Pretest analysis of shake table response of a two-span steel girder bridge incorporating accelerated bridge construction connections

Abstract

This paper presents pretest analysis of a shake table test model of a 0.35-scale, two-span, steel plate girder bridge. The objective of pretest analysis was to obtain an insight on the seismic response of the bridge model during the shake table tests. The bridge included seat type abutments, full-depth precast deck panels, and a two-column bent in which columns were pinned to the footing and integral with superstructure. Six accelerated bridge construction connections were incorporated in the bridge model. An analytical model was developed in OpenSees and was subjected to ten input bi-directional earthquake motions including near-fault and far-field records. The overall seismic response of the bridge was satisfactory for all the earthquake records at 100%, 150%, and 200% design level. All connections and capacity-protected components remained elastic, and the average ductility capacity surpassed the ductility demand even at 200% design level. Using experimental fragility curves developed for RC bridge columns, it was predicted that there was a probability of 45% that columns would undergo the imminent failure in the last run and a probability of 30% for their failure.

Keywords

shake table test / accelerated bridge construction / steel girder bridge / OpenSEES / UHPC / simple for dead continuous for live

Introduction

Bridge construction often leads to traffic delays, compromises the safety of highway workers and the traveling public, and could affect the regional economy and psychological health. Accelerated bridge construction (ABC) is a technique which utilizes prefabricated bridge elements to limit the onsite construction time. Because precast members are built offsite and under controlled environmental conditions, ABC provides an opportunity to use novel materials. In moderate and high seismic regions, it is of great importance to make sure that prefabricated elements are connected properly to guarantee bridge integrity, adequate load path, and constructability.

Several researchers have investigated the seismic performance of various connections appropriate for ABC (hereby referred to as ABC connections) in the past [¹–⁶]. These previous studies have been on components consisting of single or a subassembly of part of the bridge rather than the entire bridge system. Moreover, the unrealistic type of loading (such as uni-directional loading) and test setup (such as inverse test set up) could impose unrealistic demands on the connections. Integrating different promising ABC connections in a single bridge under realistic seismic loading could reveal issues with construction and seismic performance that would not be otherwise known. The focus of the study discussed in this article was on combining in one bridge model six ABC connection types used in different parts of the bridge: rebar hinge pocket connection (used for connecting columns to the footing); grouted duct connection (used for connecting columns to the cap beam); seismic simple for dead continuous for live (SDCL) girder-to-cap beam connection; girder-to-deck grouted pocket connection; joints filled with ultra-high performance concrete (UHPC) between deck panels; and deck panel UHPC-filled connection above the cap beam.

A rebar hinge connection is a two-way hinge detail typically used at the base of columns in multi-column bents. Two-way hinge connections reduce the transferred moment to the foundations leading to smaller and less expensive foundations. Rebar hinge connections comprise a cluster of bars placed in a pattern with a smaller diameter compared to the column diameter. Mehrsoroush et al. [⁷] and Mohebbi et al. [⁸] performed shake table tests on two-column bents incorporating rebar hinge connections that were embedded in the pockets left in the footing or cap beam.

A two-stage hybrid cap beam (consisting of a precast and cast-in-place segment), as part of a proposed precast bent system aimed for integral bridges with prestressed girders, was tested under cyclic lateral loading [⁹]. The cap beam included a lower precast cap beam installed first to support the girders and a cast-in-place upper portion to integrate the pier and superstructure. Column longitudinal bars extended into grouted ducts incorporated in the precast cap beam.

A seismic detail of cap beam to girder connection for integral steel bridges was developed at Florida International University in which girders were simply supported for the dead load and continuous for the live load. Seismic performance of the connection was experimentally investigated under cyclic lateral loading, confirming that the connection was well suited for seismic applications [¹⁰].

To provide composite action between full-depth precast deck panels and steel girders, shear studs need to be clustered in groups, and pockets need to be left in the panels to accommodate studs [¹¹]. Shreshta et al. [¹²] used different materials in the pockets connecting deck to precast girders. Authors reported that the type of grout used in the pockets does not affect shear and axial capacity of studs.

UHPC is a cementitious material with water-to-cementitious material ratio of less than 25%, and a high percentage of steel fibers. Several researchers have used UHPC in joints connecting prefabricated deck panels because of its superior bond strength to reduce the required lap splice length for deck longitudinal reinforcement, thereby enabling the use of narrower joints [¹³,¹⁴].

Component studies have provided invaluable information on the local behavior of connections, which helps formulating seismic design guidelines for ABC connections. However, to confidently recommend ABC bridges for adoption in routine bridge design and construction in high seismic regions, a comprehensive study of ABC bridge systems and the effect of interaction and load distribution among components is essential. For example, it is not known how SDCL connections behave under seismic loading when the girders are integrated with a hybrid cap beam and column grouted duct connection. Another example is possible in-plane rotations of the superstructure when columns are pinned to the footing through rebar hinge pocket connections, and the bridge is under bi-directional loading.

A large-scale, two-span ABC bridge model with steel girders was designed, constructed, and tested on the shake tables of the University of Nevada, Reno. The study was aimed at investigating the seismic response of a bridge system integrating six ABC connections under combined gravity and bi-directional horizontal seismic loading. Another objective was to evaluate the feasibility of the construction methods and the adequacy of some of the emerging design methods for ABC connections.

This article focuses on pretest analytical studies of the model, which aimed at providing beneficial input for the design of the bridge model and the experimental program. Furthermore, seismic performance of the bridge model was investigated under a large number of input earthquake motions including near-fault and far-field records to evaluate its ductility capacity and damage potential in connections. Design, construction, testing, and measured and calculated response of the bridge model are discussed elsewhere.

Bridge model description

The elevation of a typical two-span highway bridge is shown in Fig. 1. This bridge was used as the prototype. The width of the prototype superstructure section was 7.8 m. The axial load index (ALI) for the columns, defined as the dead load divided by the product of the nominal concrete compressive strength and the gross cross-sectional area of each column was 0.057. The prototype bridge was scaled down to 0.35 to enable testing on shake tables at the University of Nevada, Reno. Figure 2 shows a 3-dimensional representation of test setup. The geometric configuration and general dimensions of the bridge model are shown in Fig. 3. The bridge model incorporated two equal spans of 10.6 m, a two-column bent, full-depth precast deck panels, and seat type abutments. The skew angle was zero at both abutments. Schematics of column-to-cap beam and column-to-footing connections are shown in Fig. 4. The columns were integral with the superstructure but hinged at the base through two-way hinges embedded in pockets formed in the footing. For girders to be simply supported for dead load and continuous for seismic loads, the cap beam was constructed in two stages, a lower and an upper part, with the former being precast and the latter being cast-in-place. The girders were supported on the precast part. Figure 5 shows cap beam details before and after casting concrete on top of stage I cap beam. The longitudinal column bars passed through grouted ducts embedded in the precast cap beam and extended into the CIP part of the cap beam. To duplicate the column axial load index and stresses of the prototype bridge, extra masses were superimposed on the superstructure in form of lead pallets and concrete mass. Current United States bridge seismic design codes, such as AASHTO [¹⁵] and Caltrans SDC [¹⁶], do not include the contribution of the backwall as well as the foundation supporting the abutment in the seismic design of the bridges. The assumption of the abutment back walls to be sacrificial and shear off even under small earthquakes is to avoid any damage to the abutment pile foundation, which is hard to inspect and repair. Furthermore, previous detailed analytical studies conducted by Sadrossadat-Zadeh and Saiidi [¹⁷] concluded that elimination of the abutment interaction does not change the seismic response significantly. There is only a slight reduction in the peak displacements while maintaining the overall displacement pattern.

All components of the bridge model were designed based on AASHTO LRFD [¹⁸], and AASHTO Guide Specifications for LRFD Seismic Bridge Design [¹⁵], and emerging design methods for ABC connection based on previous studies. The bridge model was assumed to be located in Los Angeles area, Lake Wood, with the latitude and longitude of 33.84926 N, and 118.0952 W, respectively, and site class D. Design spectrum was developed utilizing United States Geological Survey (USGS), US Seismic Design Maps web application [¹⁹]. AASHTO 2009 [²⁰] was selected in the application to provide the seismic design parameter values. The time axis of the spectrum was compressed by a factor of 1.69, corresponding to the inverse of the square root of the dimensional scale factor. Design spectrum is shown in Fig. 6. The bridge components were designed such that inelastic deformations mainly occur in columns and the superstructure and footing remain essentially elastic with no yielding or damage during shake table testing. The essentially elastic elements are referred to as “capacity protected.” For capacity protected elements, inelastic response is limited to minor cracking and/or material strains that will not significantly diminish the component’s stiffness. The columns were designed based on the force-based approach according to AASHTO LRFD [¹⁸], and the design was checked using the displacement-based approach in accordance to AASHTO Guide Specs [¹⁵].

Two-way hinge connections were designed based on the procedure developed by Saiidi et al. [²¹]. The footing incorporated two corrugated steel pipes as pockets for rebar hinge elements. Column embedment length in the footing was 1.25 times the required tension development length of the column longitudinal bars, and 1.18 times the column cross sectional dimension. The lower (precast) cap beam was designed for the construction loads. The lower cap beam incorporated 24, 51-mm diameter corrugated galvanized metal ducts that were later filled with high-strength nonshrink grout. The entire cap beam was designed for seismic loading. Table 1 lists the design properties of bridge components.

A uniform cross section was used for girders throughout the bridge length. The girders were designed for Strength I, and Service I load combination in accordance to chapter 6 of Bridge Design Specifications [¹⁸]. The connection of the girders to the cap beam was designed and detailed according to SDCL connection that was developed (Fig. 5) by Taghinezhadbilondy et al. [²²]. The tie bars were designed to resist the vertical component of the seismic forces. Two steel blocks welded to the girder bottom flanges were used to improve the negative moment capacity of the connection. Dowel bars (cap beam stirrups), as the main load carrying mechanism under reverse loading, were designed based on the established Caltrans [¹⁶] design provisions for capacity protected elements.

The deck in the prototype bridge was designed considering HL93 loading as the live load, and 2.39 kPa (50 PSF) as the wearing surface. The required reinforcement area was then scaled down for the deck panels in the test model. The bridge model included 22 precast deck panels joined together with transverse female-to-female joints. The girders were connected to the deck panels using clusters of four shear studs welded to the girder top flange and embedded in grout-filled deck pockets left in the precast deck panels. Shear studs were designed for Strength and Fatigue limit states. UHPC was used in the panel joints to decrease the required lap splice length for deck longitudinal bars. Although lap splice length was sufficient for deck reinforcement over the pier using normal strength grouts, UHPC was used in the upper 70 mm of the cap beam to match the deck thickness.

Pretest computational analyses

Pretest analytical studies were conducted to estimate design forces for preliminary design of the bridge components, determine linear and nonlinear seismic response of the bridge, verify that the capacity protected elements remain in the elastic range, and determine the suitable ground motion and the loading protocol for the shake table tests. The CSiBridge [²³] and the Open System for Earthquake Engineering Simulation (OpenSees) [²⁴], finite element packages were used in the pretest studies, with the former for linear analysis and the latter for nonlinear analysis, as explained in subsequent sections.

Modeling method

A 3-dimensional computational model of the prototype bridge was created using CSiBridge software (Fig. 7). Linear analysis under Strength I, Service I, Extreme Event I, and Fatigue I limit states (according to AASHTO [¹⁸]) was conducted for the force-based design of columns, design of steel plate girders and shear studs. Shell elements, with automatically generated meshes, were used to model the deck panels. The girders, cap beam, and columns were modeled using “Frame” elements. Since the bridge was supported on seat type abutments, translation of the abutment bearing in all directions but vertical, was unrestrained. The girder to deck shear connectors were modeled using flexible link elements. The shear and axial stiffness values of the link elements were based on the measured data obtained by Shrestha et al. [¹²] using full-scale slab-girder connection tests. Figure 8 shows the force-displacement plots of connectors. The column to cap beam connection was through “Rigid Links.” All the translational and rotational components of the cap beam link elements were fixed to represent integral connections.

OpenSees was used for nonlinear static analysis (pushover) and nonlinear dynamic response history analysis (RHA). A schematic view of the OpenSees model is presented in Fig. 9. The OpenSees model was composed of linear beam column elements combined with nonlinear column fiber section elements that connected a three-dimensional assemblage of nodes. Nodes and elements were located at center of gravity of the bridge components. All nonlinear deformations in the computer model were assumed to take place in the columns.

The superstructure was modeled using Enhanced Beam-Stick model [²⁵]. A grillage was used to represent the deck and the girders. “ElasticBeamColumn” elements were used to model deck and girder elements. The longitudinal elements representing the deck were connected through elastic transverse beams. A modification factor of 0.5 was assigned to the longitudinal and transverse beams for torsional constant. Since there is no interaction between axial force and bending moment in two perpendicular directions in the grillage, Poisson’s ratio of grillage beams was set equal to zero [²⁶]. The deck elements were treated as cracked members by assigning 40% of the gross section rigidity to the deck elements [²⁷].

To capture nonlinear effects in the columns and rebar hinges, force-based beam column elements were used, which allow for the distribution of plasticity along the length of the member. The defined “aggregator” option in OpenSees was used to add cracked section shear and torsional properties to the column fiber element sections. Rebar hinge elements were fixed to the base, and the girders were supported on rollers at the abutments. The girders were connected to the cap beam by means of rigid links. Deck to girder connection was modeled by “twoNodeLink” elements. The axial and horizontal shear properties were defined for the link elements, each representing a cluster of four studs. The axial stiffness of the link element was defined using elastic bilinear uniaxial material object and shear stiffness was defined using multilinear elastic uniaxial material object.

The superstructure mass was lumped at the nodes defined at 25 points along the length of each girder. The superimposed masses were lumped at the nodes defined at the center of each concrete block or lead pallet. The center of mass node for each superimposed load was connected with a rigid beam column element vertically to the centerline of the superstructure.

The expected material properties were used in the analysis. Grade 60 reinforcement steel [with the expected yield stress of 469 MPa per SDG [¹⁶]] was specified for mild steel reinforcement, and the specified 28-day compressive strength of concrete was 27.6 MPa with the expected compressive strength of 35.9 MPa. The concrete behavior was modeled using “Concrete02,” which is a concrete model with tensile strength and linear tension softening. “ReinforcingSteel” material was used to model the longitudinal bars of the column. Damping was specified using mass and stiffness proportional coefficients that were calculated for two percent damping. The P-delta effects were included in the analysis. “UniformExcitationPattern” command, which applies the same ground motion record at different support points, was used to apply ground motion accelerations in the transverse and longitudinal directions.

Linear analysis

Results of the linear analysis under different limit states were utilized for the design of the bridge components. Detailed description of the design procedure is discussed elsewhere. Modal analysis of the prototype bridge assuming cracked section properties for columns showed that the first three modes were in-plane rotation, longitudinal (along traffic), and transverse with periods of 3.5, 0.67, and 0.59 s, respectively. Figure 10 presents the first three mode shapes.

Nonlinear static analysis

Nonlinear static analysis (pushover) was conducted in each of the transverse and longitudinal directions of the bridge model to obtain the capacity curves. The results are shown in Fig. 11. The columns were assumed to fail when either strain in an edge fiber in the core concrete reaches 125% of calculated ultimate concrete compressive strain ({\epsilon }_{\mathrm{cu}}) obtained from the Mander’s confinement model [²⁸], or a longitudinal bar strain reaches the ultimate tensile strain ({\epsilon }_{\mathrm{su}}). Using these criteria, the calculated displacement ductility capacity of the bent was 5.7 and 6.2 in the longitudinal and transverse directions, respectively. The ultimate displacement was controlled by the core concrete failure in both directions. The capacity curves were idealized by an elastoplastic relationship to estimate the plastic shear force and the effective yield displacement. The elastic portion of the idealized curve passed through the point marking the first longitudinal bar yielding. The idealized plastic lateral force was obtained by balancing the areas between the calculated and the idealized curves beyond the first reinforcing bar yield point.

Based on the dynamic mass (64.4 Metric ton) and the effective initial stiffness of the pier, the effective natural period of the bridge model was 0.44 s and 0.41 s, in the longitudinal and transverse directions of the bridge, respectively. The slope of the first branch of the idealized bilinear pushover curve was regarded as the effective initial stiffness of the pier.

Response history analysis

Response history analyses were conducted on the bridge model using OpenSees to evaluate its ductility capacity and damage potential in connections and capacity protected members. The bridge was analyzed under a large number of near-fault and far-field ground motions (GMs) of different intensities.

Two horizontal components of 5 near-fault and 5 far-field GMs, selected from the Pacific Earthquake Engineering Research Center (PEER) strong ground motion database (NGA-West2 program), were used as the input GMs in the analyses. The parameters that were used in the selection of GMs were: 1) VS30 (average small strain shear wave velocity in the upper 30.48 m of the soil column); 2) earthquake magnitude; 3) distance to fault. The range of VS30 between 200 to 360 m/s, corresponding to site class D and earthquake magnitude greater than six was assumed in the selection of GMs. The Rjb between 0 to 15 km and 15 to 30 km was used to distinguish near-fault and far-field ground motions [²⁹]. Table 2 lists the selected GMs. In this table, NGA is the new generation attenuation number and PGA is the peak ground acceleration. For scaling considerations, the duration of the motions was shortened by a factor of 1.69.

Although the use of a large number of records may improve estimates of the average demands obtained from RHA, this approach may not be practical. To minimize the statistical dispersion and maximize the accuracy in the response parameters estimated from RHA under relatively small number of records, ground motions were scaled to the target design spectrum. The scaling was applied to the spectral acceleration at the fundamental period of the bridge S_a (T₁). The response spectra for the input motions were calculated for 5 percent damping. Each component of the records was then scaled to match the design spectral acceleration at the average of the longitudinal and transverse periods of the bridge (0.43 s) that were based on the effective stiffnesses obtained from pushover analyses. To use the same scale factor for both components of each GM, the average of the two scale factors were utilized. Figure 12 shows the response spectra for the ten scaled records superimposed on the design spectrum for each direction. The records were further multiplied by 1.5 and 2.0 to represent 150% and 200% versions of the design earthquake.

A total of 30 response history analyses were conducted on the bridge model under bi-axial horizontal excitations simultaneously in the transverse and longitudinal directions. The component with higher PGA was applied in the longitudinal direction to place relatively high demands on the superstructure-substructure connections.

The maximum and residual drift ratios for each GM at 100%, 150%, and 200% design level and in both directions are listed in Table 3. Theoretical failure occurred for N1 and N3 at 150% and 200% design level, and for F1 at 200% design level, which corresponds to the ductility demand exceeding the ductility capacity. The analyses were stopped when the theoretical failure occurred. Therefore, the residual displacement is not specified for these motions. It can be seen from the table that near-fault motions are more demanding in terms of the maximum and residual drift ratios compared to far-field motions. For instance, residual drift ratio for all the far-field motions were less than 1%, which was considered negligible [³⁰]. However, N2 and N5 led to residual drift ratios of more than 1% at 200% design level. Moreover, the effect of near-fault ground motion tended to be more severe under higher-amplitude motions (for instance the 200% versions of the design level earthquake compared to 150% and 100% design level).

Table 4 lists the maximum and average values for critical response parameters under the 10 earthquake records set in addition to associated capacities. The response parameters consisted of the maximum values of cap beam shear, positive and negative moment in the cap beam, shear in the deck to girder connectors, and positive and negative moment in the superstructure. All capacity to demand ratios were equal or more than one. which indicates that cap beam, superstructure, and deck to girder connectors remained elastic even under 200% design level. It is worth noting that the calculation of the cap beam moment capacities was based on the first reinforcing bar yield and with cap beam side reinforcement being ignored. Therefore, C/D = 1.0 (for cap beam negative moment) corresponds to the yielding of the first rebar in the cap beam. Moreover, with an overstrength factor of 1.2 to obtain the cap beam flexural demand and nominal material properties to obtain its flexural capacity [¹⁶], essentially elastic behavior is ensured by using resistance factor equal to 1.0 (C/D = 1.0).

Bridge model response prediction for the test input records

Results of the RHA under the earthquake set were examined to determine the input motion in the shake table test. The 142-degree and 52-degree horizontal components of the Sylmar convertor station ground motion record obtained during the 1994 Northridge, California earthquake (referred to N2 in this paper) was selected as the input ground motion in the shake table test. The reason for this selection was in part because this motion was one of the more critical motions among the earthquake records. Another important reason was so comparisons could be made with the response of a similar two-span ABC bridge model (Calt Bridge-1) with concrete superstructure that had been tested on a shake table, under the same motion [³¹].

The component with higher PGA (the 142-degree) was applied in the longitudinal direction. The amplitude of the design earthquake was determined so that the peak resultant displacements obtained from the nonlinear dynamic analysis and that obtained from the orthogonal combination of the design displacement demands were approximately the same. As a result, the acceleration records for each component were further scaled by a factor of 0.6 to build the target design earthquake (TDE). The time scaled acceleration, velocity, and displacement histories for the target design earthquake are shown in Fig. 13. The response spectra for the two components of the TDE and their square root of sum of squares (SRSS) resultant under 5% damping is shown in Fig. 14.

The number of earthquake runs and associated scale factors were selected so that different damage states of the bridge were captured. The desired maximum displacement in each run was such that the pushover curve can be produced based on the envelope of the hysteresis curve in each direction to represent the overall nonlinear behavior of the bridge. The loading protocol started with 0.3 × Sylmar to capture the elastic response and followed by 0.65 × Sylmar and 1.0 × Sylmar, continued to 2.0 × Sylmar with 0.25 × Sylmar increments to capture. Table 5 lists the scale factors for different motions and the associated PGA values. The target shake table accelerations in the longitudinal and transverse directions are shown in Fig. 15. Response history analysis was conducted under a spliced record that combined in sequence all the records in the loading protocol. Sufficient gap with zero amplitudes were included in the beginning and at the end of each earthquake record to assure that the test model comes to complete rest after each run. The spliced record corresponded to the motion that the bridge model would undergo during the tests.

The displacement histories of top of the columns in the longitudinal and transverse directions are illustrated in Fig. 16. To identify the maximum bent displacement demand, the resultant of longitudinal and transverse displacement histories was calculated and is also shown in Fig. 16. The peak resultant displacement (157 mm) corresponding to resultant drift ratio of 7.4%, is only about 20% higher than the peak longitudinal and transverse direction values meaning that the maximum displacements in the two directions do not occur at the same time. The displacement ductility demand was estimated by dividing the resultant displacement by the bent yield displacement obtained from the idealized pushover curves. The maximum ductility demand was approximately 10.3.

The force-displacement hysteresis curves as well as the associated backbone curves under the spliced motion in both the longitudinal and transverse directions are shown in Fig. 17. The force-displacement response of the bridge model indicated stable hysteretic behavior with ample energy dissipation. The dissipated energy increased in successive runs due to the higher displacements and insignificant strength degradation.

Expected damage states

To predict the extent and type of the apparent damage in the columns of the bridge model after each earthquake run and how these compare with that of conventional bridge columns, the damage states defined by Vosooghi and Saiidi [³²] were utilized. Their database included 32 cast-in-place bridge large-scale columns tested either on shake tables or under lateral quasi-static loading. Although only eight columns were tested under bi-directional loading, it was believed that the fragility curves could be applicable to the columns of the bridge model in the current study because resultant drift ratios were used in this part of the analysis. Furthermore, the amplitude of the motion was not considered as a parameter in developing the fragility curves, and the correlation between the selected response parameters and damage states were independent of the amplitude of the motion.

Table 6 lists the damage states and the associated extent of apparent damage. Figure 18 shows photos of the apparent damages for each damage state. One of the key response parameters that can be used to indicate the probability that a component will be damaged to a given DS is the maximum drift ratio (MDR). MDR that was used in this study was based on the resultant displacements to predict damage state after each run. Table 7 lists the predicted probability of occurrence for each damage state in each run. It can be seen that in the third run, there was a probability of 60% for the formation of the flexural cracks in the columns (DS-1) and a 10% probability for minor spalling and possible shear cracks (DS-2). For the fourth run there were 90% and 60% chance for the columns to be in the DS-1 and DS-2, respectively, and a 10% chance for extensive cracks and spalling (DS-3). In Run 5, in which the ductility demand surpassed the ductility capacity of the bridge in both directions, lateral and/or longitudinal reinforcing bars were expected to be visible by a 50% probability. Moreover, there was a 25% chance for the compressive failure for the concrete core edge (imminent failure) during this run. In Run 6, there were 95%, 70%, 40%, and 20% probability of occurrence for DS-3, DS-4, DS-5, and failure, respectively. Finally, the last run, in which the first longitudinal bar reached the ultimate tensile strength in the model, the probability of DS-4 occurrence was increased to 75%. In addition, there were 45% probability of imminent failure and 30% probability of complete column failure.

Conclusions

The following conclusions were drawn based on the information and discussions presented in this paper.

1) All the components of the two-span bridge test model exhibited satisfactory seismic performance under a large number of input earthquake motions that included near-fault and far-field records.

2) Near-fault motions were more demanding in terms of the maximum and residual drift ratios compared to far-field motions. For instance, residual drift ratio for all the far-field motions were less than 1% which was considered negligible. However, N2 and N5 led to residual drift ratios of more than 1% at 200% design level.

3) The effect of near-fault ground motion tended to be more severe under higher-amplitude motions (for instance the 200% versions of the design level earthquake compared to 150% and 100% design level).

4) The theoretical failure occurred in five out of the 30 response history analyses. These included N1 and N3 at 150% and 200% design level, and N1, N3, and F1 at 200% design level.

5) Vosooghi and Saiidi’s fragility curves were used to predict damage states of the bridge model during the shake table test. It was concluded that columns would pass DS-1 in the third run and DS-2 in the fourth run. In the last run, there were 75% and 45% chance that columns would be in DS-4 and DS-5 (imminent failure), but there was only a 30% chance that columns would fail.

Detailed description of the design methods and construction sequence of the bridge model, as well as observed damages during the shake table test were developed but will be presented elsewhere [³³]. Moreover, measured experimental results of the shake table test are presented in a separate manuscript [34].

References

[1] Matsumoto E E, Waggoner M C, Sumen G, Kreger M E. Development of a Precast Bent Cap System. FHWA Report No. FHWA/TX-0-1748-2. 2001

[2] Restrepo J I, Tobolski M J, Matsumoto E E. Development of a Precast Bent Cap System for Seismic Regions. NCHRP Report 681. 2011

[3] Tazarv M, Saiidi M. Next Generation of Bridge Columns for Accelerated Bridge Construction in High Seismic Zones. Report No. CCEER-14-06. 2014

[4] Motaref S, Saiidi M, Sanders D. Seismic Response of Precast Bridge Columns with Energy Dissipating Joints. Report No. CCEER-11-01. 2011

[5] Mehrsoroush A, Saiidi M. Cyclic response of precast bridge piers with novel column-base pipe pins and pocket cap beam connections. Journal of Bridge Engineering, 2016, 21(4): 04015080

[6] Mehraein M, Saiidi M. Seismic Performance of Bridge Column-Pile-Shaft Pin Connections for Application in Accelerated Bridge Construction. Report No. CCEER-16-01. 2016

[7] Mehrsoroush A, Saiidi M, Ryan K. Development of Earthquake-Resistant Precast Pier Systems for Accelerated Bridge Construction in Nevada. Report No. 555-14-803. 2017

[8] Mohebbi A, Saiidi M, Itani A. Development and Seismic Evaluation of Pier Systems w/Pocket Connections, CFRP Tendons, and ECC/UHPC Columns. Report No. CCEER-17-02, 2017

[9] Marsh M L, Stanton J F, Eberhard M O, Haraldsson O, Khaleghi B.A Precast Bridge Bent System for Seismic Regions. Technical Report DTFH61-09. 2010

[10] Sadeghnejad A, Taghinezhadbilondy R, Azizinamini A. Seismic performance of a new connection detail in an SDCL steel bridge system. Journal of Bridge Engineering, 2019, 24(10): 04019094

[11] Badie S S, Tadros M K. Full-Depth Precast Concrete Bridge Deck Panel Systems. Report 584 by National Cooperative Highway Research Program. 2008

[12] Shrestha G, Itani A, Saiidi M. Seismic Performance of Precast Full-Depth Decks in Accelerated Bridge Construction. Report No. CCEER-17-05. 2017

[13] Graybeal B. Behavior of Field-Cast Ultra-High Performance Concrete Bridge Deck Connections under Cyclic and Static Structural Loading. Report No. FHWA-HRT-11–023. 2010

[14] Graybeal B. Design and Construction of Field-Cast UHPC Connections. Federal Highway Administration Publication No. FHWA-HRT-14-084. 2014

[15] American Association of State Highway and Transportation Officials (AASHTO). Guide Specifications for LRFD Seismic Bridge Design. Washington, D.C.: AASHTO, 2014

[16] Caltrans. Seismic Design Criteria (SDC). Version 1.7. Sacramento, CA: California Department of Transportation, 2013

[17] Sadrossadat-Zadeh M, Saiidi M. Analytical Study of NEESR-SG 4-Span Bridge Model Using OpenSees. Report No. CCEER-07-03. 2007

[18] American Association of State Highway and Transportation Officials (AASHTO). AASHTO LRFD Bridge Design Specifications. 6th ed. Washington, D.C.: AASHTO, 2012

[19] United States Geological Survey (USGS). U.S. Seismic Design Maps. Reston, Virginia, , 2016

[20] American Association of State Highway and Transportation Officials (AASHTO). AASHTO Guide Specifications for LRFD Seismic Bridge Design. Washington, D.C.: AASHTO, 2009

[21] Saiidi M S, Cheng Z Y, Sanders D. Experimental study of two-way reinforced concrete column hinges under seismic loads. ACI Structural Journal, 2009, 106(3): 340-348

[22] Taghinezhadbilondy R, Yakel A, Azizinamini A. Extending Use of Simple for Dead Load and Continuous for Live Load (SDCL) Steel Bridge System to Seismic Areas. Miami, FL: Florida International University, 2016

[23] CSiBridge. V19.2.2. CSI Computer & Structures Inc. Integrated 3-D Bridge Analysis, Design, and Rating. Berkeley, CA: Computer & Structures, Inc., 2016

[24] McKenna F, Fenves G, Scott M. Open System for Earthquake Engineering Simulation (OpenSees). Berkeley, California: Pacific Earthquake Engineering Research Center, 2000

[25] Amirihormozaki E, Pekcan G, Itani A. Analytical modeling of horizontally curved steel girder highway bridges for seismic analysis. Journal of Earthquake Engineering, 2015, 19(2): 220–248

[26] Hambly E C. Bridge Deck Behaviour. 2nd ed. New York: CRC Press, 1990

[27] Carden L P, Itani A, Buckle I G. Seismic Load Path in Steel Girder Bridge Superstructures. Report No. CCEER-05-03. 2005

[28] Mander J B, Priestley M J, Park R. Theoretical stress-strain model for confined concrete. Journal of Structural Engineering, 1988, 114(8): 1804–1826

[29] Alavi B, Krawinkler H. Consideration of near-fault ground motion effects in seismic design. In: Proceedings of the 12th World Conference on Earthquake Engineering. Auckland, 2000

[30] Japan Road Association. Design Specifications for Highway Bridges, Part Seismic Design (English edition), Tokyo, 2000

[31] Benjumea J, Saiidi M, Itani A. Experimental and Analytical Seismic Studies of a Two-Span Bridge System with Precast Concrete Elements and ABC Connections. Report No. CCEER-19-02. 2019

[32] Vosooghi A, Saiidi M S. Experimental fragility curves for seismic response of reinforced concrete bridge columns. ACI Structural Journal, 2012, 109(6): 825–834

[33] Shoushtari E, Saiidi M, Itani A, Moustafa M. Design, construction, and shake table testing of a steel girder bridge system with ABC connections. ASCE Journal of Bridge Engineering, 2018, 24(9): 04019088

[34] Shoushtari E. Shake table studies of a steel girder bridge system with ABC connections. Dissertation for the Doctoral Degree. Reno: University of Nevada-Reno, 2019, 129–167

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag GmbH Germany, part of Springer Nature