Title: Seismic Design Recommendations for Steel Girder Bridges with Integral Abutments

G. Pekcan, A. Itani & E. Monzon 0 0 0 Title: Seismic Design Recommendations for Steel Girder Bridges with Integral Abutments Authors: Gokhan Pekcan, Ph.D. (CORRESPONDING AUTHOR) Assistant Professor Department of Civil and Environmental Engineering University of Nevada, Reno / Reno, NV Ph: -- Fx: --0 E-mail: pekcan@unr.edu Ahmad M. Itani, Ph.D., P.E., S.E., F.ASCE Professor-Steel Structures Department of Civil and Environmental Engineering University of Nevada, Reno / Reno, NV Ph: -- Fx: --0 E-mail: itani@unr.edu Eric V. Monzon Graduate Research Assistant Department of Civil and Environmental Engineering University of Nevada, Reno Ph: --00 E-mail: monzone@unr.nevada.edu Word Count: 0 TEXT + ()x0 =000 FIGURES + ()x0 = 00 TABLES =,0

G. Pekcan, A. Itani & E. Monzon 0 0 0 0 Seismic Design Recommendations for Steel Girder Bridges with Integral Abutments Gokhan Pekcan, Ahmad M. Itani, and Eric Monzon ABSTRACT This paper discusses the results of the longitudinal seismic behavior of straight bridges with integral abutments. Detailed nonlinear finite element (FE) models were utilized to establish the flexibility (translational and rotational) of the steel plate girders and the abutment connections. These connection springs were incorporated in a three dimensional global model of an integral abutment bridge to study the structural dynamics characteristics, as well as the seismic load path and distribution to piles, soil, girder elements. A procedure was demonstrated to determine embedment length of steel girders in the abutment to ensure the connection rigidity and more importantly to ensure that the piles will develop their ultimate flexural capacity. INTRODUCTION Integral abutment bridges (IABs) are jointless bridges where the deck is continuous and connected monolithically to abutment wall with a moment-resisting connection. One line (row) of vertical piles under the abutment wall is often used to carry vertical bridge loads. The rationale for integral abutments is clear as presented by Wasserman (00) and Wasserman and Walker (). Throughout the United States, IABs are becoming a design choice for mostly short to moderate spans (Civjan, et al., 00; Maruri and Petro, 00). IABs designed and constructed for a variety of configurations have typically performed well when subjected to dead, live and thermal induced cyclic loading. (e.g. Soltani and Kukreti, ; Burke, ; Kunin and Alampalli, 000; Arockiasamy et al., 00; Conboy and Stoothoff, 00). However, there is a lack of information on the seismic system response of steel bridges with integral abutments. Seismic Response Characteristics: Experimental and Analytical Studies The benefits of IABs with respect to their seismic performance compared with conventional simply supported bridges are: increased redundancy, larger damping due to nonlinear cyclic soil-pile-structure interaction (pile support and abutment backfill), smaller displacements, and elimination of unseating potential. Because of the presumed rigid connections between the bridge deck and the abutments, integral bridges are expected to have improved seismic resistance compared to jointed bridges (Hoppe and Gomez, ). The highly nonlinear soil-structure and pile interaction of integral abutment is stated as one of the most complex issues as well as one of the main sources of uncertainty. This interaction is inherently nonlinear and depends on the magnitude and nature of the abutment, soil and pile deformations (translational, rotational). Limited experimental and analytical research has been conducted (e.g.: England et al., 000; Faraji et al., 00; Burdette et al., ; 00; Fennema et al., 00; Khodair and Hassiotis, 00; Hassiotis et al., 00), however, only quasi-static loading conditions were considered in the experimental studies. The most notable studies that involved field testing of large scale abutments were by Burdette et al. (000; 00; 00), and Hassiotis et al. (00). In a two-phase experimental investigation Burdette et al. (000; 00) investigated experimentally response behavior of H-piles supporting [simulated] integral bridge abutments. Steel H-piles may be oriented for weak or strong-axis bending. In the second phase, Burdette et al. conducted additional tests with H-piles along with precast, Asst. Prof., Department of Civil and Environmental Engineering, University of Nevada, Reno, Reno, NV Professor, Department of Civil and Environmental Engineering, University of Nevada, Reno, Reno, NV Grad. Asst., Department of Civil and Environmental Engineering, University of Nevada, Reno, Reno, NV

G. Pekcan, A. Itani & E. Monzon 0 0 0 0 0 prestressed concrete piles. This study concluded that the current AASHTO and AISC column equations for the design of steel H-piles had limited applicability and in most cases the equations were inappropriate (Ingram et al, 00). There are limited analytical studies that considered explicitly the seismic response characteristics of integral abutment bridges (e.g. Goel, ; Dehne and Hassiotis, 00). The main objective of the study by Goel () was to measure the vibration properties of a two-span concrete bridge from its motions recorded during actual earthquake events. Data was used in conjunction with an analytical model to investigate how abutment participation affected the vibration properties of bridges with integral abutments. It was noted that the abutment flexibility was an important element in seismic design of bridges with integral abutments. In particular, a more flexible abutment would lead to higher deformation demands on other lateral-load resisting elements. In fact, modeling the contribution of bridge abutments to overall seismic response has been the focus of significant research in the past decade. Many studies have shown how the abutment response influences the response of short-and medium-length bridges. Realistic and accurate modeling of the abutment-soil interaction becomes even more important in case of integral abutment bridges. One of the major uncertainties in the design of IAB bridges are known to be the reaction of the soil behind the abutments and next to the foundation piles, especially during thermal expansion and seismic loads. These soil reactions along with the soil-pile interaction are inherently nonlinear. Tegos et al. (00) noted that the dynamic response of the structures in the longitudinal direction depends on the overall dynamic stiffness of the deck-abutment-foundation system. More drastic reduction of seismic demand is generally observed when the abutment stiffness is activated. The study by Dehne and Hassiotis (00) concluded that accurate soil-structure interaction modeling was required to evaluate the effects of longitudinal and transverse earthquake excitation on the response of integral abutment bridges that will allow reliable capacity and demand assessments. It is particularly noted that integral abutment design depends on the connection of the superstructure to its abutment for the transfer of horizontal forces. The connection must transfer such forces without damage to the piles or the soil behind the abutment. As such, the design (and modeling) of the soil properties is of the utmost importance. Faraji et al. (00) developed a threedimensional finite-element model of an integral abutment bridge. The model consisted of a grillage model of the deck, abutment, beam-column elements to model the piles and finally soil springs with nonlinear p- y properties to model the soil-pile-abutment-structure interaction. Although this study focused on loading cases due to differential temperature, remarks were made regarding the impact of skew alignments on the forces and moments at the abutment/superstructure joint, and the seismic response of long-span integral abutment bridges. This paper presents the preliminary findings recommendations of an on-going study that investigates the longitudinal and transverse seismic response behavior of straight and skew steel bridges with integral abutments. In order to establish realistic guidelines for the modeling of steel girder to abutment connection, first, detailed finite element (FE) models of a benchmark bridge (Wasserman and Walker, ) were developed. The connection models were incorporated in a three dimensional global model of an integral abutment bridge to study the structural dynamics characteristics, as well as the load path and distribution to piles, soil, girder elements. A procedure was demonstrated to determine embedment length of steel girders in the abutment to ensure the connection rigidity and more importantly to ensure that the piles develop their flexural capacity. FINITE ELEMENT (FE) MODELING OF AN INTEGRAL ABUTMENT CONNECTION The benchmark bridge is a three-span steel-girder that was based on the example presented by Wasserman and Walker (). The total deck width is ft and total length is ft. The end spans are 0 ft long and the main span is ft long. The superstructure is composed of four composite steel plate girders spaced at. ft and in. thick deck slab. The piers are single column with diameter of 0 in. and clear height of ft. The cap beams are in. deep and in. wide. Cross-frame spacing at end spans is ft and at main span is ft. The intermediate cross-frames are Lxx/ while the support crossframes are Lxx/. The basic geometry of the bridge is shown in Figure. A typical detail at the

G. Pekcan, A. Itani & E. Monzon integral abutment was selected as shown in Figure. A detailed continuum model of concrete deck and steel girder along with the abutment is modeled using ADINA (00) as shown in Figure. The flexibility of the girder-to-abutment connection detail is determined in terms of six (three translational, three rotational) springs lumped at the abutment face. For this purpose, twelve monotonic loading cases were conducted and the results of these cases are summarized. 0 0 Figure Plan and cross-sectional view of the benchmark bridge Wasserman and Walker () Assumptions for the FE Analysis Two models were developed; the first model is referred to as fixed, which essentially means that the connection of the embedded girder and abutment is rigid. The second model is called flexible model. in which the embedded girder has compression only springs at every point of contact with the concrete material inside the abutment. The displacements and rotations were applied at the tip of the girder (where an inflection point is assumed approximately) at the centeroid of deck and girder. The reactions and displacements were measure at this location. Then the reactions and displacements of the flexible model were subtracted from those of fixed model. This removes the flexibility of the cantilever part of the girder and deck, therefore provides the springs properties at the end of the girder. These spring elements can therefore be used to represent the stiffness of girder-to-abutment connection. It is noted that the following were assumed in the development of the FE model:. The interfaces between the concrete deck-abutment, and between the embedded steel girder-abutment are located at the intersection of nodes between the elements of deck and abutment. It is assumed that there is no concrete crushing and tension resistance is neglected. Since during an earthquake the gap can open and close, no shear resistance is assumed at this interface. Interface of the concrete deck and the abutment is defined with plastic material. Nonlinear truss (compression-only) elements are used at common nodes of deck and abutment.. Fix boundaries were used for the abutment boundaries associated with the pile and abutment soil (Figure ). Soil and pile stiffnesses will be accounted for explicitly in the global system model.

G. Pekcan, A. Itani & E. Monzon. Steel girder is modeled with linear elastic material properties. However, the stresses in the steel girder are monitored and the first yield of the steel as well as the subsequent ones is recorded.. The maximum compressive strength of concrete is assumed to be ksi with a modulus of elasticity of,00 ksi. The modulus of elasticity of steel is,000 ksi. The yield stress of steel is 0 ksi. Although the steel girder is modeled elastically, the yield stress is needed to monitor the sequence of yielding in steel girder. 0 Figure Plan and elevation view at the integral abutment

G. Pekcan, A. Itani & E. Monzon Web stiffener 0 0 0 Figure Isometric view of the FE model and girder detail inside the abutment Summary of FE Analysis Results The FE models were subjected to six monotonic loading conditions; the displacements and rotations were applied at the center-of-geometry of the free end of the deck-girder assembly. The reactions and displacements were measured at this location. The reaction-displacement of the flexible model was subtracted from those of fixed model. Therefore, the flexibility of the cantilever portion of the assembly was excluded, hence providing the spring properties at the girder-to-abutment connection. The actiondeformation curves so-obtained for the rotation about the transverse axis (parallel to the abutment wall) along with the associated Von Mises stress distributions at the first yield of the embedded steel girder are shown in Figure. There is a total of six () unsymmetrical nonlinear springs that will account for the variation of location of neutral axis in an equivalent sense. Because of asymmetry and nonlinearity in the force-deformation curves the location of neutral axis changes as the strain increases. The nonlinearity of the concrete interface is modeled as explained earlier, and during the pushover loading, the Von Mises stresses in the embedded steel girder were monitored also to capture the sequence and propagation of yielding in steel. The percentage of yielded [finite] elements is used as an indicator of the distribution of plasticity in steel. On the action-deformation relations of the equivalent springs in the embedded region, the dashed line is used to represent the force-displacement of the entire system, including girder and slab, whereas the solid line is the force-displacement in the embedded region only. This was obtained by subtracting the flexibilities of the flexible and fixed models as explained earlier. Also, the first yield of steel girder and the deformation at which 0% of the embedded steel girder yielded, are marked in Figure. THREE-DIMENSIONAL (D) GLOBAL MODELS Detailed D models of bridges with integral abutments shown in Figure are developed to facilitate the investigation of general structural dynamics characteristics. Several boundary conditions have been identified to establish the effects on the modal response of the bridges in comparison to the most realistic condition that incorporates the springs presented earlier. Deck slab, girders, cap beams and columns The bridge model geometry and components were defined using the Bridge Modeler (BrIM) option in SAP000 (CSI, 00). The deck slab was modeled as shell elements while girders, cap beams and columns were modeled as beam elements. The shell elements were located at the center of gravity of deck slab. In Figure, the girders are shown to be located at the center of top flange. This was due to internal automation in SAP000 when BrIM is used in modeling the bridge superstructure but the properties are actually transformed to the center of gravity of girder. Note that the center of gravity of the plate girder may vary along the span. Locating the girder beam element at the center of top flange makes the

G. Pekcan, A. Itani & E. Monzon modeling easier since the center of top flange will not vary along the span. The cap beams were defined with gross section properties. Effective section properties were assigned to the columns. Rotation about yneg Rotation about ypos Figure Moment-rotation response at the integral abutment and correspond stress distribution on the girder inside the embedded section

G. Pekcan, A. Itani & E. Monzon Soil springs Girder- Abutment Springs Soil-pile springs Rigid frame elements Pin supports (representing the location of pile inflection point) HP0x pile Figure D view and details of the SAP000 models Table Effective stiffnesses at first yield of girder-abutment connection DOF Unit Description K pos K neg K mod U kip/in vertical action U kip/in longitudinal action, 0,0, U kip/in transverse action R kip-in/rad rotation about vertical axis.0e+0.0e+0.0e+0 R kip-in/rad rotation about longitudinal axis.e+0.e+0.e+0 R kip-in/rad rotation about transverse axis.e+0.e+0.e+0 0 K pos = effective stiffness at positive region of backbone curve at first yield K pos = effective stiffness at negative region of backbone curve at first yield K mod = effective stiffness used for modal analysis

G. Pekcan, A. Itani & E. Monzon 0 0 0 0 Girder-abutment connection The girder-abutment connection is represented by a -DOF spring with nonlinear translational properties and nonlinear rotational properties. The spring properties were obtained from the FE analyses of the girder-abutment connection as presented earlier. For the purpose of modal analysis, the elastic stiffnesses used were the effective stiffness at first yield as summarized in Table. For DOFs where the backbone curve is asymmetric (U, U and R), the smaller stifnesses were used in modal analysis. Abutment soil passive resistance The stiffness property of the soil behind the abutment was obtained from Caltrans Seismic Design Criteria (SDC, Caltrans 00). It assumes a uniform soil passive pressure distribution along the abutment. The initial abutment stiffness is given by kip / in h K abut 0 w () ft. where, w is the abutment width and h is the abutment height. The initial stiffness of 0 kip/in/ft was assumed based on the results of large-scale abutment testing at University of California, Davis (UC Davis). This stiffness is scaled by the ratio (h/.) because the height of the abutment tested at UC Davis was. ft. The maximum soil passive pressure that can be developed behind the abutment is.0 ksf. Thus, the total static passive force is given by h P Ae.0 ksf (). where, A e is the effective abutment area. Similar to initial stiffness, the maximum soil passive force is scaled by (h/.). The soil resistance is distributed to the four girders, the stiffnesses of the soil springs connected to exterior and interior girders are 00 kip/in and 0 kip/in respectively with yield deformation at. in. The difference between the soil properties at exterior and interior girder is due to difference in tributary area. The ultimate soil displacement in the figure is. in. which is 0% of the abutment height. This is the code recommended displacement for cohesive soils. In AASHTO Guide Specification (00), the soil passive resistance is dependent on the type of backfill used. For cohesionless, non-plastic backfill with fines content < 0%, the soil passive pressure is p p = H w / ksf per foot of wall length. However, for cohesive backfill with clay fraction > % and estimated undrained shear strength > ksf, the soil passive pressure is p p = ksf. If the stress and strength parameters (c and ) are known, the passive force may be computed using accepted analysis procedures. The above parameters were not readily available thus the Caltrans SDC provisions were used in defining the soil stiffness properties. For modal analysis, only half of the initial stiffnesses was assigned to the abutment soil springs. This is due to the fact that the soil springs are compression only springs. Soil-pile interaction The soil-pile properties used in this study were also adopted from Wasserman and Walker (). Each abutment is supported on seven HP0x steel piles. In the analytical model, beam elements with lengths equal to. ft were used to represent these piles. The.-ft length was based on the location of zero moment in the pile which was measured from the pile head. The soil-pile interaction was represented by -DOF springs located at 0 ft, ft, and 0 ft below the pile head. Figure shows the p-y curves that defines these springs. In modal analysis, the elastic stiffnesses used are kip/in, 0 kip/in, and kip/in for springs located at 0 ft, ft, and 0 ft, respectively. Due to lack of data, it was assumed that the p-y curves in the transverse direction are the same as that in the longitudinal direction.

G. Pekcan, A. Itani & E. Monzon 0 0 0 Figure p-y Curves for Soil-Pile Interaction MODAL ANALYSES Various models were developed to study the effect of connection flexibility at the integral abutment on the global system behavior in relation to the flexibilities associated with the surrounding soil and supporting piles. For this purpose, a total of seven different cases were identified (Table ). In the table, rigid implies that the component was assigned a large stiffness, free means zero stiffness, and flexible means actual (realistic) stiffness values were used for the components listed. It is noted that, for Case and, steel piles are modeled such that they are in strong axis bending with respect to the longitudinal direction of the bridge. The mode shapes were identified and modal participating mass ratios for each mode in all of the cases were recorded. Among the seven cases, Case and, Case and, and Case and are to isolate the effect of the connection flexibility on the overall system dynamics. Particularly the second pair establishes those effects in the absence of abutment soil passive resistance, and the first pair assumes infinite resistance from the abutment back soil. It is important to note that in Case and Case where infinitely rigid connection is assumed, the first modes of vibration were in fact not significant modes both occurring with T = 0. sec. When the connection flexibility is introduced, however, Case and Case result in remarkably similar modal response. Further comparison of modal responses between Case and Case confirm the fact that the abutment back soil contribution to the overall global response of integral abutment bridges is negligible. In other words, among the cases where abutment back soil resistance is considered as rigid, free, or flexible where other conditions were kept the same, the modal response did not change significantly. However, comparisons between Case Case, and Case Case clearly indicate the significant effect of the consideration of connection flexibilities on the global response behavior, however when no soil-pile interaction is modeled.

G. Pekcan, A. Itani & E. Monzon 0 0 0 0 The most striking observation is the change in the global system dynamics due to soil-pile interaction compared to all other cases. Significant modal coupling occurs in Mode # at a much longer period, which involves predominantly transverse displacements and deck rotations. While the longitudinal mode occurs at a very small period in Case (no soil-pile interaction, Mode # with 0.0 sec), it takes place at a significantly longer period of vibration when soil-pile interaction is considered in Case (Mode # with 0. sec). Finally, the comparison of Case and show that there is essentially no difference between modal periods and participation ratios due to connection flexibility. Therefore, the effect of soilpile interaction on the overall system response characteristics clearly is the most significant and governs over the effect of connection flexibility. Table Various boundary conditions investigated Case Abutment-Soil Girder-Abutment Connection Soil-Pile Interaction rigid rigid rigid rigid flexible rigid free rigid rigid free flexible rigid flexible flexible rigid flexible flexible flexible flexible rigid flexible ANALYSIS OF ABUTMENT FORCES Nonlinear pushover analyses were conducted to assess primarily the load path, hence the distribution of abutment force to various components. For this purpose, the abutment-soil springs were modeled at four locations along the abutment height. Previously, only one abutment-soil spring was used and it was aligned with the girder beam element. Comparisons of the modal properties of these two models show that the effect of the number of abutment-soil springs on global behavior is minimal but will have considerable effect on local distribution of forces. The difference between modal periods and modal participation ratios is in the order of % to %. The moment demand on the bridge superstructure, however, was observed to be sensitive on the number of abutment-soil springs used. The four abutmentsoil springs was deemed appropriate for the purpose of obtaining the accurate superstructure moments. The Mode # (longitudinal translation) was used as the pushover load for both Case and Case. Since the models were pushed only in the longitudinal direction, full compression stiffness was assigned to the abutment-soil springs in the direction of the push and zero stiffness was assigned at the other abutment. The analyses suggest that the weak link in terms of longitudinal capacity are the piles. Therefore, a procedure is demonstrated to determine the required embedment length of the steel girders in to the abutment, calculated based on the corresponding bridge superstructure moments and that will ensure the development of plastic moment capacity of the piles. Girder-to-abutment embedment length The embedment length of steel girders in to was calculated based on the mechanism (Figure ) proposed by Shama et al. (00). This mechanism assumes a simplified stress mechanism and was developed for steel-pile embedment in to concrete cap beam or abutment. The resistance to external moment M is to come from the couple created by bearing stresses on concrete. The concrete stress block force C m can be evaluated as Cm 0. f ' cb f lemb () where, is a factor applied to f c ; is a factor for the depth of stress block; f c is the concrete compressive strength; b f is the flange width of the steel section (girder flange width in this case); and l emb is the embedment length of steel section into the concrete. The couple lever arm is

G. Pekcan, A. Itani & E. Monzon 0 jd l emb 0. () Assuming =, jd can be taken as 0.l emb. The moment capacity of the connection is then M c Cm jd () where is the strength reduction factor and is equal to 0.. By substituting () and (), () can be rewritten as M 0. ' c fcb f l emb () To maintain integrity of the connection, the M c M criteria must be satisfied. Thus, the equation for embedment length is given by M lemb () f ' cb f For a maximum concrete strain cu of 0.00, the stress block factors and can be taken as both equal to 0.. Applying the factor, () can be further simplified as M lemb. () f ' cb f lemb F P M Cm f'c 0 jd Cm Figure Simplified mechanism for steel section embedded in concrete Case Realistic girder-abutment connection stiffness The total base shear was,0 kips as illustrated in Figure a. This base shear corresponds to instance where the pile plastic moment was reached at Abutment. The plastic moment capacity of each HP0x pile is, kip-in. The maximum pile moments were observed at Abutment but the maximum superstructure moments were observed at Abutment. As shown in Figure a, % of the total base shear is at the Abutment. Of this base shear, % is due to the abutment-soil passive resistance and % is due to the soil-pile interaction. The bents are taking a small portion only of the total base shear which is % at each bent. The remaining % of the total base shear is taken by Abutment which is primarily due to the soil-pile resistance. The girder bending moments due to dead load and pushover along the bridge length and on individual girders were recorded. At Abutment, the bending moment on the entire bridge section was,0 kip-in (positive moment), on exterior girders it was 0, kip-in, and on interior girders it was, kip-in. At Abutment, the bending moment on the entire bridge was,

G. Pekcan, A. Itani & E. Monzon kip-in (negative moment), on exterior and interior girders, it was, kip-in, and, kip-in, respectively. Note that, as mentioned above, the pile plastic moment was reached at Abutment. However, the design moments to be used for calculating the girder-to-abutment embedment lengths are the moments at Abutment as they are considerably larger than the moments at Abutment. The maximum of exterior and interior girder moments is used to determine the embedment lengths. Abutment Abutment (a) kips kips (%) (%),0 kips (b) kips kips (%) (%), kips Figure Distribution of total base shear; (a) Case, (b) Case kips (%) kips (%), kips (%), kips (%) 0 0 0 0 Case Infinitely rigid girder-abutment connection stiffness The total base shear was, kips as illustrated in Figure b which implies that the effect of girderabutment connection flexibility is insignificant as the difference in modal properties and base shears is very small (in the order of % to %). The distribution of base shear among the abutments and bents is similar to that in Case where Abutment is taking most of the applied loads. Also similar to Case, the pile plastic moment was reached at Abutment but the maximum superstructure moments were observed at Abutment. At Abutment, the total bending moment on the entire bridge section was,0 kip-in (positive moment), on exterior and interior girders it was 0, kip-in, and 0,0 kip-in, respectively. At Abutment, the corresponding bending moments were,0 kip-in (negative moment),, kipin, and. kip-in. Calculated embedment lengths The girder-to-abutment embedment length is determined based on the superstructure moments observed at Abutment which are negative moments. It is noted that these moments corresponds to the instance at which the flexural capacity of piles are reached due to modal pushover in the critical longitudinal direction. For the bridge considered in this study, the girder bottom flange width b f is inches and the concrete compressive strength f c is ksi. Thus, the required embedment length (equation () for Case is inches. For Case, the required embedment length is 0 inches. These are about 0% larger than the provided embedment length of 0 inches which was also used in the finite element analysis of the girderabutment connection. SUMMARY, RECOMMENDATIONS AND FUTURE RESEARCH This paper discusses the results of the longitudinal seismic behavior of straight bridges with integral abutments. Detailed nonlinear finite element (FE) models were utilized to establish the flexibility (translational and rotational) of the steel plate girders and the abutment connections. These connection springs were incorporated in a three dimensional global model of an integral abutment bridge to study the structural dynamics characteristics, as well as the seismic load path and distribution to piles, soil, girder elements. A procedure was demonstrated to determine embedment length of steel girders in the abutment to ensure the connection rigidity and more importantly to ensure that the piles will develop their ultimate flexural capacity. Nonlinear pushover analyses were conducted to assess primarily the load path, hence the distribution of abutment force to various components. The results of these analyses indicated that the piles play a detrimental role in the longitudinal response of bridges with integral abutments. The piles

G. Pekcan, A. Itani & E. Monzon 0 0 0 0 represent the weak link. Therefore, a procedure is demonstrated to determine the required embedment length of the steel girders in to the abutment, calculated based on the corresponding bridge superstructure moments and that will ensure the development of plastic moment capacity of the piles. Further analysis will investigate the load path and the seismic response in the transverse direction of the bridge. The study will also investigate the seismic behavior and response of straight and skew bridges and compare it to the behavior of bridges using seat type abutments. ACKNOWLEDGEMENTS Partial funding for the study presented in this paper was provided by the Federal Highway Administration (FHWA) through a contract (DTFH-0-D-0000 Task Order 00) with the HDR Engineering, Inc. The authors would also like to acknowledge Mr. Edward Wasserman, and Mr. Vasant Mistry of FHWA, and Mr. John Yadlosky of HDR Engineering, Inc. for their valuable comments and feedback. Any opinions, findings, and conclusions expressed in this paper are those of the authors and do not necessarily reflect the view of the sponsor. REFERENCES Adina Reference Manual (00), www.adina.com. Arockiasamy, M., Butrieng, N.; and Sivakumar, M. (00). State-of-the-Art of Integral Abutment Bridges: Design and Practice, ASCE Journal of Bridge Engineering, (), -0. Burke, M. P., Jr. (). Integral bridges: Attributes and limitations. Transportation Research Record., National Research Council, Washington, D.C.,. Civjan, S.A., Bonczar, C., Breña, S.F., DeJong, J., and Crovo, D. (00). Integral Abutment Bridge Behavior: Parametric Analysis of a Massachusetts Bridge. ASCE Journal of Bridge Engineering, (), -. Conboy, D. W., and Stoothoff, E. J. (00). Integral abutment design and construction: The New England experience. Proceedings of the Integral Abutment and Jointless Bridges 00 Conference, Federal Highway Administration, 0 0, Baltimore, Mar. 00. Computers and Structures, Inc. (00). SAP000, Version.0., Integrated Structural Analysis and Design Software, Berkeley, CA. Kunin, J., and Alampalli, S. (000). Integral abutment bridges: Current practice in United States and Canada. Journal of Performance of Constructed Facilities, (), 0. Maruri, R., and Petro, S. (00). Survey Results. Proceedings of the Integral Abutment and Jointless Bridges 00, Federal Highway Administration, -, Baltimore, Mar. 00. Shama, A. A., Mander, J. B., and Chen, S. S. (00). Seismic investigation of steel pile bents: II. retrofit and vulnerability analysis, Earthquake Spectra (), 0. Soltani, A. A., and Kukreti, A. R. (). Performance evaluation of integral abutment bridges. Transportation Research Record., National Research Council, Washington D.C.,. Tegos, I., Sextos, A., Mitoulis, S, and Tsitotas, M. (00) 'Contribution to the improvement of the seismic performance of Integral Bridges', th European Workshop on the Seismic Behaviour of Irregular and Complex Structures, Thessaloniki, CD-ROM Volume, Paper No. Wasserman, E. P. (00). Integral Abutment Design (Practices in the United States). st U.S.- Italy Seismic Bridge Workshop. Pavia, Italy. Wasserman, E. P., and Walker, J. J. (). Integral Abutments for Steel Bridges. In Highway Structures Design Handbook, Vol. Chicago: American Iron Steel Institute.