You are on page 1of 12

19th Analysis & Computation Specialty Conference 2010 ASCE

Sensitivity of Dynamic Response of Bridges under Multiple Hazards to Aging Parameters J. E. Padgett1, J. Ghosh2, N. Ataei3
1

Assistant Professor, Department of Civil and Environmental Engineering, Rice University, 6100 Main Street, MS-318, Houston, TX 77005; PH: 713-348-2325; E-mail: jamie.padgett@rice.edu
2

Graduate Research Assistant, Department of Civil and Environmental Engineering, Rice University, 6100 Main Street, MS-318, Houston, TX 77005; E-mail: jg17@rice.edu

Graduate Research Assistant, Department of Civil and Environmental Engineering, Rice University, 6100 Main Street, MS-318, Houston, TX 77005; E-mail: navid.ataei@rice.edu

ABSTRACT Continued aging and deterioration of bridges poses a threat to bridge performance not only under regular service loads, but also results in pronounced vulnerability under extreme dynamic loads, such as seismic or hurricane induced surge and wave loading. In fact, aging of bridges in the form of corrosion attacks load paths critical under dynamic loads, including the superstructuresubstructure connection elements and the reinforcing steel in column plastic hinge zones. This paper investigates the effect of aging on the dynamic response of multiple span concrete girder bridges when subjected to seismic as well as coupled surge and wave loading induced by hurricanes. The paper highlights the key differences and similarities in the nature of loads under the two natural hazards, the demand placed on key components, and the resulting dynamic response and failure modes of aging bridges. Nonlinear dynamic analysis is conducted using 3dimensional bridge models with time-varying model parameters due to corrosion of reinforcing bars in decks and columns and degradation of elastomeric bearings with steel dowels. The sensitivity of component response, such as column demands, bearing deformations, or deck displacement, to variation in aging parameters is investigated in the study. Findings indicate that while the nonlinear dynamic behavior and select failure modes of the bridges may differ between the seismic and surge/wave loading cases, there is some consistency in the impact and criticality of aging parameters affecting dynamic response under the extreme loading cases, such as corrosion of bearing dowels and column reinforcement. These results form the foundation for multi-hazard vulnerability assessment of bridges considering the present in field condition.

INTRODUCTION The American Society of Civil Engineers recently estimated that over half of the 599,766 bridges in the US are approaching the end of their design life, and nearly a quarter need significant retrofit or replacement to eliminate deficiencies (ASCE 2009). This deficient state underscores the aged and deteriorated condition of bridges, which may reveal itself in a number of forms, such as spalling of reinforced concrete members, buildup of debris leading to corrosion of steel bearings, and corrosion of steel reinforcement in reinforced concrete components, among others (Figure 1a). Many of these same components comprise the primary lateral force resisting system

19th Analysis & Computation Specialty Conference 2010 ASCE

of bridges under such dynamic loading cases as seismic or hurricane induced surge and wave loading. Aging due to environmental stressors, corrosion, and member deterioration affect the structures capacity (Enright and Frangopol 1998; Gardoni et al. 2002; Stewart and Rosowsky 1998), while potential live load increase due to population shift, growth, or changing transport patterns affect the service demands placed on the structure (Ghosn 1999; Nowak 1993). In past decades, advances have been made in assessing bridge service load reliability considering aging and deterioration (Akgul and Frangopol 2004; DeStefano and Grivas 1998; Park et al. 1998; Stewart 2001). However, despite the potential effects of deterioration across a large population of aged bridges on performance during natural hazards, there has been a lack of historic consideration of the joint effect of natural hazard and aging threats, particularly when evaluating the dynamic behavior and subsequent reliability of bridge systems under natural hazard loading. This paper investigates the effect of aging on the dynamic response of multiple span concrete girder simply supported bridges when subjected to either seismic or hurricane induced storm surge and wave loading. The paper highlights the key differences and similarities in the nature of loads under the two natural hazards, the demand placed on key components, and the resulting dynamic response and failure modes of aging bridges. Nonlinear dynamic analysis is conducted using 3-dimensional analytical bridge models with time-varying model parameters due to corrosion of reinforcing bars, degradation of elastomeric bearings with steel dowels, among other parameters. The paper emphasizes investigation of the sensitivity of component response, such as column demands, bearing deformations, or deck displacements, to variation in these aging parameters. The sensitivity analysis aims to provide new insight into the impacts of simultaneously occurring threats and guide future multi-threat reliability analyses. Furthermore, it offers direction as to which parameters of in-field bridge condition solicited through future inspection or health monitoring systems are most critical for informing more reliable bridge vulnerability models under dynamic loading. CASE STUDY BRIDGE AND LOAD MODELS Loads and Failure Modes from Earthquakes and Hurricane-Induced Storm Surge/Wave While the characteristics of earthquake and hurricane induced loading are unique in nature, some similarities in terms of failure modes and critical components have been illustrated in past events. For example past observations of damage patterns from Hurricane Katrina indicate that there are similarities in the types and patterns of bridge damage resulting from the two hazards (Padgett et al. 2008). The unseating of bridge spans during Katrina resulted in a considerable amount of damage, disruption, and losses. This has also been a common failure mode in earthquake events (Jennings 1971; Moehle 1995; Comartin 2004). During a seismic event, superstructure displacements are induced by inertial loads from the earthquake. Collapse of the deck spans occurs if the seat width is not adequate to accommodate the excessive displacements, as shown in Figure 1b. In addition, associated modes of failure for seismic damage of bridges are often associated with vulnerable bearings and limited connectivity, inadequate seismic detailing of columns leading to limited ductility capacity or shear strength, as well as potential liquefaction or poor foundation behavior among others. Such deficiencies are particularly of concern for non-seismically designed bridges, such as pre-1990 central US bridges. During a hurricane, the loads from storm surge and wave action results in forces which could displace the bridge deck, both longitudinal and transverse to the bridge centerline (Figure

19th Analysis & Computation Specialty Conference 2010 ASCE

(a) (b) (c) Figure 1. Bridge damage from (a) corrosion and deterioration; (b) unseating in 1999 Chi-Chi (Taiwan) Earthquake; (c) collapse in 2005 Hurricane Katrina at Bay St. Louis.

1c). In general these loads have not been considered in the historic design of coastal bridges. However, empirical evidence reveals their potential to yield bearings, damage guardrails, shift and unseat spans, among other component damage. Subsequently, a recent body of research has focused on hazard modeling and estimation of surge and wave forces on bridge superstructures (Chen et al. 2007; Cuomo et al. 2009; Douglass et al. 2006; Marin and Sheppard 2009). In this paper, the load estimates used to illustrate bridge response to deck wave forces is derived from the new AASHTO specifications for bridges vulnerable to coastal storms (Modjeski and Masters 2008), and extended with basic assumptions of the time variation of forces. The load models were derived from the extensive studies of Sheppard et al. and have been selected due to their accuracy with experimental and field data. The complete equation of peak wave forces are presented in (Modjeski and Masters 2008). Figure 2 illustrates a force time history estimate for a sample case of structure and load parameters which will be detailed in subsequent sections. The vertical force is composed of a drag force, inertial force, and buoyant force, often collectively termed the quasi-static force, as well as the impact force. This impact force is a result of the wave contacting the girders of the bridge, trapping pockets of air in between the wave and the bottom of the bridge deck, and yielding a sudden force on the structure. These impact forces are taken to be the same magnitude in the horizontal direction as the vertical impact forces in this study, and are assumed to have an effect on the bridge for approximately 5/8 the period of the wave, based on visual observation of results given by Marin and Sheppard (Marin and Sheppard 2009). The negative portion of the force is caused by the suction force that arises from the wave pulling down on the air pocket and the force from the water mass on top of the bridge deck. Since during the wave passage the forces imparted change dramatically with time, moments are induced on the bridge deck, which are estimated at each time step via the moment arm method and applied to the spline bridge model described in the following section. Description of Bridge Geometry and Finite Element Model Given the potential similarity in critical bridge components and viable failure modes, the question surfaces as to whether or not it is important to have refined analytical models of in-field condition for both dynamic load cases, or if various aging parameters have different levels of significance for each hazard. For example, in a region of strong seismicity, is it more critical to update modeling assumptions of bearing stiffness or component corrosion? Do the significant aging parameters differ in the bridge is in a region of susceptibility to hurricane induced storm surge or wave loading? To address these questions a sensitivity study will be conducted with a case study bridge as described in the subsequent section.

19th Analysis & Computation Specialty Conference 2010 ASCE

Figure 3. Case study multi-span simply supported concrete girder bridge.

The class of bridges that will serve as a case study for the sensitivity study is the multispan simply supported concrete girder bridge class--a common bridge type found in the Central and Southeastern US (CSUS) (FHWA 2002). This bridge class was the most prevalently damaged bridge during past hurricane events, such as Hurricane Katrina (Padgett et al. 2009). Moreover, past studies have revealed the susceptibility to damage during seismic events, including potential damage to non-seismically detailed columns or concerns over deck displacements due to short seat widths (Nielson and DesRoches 2007). Figure 3 provides an overview of the bridge geometry of the generic bridge considered in the sensitivity study, identified in past studies as representative of typical non-seismically designed CSUS bridges. The substructure consists of two bent caps, each with three circular concrete columns. The outmost spans of the bridge supported on seat type abutments. The reinforced concrete columns have limited reinforcement (e.g. roughly 1% longitudinal steel and widely spaced transverse ties). Alternating fixed and expansion elastomeric pads with steel dowels serve as the bridge bearings, and continuity is provided between the deck and girders of the three spans. However limited resistance to uplift is provided only by the deck weight and the limited resistance of the bearing dowels integrated in the elastomeric pads, which are primarily targeted at mitigating longitudinal or transverse motion under traffic loads. The numerical model for the bridge is developed in OpenSees (McKenna and Fenves 2001). The general modeling approach based largely upon past recommendations for typical CSUS concrete girder bridges (Nielson 2005). Unique consideration, however, is made in the connection modeling to capture the response of the bridge under the coupled vertical andhorizontal forces of waves appropriately. The simple supports for the decks are modeled with nonlinear elements representative of the elastomeric bearings, both fixed and expansion, with dowels that yield upon excess loading as anticipated in potential seismic or surge events. A vertical contact element has been incorporated to allow the deck to be uplifted from the supports, and settle back on the bent beam when the vertical force reduces during wave passage. This

19th Analysis & Computation Specialty Conference 2010 ASCE

vertical contact element is comprised of both a bar slip model to capture the dowel pullout or yield, as well as a pounding element to reflect the impact loads and energy dissipation upon slamming of the deck back on the supports after uplift. The columns and bent caps are modeled with nonlinear beam column elements having fiber sections. AGING PARMETERS AND IMPACT ON DYNAMIC RESPONSE Aging Paramters of Interest for Study A set of six aging parameters are explored for their impact on dynamic response of MSSS concrete bridges under either seismic or storm surge loading. These parameters are intended to reflect the evolving field condition of bridges along their service life. These include such parameters as concrete cover spalling, which may be inferred from visual inspection, or bearing stiffness change, which rely upon recent advances in the monitoring of bridges coupled with global and local parameter identification. Potential upper and lower levels of the aging parameters are shown in Table 1 and will be discussed in the subsections that follow. These levels correspond to estimates of mean value parameters at time zero and after 75 years of aging with assumed exposure conditions discussed below.
Table 1. Upper and lower levels of parameters considered in the sensitivity analysis. Lower Upper Parameter level level Column Concrete Cover (mm) 0 63.50 2 Rebar Area (mm /bar) 335.48 645.16 Bearing Dowel Lateral Strength (kN) 58.76 111.60 Bearing Stiffness (kN/mm) 3.36 4.08 4 Deck Moment of Inertia (m ) 0.09 0.12 Vertical Dowel Strength (kN) 152.66 293.57

Corroding Reinforced Concrete Columns. Several of the aging parameters considered in this study can be attributed to corrosion of bridge elements, such as reinforced concrete columns or decks. This corrosion deterioration can be primarily attributed to chloride ingress through the reinforced concrete components which eventually leads to reduction in effective bar diameter of reinforcing steel in concrete columns or decks, or attack on exposed steel dowels in the elastomeric bearings. The process initiates following exceedance of the corrosion initiation time, which can be modeled following Ficks second law of diffusions (Enright and Frangopol 1998), depending upon such parameters as the diffusion coefficient, surface and critical choloride concentration. Beyond initiation of the corrosion mechanism, the rate of corrosion has a strong influence on the loss of area of reinforcing steel. For the present case study the corrosion parameters as presented in Table 2 are assumed to be lognormally distributed. Selection of these random variables that affect the corrosion deterioration of the bridge components and their respective distributions are based on the in-field corrosion related studies of existing bridge components in the United States (Whiting et al. 1990; Weyers et al. 1994; Enright and Frangopol 1998). It should be noted that the descriptors of these corrosion deterioration random variables and their potential to degrade the bridge components can vary significantly from one bridge to another depending upon the bridge type and environmental exposure. The base environmental exposure condition leading to corrosion considered in this study is the case of deicing salt

19th Analysis & Computation Specialty Conference 2010 ASCE

exposure. Further details of the full probabilistic modeling can be found elsewhere (Ghosh and Padgett 2009), but for the purposes of the sensitivity study, the lower level of the corroded reinforced concrete column rebar area corresponds to the mean value for the case study bridge after 75 years of exposure, given statisics of lognormally distributed corrosion parameters listed in Table 2. However, field data collected regarding such parameters as corrosion rate from corrosion rate sensors could help to identify structure specific models. The sensitivity of the dynamic response to such paramers will be investigated. The second aging parameter considered in the sensitivity study is the spalling of the reinforced concrete columns, which either may or may not be associated with the corrosion process. The assumed lower and upper levels for the sensitivity study correspond to no spalling and full loss of cover concrete, respectively. Hence the influence of concrete cover loss on the dynamic response of the bridge under seismic or hurricane induced surge/wave will be explored at the extreme case of zero cover concrete.
Table 2. Modeling parameters affecting corrosion process modeling used to determine probability distribution of corroded steel elements in the sensitivity study. Note that the upper and lower levels in the sensitivity study correspond to the mean at 0 and 75 years, respectively. Descriptor Mean COV Cover Depth, x (cm) 3.81 0.20 2 Diffusion Coefficient, Dc (cm /year) 1.29 0.10 Surface Chloride Concentration, C0 (wt % concrete) 0.10 0.10 Critical Chloride Concentration, Ccr (wt % concrete) 0.040 0.10 Rate of Corrosion, rcorr (mm/year) 0.127 0.30

Aging Elastomeric Bearings. The bearings commonly used in the MSSS concrete bridges are elastomeric pads with steel dowels. The intial stiffness of the elastomeric pads can be estimated (1) where G is the shear modulus, A is the area of the bearing, and h is the height of the elastomeric pad. Elastomers have been found to be susceptible to aging due to environmental exposure that increases the stiffness of the bearings. This stiffness can be attributed to an increase in shear modulus of the elastomer, which has been shown in past studies to cause a 21% increase in the modulus (Gu et al. 2005). Hence the sensitivity study explores upper and lower levels of the bearing initial stiffness of 160 MPa and 131 MPa to consider the potential aging of the bearing. In addition to the elastomeric pad, the response of the bearings are influenced by the steel dowel, which prevent excessive motion of the deck relative to the bent or abutment primarily under the consideration of service load. These steel dowels were considered to have a reduced area due to corrosion of the exposed steel, therefore affecting both the lateral strength as well as vertical resistance of the dowel. The area of the steel dowel varies 645 mm2 for the pristine bridge to 335 mm2 for the deteriorated bridge. The reduction in bearing dowel results in a lateral strength reduction from 112 kN to 59 kN and vertical strength from 294 kN to 153 kN. Corroding Reinforced Concrete Decks. Similar to the consideration of corroding reinforcement in the RC columns, the concrete decks are susceptible to corrosion of the reinforcement. The same corrosion parameters are assumed for the deck model as previously noted for the columns, with the exception of the initial cover. The primary impact on the finite element model considered in the sensitivity study is a change in the deck stiffness as modeled by

19th Analysis & Computation Specialty Conference 2010 ASCE

the moment of intertia for the beam column elements. Monitoring this evolution of component stiffness is also the emphasis of current bridge monitoring systems, which is expected offer future field measurements of the potential change in this aging parameter (Huth et al. 2005; Soyoz and Feng 2009; Zanardo et al. 2006). Table 1 presents the upper and lower levels for the moment of inertia (0.12 and 0.09 m2) corresponding to predictive mean value estimates at time zero and time 75 years, respectively. Seismic Loads: Example Dynamic Response of Aged Bridge A sample deterministic analysis is conducted to compare the dynamic response of the as-built, or pristine, and degraded bridge. For the aged bridge case, the parameters are set to the lower levels for column cover, rebar area in columns, bearing dowel strength, deck moment of inertia, and vertical dowel strength, while the elastomer stiffness in the bearing is set at the upper level. The fundamental period of the pristine bridge is 0.49 sec which shifts to 0.54 sec for the extreme case aged bridge. In both cases the first two modes correspond to a longitudinal followed by transverse mode. Figure 4 illustrates a sample of the hysteretic response of two critical components in the multi-span simply supported concrete girder bridge under seismic loading with a ground motion having 0.5g PGA and a duration of 29 sec from the Rix and Fernandez suite of synthetic records (Rix and Fernandez 2004). The figure reveals the increase in fixed bearing deformation in the longitudinal direction including the effects of the reduction of dowel strength due to aging. Furthermore, the column cuvature ductility demands increase from 0.82 to 1.28 due to the changing in dynamic characteristics of the bridge. A rigorous experimental design will be utilized to determine which aging parameters have the most statistically significant effect on the peak demands placed on these components, among others.

Figure 4. Comparison of bearing deformation and column curvature demands for deterministic seismic loading case for the as built and degraded bridge.

Surge/Wave Loads: Example Dynamic Response of Aged Bridge A case study surge and wave parameter combination is adopted for the deck load model to assess the dynamic response of the aged bridge under hurricane induced loading. This hazard condition corresponds to typical Hurricane Katrina conditions and yields the wave loads on the case study MSSS concrete bridge as previously shown in Figure 2. The loading is sufficient to uplift the bridge deck off of its supports including yielding the connection elements and displace the deck transversely. The time history of the deck response in the vertical and transverse directions are shown in Figure 5. Due to the lower vertical and transverse strength of the bearing dowels, the

19th Analysis & Computation Specialty Conference 2010 ASCE

aged bridge has a larger displacement of the deck following bearing yield and accumulated deformation. Also shown in Figure 5 is the moment curvature of the column on the waveward side of the bridge. The peak demands are significantly less than shown in the seismic load case. However, the also plot reveals the influence of changing axial loads due to uplift and pounding of the deck. The reduction peak demands in the degraded bridge case reduce relative to the asbuilt bridge due to the lack of force transmission after uplift of the deck and damage of the superstructure-substructure connection. This illustrates the potential benefit of sacrificial connections under severe loading to mitigate forces transmitted to the substructure. In this example, the aging of the bearings primarily in the form of corrosion of dowels rendered the connections less capable of transmitting large loads to the substructure, yet also resulted in poorer performance in terms of mitigating deck displacements.

Figure 5. Comparison of the deck displacement for a sample deterministic surge/wave loading case for the as built and degraded bridge, and resulting impact on column demands considering change in axial load as well as lateral forces transmitted from deck.

SENSITIVITY STUDY Experimental Design A screening experiment using a two-level fractional factorial design is adopted for the sensitivity analysis. This experimental design takes advantage of a reduced number of simulations, or runs, from to achieve a design in which all main effects and some two factor interactions are clear. The design requires 32 runs with various combinations of upper and lower levels of the aging parameters. Since a single ground motion, or wave load, is used in each of the 32 runs of the experiment, two replicates of the experiment are conducted with addition ground motions or surge/wave parameters to avoid bias in the results yielding a total of 96 runs for the seismic study and 96 runs for the hurricane study. Table 3 shows the seismic and surge/wave parameters for each replicate. Peak component responses are monitored for every run of the experiment, considering the bearings, columns, abutments, and deck displacements.
Table 3. Hazard parameters for the seismic and surge/wave cases used in the sensitivity study.
Case 1 2 3 Earthquake parameters Peak ground Duration accel., pga (g) (sec) 0.07 9 0.3 13 0.5 29 Parameters of Hurricane Induced Storm Surge and Wave Cases Wave Max wave Crest ht, Wave Relative length, ht., Hmax max period, T surge, Zc (m) (m) (m) (sec) (m) 36.9 2.4 1.7 5.5 1.4 36.9 3.7 2.6 6 1.2 42.4 5.3 3.8 6.5 0.6

19th Analysis & Computation Specialty Conference 2010 ASCE

Results of ANOVA An analysis of variance (ANOVA) is conducted to test the statistical significance of varying each aging parameter on the bridge response. While the general formulation for ANOVA tables can be found in most statistical analysis texts (Hines et al. 2003), the result of the hypothesis test is presented herein. If the p-values from the ANOVA are small, there is more evidence that the effect of aging parameter variation on peak dynamic response of the component is indeed significant (i.e. the null hypothesis is rejected). A separate ANOVA is conducted for each hazard (seismic or surge/wave) and each component response. While two factor interactions were assessed in the full study only main effects will be presented for brevity. Seismic. The results of the sensitivity study are shown in Table 4 for the dynamic response of the bridge under seismic loading. The boldfaced p-values indicate those under the cutoff =0.05, which imply that the effect of varying the aging parameter have a statistically significant effect on the component response of interest. The table reveals that the potential variation in lateral strength of the dowel is the most significant parameter, influencing the transverse displacement of the fixed bearings (fxd trans), expansion bearings (exp trans), and abutment in active action (abut-act). Though the abutments in passive action and in the transverse direction were also analyzed, they are not shown in the table since none of the parameters had a significant impact on the response. The area of corroded rebar is also shown to have a statistically significant impact on the curvature ductility demand of the columns (column curvature), as anticipated, and the peak total deck displacement in the transverse direction (deck-disp trans), which includes the displacements associated with both bearing and column deformations. Table 4. p-values from ANOVA indicating significant aging parameters affecting component response under seismic loading.
Parameters Column Cover Rebar Area Dowel Strength Bearing Pad Stiffness Deck Moment of Inertia Vertical Dowel Strength Column Curvature 0.391 0.012 0.288 0.970 0.864 0.874 FxdLong 0.796 0.671 0.061 0.891 0.993 0.998 Fxd Trans 0.664 0.808 0.002 0.722 0.876 0.883 Exp Long 0.932 0.898 0.698 0.905 0.991 0.983 Exp Trans 0.671 0.800 0.002 0.481 0.943 0.942 AbutAct 0.914 0.952 0.000 0.920 0.986 0.985 Deck-Disp Trans 0.528 0.025 0.158 0.419 0.940 0.888

Surge and Wave. The p-values summarizing the results of the sensitivity study under hurricane induced surge and wave loading are shown in Table 5. This study indicates that for the case study MSSS concrete girder bridge, the most significant aging parameters (for =0.05) affecting dynamic response of the bridge components include the lateral dowel strength, followed by the column rebar area and cover. While the longitudinal bearing responses were also investigated they are not presented in the table for brevity, since no parameter variation had a statistically significant impact on the peak responses. It is noted that the responses termed fixed and expansion bearings in the transverse direction represent the deformation of the bearing springs prior to uplift. Beyond uplift of the deck and vertical failure of the bearings, the elastomeric bearings no longer engage in the transverse direction and the critical parameter of interest is the deck displacement itself. Hence while the response of the bearing elements are not critically affected by aging parameter variation, the change in dowel strength in the bearing has a

19th Analysis & Computation Specialty Conference 2010 ASCE

10

significant impact on the peak deck displacement as well as the abutment deformations and column demands. Table 5. p-values from ANOVA indicating significant aging parameters affecting component response under hurricane induced storm surge and wave loading.
Parameters Column Cover Rebar Area Dowel Strength Bearing Pad Stiffness Deck Moment of Inertia Vertical Dowel Strength Column Curvature 0.037 0.004 0.007 0.721 0.227 0.785 Fxd Trans 0.802 0.712 0.797 0.870 0.870 0.789 Exp Trans 0.793 0.699 0.781 0.864 0.873 0.772 AbutAct 0.303 0.777 0.001 0.900 0.466 0.766 AbutPass 0.303 0.777 0.001 0.900 0.466 0.766 AbutTrans 0.375 0.104 0.000 0.778 0.452 0.758 Deck-Disp Vert 0.908 0.908 0.920 0.697 0.854 0.060 Deck-Disp Trans 0.875 0.765 0.023 0.983 0.924 0.755

Comparison between the p-values presented in Table 4 and Table 5 reveals that relatively consistent results in terms of the importance of dowel strength in affecting the largest number of component responses, as well as the intuitive significance of rebar area affecting the column curvature ductility demands under lateral loading, regardless of loading type. However, there are several differences between the significant aging parameters and component responses affected between the two types of extreme events. This can be attributed to the inherent differences in the nature of the two hazards and dynamic response induced, as well as in the loading and modeling assumptions. For example, wave and surge induced loads are applied to the bridge superstructure, while seismic ground motions excite the structure through its supports. The wave loads impact the bridge deck in both vertical and horizontal direction, which yields a significant difference in the dynamic response of structure. Taking into account the above mentioned facts, explanation of results becomes clearer. For example, the transverse displacements of the deck under earthquake loading are primarily a result of inertial loading of the superstructure and column drift, affected by the reinforcement of the concrete columns (see Table 4). However, under surge/wave load, the deck is directly loaded and the capability of the dowels to resist the lateral motion emerges as a more significant parameter. These dowels play a significant role on the dynamic behavior of bridge during hurricane event, since they limit the deck displacement, and in addition, transfer the forces from decks to supports. Also, due to the existence of vertical and transverse loads under the surge/wave load condition, the axial forces and moments in columns change, contributing to the significance of gross area of concrete column section, and hence concrete cover significance. CONCLUSIONS This study reveals the importance of considering in field condition, including corrosion, aging, and deterioration of bridge components, when assessing dynamic response under extreme loading. A case study multi-span simply supported concrete girder bridge is evaluated when subjected to seismic loading as well as hurricane induced storm surge and wave loading. The paper compares and contrasts the nature of the loading and failure of bridges under the two natural hazards. Hurricane damage may be attributed to uplift and lateral wave loading (including buoyant, drag, added mass, and impact) while the forces induced under seismic loading are a result of inertial loading given ground shaking. However, some similarities in component damage and failure modes exist including bearing damage and potential span unseating. A sensitivity analysis of the dynamic response to variation in aging parameters is

19th Analysis & Computation Specialty Conference 2010 ASCE

11

conducted given both hazards. Potential ranges in aging parameters for critical bridge components are identified based on a case study environmental exposure to chlorides as well as past experimental testing and field surveys. The aging parameters of interest include loss of concrete cover from spalling, corrosion of dowels in the bearing connections, increase in elastomer stiffness of bearings, and decrease in moment of inertia of bridge deck due to corrosion as well as reduction in vertical strength of dowel. A deterministic analysis reveals that the bridge with extreme aged conditions (e.g. mean parameters at year 75) would be expected to have larger column curvature ductility demands and damage under seismic loading as well as increased deformation demands on the bearings. Similarly, under surge and wave loading the deck displacements can be expected to increase due to the loss of connection strength, yet excess force transfer to the substructure may be mitigated. The sensitivity study presented in this paper investigated which aging parameters have the most statistically significant effect on the dynamic response of the bridge. Under seismic loading, the ANOVA revealed the importance of change in lateral strength of the bearing dowels due to corrosion of the steel components. The second parameter of interest is the loss of area of reinforcement in the reinforced concrete columns which significantly influences the peak column curvature ductility demands. While different component response quantities may be affected under surge/wave loading, the same two parameters are of importance, in addition to column cover. These findings reveal that corrosion of bearing dowels and column reinforcement are two of the most critical parameters to carefully treat in a probabilistic analysis when assessing the reliability of aging bridge under seismic or hurricane threats. Furthermore, this study offers insight into the most critical aging parameters to be monitoring in the field in order to better improve our dynamic response and reliability estimates regardless of which natural hazard is the most prevalent threat in the region. ACKNOWLEDGEMENTS The authors would like to acknowledge the support of the National Science Foundation under award #CMMI-0928493 as well as the Houston Endowment. REFERENCES
Akgul, F., and Frangopol, D. M. (2004). "Time-dependent interaction between load rating and reliability of deteriorating bridges." Engineering Structures, 26(12), 1751-1765. ASCE. (2009). "Report Card for America's Infrastructure." American Society of Civil Engineers, Reston, VA. Chen, Q., Wang, L., Zhao, H., and Douglass, S. L. ( 2007). "Predictions of storm surges and wind waves on coastal roadways in hurricane-prone areas. ." Journal of Coastal Research. , 23, 1304-1317. Cuomo, G., Shimosako, K.-I., and Takahashi, S. (2009). "Wave-in-deck loads on coastal bridges and the role of air." Coastal Engineering, 56(8), 793-809. DeStefano, P. D., and Grivas, D. A. (1998). "Method for estimating transition probability in bridge deterioration models." Journal of Infrastructure Systems, 4(2), 56-62. Douglass, S. L., Chen, J., Olsen, J. M., Edge, B. L., and Brown, D. (2006). "Wave Loads on Bridge Decks." Office of Bridge Technology, Washington, TC. Enright, M. P., and Frangopol, D. M. (1998). "Probabilistic analysis of resistance degradation of reinforced concrete bridge beams under corrosion." Engineering Structures, 20(11), 960-971. Enright, M. P., and Frangopol, D. M. (1999). "Condition prediction of deteriorating concrete bridges using Bayesian updating." Journal of Structural Engineering, 125(10), 1118-1125.

19th Analysis & Computation Specialty Conference 2010 ASCE

12

FHWA. (2002). "National Bridge Inventory Data." Office of Bridge Technology, Federal Highway Administration, McLean, VA. Gardoni, P., Der Kiureghian, A., and Mosalam, K. M. (2002). "Probabilistic capacity models and fragility estimates for reinforced concrete columns based on experimental observations." Journal of Engineering Mechanics, 128(10), 1024. Ghosh, J., and Padgett, J. E. (2009). "Aging Considerations in the Development of Time-Dependent Seismic Fragility Curves " Journal of Structural Engineering, In Review. Ghosn, M. (1999). "Modeling of bridge dead and live loads." Bridge Safety and Reliability, Reston, VA, USA, 2357. Gu, H., Itoh, Y., and Satoh, K. "Effect Of Rubber Bearing Ageing On Seismic Response Of Base-Isolated Steel Bridges." 4th International Conference on Advances in Steel Structures, Shanghai, China, 627-1632. Hines, W. W., Montgomery, D. C., Goldsman, D. M., and Borror, C. M. (2003). Probability and statistics in engineering, John Wiley and Sons, Hoboken. Huth, O., Feltrin, G., Maeck, J., Kilic, N., and Motavalli, M. (2005). "Damage Identification Using Modal Data: Experiences on a Prestressed Concrete Bridge." Journal of Structural Engineering, 131(12), 1898-1910. Marin, J., and Sheppard, D. M. "Storm Surge and Wave Loading on Bridge Superstructures." ASCE Structures Congress, Austin, TX. McKenna, F., and Fenves, G. L. (2001). "OpenSees Command Language Manual." Modjeski and Masters, I., et al. (2008). "Guide Specifications for Bridges Vulnerable to Coastal Storms." TASK ORDER DTFH61-06-T-70006, AASHTO. Nielson, B. (2005). "Analytical Fragility Curves for Highway Bridges in Moderate Seismic Zones," Georgia Institute of Technology, Atlanta. Nielson, B., and DesRoches, R. (2007). "Seismic Fragility Curves for Typical Highway Bridge Classes in the Central and Southeastern United States." Earthquake Spectra, 23(3), 615-633. Nowak, A. S. (1993). "Live load model for highway bridges." Structural Safety, 13(1-2), 53-66. Padgett, J., Desroches, R., Nielson, B., Yashinsky, M., Kwon, O.-S., Burdette, N., and Tavera, E. (2008). "Bridge damage and repair costs from Hurricane Katrina." Journal of Bridge Engineering, 13(1), 6-14. Padgett, J. E., Spiller, A., and Arnold, C. (2009). "Statistical analysis of coastal bridge vulnerability based on empirical evidence from Hurricane Katrina." Structure and Infrastructure Engineering(DOI:10.1080/15732470902855343). Park, C. H., Nowak, A. S., Das, P. C., and Flint, A. R. (1998). "Time-varying reliability model of steel girder bridges." Proceedings of the Institution of Civil Engineers, Structures and Buildings, 128(4), 359-367. Rix, G. J., and Fernandez, J. A. (2004). "Earthquake Ground Motion Simulation." Mid-America Earthquake Center, www.ce.gatech.edu/research/mae_ground_motion/. Soyoz, S., and Feng, M. Q. (2009). "Long-Term Monitoring and Identification of Bridge Structural Parameters." Computer-Aided Civil and Infrastructure Engineering, 24(2), 82-92. Stewart, M. G. (2001). "Reliability-based assessment of ageing bridges using risk ranking and life cycle cost decision analyses." Reliability Engineering and System Safety, 74(3), 263-273. Stewart, M. G., and Rosowsky, D. V. (1998). "Structural safety and serviceability of concrete bridges subject to corrosion." Journal of Infrastructure Systems, 4(4), 146-155. Zanardo, G., Hao, H., Xia, Y., and Deeks, A. J. (2006). "Stiffness Assessment through Modal Analysis of an RC Slab Bridge before and after Strengthening." Journal of Bridge Engineering, 11(5), 590-601.

You might also like