Three-dimensional nonlinear seismic performance ... - Rice University
Engineering Structures 30 (2008) 1869–1878 www.elsevier.com/locate/engstruct
Three-dimensional nonlinear seismic performance evaluation of retrofit measures for typical steel girder bridges Jamie E. Padgett a,∗ , Reginald DesRoches b,1 a Department of Civil and Environmental Engineering, Rice University, 6100 Main Street, MS-318, Houston, TX 77005, USA b School of Civil and Environmental Engineering, Georgia Institute of Technology, 790 Atlantic Dr., Atlanta, GA 30332-0355, USA
Received 3 August 2006; received in revised form 11 December 2007; accepted 12 December 2007 Available online 22 January 2008
Abstract Steel girder bridges are common bridges in North America which have exhibited considerable vulnerabilities in past earthquake events. This paper conducts nonlinear time history analysis using detailed three-dimensional models of typical multi-span simply supported and multispan continuous steel girder bridges to evaluate the effectiveness of various retrofit strategies. Restrainer cables, steel jackets, shear keys, and elastomeric isolation bearings are assessed for their influence on the variability and peak longitudinal and transverse response of critical components in the bridges. The results indicate that different retrofit measures may be more effective for each class of bridges. For example, the restrainer cables are effective for the multi-span simply supported (MSSS) bridge, yet not the multi-span continuous (MSC) bridge; shear keys improve the transverse bearing response in the MSC bridge but are not effective in the MSSS bridge which exhibits less transverse vulnerability; and elastomeric bearings improve the response of the vulnerable columns in both the MSSS and MSC bridges yet lead to increased abutment demands in the MSC bridge. The study reveals that while a retrofit may have a positive influence on the targeted component, other critical components may be unaffected or negatively impacted. This lends support to vulnerability assessments that consider the impact of retrofit on system vulnerability reflecting the contribution of multiple components. c 2007 Elsevier Ltd. All rights reserved.
Keywords: Retrofit; Steel bridges; Seismic response; Steel jackets; Restrainer cables; Isolation; Shear keys
1. Introduction Steel bridges are often potentially vulnerable highway bridges found across various seismic regions of North America. Tens of thousands of steel bridges exist in the current North American bridge inventory, most of which have not been designed considering earthquake loading [1]. More than onethird of the highway bridges in the Central and Southeastern United States (CSUS) are slab-on-steel girder bridges and were constructed during the 1950s through the 1980s, prior to the initiation of seismic design practices in the region [2]. Concrete slab-on-steel girder bridges are also the most common steel bridges found in Canada, which like the CSUS are often supported on non-seismically designed concrete piers and ∗ Corresponding author. Tel.: +1 713 348 2325.
E-mail addresses: [email protected] (J.E. Padgett), [email protected] (R. DesRoches). 1 Tel.: +1 404 385 0826. c 2007 Elsevier Ltd. All rights reserved. 0141-0296/$ - see front matter doi:10.1016/j.engstruct.2007.12.011
abutments [3]. Multi-span simply supported (MSSS) and multispan continuous (MSC) slab-on-girder bridges account for a majority of the steel bridges in North American. While these bridge types are less common on the West Coast, there have been examples of damage to steel bridges in past earthquake events in this part of the US (Fig. 1). The state of California has hundreds of steel girder bridges, several of which experienced damage during the 1994 Northridge earthquake [5]. Typical damage included pounding damage, damage to bearings and fracture of anchor bolts, and damage to piers and abutments [6]. A number of steel bridges were damaged in the 1995 Hyogoken-Nanbu (Kobe, Japan) earthquake. These bridges have details and characteristics that are similar to the steel girder bridges found in North America and particularly the Central and Eastern US [7]. Damage to non-seismically detailed reinforced concrete substructures included flexural failure of columns due to inadequate confinement and shear failures due to inadequate
1870
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
Fig. 1. Damage to steel girder bridges in past earthquake events: (a) pounding damage in Northridge, (b) column damage and (c) shifted superstructure in Kobe [4], and (d) bearing damage in Nisqually (courtesy of WDOT).
shear reinforcement [3]. There was a significant amount of observed damage to steel bearings, as well as excessive deck displacements as a consequence of bearing failure and the loss of connection between the superstructure and substructure. Approximately 8 steel girder bridges were damaged in the 2001 Nisqually Earthquake in the state of Washington. These bridges were often constructed prior to 1975 and suffered bearing damage, spalling of concrete columns, and cracking at the abutments [8]. A number of studies have analytically evaluated the seismic response of typical MSSS and MSC steel girder bridges in order to better understand the seismic behavior and impact of modeling fidelity on the performance of these bridges [1,2,9, 10]. Using a linear elastic analysis, Dicleli and Bruneau [1] found that bearing stiffness significantly affects the response of MSSS steel girder bridges, and indicated that if pounding were considered in the longitudinal direction, there could be a large potential for shearing of bearings and span unseating. They stated that for MSC bridges, damage to steel bearings is probable, but may serve as an effective way of isolating the superstructure and preventing further column damage [9]. Nielson and DesRoches [2] performed nonlinear time history analyses with three-dimensional models and concluded that at a hazard level of 2% in 50 years, the MSSS steel girder bridge could be expected to have fractured or toppled steel fixed and rocker bearings, as well as excessive demands placed on the columns leading to cracking, spalling, and potential lapsplice failure. They concluded that considerable damage to the abutments was not expected. In a study evaluating the effect of steel bearing stiffness on the seismic response of MSSS steel girder bridges, Ala Saadeghvaziri and Rashidi [10] concluded that even in moderate earthquakes the bearing capacity would often be exceeded. As a result they recommended bearing replacement, increased seat lengths, or restrainer cables as potential seismic retrofits. In general, there have been numerous studies evaluating the efficacy of retrofits for other types of bridges, such as multi-frame concrete bridges, or the impact of retrofit on improving a particular component response. These approaches include strengthening, such as the use of steel restrainer cables [11,12]; capacity improvement, for example steel jacketing of columns [13,14]; force limitation or response
modification, such as implementing an isolation strategy [15, 16]; and other bridge retrofit approaches [17,18]. However, there have been relatively few studies that have explored the impact of various retrofit measures on the seismic response of multi-span simply supported or continuous steel girder bridges, which have been revealed to have potentially significant seismic vulnerabilities. Saiidi et al. [19] evaluated different restrainer design methods for MSSS bridges using two- and five-span steel girder bridges. They concluded that while the use of restrainers tended to decrease the critical relative displacements, the reduction was not necessarily proportional to the number of restrainers used, and that the bearing strength has a significant impact on the response. Maleki [20] considered the effects of side retainers (or keeper plates) on single span steel girder bridges with non-seismic elastomeric bearings, and concluded that the gap between the bearings and keeper should be considered in order to avoid underestimating the forces transferred to the substructure. Ductile end-diaphragm systems were investigated as retrofits which help us to provide energy dissipation and limit the forces transferred to vulnerable substructure elements in work by Zahrai and Bruneau [21]. They found that the retrofits showed promise, particularly for bridges having long spans and few steel girders. Even fewer studies have provided comparisons of the impact of different retrofit measures on the seismic performance of typical steel girder bridges. One study by DesRoches et al. [22] examined the response of steel girder bridges retrofit with elastomeric bearings, lead-rubber bearings, and restrainer cables. Detailed two-dimensional analytical models were evaluated using nonlinear time history analysis to assess the impact of the retrofits in the longitudinal direction. However, past studies have indicated that classes of steel girder bridges may potentially be vulnerable in the transverse direction [9]. To the authors’ knowledge, comparative studies have not been performed using detailed three-dimensional nonlinear models to assess the longitudinal and transverse seismic response of MSSS and MSC steel girder bridges with an array of retrofit measures—ranging from strengthening, to response modification, to increasing capacity. These types of comparative studies are essential to enhance our understanding
1871
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
of the seismic response of steel girder bridges with different retrofit strategies, and assist in identifying the relative impact of various retrofits on the response of critical components. Such insight supports decision making regarding retrofit of these classes of vulnerable bridges common in North America. To address this need, the study presented herein assesses and compares the impact of four commonly used retrofit measures, which target previously identified vulnerabilities, on the longitudinal and transverse seismic response of MSSS and MSC steel girder bridges. The three-dimensional models are subjected to nonlinear time history analysis considering the effects of pounding and the nonlinear behavior of critical components, such as the steel bearings and columns. This facilitates a comparison of the impact of steel jackets, shear keys, elastomeric isolation bearings, and restrainer cables on the seismic performance of each class of steel bridges. 2. Three-dimensional analytical modeling and retrofit measures for typical steel girder bridges 2.1. MSSS and MSC bridge characteristics and model The multi-span simply supported and multi-span continuous steel girder bridges examined in this study have characteristically non-seismic detailing, including multi-column bents having approximately a 1% longitudinal reinforcement ratio with widely spaced transverse ties providing limited confinement, and high-type steel fixed and expansion (rocker) bearings. The details and geometry for the bridges are based on past studies which have examined bridge plans from over 150 bridges and presented typical representative configurations for MSSS and MSC steel bridges found in the CSUS [23,24]. Fig. 2 shows a photo of a typical MSC steel girder bridge in Illinois, similar to the bridges analyzed in this study. The bridges considered are zero-skew, three-span concrete slab-on-girder bridges with 15 m wide decks consisting of eight steel plate girders. Each bridge is supported on three column bents, having 915-mmdiameter columns that are 4.6 m tall, and seat-type pile-bent abutments. The primary difference in the representative bridges for the MSSS and MSC bridges are the lengths and weights of typical spans, presence and size of gaps, and bearing configuration. The continuous bridge has expansion bearings at each abutment and fixed bearings between the continuous deck and pier, while the simply supported bridge has alternating fixed and expansion bearings supporting each deck. The span lengths, weights, and gaps for each bridge are listed in Table 1. Further details can be found in work by Nielson and DesRoches [2,24]. While modeling details for the as-built bridge are presented elsewhere [2,24], the general modeling scheme for both the asbuilt and the retrofitted bridge developed in the finite element analysis package OpenSees [25] is presented in Fig. 3. The deck is modeled as elastic beam–column elements, and the columns and bent caps are modeled using fiber elements. The steel fixed and expansion bearings are modeled using nonlinear translational springs, based on previous large-scale testing of similar bearings [26]. The pile foundation stiffness is considered using translational and rotational linear springs,
Fig. 2. Example multi-span continuous (MSC) slab-on-steel girder bridge found in Illinois, with location of rocker and fixed steel bearings identified. Table 1 Properties of MSSS and MSC bridge spans Length (m)
Weight (kN/m)
Gap (mm)
MSSS Bridge End Span Mid Span
12.2 24.4
39.0 52.0
38.1 25.4
MSC Bridge End Span Mid Span
30.3 30.3
68.3 68.3
76.2 N/A
and the abutments are represented in the active (tension), passive (compression), and transverse directions by nonlinear inelastic springs [27]. Pounding between decks and the deck and abutment is accounted for using a trilinear contact element. 2.2. Retrofit measures The retrofit measures evaluated for the classes of typical MSSS and MSC steel girder bridges include steel jacketing of columns, transverse shear keys, elastomeric isolation bearings, and steel restrainer cables. Examples of these retrofits installed in steel girder bridges are shown in Fig. 4. 2.2.1. Steel jackets Steel jacketing has been used as a retrofit measure to enhance the flexural ductility, shear strength, or performance of lap splices in reinforced concrete bridge columns. The steel jacket encasement provides enhanced confinement, which ultimately improves the ductility capacity through increased compressive strength and ultimate strain in the concrete [13]. For the MSSS and MSC concrete bridge columns, a grouted gap of 19.1 mm is assumed, and a minimum jacket thickness of 10 mm is sufficient to provide for lap-splice confinement and a significant increase in ductility capacity (ultimate curvature ductility demand of over µφ = 30). The steel jacketing is captured in the analytical bridge model by altering the fiber model for the columns to reflect the enhanced compressive strength and ultimate strain in the concrete fibers, estimated following Chai’s procedure [13]. Though it is not a desired effect, past testing has found that full height steel jackets also lead to an increase in column stiffness in the range of 20%–40% [18]. The elastic modulus of the section is increased in the analysis such that the steel jacketed column stiffness
1872
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
Fig. 3. Three-dimensional nonlinear analytical model of MSSS bridge focusing on modeling of retrofitted components: elastomeric bearings, restrainer cables, steel jackets, and shear keys, as well as select as-built components.
Fig. 4. Steel girder bridges retrofitted in CSUS with (a) steel jackets, (b) elastomeric bearings, (c) shear keys, and (d) restrainer cables.
is 30% larger than the as-built concrete columns. In general, while the analytical model is slightly affected by the use of steel jacketing, the primary impact of the retrofit is to increase the column ductility capacity. 2.2.2. Shear keys Shear keys serve to restrain the deck motion when a bridge is excited in the transverse direction and facilitate shear force transfer to the substructure. These devices are often concrete blocks provided at each bearing location, as the example in Fig. 4c illustrates. The assumed shear key design for this study follows a shear friction approach and limits the shear forces transferred to the substructure to half of the shear strength of the concrete columns, thus preventing excessive column demands. Shear keys are provided at every bearing (girder) location, as illustrated in Fig. 3 for one section of the bridge, and are
provided on at each abutment and bent beam. The shear keys are represented by a coulomb friction model, as shown in Fig. 3. 2.2.3. Elastomeric isolation bearings Elastomeric bearings are a form of isolation bearings that have been used in bridge and building construction for over 35 years [28]. The general concept of isolation is to shift the natural period of the structure out of the region of dominant earthquake energy, to increase the damping, and to limit the forces transferred from the superstructure to the substructure. It is often adopted as a retrofit scheme because isolation systems tend to reduce the need for costly retrofit of deficient pier and foundation elements. Koh and Kelly [29] have identified elastomeric bearings (or laminated-rubber bearings) as the simplest method of isolation, making them prime candidates for retrofit of typical MSSS and MSC steel girder bridges.
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
1873
The elastomeric isolation bearings are designed in this study targeting a fundamental period increase of 2–3 times the initial period for the MSSS and MSC steel bridges. The design results in a fundamental period increase (of the longitudinally dominant mode) from 0.29 to 0.72 s for the MSSS bridge, and from 0.44 to 1.06 s for the MSC bridge using 76.2 mm and 82.6 mm total height of elastomer, respectively. The analytical model of the bearings is shown in Fig. 3, following Kelly’s [16] work which showed that elastomeric bearings can be modeled using a bilinear element. Keeper plates are provided in the transverse direction, which have an approximate yield force of 140 kN and a gap of 12.7 mm between the keeper and bearing. 2.2.4. Restrainer cables Restrainer cables are devices which serve to limit the deck displacement and relative hinge openings in order to prevent unseating of bridge spans, or as fail-safe mechanisms to support bridge decks in case of unseating. They are often employed in bridges with insufficient seat widths, which has been found to be common in many MSSS and MSC steel girder bridges [10]. The restrainers are placed at the hinge locations at the deckabutment and deck-bent cap interfaces in the MSSS bridge, and only at the deck-abutment location in the MSC bridge. The restrainer cables are designed in such a way that they can support half of the weight of their adjacent deck as a simplistic restrainer design procedure, which is often employed in regions of low-to-moderate seismicity where steel girder bridges are common. This results in a total cable area of 2.0 cm2 at the deck-abutment and deck-bent cap location of the end spans and 5.2 cm2 at the deck-bent cap location of the center span of the MSSS bridge. For the MSC bridge a total area of 8.5 cm2 is provided at the end of the MSC bridge at the deck-abutment location. The restrainers are modeled as bilinear components that act only in tension after exhausting the initial 9.5 mm slack in the 228.6-cm-long cable, which is represented by a gap in the analytical model. This bilinear stress–strain model for the steel restrainer cables is shown in Fig. 3. 3. Seismic response of retrofitted MSSS & MSC steel girder bridges The seismic response of the retrofitted multi-span simply supported and multi-span continuous steel girder bridges is assessed through nonlinear time history analysis using a suite of synthetic ground motions that were developed by Wen and Wu [30]. They simulated ground motions for three cities in mid-America for use in seismic performance assessments of structures, such that the median of the response spectra match the uniform hazard response spectra corresponding to a probability of exceedance of 2% and 10% in 50 years. The suite of 10 ground motions representative of a 2% in 50 year event in Memphis, TN are used to assess the seismic response of the retrofitted MSSS and MSC bridges. Fig. 5 shows the acceleration response spectra for each ground motion in the suite along with a plot of the mean and mean plus or minus one standard deviation response spectra.
Fig. 5. Response spectra for the 2% in 50 year Memphis ground motion suite used in the analysis. Ground motion #5 is used to illustrate typical component responses.
3.1. Typical seismic responses of bridge components In order to illustrate the impact of the retrofit measures on the performance of various components in the bridge, typical nonlinear responses are shown for the as-built and retrofitted MSC and MSSS steel girder bridges subjected to ground motion #5, which has a peak ground acceleration of 0.47g. Note that the acceleration response spectrum for this ground motion is highlighted in Fig. 5. The ground motion is applied separately along the principle orthogonal axes of the bridges to facilitate a comparison of how the retrofits affect both longitudinal and transverse motion. 3.1.1. Columns As indicated in past studies, the ductility demands placed on the columns in both the as-built MSSS and MSC steel girder bridges is significant, and could be expected to result in considerable damage to the columns in the form of cracking, spalling, and lap-splice failure. Fig. 6 shows a plot of the moment-curvature response of the left column in the MSC bridge when loaded longitudinally. It is seen that through the use of elastomeric isolation bearings, the peak curvature ductility demands reduce from µ = 2.7 to µ = 0.6, and that with restrainer cables demands reduce to µ = 1.8. Hence the elastomeric bearings effectively isolate the superstructure from the substructure and reduce column demands, and in limiting the deck displacements the restrainers also improve the column response. Similar results are seen in the MSSS bridge, however the initial demands are of the order of µ = 1.3. 3.1.2. Expansion bearings The expansion bearing deformations are a direct result of the induced motion of the bridge decks, and the relative displacement between the bridge decks and abutments in the MSC bridge, as well as the decks and bent caps in the MSSS bridge. The longitudinal response of the steel rocker bearings in the MSSS steel girder bridge are shown in Fig. 7a. The restrainer cables reduce the peak deformations from 59.5 to 46.8 mm for the case examined. It is noted that the elastomeric isolation bearing at this location has a 36.2 mm deformation, though this is an altogether different component. The transverse
1874
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
Fig. 6. Curvature ductility demands placed on the columns in the as-built MSC bridge compared to the bridge retrofit with elastomeric bearings and restrainer cables (longitudinal response and loading).
(a) MSSS longitudinal response under longitudinal loading.
(b) MSC transverse response under transverse loading. Fig. 7. Effect of retrofit on the bearing deformations at (a) expansion bearing 1 in the longitudinal direction of the MSSS bridge, and (b) expansion bearing 2 in the transverse direction of the MSC bridge.
deformations are relatively small (4.9 mm) and tend to be unaffected by the various retrofits, other than replacement with elastomeric bearings. In the continuous bridge, the deformation of the expansion bearings located at the abutments are in excess of 100 mm and reduced to 80 mm by using the restrainer cables, though yielding of the cables occurs at this level. In the transverse direction, these bearings exhibit highly nonlinear behavior, with deformations of 47.7 mm reduced to 13.5 mm through the use of shear keys which engaged to limit out-ofplane motion of the deck and bearings (Fig. 7b). 3.1.3. Fixed bearings The fixed bearings also exhibit nonlinear behavior in the transverse direction for the MSC steel girder bridge, with deformations of 16.6 mm. As shown in Fig. 8, the deformations are reduced to 5.3 mm with shear keys. In the longitudinal direction, the fixed bearings on the MSC bridge deform very
little, as the bridge deck, fixed bearings, and columns tend to respond like a single degree of freedom system. However, for the MSSS bridge, the longitudinal deformations are of the order of 20 mm and transverse are 10 mm, yet are primarily unaffected by the retrofits. 3.1.4. Abutments The deformation of the abutment in active action (tension), passive action (compression), and in the transverse direction are relatively low for both the as-built MSSS and the MSC steel bridges. However, the use of elastomeric bearings in the MSC bridge increases the passive deformations from 17.4 to 29.0 mm due to pounding, while the restrainers reduce them to 1.0 mm. Fig. 9 illustrates that the initial transverse deformations are 6.0 mm, yet they increase to 10.0 mm with shear keys and 34.0 mm with elastomeric bearings having the keeper plate detail. This is a result of the physical mechanisms enabling
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
1875
Fig. 8. Transverse deformation of fixed bearings in the MSC bridge loaded transversely reduced by shear keys.
Fig. 9. Abutment response in the transverse direction when loaded in the transverse direction, increased due to shear keys and elastomeric bearings in the MSC steel girder bridge.
force transfer to the abutments in the transverse direction while they are limiting the displacement and deformation of other bridge components. 3.2. Comparison of retrofit impact for MSSS and MSC steel girder bridges Composite results of the peak component responses (bearing deformations, abutments deformations, column ductility demands) using the suite of 10 ground motions provide further insight on the effect of retrofit on the response statistics for each of the bridge types. Figs. 10 and 11 provide box plots of the peak component responses for the MSSS and MSC bridges in the as-built condition, and retrofit with restrainer cables, steel jackets, elastomeric bearings, and shear keys. A box plot presents a visual summary of several key statistical quantities of the responses and allows for comparison of how retrofitting the bridges affect the distribution of the peak component responses [31]. The boundary of the box plots represents the 25th and 75th percentiles of the peak component responses, while the whiskers plotted above and below the box represent the 5th and 95th percentile with outliers indicated. A black line is plotted in the box at the 50th percentile, or median value, and the dashed white line indicates the mean of the peak responses for the suite of time history analyses. Each bar in the figure represents the statistical quantities for a particular bridge in its asbuilt or retrofitted condition. 3.2.1. Multi-span simply supported steel girder bridge It is observed for the MSSS steel girder bridge in Fig. 10 that the mean peak expansion bearing deformations in the longitudinal direction are reduced by nearly 28% with the
restrainer cable retrofit, yet the other retrofits tend to have little effect (with the exception of the elastomeric bearings which actually replace the vulnerable steel bearings). The relatively small transverse expansion bearing deformations (4.1 mm), longitudinal (22.4 mm) and transverse (10.8 mm) fixed bearing deformations are not impacted by retrofit. However, these small levels of deformations may still be sufficient to produce moderate damage. For example, past tests of similar bearings revealed that longitudinal deformations of over 20 mm in the fixed bearings may be associated with cracking in the concrete pier, prying of the bearings, and severe deformations in the anchor bolts [26]. Replacement of these bearings with elastomeric bearings may be a viable retrofit option, where damage to the vulnerable steel bearings is altogether avoided. While this option leads to larger variability in the bearing response, as illustrated in Fig. 10, the level of deformation in the elastomeric isolation bearings is well under strain levels of 100% and within acceptable deformation ranges for these flexible bearings [32]. The demands placed on the columns under transverse loading are not anticipated to compromise the columns, but may lead to potential damage in the longitudinal direction. Although the restrainer cables reduce the mean peak demands by roughly 17% to 1.0, the upper quartile have exceeded the yield and near the level of expected cracking. These levels of damage may be considered acceptable, however, and the restrainers are found to be effective in both reducing the expansion bearing deformations and column demands for the MSSS steel bridge. The elastomeric bearings significantly reduce the column demands when loaded in each of the orthogonal directions, and effectively isolate the superstructure. Finally, although the
1876
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
Fig. 10. Box plot of component responses for retrofitted MSSS steel girder bridge. (AB = as-built; RC = restrainer cables; SJ = steel jackets; EB = elastomeric isolation bearings; SK = shear keys).
response of the bridge is not significantly affected by the steel jackets, the observed demands placed on the columns even at the 95th percentile are well below the enhanced ductility capacity of the columns. As found in past studies, the level of abutment deformation in the longitudinal and transverse directions of the as-built MSSS bridge are not likely to result in abutment damage [2]. While most of the retrofits do not impact the abutment deformations, Fig. 10 illustrates that the elastomeric bearings increase the passive and transverse action of the abutment due to pounding, as well as increase the variability in response. 3.2.2. Multi-span continuous steel girder bridge The mean peak transverse deformations of the steel fixed and expansion bearings in the MSC bridge (Fig. 11) reach levels of potential damage, at 13.1 mm and 41.7 mm respectively. At these levels, the shear keys become effective in restricting transverse motion and reducing the bearing deformations. Toppling of the rocker bearings in the longitudinal direction is a particular concern for this bridge type, because past studies have indicated potential instability at a level of 94 mm, which is exceeded on average [2,10]. While the restrainers tend to reduce the deformation and variability in response, the mean peak deformations are only decreased to 82.8 mm because the restrainers often yield at such levels. This indicates that seat extenders may be a viable option to provide additional support
length at the abutments in the case that the deck falls off of the supporting bearings. Alternatively, the use of elastomeric bearings is a feasible approach, since there is less potential for bearing damage. As seen in Fig. 11, the demands placed on the columns of the MSC steel girder bridge are excessive, particularly in the longitudinal direction. The elastomeric bearings effectively decrease the ductility demands to well below µ = 0.8, even at the 95th percentile, and reduce the variability in column demands. Additionally, the peak demands placed on the columns retrofit with steel jackets are within an acceptable level, ranging from roughly µ = 1.6 to µ = 2.6 at the 5th and 95th percentiles, which are well below the ultimate capacity estimated at over µ = 30. In passive action, the use of elastomeric bearings increases the mean peak abutment deformations by nearly 64% from 16.0 to 26.3 mm—again a result of pounding between the continuous deck and the abutments. The variability in response is also increased, with some responses exceeding 50 mm. The passive deformations are actually decreased through the use of the restrainer cable retrofit. The elastomeric bearings also increase the transverse deformations significantly to a mean of 20.4 mm, and the shear keys nearly double the deformations to 9.0 mm. Both increases can be attributed to the load transfer from the deck to the abutment through contact with either the shear key or bearing keeper plate.
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
1877
Fig. 11. Box plot of component responses for retrofitted MSC steel girder bridge. (AB = as-built; RC = restrainer cables; SJ = steel jackets; EB = elastomeric isolation bearings; SK = shear keys) (*Note: The longitudinal Elastomeric Bearing (EB) deformations at the location of FB 1 are not shown for scaling purposes but are the same as those plotted at EX 1.).
4. Conclusions High-fidelity three-dimensional nonlinear analytical models of typical multi-span simply supported and multi-span continuous steel girder bridges are analyzed with various retrofit measures to provide a comparative seismic performance evaluation. The use of restrainer cables, steel jackets, elastomeric isolation bearings, and shear keys, which target the identified deficiencies of these non-seismically designed bridges, are assessed for their impact on typical critical component responses. A synthetic ground motion suite representative of the 2% in 50 years hazard level for Memphis, TN is used to evaluate the response under different retrofit measures and support the discernment of a number of significant conclusions. The analyses reveal that shear keys are more effective for the MSC steel girder bridge which exhibits bearing vulnerability in the transverse direction. However, there is still a need to address the potential for bearing damage in the longitudinal direction as well as other components in the MSC bridge. Since the typical MSSS steel girder bridge exhibits predominant response and vulnerability in the longitudinal direction, the shear keys are of little impact. While the initial column demands in the longitudinal and transverse directions are considerably higher in the typical
MSC bridge, which responds similar to a single degree of freedom system with larger inertial loading, there is risk of column damage in both bridge types. The steel jackets provide ample capacity improvement in the columns and significantly reduce the likelihood for column damage in both the MSSS and MSC steel girder bridges, yet do not impact the response of other vulnerable components, such as the bearings. The elastomeric bearings are highly effective in reducing potential component damage in both bridges, and offer a viable retrofit option. They considerably reduce the column demands by isolating the superstructures, and diminish the likelihood for bearing damage by replacing the vulnerable steel bearings with flexible isolation bearings. It is noted, however, that their use increases the demands placed on the abutments in the transverse direction and in passive action due to pounding, which requires careful consideration, particularly for the MSC bridge. A less invasive and less costly retrofit measure using steel restrainer cables may be an acceptable alternative for the MSSS bridge, but is not as effective for the MSC bridge. In the MSSS steel bridge, the restrainer cables cut the mean peak expansion bearing deformations below the levels of expected bearing toppling, and reduce the column demands to levels nearing yield or potential cracking, which may be tolerable. It is noted that the deformation demands placed on the steel fixed bearings are not improved with the restrainers because
1878
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
of their magnitude relative to the restrainer slack. Although the deformations are relatively small and not indicative of unseating (10.8 mm longitudinal, 22.4 mm transverse) there is still potential for bearing damage. The restrainer cables are less effective for the MSC bridge which has higher bearing deformations and inertial loads which tend to yield the cables. This study provides insight on the impact of retrofit on the seismic performance of the two classes of steel girder bridges. It is exemplified that different retrofits may be more appropriate for either the MSSS or MSC steel girder bridge. Owing to the nature of the system response, the demands placed on the components in a typical MSC steel girder bridge in both the transverse and longitudinal direction pose a larger threat to the post-earthquake bridge performance, and present a challenge in identifying the most effective retrofit measure. In general, it is illustrated for both bridge types that while the performance of some components is improved through retrofit, others may be negatively impacted or not affected. This lends support to system-wide vulnerability assessments which consider the contribution of multiple components when assessing retrofit impact. The results of the analysis presented support decision making for future evaluations of the seismic risk to these bridges and identification of potential retrofit measures. It offers an enhanced understanding of the seismic response of these classes of retrofitted bridges and indicates key considerations for retrofit impact on critical components in the bridges. Acknowledgement This study has been supported by the Earthquake Engineering Research Centers program of the National Science Foundation under Award Number EEC-9701785 (Mid-America Earthquake Center). References [1] Dicleli M, Bruneau M. Seismic performance of multispan simply supported slab-on-girder steel highway bridges. Engineering Structures 1995;17(1):4–14. [2] Nielson B, Desroches R. Influence of modeling assumptions on the seismic response of multi-span simply supported steel girder bridges in moderate seismic zones. Engineering Structures 2006;28(8):1083–92. [3] Bruneau M. Performance of steel bridges during the 1995 HyogokenNanbu (Kobe, Japan) earthquake—a North American perspective. Engineering Structures 1998;20(12):1063–78. [4] NISEE Karl V. Steinbrugge Collection. Earthquake Engineering Research Center. University of California: Berkeley. [5] Housner GW, Thiel CC. The continuing challenge: Report on the performance of state bridges in the Northridge earthquake. Earthquake Spectra 1995;11(4):607–36. [6] Moehle J, et al. Highway bridges and traffic management. Earthquake Spectra 1995;11(S2):287–372. [7] Ghasemi H. Seismic protection of bridges. Public Roads 1999;62(5). [8] Ranf RT, Eberhard MO, Berry MP. Damage to bridges during the 2001 Nisqually earthquake. Tech. Rep. PEER 2001/15. Berkeley (CA): Pacific Earthquake Engineering Research Center; 2001.
[9] Dicleli M, Bruneau M. Seismic performance of single-span simply supported and continuous slab-on-girder steel highway bridges. Journal of Structural Engineering 1995;121(10):1497–506. [10] Ala Saadeghvaziri M, Rashidi S. Effect of steel bearings on seismic response of bridges in Eastern United States. In: Proc. 6th US national conference on earthquake engineering. EERI. 1998. [11] Selna LG, Malvar LJ. Bridge retrofit testing: Hinge cable restrainers. Journal of Structural Engineering 1989;115(4):920–34. [12] Saiidi M, Maragakis EA, Feng S. Parameters in bridge restrainer design for seismic retrofit. Journal of Structural Engineering 1996;122(1):61–8. [13] Chai YH, Priestley MN, Seible F. Seismic retrofit of circular bridge columns for enhanced flexural performance. ACI Structural Journal 1991; 88(5):572–84. [14] Zhang Y, Cofer WF, McLean DI. Analytical evaluation of retrofit strategies for multicolumn bridges. Journal of Bridge Engineering 1999; 4(2):143–50. [15] Ghobarah A, Ali HM. Seismic performance of highway bridges. Engineering Structures 1988;10(3):157–66. [16] Kelly JM. The analysis and design of elastomeric bearings for application in bridges. In: Proc. U.S.–Italy workshop on seismic protective systems for bridges. 1998. [17] FHWA. Seismic retrofitting manual for highway structures: Part 1 bridges. FHWA-RD-04-XXX. October 2005; Draft. [18] Priestley MJN, Seible F, Calvi GM. Seismic design and retrofit of bridges. New York: John Wiley and Sons; 1996. [19] Saiidi M, Randall M, Maragakis EA, Isakovic T. Seismic restrainer design methods for simply supported bridges. Journal of Bridge Engineering 2001;6(5):307–15. [20] Maleki S. Effect of side retainers on seismic response of bridges with elastomeric bearings. Journal of Bridge Engineering 2004;9(1):95–100. [21] Zahrai SM, Bruneau M. Ductile end-diaphragms for seismic retrofit of slab-on-girder steel bridges. Journal of Structural Engineering 1999; 125(1):71–80. [22] DesRoches R, Choi E, Leon RT, Pfeifer T. Seismic response of multiple span steel bridges in central and southeastern united states—II: Retrofitted. Journal of Bridge Engineering 2004;9(5):473–9. [23] Choi E. Seismic analysis and retrofit of mid-American bridges. Ph.D. thesis. Georgia Institute of Technology; 2002. [24] Nielson BG. Analytical fragility curves for highway bridges in moderate seismic zones. Ph.D. Georgia Institute of Technology; 2005. [25] McKenna F, Fenves GL. Open system for earthquake engineering simulation. Pacific earthquake engineering research center, version 1.6.2.c. 2005. [26] Mander JB, Kim DK, Chen SS, Premus GJ. Response of steel bridge bearings to the reversed cyclic loading. Tech. Rep. NCEER 96-0014. Buffalo (NY): National Center for Earthquake Engineering Research; 1996. [27] Martin GR, Yan L. Modeling passive earth pressure for bridge abutments. In: Earthquake-induced movements and seismic remediation of existing foundations and abutments. ASCE 1995 annual national convention, vol. Geotechnical special publication 55, San Diego (CA): ASCE; 1995. [28] Stanton JF, Roeder CW. Elastomeric bearings: An overview. Concrete International: Design and Construction 1992;14(1):41–6. [29] Koh CG, Kelly JM. Viscoelastic stability model for elastomeric isolation bearings. Journal of Structural Engineering 1989;115(2):285–302. [30] Wen YK, Wu CL. Uniform hazard ground motions for mid-American cities. Earthquake Spectra 2001;17(2):359–84. [31] Hines WW, Montgomery DC, Goldsman DM, Borror CM. Probability and statistics in engineering. 4th ed. Hoboken: John Wiley and Sons; 2003. [32] Mori A, Moss PJ, Cooke N, Carr AJ. The behavior of bearings used for seismic isolation under shear and axial load. Earthquake Spectra 1999; 15(2):199–224.
Three-dimensional nonlinear seismic performance evaluation of retrofit measures for typical steel girder bridges Jamie E. Padgett a,∗ , Reginald DesRoches b,1 a Department of Civil and Environmental Engineering, Rice University, 6100 Main Street, MS-318, Houston, TX 77005, USA b School of Civil and Environmental Engineering, Georgia Institute of Technology, 790 Atlantic Dr., Atlanta, GA 30332-0355, USA
Received 3 August 2006; received in revised form 11 December 2007; accepted 12 December 2007 Available online 22 January 2008
Abstract Steel girder bridges are common bridges in North America which have exhibited considerable vulnerabilities in past earthquake events. This paper conducts nonlinear time history analysis using detailed three-dimensional models of typical multi-span simply supported and multispan continuous steel girder bridges to evaluate the effectiveness of various retrofit strategies. Restrainer cables, steel jackets, shear keys, and elastomeric isolation bearings are assessed for their influence on the variability and peak longitudinal and transverse response of critical components in the bridges. The results indicate that different retrofit measures may be more effective for each class of bridges. For example, the restrainer cables are effective for the multi-span simply supported (MSSS) bridge, yet not the multi-span continuous (MSC) bridge; shear keys improve the transverse bearing response in the MSC bridge but are not effective in the MSSS bridge which exhibits less transverse vulnerability; and elastomeric bearings improve the response of the vulnerable columns in both the MSSS and MSC bridges yet lead to increased abutment demands in the MSC bridge. The study reveals that while a retrofit may have a positive influence on the targeted component, other critical components may be unaffected or negatively impacted. This lends support to vulnerability assessments that consider the impact of retrofit on system vulnerability reflecting the contribution of multiple components. c 2007 Elsevier Ltd. All rights reserved.
Keywords: Retrofit; Steel bridges; Seismic response; Steel jackets; Restrainer cables; Isolation; Shear keys
1. Introduction Steel bridges are often potentially vulnerable highway bridges found across various seismic regions of North America. Tens of thousands of steel bridges exist in the current North American bridge inventory, most of which have not been designed considering earthquake loading [1]. More than onethird of the highway bridges in the Central and Southeastern United States (CSUS) are slab-on-steel girder bridges and were constructed during the 1950s through the 1980s, prior to the initiation of seismic design practices in the region [2]. Concrete slab-on-steel girder bridges are also the most common steel bridges found in Canada, which like the CSUS are often supported on non-seismically designed concrete piers and ∗ Corresponding author. Tel.: +1 713 348 2325.
E-mail addresses: [email protected] (J.E. Padgett), [email protected] (R. DesRoches). 1 Tel.: +1 404 385 0826. c 2007 Elsevier Ltd. All rights reserved. 0141-0296/$ - see front matter doi:10.1016/j.engstruct.2007.12.011
abutments [3]. Multi-span simply supported (MSSS) and multispan continuous (MSC) slab-on-girder bridges account for a majority of the steel bridges in North American. While these bridge types are less common on the West Coast, there have been examples of damage to steel bridges in past earthquake events in this part of the US (Fig. 1). The state of California has hundreds of steel girder bridges, several of which experienced damage during the 1994 Northridge earthquake [5]. Typical damage included pounding damage, damage to bearings and fracture of anchor bolts, and damage to piers and abutments [6]. A number of steel bridges were damaged in the 1995 Hyogoken-Nanbu (Kobe, Japan) earthquake. These bridges have details and characteristics that are similar to the steel girder bridges found in North America and particularly the Central and Eastern US [7]. Damage to non-seismically detailed reinforced concrete substructures included flexural failure of columns due to inadequate confinement and shear failures due to inadequate
1870
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
Fig. 1. Damage to steel girder bridges in past earthquake events: (a) pounding damage in Northridge, (b) column damage and (c) shifted superstructure in Kobe [4], and (d) bearing damage in Nisqually (courtesy of WDOT).
shear reinforcement [3]. There was a significant amount of observed damage to steel bearings, as well as excessive deck displacements as a consequence of bearing failure and the loss of connection between the superstructure and substructure. Approximately 8 steel girder bridges were damaged in the 2001 Nisqually Earthquake in the state of Washington. These bridges were often constructed prior to 1975 and suffered bearing damage, spalling of concrete columns, and cracking at the abutments [8]. A number of studies have analytically evaluated the seismic response of typical MSSS and MSC steel girder bridges in order to better understand the seismic behavior and impact of modeling fidelity on the performance of these bridges [1,2,9, 10]. Using a linear elastic analysis, Dicleli and Bruneau [1] found that bearing stiffness significantly affects the response of MSSS steel girder bridges, and indicated that if pounding were considered in the longitudinal direction, there could be a large potential for shearing of bearings and span unseating. They stated that for MSC bridges, damage to steel bearings is probable, but may serve as an effective way of isolating the superstructure and preventing further column damage [9]. Nielson and DesRoches [2] performed nonlinear time history analyses with three-dimensional models and concluded that at a hazard level of 2% in 50 years, the MSSS steel girder bridge could be expected to have fractured or toppled steel fixed and rocker bearings, as well as excessive demands placed on the columns leading to cracking, spalling, and potential lapsplice failure. They concluded that considerable damage to the abutments was not expected. In a study evaluating the effect of steel bearing stiffness on the seismic response of MSSS steel girder bridges, Ala Saadeghvaziri and Rashidi [10] concluded that even in moderate earthquakes the bearing capacity would often be exceeded. As a result they recommended bearing replacement, increased seat lengths, or restrainer cables as potential seismic retrofits. In general, there have been numerous studies evaluating the efficacy of retrofits for other types of bridges, such as multi-frame concrete bridges, or the impact of retrofit on improving a particular component response. These approaches include strengthening, such as the use of steel restrainer cables [11,12]; capacity improvement, for example steel jacketing of columns [13,14]; force limitation or response
modification, such as implementing an isolation strategy [15, 16]; and other bridge retrofit approaches [17,18]. However, there have been relatively few studies that have explored the impact of various retrofit measures on the seismic response of multi-span simply supported or continuous steel girder bridges, which have been revealed to have potentially significant seismic vulnerabilities. Saiidi et al. [19] evaluated different restrainer design methods for MSSS bridges using two- and five-span steel girder bridges. They concluded that while the use of restrainers tended to decrease the critical relative displacements, the reduction was not necessarily proportional to the number of restrainers used, and that the bearing strength has a significant impact on the response. Maleki [20] considered the effects of side retainers (or keeper plates) on single span steel girder bridges with non-seismic elastomeric bearings, and concluded that the gap between the bearings and keeper should be considered in order to avoid underestimating the forces transferred to the substructure. Ductile end-diaphragm systems were investigated as retrofits which help us to provide energy dissipation and limit the forces transferred to vulnerable substructure elements in work by Zahrai and Bruneau [21]. They found that the retrofits showed promise, particularly for bridges having long spans and few steel girders. Even fewer studies have provided comparisons of the impact of different retrofit measures on the seismic performance of typical steel girder bridges. One study by DesRoches et al. [22] examined the response of steel girder bridges retrofit with elastomeric bearings, lead-rubber bearings, and restrainer cables. Detailed two-dimensional analytical models were evaluated using nonlinear time history analysis to assess the impact of the retrofits in the longitudinal direction. However, past studies have indicated that classes of steel girder bridges may potentially be vulnerable in the transverse direction [9]. To the authors’ knowledge, comparative studies have not been performed using detailed three-dimensional nonlinear models to assess the longitudinal and transverse seismic response of MSSS and MSC steel girder bridges with an array of retrofit measures—ranging from strengthening, to response modification, to increasing capacity. These types of comparative studies are essential to enhance our understanding
1871
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
of the seismic response of steel girder bridges with different retrofit strategies, and assist in identifying the relative impact of various retrofits on the response of critical components. Such insight supports decision making regarding retrofit of these classes of vulnerable bridges common in North America. To address this need, the study presented herein assesses and compares the impact of four commonly used retrofit measures, which target previously identified vulnerabilities, on the longitudinal and transverse seismic response of MSSS and MSC steel girder bridges. The three-dimensional models are subjected to nonlinear time history analysis considering the effects of pounding and the nonlinear behavior of critical components, such as the steel bearings and columns. This facilitates a comparison of the impact of steel jackets, shear keys, elastomeric isolation bearings, and restrainer cables on the seismic performance of each class of steel bridges. 2. Three-dimensional analytical modeling and retrofit measures for typical steel girder bridges 2.1. MSSS and MSC bridge characteristics and model The multi-span simply supported and multi-span continuous steel girder bridges examined in this study have characteristically non-seismic detailing, including multi-column bents having approximately a 1% longitudinal reinforcement ratio with widely spaced transverse ties providing limited confinement, and high-type steel fixed and expansion (rocker) bearings. The details and geometry for the bridges are based on past studies which have examined bridge plans from over 150 bridges and presented typical representative configurations for MSSS and MSC steel bridges found in the CSUS [23,24]. Fig. 2 shows a photo of a typical MSC steel girder bridge in Illinois, similar to the bridges analyzed in this study. The bridges considered are zero-skew, three-span concrete slab-on-girder bridges with 15 m wide decks consisting of eight steel plate girders. Each bridge is supported on three column bents, having 915-mmdiameter columns that are 4.6 m tall, and seat-type pile-bent abutments. The primary difference in the representative bridges for the MSSS and MSC bridges are the lengths and weights of typical spans, presence and size of gaps, and bearing configuration. The continuous bridge has expansion bearings at each abutment and fixed bearings between the continuous deck and pier, while the simply supported bridge has alternating fixed and expansion bearings supporting each deck. The span lengths, weights, and gaps for each bridge are listed in Table 1. Further details can be found in work by Nielson and DesRoches [2,24]. While modeling details for the as-built bridge are presented elsewhere [2,24], the general modeling scheme for both the asbuilt and the retrofitted bridge developed in the finite element analysis package OpenSees [25] is presented in Fig. 3. The deck is modeled as elastic beam–column elements, and the columns and bent caps are modeled using fiber elements. The steel fixed and expansion bearings are modeled using nonlinear translational springs, based on previous large-scale testing of similar bearings [26]. The pile foundation stiffness is considered using translational and rotational linear springs,
Fig. 2. Example multi-span continuous (MSC) slab-on-steel girder bridge found in Illinois, with location of rocker and fixed steel bearings identified. Table 1 Properties of MSSS and MSC bridge spans Length (m)
Weight (kN/m)
Gap (mm)
MSSS Bridge End Span Mid Span
12.2 24.4
39.0 52.0
38.1 25.4
MSC Bridge End Span Mid Span
30.3 30.3
68.3 68.3
76.2 N/A
and the abutments are represented in the active (tension), passive (compression), and transverse directions by nonlinear inelastic springs [27]. Pounding between decks and the deck and abutment is accounted for using a trilinear contact element. 2.2. Retrofit measures The retrofit measures evaluated for the classes of typical MSSS and MSC steel girder bridges include steel jacketing of columns, transverse shear keys, elastomeric isolation bearings, and steel restrainer cables. Examples of these retrofits installed in steel girder bridges are shown in Fig. 4. 2.2.1. Steel jackets Steel jacketing has been used as a retrofit measure to enhance the flexural ductility, shear strength, or performance of lap splices in reinforced concrete bridge columns. The steel jacket encasement provides enhanced confinement, which ultimately improves the ductility capacity through increased compressive strength and ultimate strain in the concrete [13]. For the MSSS and MSC concrete bridge columns, a grouted gap of 19.1 mm is assumed, and a minimum jacket thickness of 10 mm is sufficient to provide for lap-splice confinement and a significant increase in ductility capacity (ultimate curvature ductility demand of over µφ = 30). The steel jacketing is captured in the analytical bridge model by altering the fiber model for the columns to reflect the enhanced compressive strength and ultimate strain in the concrete fibers, estimated following Chai’s procedure [13]. Though it is not a desired effect, past testing has found that full height steel jackets also lead to an increase in column stiffness in the range of 20%–40% [18]. The elastic modulus of the section is increased in the analysis such that the steel jacketed column stiffness
1872
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
Fig. 3. Three-dimensional nonlinear analytical model of MSSS bridge focusing on modeling of retrofitted components: elastomeric bearings, restrainer cables, steel jackets, and shear keys, as well as select as-built components.
Fig. 4. Steel girder bridges retrofitted in CSUS with (a) steel jackets, (b) elastomeric bearings, (c) shear keys, and (d) restrainer cables.
is 30% larger than the as-built concrete columns. In general, while the analytical model is slightly affected by the use of steel jacketing, the primary impact of the retrofit is to increase the column ductility capacity. 2.2.2. Shear keys Shear keys serve to restrain the deck motion when a bridge is excited in the transverse direction and facilitate shear force transfer to the substructure. These devices are often concrete blocks provided at each bearing location, as the example in Fig. 4c illustrates. The assumed shear key design for this study follows a shear friction approach and limits the shear forces transferred to the substructure to half of the shear strength of the concrete columns, thus preventing excessive column demands. Shear keys are provided at every bearing (girder) location, as illustrated in Fig. 3 for one section of the bridge, and are
provided on at each abutment and bent beam. The shear keys are represented by a coulomb friction model, as shown in Fig. 3. 2.2.3. Elastomeric isolation bearings Elastomeric bearings are a form of isolation bearings that have been used in bridge and building construction for over 35 years [28]. The general concept of isolation is to shift the natural period of the structure out of the region of dominant earthquake energy, to increase the damping, and to limit the forces transferred from the superstructure to the substructure. It is often adopted as a retrofit scheme because isolation systems tend to reduce the need for costly retrofit of deficient pier and foundation elements. Koh and Kelly [29] have identified elastomeric bearings (or laminated-rubber bearings) as the simplest method of isolation, making them prime candidates for retrofit of typical MSSS and MSC steel girder bridges.
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
1873
The elastomeric isolation bearings are designed in this study targeting a fundamental period increase of 2–3 times the initial period for the MSSS and MSC steel bridges. The design results in a fundamental period increase (of the longitudinally dominant mode) from 0.29 to 0.72 s for the MSSS bridge, and from 0.44 to 1.06 s for the MSC bridge using 76.2 mm and 82.6 mm total height of elastomer, respectively. The analytical model of the bearings is shown in Fig. 3, following Kelly’s [16] work which showed that elastomeric bearings can be modeled using a bilinear element. Keeper plates are provided in the transverse direction, which have an approximate yield force of 140 kN and a gap of 12.7 mm between the keeper and bearing. 2.2.4. Restrainer cables Restrainer cables are devices which serve to limit the deck displacement and relative hinge openings in order to prevent unseating of bridge spans, or as fail-safe mechanisms to support bridge decks in case of unseating. They are often employed in bridges with insufficient seat widths, which has been found to be common in many MSSS and MSC steel girder bridges [10]. The restrainers are placed at the hinge locations at the deckabutment and deck-bent cap interfaces in the MSSS bridge, and only at the deck-abutment location in the MSC bridge. The restrainer cables are designed in such a way that they can support half of the weight of their adjacent deck as a simplistic restrainer design procedure, which is often employed in regions of low-to-moderate seismicity where steel girder bridges are common. This results in a total cable area of 2.0 cm2 at the deck-abutment and deck-bent cap location of the end spans and 5.2 cm2 at the deck-bent cap location of the center span of the MSSS bridge. For the MSC bridge a total area of 8.5 cm2 is provided at the end of the MSC bridge at the deck-abutment location. The restrainers are modeled as bilinear components that act only in tension after exhausting the initial 9.5 mm slack in the 228.6-cm-long cable, which is represented by a gap in the analytical model. This bilinear stress–strain model for the steel restrainer cables is shown in Fig. 3. 3. Seismic response of retrofitted MSSS & MSC steel girder bridges The seismic response of the retrofitted multi-span simply supported and multi-span continuous steel girder bridges is assessed through nonlinear time history analysis using a suite of synthetic ground motions that were developed by Wen and Wu [30]. They simulated ground motions for three cities in mid-America for use in seismic performance assessments of structures, such that the median of the response spectra match the uniform hazard response spectra corresponding to a probability of exceedance of 2% and 10% in 50 years. The suite of 10 ground motions representative of a 2% in 50 year event in Memphis, TN are used to assess the seismic response of the retrofitted MSSS and MSC bridges. Fig. 5 shows the acceleration response spectra for each ground motion in the suite along with a plot of the mean and mean plus or minus one standard deviation response spectra.
Fig. 5. Response spectra for the 2% in 50 year Memphis ground motion suite used in the analysis. Ground motion #5 is used to illustrate typical component responses.
3.1. Typical seismic responses of bridge components In order to illustrate the impact of the retrofit measures on the performance of various components in the bridge, typical nonlinear responses are shown for the as-built and retrofitted MSC and MSSS steel girder bridges subjected to ground motion #5, which has a peak ground acceleration of 0.47g. Note that the acceleration response spectrum for this ground motion is highlighted in Fig. 5. The ground motion is applied separately along the principle orthogonal axes of the bridges to facilitate a comparison of how the retrofits affect both longitudinal and transverse motion. 3.1.1. Columns As indicated in past studies, the ductility demands placed on the columns in both the as-built MSSS and MSC steel girder bridges is significant, and could be expected to result in considerable damage to the columns in the form of cracking, spalling, and lap-splice failure. Fig. 6 shows a plot of the moment-curvature response of the left column in the MSC bridge when loaded longitudinally. It is seen that through the use of elastomeric isolation bearings, the peak curvature ductility demands reduce from µ = 2.7 to µ = 0.6, and that with restrainer cables demands reduce to µ = 1.8. Hence the elastomeric bearings effectively isolate the superstructure from the substructure and reduce column demands, and in limiting the deck displacements the restrainers also improve the column response. Similar results are seen in the MSSS bridge, however the initial demands are of the order of µ = 1.3. 3.1.2. Expansion bearings The expansion bearing deformations are a direct result of the induced motion of the bridge decks, and the relative displacement between the bridge decks and abutments in the MSC bridge, as well as the decks and bent caps in the MSSS bridge. The longitudinal response of the steel rocker bearings in the MSSS steel girder bridge are shown in Fig. 7a. The restrainer cables reduce the peak deformations from 59.5 to 46.8 mm for the case examined. It is noted that the elastomeric isolation bearing at this location has a 36.2 mm deformation, though this is an altogether different component. The transverse
1874
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
Fig. 6. Curvature ductility demands placed on the columns in the as-built MSC bridge compared to the bridge retrofit with elastomeric bearings and restrainer cables (longitudinal response and loading).
(a) MSSS longitudinal response under longitudinal loading.
(b) MSC transverse response under transverse loading. Fig. 7. Effect of retrofit on the bearing deformations at (a) expansion bearing 1 in the longitudinal direction of the MSSS bridge, and (b) expansion bearing 2 in the transverse direction of the MSC bridge.
deformations are relatively small (4.9 mm) and tend to be unaffected by the various retrofits, other than replacement with elastomeric bearings. In the continuous bridge, the deformation of the expansion bearings located at the abutments are in excess of 100 mm and reduced to 80 mm by using the restrainer cables, though yielding of the cables occurs at this level. In the transverse direction, these bearings exhibit highly nonlinear behavior, with deformations of 47.7 mm reduced to 13.5 mm through the use of shear keys which engaged to limit out-ofplane motion of the deck and bearings (Fig. 7b). 3.1.3. Fixed bearings The fixed bearings also exhibit nonlinear behavior in the transverse direction for the MSC steel girder bridge, with deformations of 16.6 mm. As shown in Fig. 8, the deformations are reduced to 5.3 mm with shear keys. In the longitudinal direction, the fixed bearings on the MSC bridge deform very
little, as the bridge deck, fixed bearings, and columns tend to respond like a single degree of freedom system. However, for the MSSS bridge, the longitudinal deformations are of the order of 20 mm and transverse are 10 mm, yet are primarily unaffected by the retrofits. 3.1.4. Abutments The deformation of the abutment in active action (tension), passive action (compression), and in the transverse direction are relatively low for both the as-built MSSS and the MSC steel bridges. However, the use of elastomeric bearings in the MSC bridge increases the passive deformations from 17.4 to 29.0 mm due to pounding, while the restrainers reduce them to 1.0 mm. Fig. 9 illustrates that the initial transverse deformations are 6.0 mm, yet they increase to 10.0 mm with shear keys and 34.0 mm with elastomeric bearings having the keeper plate detail. This is a result of the physical mechanisms enabling
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
1875
Fig. 8. Transverse deformation of fixed bearings in the MSC bridge loaded transversely reduced by shear keys.
Fig. 9. Abutment response in the transverse direction when loaded in the transverse direction, increased due to shear keys and elastomeric bearings in the MSC steel girder bridge.
force transfer to the abutments in the transverse direction while they are limiting the displacement and deformation of other bridge components. 3.2. Comparison of retrofit impact for MSSS and MSC steel girder bridges Composite results of the peak component responses (bearing deformations, abutments deformations, column ductility demands) using the suite of 10 ground motions provide further insight on the effect of retrofit on the response statistics for each of the bridge types. Figs. 10 and 11 provide box plots of the peak component responses for the MSSS and MSC bridges in the as-built condition, and retrofit with restrainer cables, steel jackets, elastomeric bearings, and shear keys. A box plot presents a visual summary of several key statistical quantities of the responses and allows for comparison of how retrofitting the bridges affect the distribution of the peak component responses [31]. The boundary of the box plots represents the 25th and 75th percentiles of the peak component responses, while the whiskers plotted above and below the box represent the 5th and 95th percentile with outliers indicated. A black line is plotted in the box at the 50th percentile, or median value, and the dashed white line indicates the mean of the peak responses for the suite of time history analyses. Each bar in the figure represents the statistical quantities for a particular bridge in its asbuilt or retrofitted condition. 3.2.1. Multi-span simply supported steel girder bridge It is observed for the MSSS steel girder bridge in Fig. 10 that the mean peak expansion bearing deformations in the longitudinal direction are reduced by nearly 28% with the
restrainer cable retrofit, yet the other retrofits tend to have little effect (with the exception of the elastomeric bearings which actually replace the vulnerable steel bearings). The relatively small transverse expansion bearing deformations (4.1 mm), longitudinal (22.4 mm) and transverse (10.8 mm) fixed bearing deformations are not impacted by retrofit. However, these small levels of deformations may still be sufficient to produce moderate damage. For example, past tests of similar bearings revealed that longitudinal deformations of over 20 mm in the fixed bearings may be associated with cracking in the concrete pier, prying of the bearings, and severe deformations in the anchor bolts [26]. Replacement of these bearings with elastomeric bearings may be a viable retrofit option, where damage to the vulnerable steel bearings is altogether avoided. While this option leads to larger variability in the bearing response, as illustrated in Fig. 10, the level of deformation in the elastomeric isolation bearings is well under strain levels of 100% and within acceptable deformation ranges for these flexible bearings [32]. The demands placed on the columns under transverse loading are not anticipated to compromise the columns, but may lead to potential damage in the longitudinal direction. Although the restrainer cables reduce the mean peak demands by roughly 17% to 1.0, the upper quartile have exceeded the yield and near the level of expected cracking. These levels of damage may be considered acceptable, however, and the restrainers are found to be effective in both reducing the expansion bearing deformations and column demands for the MSSS steel bridge. The elastomeric bearings significantly reduce the column demands when loaded in each of the orthogonal directions, and effectively isolate the superstructure. Finally, although the
1876
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
Fig. 10. Box plot of component responses for retrofitted MSSS steel girder bridge. (AB = as-built; RC = restrainer cables; SJ = steel jackets; EB = elastomeric isolation bearings; SK = shear keys).
response of the bridge is not significantly affected by the steel jackets, the observed demands placed on the columns even at the 95th percentile are well below the enhanced ductility capacity of the columns. As found in past studies, the level of abutment deformation in the longitudinal and transverse directions of the as-built MSSS bridge are not likely to result in abutment damage [2]. While most of the retrofits do not impact the abutment deformations, Fig. 10 illustrates that the elastomeric bearings increase the passive and transverse action of the abutment due to pounding, as well as increase the variability in response. 3.2.2. Multi-span continuous steel girder bridge The mean peak transverse deformations of the steel fixed and expansion bearings in the MSC bridge (Fig. 11) reach levels of potential damage, at 13.1 mm and 41.7 mm respectively. At these levels, the shear keys become effective in restricting transverse motion and reducing the bearing deformations. Toppling of the rocker bearings in the longitudinal direction is a particular concern for this bridge type, because past studies have indicated potential instability at a level of 94 mm, which is exceeded on average [2,10]. While the restrainers tend to reduce the deformation and variability in response, the mean peak deformations are only decreased to 82.8 mm because the restrainers often yield at such levels. This indicates that seat extenders may be a viable option to provide additional support
length at the abutments in the case that the deck falls off of the supporting bearings. Alternatively, the use of elastomeric bearings is a feasible approach, since there is less potential for bearing damage. As seen in Fig. 11, the demands placed on the columns of the MSC steel girder bridge are excessive, particularly in the longitudinal direction. The elastomeric bearings effectively decrease the ductility demands to well below µ = 0.8, even at the 95th percentile, and reduce the variability in column demands. Additionally, the peak demands placed on the columns retrofit with steel jackets are within an acceptable level, ranging from roughly µ = 1.6 to µ = 2.6 at the 5th and 95th percentiles, which are well below the ultimate capacity estimated at over µ = 30. In passive action, the use of elastomeric bearings increases the mean peak abutment deformations by nearly 64% from 16.0 to 26.3 mm—again a result of pounding between the continuous deck and the abutments. The variability in response is also increased, with some responses exceeding 50 mm. The passive deformations are actually decreased through the use of the restrainer cable retrofit. The elastomeric bearings also increase the transverse deformations significantly to a mean of 20.4 mm, and the shear keys nearly double the deformations to 9.0 mm. Both increases can be attributed to the load transfer from the deck to the abutment through contact with either the shear key or bearing keeper plate.
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
1877
Fig. 11. Box plot of component responses for retrofitted MSC steel girder bridge. (AB = as-built; RC = restrainer cables; SJ = steel jackets; EB = elastomeric isolation bearings; SK = shear keys) (*Note: The longitudinal Elastomeric Bearing (EB) deformations at the location of FB 1 are not shown for scaling purposes but are the same as those plotted at EX 1.).
4. Conclusions High-fidelity three-dimensional nonlinear analytical models of typical multi-span simply supported and multi-span continuous steel girder bridges are analyzed with various retrofit measures to provide a comparative seismic performance evaluation. The use of restrainer cables, steel jackets, elastomeric isolation bearings, and shear keys, which target the identified deficiencies of these non-seismically designed bridges, are assessed for their impact on typical critical component responses. A synthetic ground motion suite representative of the 2% in 50 years hazard level for Memphis, TN is used to evaluate the response under different retrofit measures and support the discernment of a number of significant conclusions. The analyses reveal that shear keys are more effective for the MSC steel girder bridge which exhibits bearing vulnerability in the transverse direction. However, there is still a need to address the potential for bearing damage in the longitudinal direction as well as other components in the MSC bridge. Since the typical MSSS steel girder bridge exhibits predominant response and vulnerability in the longitudinal direction, the shear keys are of little impact. While the initial column demands in the longitudinal and transverse directions are considerably higher in the typical
MSC bridge, which responds similar to a single degree of freedom system with larger inertial loading, there is risk of column damage in both bridge types. The steel jackets provide ample capacity improvement in the columns and significantly reduce the likelihood for column damage in both the MSSS and MSC steel girder bridges, yet do not impact the response of other vulnerable components, such as the bearings. The elastomeric bearings are highly effective in reducing potential component damage in both bridges, and offer a viable retrofit option. They considerably reduce the column demands by isolating the superstructures, and diminish the likelihood for bearing damage by replacing the vulnerable steel bearings with flexible isolation bearings. It is noted, however, that their use increases the demands placed on the abutments in the transverse direction and in passive action due to pounding, which requires careful consideration, particularly for the MSC bridge. A less invasive and less costly retrofit measure using steel restrainer cables may be an acceptable alternative for the MSSS bridge, but is not as effective for the MSC bridge. In the MSSS steel bridge, the restrainer cables cut the mean peak expansion bearing deformations below the levels of expected bearing toppling, and reduce the column demands to levels nearing yield or potential cracking, which may be tolerable. It is noted that the deformation demands placed on the steel fixed bearings are not improved with the restrainers because
1878
J.E. Padgett, R. DesRoches / Engineering Structures 30 (2008) 1869–1878
of their magnitude relative to the restrainer slack. Although the deformations are relatively small and not indicative of unseating (10.8 mm longitudinal, 22.4 mm transverse) there is still potential for bearing damage. The restrainer cables are less effective for the MSC bridge which has higher bearing deformations and inertial loads which tend to yield the cables. This study provides insight on the impact of retrofit on the seismic performance of the two classes of steel girder bridges. It is exemplified that different retrofits may be more appropriate for either the MSSS or MSC steel girder bridge. Owing to the nature of the system response, the demands placed on the components in a typical MSC steel girder bridge in both the transverse and longitudinal direction pose a larger threat to the post-earthquake bridge performance, and present a challenge in identifying the most effective retrofit measure. In general, it is illustrated for both bridge types that while the performance of some components is improved through retrofit, others may be negatively impacted or not affected. This lends support to system-wide vulnerability assessments which consider the contribution of multiple components when assessing retrofit impact. The results of the analysis presented support decision making for future evaluations of the seismic risk to these bridges and identification of potential retrofit measures. It offers an enhanced understanding of the seismic response of these classes of retrofitted bridges and indicates key considerations for retrofit impact on critical components in the bridges. Acknowledgement This study has been supported by the Earthquake Engineering Research Centers program of the National Science Foundation under Award Number EEC-9701785 (Mid-America Earthquake Center). References [1] Dicleli M, Bruneau M. Seismic performance of multispan simply supported slab-on-girder steel highway bridges. Engineering Structures 1995;17(1):4–14. [2] Nielson B, Desroches R. Influence of modeling assumptions on the seismic response of multi-span simply supported steel girder bridges in moderate seismic zones. Engineering Structures 2006;28(8):1083–92. [3] Bruneau M. Performance of steel bridges during the 1995 HyogokenNanbu (Kobe, Japan) earthquake—a North American perspective. Engineering Structures 1998;20(12):1063–78. [4] NISEE Karl V. Steinbrugge Collection. Earthquake Engineering Research Center. University of California: Berkeley. [5] Housner GW, Thiel CC. The continuing challenge: Report on the performance of state bridges in the Northridge earthquake. Earthquake Spectra 1995;11(4):607–36. [6] Moehle J, et al. Highway bridges and traffic management. Earthquake Spectra 1995;11(S2):287–372. [7] Ghasemi H. Seismic protection of bridges. Public Roads 1999;62(5). [8] Ranf RT, Eberhard MO, Berry MP. Damage to bridges during the 2001 Nisqually earthquake. Tech. Rep. PEER 2001/15. Berkeley (CA): Pacific Earthquake Engineering Research Center; 2001.
[9] Dicleli M, Bruneau M. Seismic performance of single-span simply supported and continuous slab-on-girder steel highway bridges. Journal of Structural Engineering 1995;121(10):1497–506. [10] Ala Saadeghvaziri M, Rashidi S. Effect of steel bearings on seismic response of bridges in Eastern United States. In: Proc. 6th US national conference on earthquake engineering. EERI. 1998. [11] Selna LG, Malvar LJ. Bridge retrofit testing: Hinge cable restrainers. Journal of Structural Engineering 1989;115(4):920–34. [12] Saiidi M, Maragakis EA, Feng S. Parameters in bridge restrainer design for seismic retrofit. Journal of Structural Engineering 1996;122(1):61–8. [13] Chai YH, Priestley MN, Seible F. Seismic retrofit of circular bridge columns for enhanced flexural performance. ACI Structural Journal 1991; 88(5):572–84. [14] Zhang Y, Cofer WF, McLean DI. Analytical evaluation of retrofit strategies for multicolumn bridges. Journal of Bridge Engineering 1999; 4(2):143–50. [15] Ghobarah A, Ali HM. Seismic performance of highway bridges. Engineering Structures 1988;10(3):157–66. [16] Kelly JM. The analysis and design of elastomeric bearings for application in bridges. In: Proc. U.S.–Italy workshop on seismic protective systems for bridges. 1998. [17] FHWA. Seismic retrofitting manual for highway structures: Part 1 bridges. FHWA-RD-04-XXX. October 2005; Draft. [18] Priestley MJN, Seible F, Calvi GM. Seismic design and retrofit of bridges. New York: John Wiley and Sons; 1996. [19] Saiidi M, Randall M, Maragakis EA, Isakovic T. Seismic restrainer design methods for simply supported bridges. Journal of Bridge Engineering 2001;6(5):307–15. [20] Maleki S. Effect of side retainers on seismic response of bridges with elastomeric bearings. Journal of Bridge Engineering 2004;9(1):95–100. [21] Zahrai SM, Bruneau M. Ductile end-diaphragms for seismic retrofit of slab-on-girder steel bridges. Journal of Structural Engineering 1999; 125(1):71–80. [22] DesRoches R, Choi E, Leon RT, Pfeifer T. Seismic response of multiple span steel bridges in central and southeastern united states—II: Retrofitted. Journal of Bridge Engineering 2004;9(5):473–9. [23] Choi E. Seismic analysis and retrofit of mid-American bridges. Ph.D. thesis. Georgia Institute of Technology; 2002. [24] Nielson BG. Analytical fragility curves for highway bridges in moderate seismic zones. Ph.D. Georgia Institute of Technology; 2005. [25] McKenna F, Fenves GL. Open system for earthquake engineering simulation. Pacific earthquake engineering research center, version 1.6.2.c. 2005. [26] Mander JB, Kim DK, Chen SS, Premus GJ. Response of steel bridge bearings to the reversed cyclic loading. Tech. Rep. NCEER 96-0014. Buffalo (NY): National Center for Earthquake Engineering Research; 1996. [27] Martin GR, Yan L. Modeling passive earth pressure for bridge abutments. In: Earthquake-induced movements and seismic remediation of existing foundations and abutments. ASCE 1995 annual national convention, vol. Geotechnical special publication 55, San Diego (CA): ASCE; 1995. [28] Stanton JF, Roeder CW. Elastomeric bearings: An overview. Concrete International: Design and Construction 1992;14(1):41–6. [29] Koh CG, Kelly JM. Viscoelastic stability model for elastomeric isolation bearings. Journal of Structural Engineering 1989;115(2):285–302. [30] Wen YK, Wu CL. Uniform hazard ground motions for mid-American cities. Earthquake Spectra 2001;17(2):359–84. [31] Hines WW, Montgomery DC, Goldsman DM, Borror CM. Probability and statistics in engineering. 4th ed. Hoboken: John Wiley and Sons; 2003. [32] Mori A, Moss PJ, Cooke N, Carr AJ. The behavior of bearings used for seismic isolation under shear and axial load. Earthquake Spectra 1999; 15(2):199–224.
