains monitored during a live-load test on the California Bridge and the changes of rotation observed from the thermal gradients of Utah Bridge. ese results were used for the prediction of the bridge's internal stresses produced by temperature variations in the full cross section of the modeled bridge. Due to the measurements made on the end restraint of the California Bridge, it was found a very accurate prediction of its behavior at service conditions. e temperature differential and tensile stresses were obtained and correlated with the calculated values in concordance with design specifications.
Russell and Gerken [7] suggested design considerations for integral abutment bridges. It was noted that the movements occurred by changes in the bridge temperature and concrete creep and shrinkage, among others. ey also suggested that the seasonal temperature changes affect the bridge length primarily, while the daily temperature affects the thermal gradients from the depth of the bridge structure.
Holombo et al. [1] conducted an experimental research on an integral bridge connection along its longitudinal direction. He prototyped two 40% scaled modeled bridges which represent the typical bridge construction in California. e columns were defined with an inelastic behavior, and the superstructure was designed to perform elastically. e test results showed a good ductility performance of the integral connection with a small strength degradation when subjected to seismic loads. e superstructure remained elastic with few small cracks, which closed after the removal of the corresponding loads. e author recommended extending the longitudinal reinforcement in the column as much as possible inside the bent cap to provide a better distribution of the seismic forces.
Chen et al. [8] conducted shake table tests on two 1/7scale tall-pier models to evaluate the impacts of higher modes on the seismic performance of bridges with piers heights of over 40 m. ey proposed two ways to improve the seismic performance: (1) eliminating the mid-height plastic response by including more longitudinal steel, and (2) using more confinement in the mid-height region to improve pier ductility and avoid shear.
Mitoulis [9] conducted a review on the challenges and opportunities for the application of integral abutment bridges in earthquake-prone areas. e findings showed that backfill soil benefits the bridge mostly by reducing bending moments and pier drifts, which could lead to more cost-effective designs. However, because the IAB-backfill interaction was very case-dependent, careful and precise modeling of the backfill soil was suggested to avoid underestimating the bridge stresses.
Argyroudis et al. [10] conducted a research on the use of rubberized backfills for improving the seismic response of integral abutment bridges. e findings showed that using rubberized backfill reduces backfill settlements, bridge deck horizontal displacements, residual horizontal displacements of the top of the abutment, and abutment pressures by up to 55 percent, 18 percent, 43 percent, and 47 percent, respectively, when compared to a conventional backfill. On the abutment wall, there was also a noticeable reduction in bending moments and shear forces. As a result, rubberized backfills appeared to be a feasible alternative for reducing earthquake risk and achieving economic design goals for transportation network resilience with minimal damage.
Sritharan et al. [11] conducted an experimental investigation on an integral bridge pier consisting of a concrete column having I-shaped precast concrete girders with an inverted T-beam concrete cap, which accelerated the construction methods of the bridge in a seismic zone. e behavior of the overall system and girder-to-cap connections was evaluated. In the research, two girder-tocap connections were analyzed: one was already implemented in practice and the other was a new proposed model. e research was conducted in a large-scale set-up, which helped to exhibit a good seismic response on the columns with successful plastic hinge formation. In the experimental study, the as-built connection with an inverted T-pier cap on an I-shaped precast girder behaved as a fully continuous connection rather than acting as a pinned connection. In the experiments, the improved girder-to-cap connections performed well, which showed a higher dependable response better than the as-built connection. Wang et al. [12] made an experimental investigation on the prefabricated bridge piers with grouted splice connections subjected to seismic loading. In his research, he found that the grouted splice piers performed better when compared to cast-in places. Grouted splice displacement capacity was lower than cast-in-place. Jia et al. [13] did their research on the precast bridge pier with elastomeric pads built-in subjected to seismic evaluation. In the research, elastomeric pads were kept at a plastic hinge area zone of the bridge pier. When subjected to seismic loads, the pier with the elastomeric pad showed better results. Usage of elastomeric pad provided stiffness and energy dissipation capacity. Zakeri and Zareian [14] made a design for a bridge with a skewed abutment subjected to minor earthquake loads. In his research, he found a negative trend present in between abutment skew angle and reliability index. He found that as the ratio of the column height to diameter increases, the reliability index becomes lower. His proposed method helped to design bridge piers for the target performance.

Experimental Program
e dimensions of a real bridge are taken and are reduced to one-fourth scale as per Cauchy's similitude law for fabrication of the experimental specimen. e bridge is modeled in STAAD. Pro and dead load, live load, and moments are determined as illustrated in Figure 1. Using the values obtained, the design and detailing of the bridge were carried out.
en, the materials are procured and the material properties are determined. e design mix for the M35 grade of concrete is carried out. en, the reinforcement cage is prepared and strain gauges are attached to the specimens. Concreting and curing of the specimens are carried out and are then tested for thermal and seismic loads. Simultaneously, the structures are modeled in ANSYS 14.5, and the results of the loadings are obtained. Finally, the experimental and analytical results are compared and validated [15,16].

Experimental Studies on Pier and Pier Cap Connections.
e integral bridges are subjected to a variety of forces. ese forces can be broadly classified into primary and secondary loads.
e physical primary loads include live load, dead load, and seismic and wind loads, while the secondary loads are due to the sinking of supports, creep, and shrinkage, etc. Only a few secondary loads are significant in designing the structure. e cross section of deck and girders and IRC class-70R tracked and wheeled vehicle loading are illustrated in Figures 2 and 3, respectively [17][18][19].
To calculate the live and dead load moments and forces acting on the structure, the prototype is modeled and analyzed in STAAD.Pro. Loads are considered as per IRC class-70R loading. e analysis of the prototype in STAAD.Pro is exhibited in Figure 4. e results obtained are as follows. e total dead load and live load on the structure were 4300 kN and 980 kN, respectively. e pier cap is subjected to a factored bending moment of 6300 kN m and a factored shear of 2400 kN. e pier is subjected to a factored load of 7335 kN and a longitudinal moment of 1000 kNm and a transverse moment of 2740 kNm.

Description of the Bridge.
e span of the prototype bridge is 60 m, in which the girder to girder distance is 3 m. e size of the pier cap is 1000 mm × 1600 mm and the size of the pier is 2000 mm × 800 mm. e taken model is 15 m, the girder to girder distance is 0.75 m, the size of the pier cap is 250 mm × 400 mm, and the size of the pier is 500 mm × 200 mm. e integral bridge has two lanes with a footpath on one end and a median on the other end. e specifications of the bridge and detailing of members of the bridge are presented in Tables 1 and 2, respectively. e dimensions of the bridge model and detailing of the pier, pier cap, and deck system of the bridge are illustrated in Figures 5 and 6, respectively.

Detailing.
e bridge pier model that was used in the experiment is shown in Figure 6 [20], and the mix proportion is shown in Table 4. Sample cubes were cast and tested for their 28-day compressive

Specimen Casting
Based on the bar bending details, two reinforcement cages were prepared to test the behavior of pier-pier cap connections when subjected to thermal load and seismic loading, respectively. e strain gauges were positioned at the critical points in the reinforcement cage to measure the strain which will be experienced by the reinforcement at the time of loading. e critical sections were found from analysis of the structure in ANSYS 14.5 commonly used strain gauge consisting of an insulating flexible backing with a metallic foil pattern is used. e strain gauges attached to the reinforcement using adhesive are illustrated in Figures 7  and 8, respectively. When the reinforcement gets deformed, the foil also gets deformed, which makes the electrical resistance to change, which provides us the strain in a constant amount. e reinforcement cages are kept inside the mold, and the concrete is poured to obtain the desired shape and size of the specimen. e molds are made by using waterproof plywood (2 numbers of 7 × 4 feet and 1 number of 4 × 3 feet), and adequate supports are given at its sides which prevents the mold from deforming when concrete is poured. Runners were provided which avoids the bulging of the mold and the cover blocks of 20 mm are kept at its sides and bottom [21,22]. e concrete mixture was mixed well in a batch mixer. e compaction was carried out using a needle vibrator, and the clogging was prevented by tamping of aggregates between the rebars to avoid honeycombing of structure. Segregation and bleeding of concrete have also been prevented while concreting the specimen. In each batch of specimen casting, 3 cubes were cast simultaneously. e specimen's demolding was carried out after 24 hours of curing. Water curing was done for 28 days by covering the specimen with fully soaked gunny bags. e details reinforcement and formwork for specimen are illustrated in Figures 9 and 10, respectively.

Experimental Set-Up and Loading.
e experimental process was carried out in the Structural Dynamics Laboratory, Structural Engineering Division, Anna University. e prototype specimens were tested in a well-equipped setup which are subjected to thermal load and reverse cyclic loading. When applying the lateral loads, 50 kN capacity manual hand-controlled jacks were connected to the reaction frame, and the axial load was applied by using a hydraulic jack. e schematic view of the experimental set-up and testing of specimens for thermal load is illustrated in Figures 11 and 12, respectively. e axial load was applied at the center of the pier. e slab was fixed to the frame to prevent rotation, and three plates were provided at the bottom of the slab, which induced internal stresses on the slab. For thermal analysis, the entire specimen was subjected to an axial load of 330 kN, and a lateral load of 11 kN was applied. For seismic analysis, an axial load of 280 kN (50% of the live load was considered as per codal provision) and a lateral load of 33 kN were applied.    6 Advances in Civil Engineering   Advances in Civil Engineering 7 e lateral load of 11 KN was calculated from Hooks law ΔL/L � α L ΔT. is expression gives the stress for thermal analysis.

Testing of the Specimen.
After the specimens were cast and cured for 28 days, they were moved to the strong floor using the gantry crane. e specimens were then centered and placed over three steel plates. en, the specimens were tested for thermal and seismic loading.

Analytical Studies on Pier-Pier Cap Connections.
Reinforced concrete structures have been largely deployed in various disciplines, especially in bridges. ese structures are designed based on experimental data. Finite element analysis has provided a path for a more realistic analysis in various civil engineering disciplines including bridge structures. For instance, FEM provides a clear visualization of where the structure will twist or bend and indicates the distribution of displacements and stresses. Due to the complexity of the composite nature of the material, modeling structures is a challenging task. e pier-pier cap joints of the integral bridge were modeled, analyzed, and designed in the previous sections. A scaled model was adopted for experimental investigation considering the testing facilities at the laboratory, and 1 : 4 scale was adopted. e same model was used for the analytical investigation.

Finite Element Modeling
e pier-pier cap model of 1/4th scale was analyzed using ANSYS software (version 14.5). Element type SOLID65 was selected to model the concrete. BEAM188 element was chosen to represent the reinforcement. SOLID65 was opted for the three-dimensional modeling of solids and reinforcing bars (rebars). e models of steel reinforcement and pier-pier cap are illustrated in Figures 13 and 14, respectively [23]. e solid65 has the capability of crushing in compression and cracking in tension. e elements were defined by eight nodes, each having 3 degrees of freedom with translations in the x, y, and z node directions. e BEAM188 element is relevant for analyzing the slender to moderately stubby/thick beam structures. e option BEAM188 is a 2-node linear element, whereas in 3D, it has 6 degrees of freedom at every node. e degrees of freedom in every node have the translations in x, y, and z directions and rotations in x, y, and z directions. e elements in the beams are suited for linear, large rotation, and large strain nonlinear applications. e real constants which are considered in SOLID65 elements are the volume ratio and the orientation angles in X and Y directions.
e pier-pier cap model with discrete reinforcement was considered in the present study. As for the reinforcement, circular sections with the respective diameters were created, which were then imparted to the reinforcing element. BEAM188 and the steel properties were added in the element attributes window. e elements were then generated by creating and connecting the nodes.

Material Properties.
e average cube strength (fck) at 28-day was 37.2 MPa, and it was used in modeling as obtained from the experiment. e elasticity modulus of concrete was considered as 5000√fck as per [24] which is 3.0414 × 1010 N/m 2 . Typical shear transfer range coefficients are from 0 to 1, in which 0 represents a smooth crack and 1 represents a rough crack. e uniaxial crushing stress used in the model refers to the uniaxial unconfined compressive strength. e uniaxial tensile cracking stress in concrete was determined by using the equation f � 0.623√fck where fck is the cube strength of the experimentally tested specimen [25]. e material properties defined in the model are presented in Table 5.

Displacement Boundary Conditions.
e pier was assumed to be fixed at the slab in 3 lines; that is, all DOFs were constrained to simulate the presence of girder loads.

Load Boundary Conditions.
In thermal analysis, the pier-pier cap connection was subjected to a pressure load of 330 kN at the base of the pier and the thermal lateral load was applied in increments up to 11 kN along the longitudinal direction of the pier. In the seismic analysis, the connection was subjected to an axial load of 280 kN at the base of the pier, and seismic loading is applied using reverse cyclic loading, applied in increments up to 33 kN along the longitudinal direction of the pier. e corresponding boundary conditions are illustrated in Figure 15. e method of finite element analysis was carried out for pier-pier cap connection. e standard Newton-Raphson method was adopted for static analysis.

ermal Results.
e bridge pier system in the inverted position was subjected to an increasing axial load up to 330 kN. e axial compression was simultaneously measured using LVDT. After the axial load reached 330 kN, an  Advances in Civil Engineering   increasing load up to 20 kN was applied by a manual hydraulic Jack fitted to the load frame. e lateral displacements were also measured for each 1 kN increment in load. e load was increased to 20 kN, to check for its lateral loadcarrying capacity. e load versus displacement and axial stress versus axial strain are illustrated in Figures 16 and 17, respectively. e axial stress versus strain of the specimen when subjected to thermal load is shown in Figure 17.

Seismic Type (Cyclic Loading) Results.
For simulating the seismic effects, the specimen was subjected to the reverse cyclic loading of incremental loading up to 33 kN in an increment of 3 kN. e loading pattern is shown in Figure 18. e lateral displacement was measured for each 3 kN increment in load. e lateral displacement with respect to load cycles is shown in Figure 19. e lateral displacement versus lateral load graph is shown in Figure 20. It is seen from the graph that the curve is linear and the behavior is elastic.

ermal Results.
With the application of thermal lateral load, the following displacement results are obtained. e displacement is found to be minimum at the slab and progressively increases toward the base of the column. e maximum displacement is found to be 3.67 mm. e displacement due to thermal loads is shown in Figure 21. e stress accumulations due to the application of thermal loads are shown in Figure 22. e stresses are found to be maximum at the pier-pier cap connections. e strain obtained due to the application of thermal load is shown in Figure 23. Blue indicates compressive strain and red indicates tensile strain.

Seismic Type (Cyclic Loading) Results.
With the application of reverse cyclic load, the following displacement values are obtained as shown in Figure 24. e maximum displacement is found to be 17.76 mm. e stresses developed due to reverse cyclic loads are as shown in Figure 25. e stresses developed are found to be maximum at the pier-pier cap connections.
e strain values obtained due to the application of seismic load are as displayed in Figure 26. It is found that the corners of the pier-pier cap junction experience maximum strain.

Validation and Discussion of Results
e variation of lateral load with respect to lateral displacement as determined from the experimental test was found to be similar to that value obtained from the computational model. e displacement value obtained from the numerical study was less than the experimental value by 9%. is might be due to the assumed perfect bonding between steel and concrete and the perfect fixity condition at the base. In the experimental testing, the specimen adjusts to its position during the initial stages of loading. e rate of loading of the manual hydraulic jack is not constant. is might have resulted in higher displacements during the experimental test. e load displacement variation with thermal load, strain versus load, and load displacement variation for seismic type loading are illustrated in Figures 27 -29, respectively. e lateral load versus lateral displacement graph shows that the specimen is within the elastic limit and hence the behavior is linear.
From the axial stress versus axial strain graph, the average modulus of elasticity was found to be 32500 MPa, which is similar to the modulus of elasticity as computed by formula 5000 (fck) 0.5 which is around 30414 MPa. e strain measures from the strain gauges at the initial loading is might have been due to the rearrangement and shape changes of the silicon cover over the strain gauges.
Response of pier-pier cap system subjected to seismic loading. e load versus displacement curve for experimental tests showed that the specimen is in its elastic state and there is no indication of yielding of the rebars and no cracks were observed. us, the structure is safe for the design of seismic load.

12
Advances in Civil Engineering

Conclusions
As discussed previously, two ¼ scaled-down models of an integral bridge were considered for the present study. Using the load values obtained from STAAD.Pro, the design and bridge detailing are developed. Based on the material properties, the mix design of concrete was conducted. e reinforcement cage was then prepared on which the strain gauges were fixed. Casting and curing of the specimens were carried out. e specimens were then moved to the strong floor and tested for thermal and seismic loads, respectively. en the models of the specimens were generated in ANSYS 14.5 and analyzed. e experimental results obtained were validated with the numerical results. e main novelties of this research were the computational modeling and experimental investigation of pier-pier cap connections. Little research work has been carried out in this context, particularly on this type of connection subjected to thermal and seismic loads. e following conclusions have been inferred from the study of the behavior of pier-pier cap connections of an integral bridge.
(1) e numerical results were in accordance with the experimental results. e maximum displacements of the experimental tests were found to be higher than the numerical predictions by 9%. (2) e connection, when subjected to thermal loads, developed stresses in the range of 3-5 MPa which is within the elastic range. (3) For seismic loads, the stresses developed were about 8 MPa which is also within the elastic limit. e design of the reinforcement is safe for this type of load.    Advances in Civil Engineering (4) It is observed from the numerical and experimental investigation that the stresses developed at the pierpier cap connections are much higher than the stresses at the base of the pier. (5) ere was no visual evidence of cracking. e reinforcement provided and the cross section were sufficient for thermal stresses as well as seismic loading. Hence, under service loads, the thermal effect is negligible and the provided section is safe under seismic loads.
Future studies can include further enhancements such as the optimization of the reinforcement area. e scale of the model can be modified as desired. e load can be applied along the transverse direction, and its magnitude with the number of loading cycles can also be modified to investigate the postelastic behavior.
Data Availability e datasets generated during the current study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that they have no conflicts of interest.