seismic assessment of two high risk multi-span bridges · pdf fileseismic assessment of two...

Seismic Assessment of Two High Risk Multi-span Bridges in SA – Q. Yang & P.Molloy, DTEI 1

Seismic Assessment of Two High Risk Multi-span Bridges in SA

Qi Yang, Structural Engineer, Department for Transport, Energy and

Infrastructure SA Phil Molloy, Supervising Engineer, Department for Transport, Energy and

Infrastructure SA

SYNOPSIS The Department for Transport, Energy & Infrastructure (DTEI) in South Australia recently undertook a seismic risk study of all of their bridges to identify those at risk from a large earthquake by considering the earthquake hazard, the structural risk, and the bridge importance. Results of the study identified and ranked a number of bridges to be assessed in more depth. A comprehensive assessment was then carried out to investigate the seismic vulnerability of two high risk multi-span bridges on National Highway One at Port Augusta. Bridge #1 is a long square bridge with a slightly curved alignment and consists of 20 No. 26m long simply-supported spans supported by pile foundations. Bridge #2 is a short skewed bridge and consists of 3 No. 12 m long simply-supported spans and multiple circular columns piers with spread footings. This paper will discuss the risk analysis process, and then focus on detailed seismic assessment of the two bridges. A comparison will be made between the two bridges to highlight the effects of the bridge skew, soil-structure-interaction, pier type and earthquake wave travelling effects on the bridge response. Finally, the paper will summarise the seismic deficiencies for each bridge. Seismic retrofitting required to address performance deficiencies will be described and discussed. INTRODUCTION Bridges are critical structures in the transportation system. While new bridges are designed with improved seismic standards and details, many existing bridges pose a risk of failure in a large earthquake. Several recent destructive earthquakes, particularly the 1989 Loma Prieta and 1994 Northridge earthquakes in California, and the 1995 Kobe earthquake in Japan, have caused significant damage to highway structures. These events gave impetus to a serious review and seismic assessment of existing bridges. Investigations have indicated that bridges designed and constructed prior to the development of modern seismic design guidelines are vulnerable to damage, in particular due to deficiencies in the strength and ductility of bridge piers, shear capacity of bearings, and inadequate girder seat lengths at the abutments or piers. The Department for Transport, Energy and Infrastructure (DTEI) in SA recently undertook an earthquake risk study of all of their bridges to identify those at risk from a large seismic event. The study considered the earthquake hazard (magnitude of earthquake & foundation response), the structural risk (susceptibility to damage,


dependant on structure type & configuration), and the bridge importance (strategic significance to freight movement and/or post-disaster response). Results of the study identified and ranked a number of bridges that were to be analysed in more depth to assess their performance in a 2000 year return period earthquake. Two of these structures ranked as the highest risk were major bridges on National Route One at Port Augusta. As shown in figure 1, Bridge #1 is a long squared bridge crossing the Spencer Gulf and Bridge #2 is a short skewed bridge crossing two railway tracks. They were both constructed in 1970. Total replacement cost is estimated at approximately $35m for bridge #1 and $5.5m for bridge #2.

Figure 1 Locality Plan of the Two High-risk Bridges The main objectives of the assessment were to:

o determine the dynamic response of elements of the two bridges to earthquake ground motions with an annual probability of exceedance of 0.0005;

o determine the capacity of the bridges to resist these motions; o identify any deficiencies and design appropriate retrofitting schemes; o study the effects of the bridge skew, soil-structure-interaction and

earthquake wave travelling effects on the responses of bridge seismic modelling.

The analysis applied to the assessment conformed to the Australian Standard for Earthquake loads (AS1700.4-2007), Australian bridge design standard Part 2: Design loads (AS5100.2-2004) and other relevant codes, standards and specifications. The program Ruaumoko3D (Carr, 2007), nonlinear time-history analysis software, was used to model and analyse the bridges as grillage models. Three artificial ground motions were produced by the program SIMQKE (Carr, 2007), which generated statistically independent artificial acceleration time histories to match the specified response spectrum. Different models were set up to investigate the effects on the response of bridges. Conclusions were then drawn by comparing the results of various models. GENERAL DESCRIPTION OF THE BRIDGES BRIDGE #1

Spencer Gulf

Bridge #2

Highway A1

Bridge #1


Bridge #1 is a 20 span simply supported bridge carrying 2 traffic lanes and pedestrians. The overall length is about 530 m and the deck is approximately 11 m wide. It consists of a slightly curved portion from the Adelaide abutment to pier No. 7 and a straight portion from pier No. 8 to the Whyalla abutment. Each span is approximately 26 m long and shares a common support with the adjacent span via notched-end girders. This type of support is particularly vulnerable to earthquake damage. Loss of any one span could potentially trigger a progressive collapse of all spans. The superstructure of each span is comprised of 5 No. precast prestressed concrete (PSC) I-beams with cast-in-place decks. The bridge elevation is shown schematically in Figure 2(a). Steel rocker bearings, shown in figure 3, are used at each end of the girders. Expansion bearings are used at girder-to-girder connections within the curved portion and also at pier No. 11 and 17. The remaining girder-to-girder connections are made through fixed bearings. The girder supports at the piers and Whyalla abutment are through fixed bearings and expansion bearings are used at the Adelaide abutment. Wall piers are used from pier No.1 to No.8 and No.14, supported by 2.4 m wide x 0.91 m deep pile caps and a group of sixteen 457 mm octagonal PSC piles. Six 610 mm octagonal PSC piles with crossheads are used for the other piers. The site consists of recently deposited shelly sands and clays up to several meters in depth overlying sands and clayey sands of Hindmarsh formation approximately 25 metres thick, which in turn overlies sandstone of Tent Hill formation.

(a) Profile of Bridge #1

(b) Profile of Bridge #2

Figure 2 Bridge Profiles

Figure 3 Steel Rocker Bearings (side view)

ADELAIDE

ABUTMENT

WHYALLA

ABUTMENT PIER 1 PIER 2


BRIDGE #2 Bridge #2 is a three span simply supported structure with PSC I-beams, multiple RC column piers and RC wall abutments constructed on spread footings. Each span is approximately 12 m long and is skewed at 33 degrees, and the deck is approximately 21 m wide overall in order to accommodate 4 lanes of traffic and pedestrians. The bridge elevation is schematically shown in Figure 2(b). Expansion bearings were used at abutments and at the south end of the girders at the middle span. Similar steel rocker bearings are used as shown in Figure 3. The superstructure of each span consists of 15 No. precast pre-tensioned concrete I-beams made composite with an in-situ concrete deck. Each pier consists of 10 No. 457 mm diameter and 6.1 m high RC columns constructed on a 2.33 m wide shallow spread footing. The RC wall abutments are 381 mm thick and 4.3 m high constructed on a 3.3 m wide spread footing. The piers and abutments are founded on a weakly cemented sand of low relative density. This extends to a depth of three metres below the pier footing and is underlain by layers of sand, clayey sand and clay. Groundwater level at the time of investigation was three metres below underside of the pier footing. The existing multi-column piers are to be strengthened against rail traffic collision by casting a wall wrapping around the existing columns. A wider and thicker spread footing will be built on and connected into the top of the existing footing.

STRUCTURAL MODELLING The bridges were modelled with the following assumptions:

o the bridges are straight; o the minimum earthquake wave travelling speed is 125 m/s (Singh &

Fenves,1994); o the superstructure, including girders, diaphragms, and deck, remains elastic

during the applied earthquake (Priestley et al., 1996); o the two abutments can be considered as two fixed points for modelling; o the interaction between the structure and the surrounding soil can be modelled

by displacement spring elements (Carr, 2006).

There are typically two ways to model the superstructure of a long bridge: a spine model with beam elements following the centre of gravity of the cross section along the length of the bridge, and a grillage of beam elements. For superstructures carrying wide roadways, the spine model may produce erroneous results, particularly when combinations of earthquake forces with gravity loads need to be investigated (Priestley et al., 1996). A simple spine model may not capture even gravity-load distributions to individual bearings and piers, since loads in the spine model are typically applied along the spine axis only. Therefore, a grillage model was employed as shown in Figure 4. The general structural modelling is similar for the two bridges. According to AS5100.2-2004, only the permanent effects are required to be considered when combined with earthquake load for ultimate limit state design. The mass of the superstructure was lumped at each node of the bridge grillage model except for pier walls and piles, whose top half masses were lumped on the top end of


each member and zero mass at the fixed base nodes. The Rayleigh damping model was used to model the damping exhibited by the structure in which the fractions of critical damping were assumed to be 5%. Earthquake motions in three directions - longitudinal, transverse and vertical - were analysed.

(a) Bridge#1 Model (b) Bridge#2 Model

Figure 4 Grillage Model for a Span of TheTwo Bridges Superstructures The longitudinal I-beams for each span were modelled by element 1, while element 2, applied at the centre and quarter points of each span, was used to model the diaphragm/deck member. Element 2 was fixed to element 1 at each intersectional node. Although detailed cracked section stiffness analyses can be performed for each superstructure element to determine the effective stiffness, it is recommended by Priestley et al. (1996) to calculate the gross-section stiffness Ig and reduce it to Ie = 0.5 Ig for reinforced concrete and assume no reduction (i.e. Ie = Ig) for uncracked PSC superstructures. The torsional rigidity J for grillage elements can also be determined from standard mechanics principles and can be considered as fully effective provided the torsional cracking moment is not exceeded.

Bearings

Bearings were modelled by 3D Compound-spring elements (Carr, 2007), which are capable of different spring properties in three different directions. Different properties were assigned for fixed and expansion bearings. For a fixed bearing, horizontal movements are restrained by a central dowel; vertical movement is only possible when uplift is greater than the dead load applied on the bearing; and rotation is free about both the vertical & transverse axes, and restrained about the longitudinal axis. For an expansion bearing, the longitudinal movement is free within a limited distance, transverse movement is restrained by both a central dowel and side bars, and the rotational fixity is the same as that of the fixed bearing.

Element Name: 1.Girder, 2.Diaphragm, 3.Girder-Girder bearing, 4.Girder-Pier bearing, 5.Crosshead, 6.Pier wall, 7.Piles 8.Pier column, 9.Edge beam, 10.Pile cap, 11.Pier footing, 12.Soil spring element


The movement restraint was modelled by the stiffness of the spring in that direction. For instance, the spring stiffness for a fixed bearing in the longitudinal direction was estimated based on the idealised shearing deformation of the central stainless dowel as kdowel = GsA/d, where Gs is the steel shear modulus, and A & d are cross sectional area and diameter of the dowel respectively. In the transverse direction, the same stiffness was used for both fixed and expansion bearings. An elastic hysteresis rule was assumed for fixed bearings after Singh & Fenves (1994). For expansion bearings, a hertzian contact spring with a slackness hysteresis rule was used in the longitudinal direction, to model the limited distance of the movement, as shown in Figure 5. This hysteresis rule permitted the definition of two different contact lengths or gaps for both positive and negative directions, beyond which a spring stiffness model could also be defined to simulate the situation when the expansion bearing travels beyond its available length. For example, a positive gap between two adjacent girders was set as the positive gap, beyond which girder pounding will occur. A spring stiffness derived from the concrete compressive stiffness was then used to model the case when one girder drifted beyond the positive gap and hit the adjacent girder. In the negative direction, the expansion bearing movement limit of negative gap was applied. The values of the actual available gap distances were measured on site. In order to prevent girders “dropping off” during the analysis, an idealised restrainer cable was applied to connect the two adjacent girders, and the slackness length of the cable was set equal to the negative gap. The tensile stiffness of the restrainer cable was then used for the bearing stiffness beyond the negative gap. In this way, the program could calculate the relative displacements in the joint, the pounding forces of two adjacent girders, and the tensile force in restrainer cables if activated during the analysis. The rotational stiffness of bearings was set at zero about the vertical and transverse axes, and set at full fixity about the longitudinal axis for simplicity, because the girders are effectively restrained against rotation about the longitudinal axis.

Figure 5 Hertzian contact spring with slackness hysteresis rule

Piers Pier columns/piles were modelled by beam-column members (Carr, 2007). Columns/piles were modelled as 3D concrete beam-column members using a one-component model (Carr, 2004), which idealised a reinforced concrete beam as a perfectly elastic mass-less line element with non-linear rotational springs at the two ends to model potential plastic hinges. The bi-linear hysteresis rule (Carr, 2007) was employed for the hinge spring stiffness to represent the elastic and inelastic behaviour of the member.

The effective column/pile stiffness EIe was determined from section moment-curvature analyses as EIe = My/Φy, where My and Φy represent the ideal yield


moment and curvature for a bilinear moment-curvature approximation (Priestley et al., 1996). The effect of concrete confinement was also considered using Mander et al.’s model (1989). The results of the section moment-curvature analysis of the pier columns under a static eccentric axial load are shown in Figure 6. The effective stiffness reduction in shear was considered proportional to the effective stiffness reduction in flexure (Priestley et al., 1996). The torsional moment of inertia was multiplied by a factor of 0.3 to obtain the effective torsional moment of inertia for these bridge piles after Singh and Fenves (1994).

0

100

200

300

400

500

600

700

800

0 0.01 0.02 0.03 0.04 0.05

Curvature (rad/m)

Mom

ent (k

N.m

)

0

50

100

150

200

250

300

350

0 0.005 0.01 0.015 0.02 0.025

Curvature (rad/m)

Mo

me

nt

(kN

.m)

(a) 610 mm dia PSC Pile of Bridge #1 (b) 457 mm dia Pier Column of Bridge #2

Figure 6 Example of Moment-curvature relationship for columns

Pier walls were modelled as quadrilateral finite elements (Carr, 2007) and assumed to be fixed along the base joint with pile caps. Foundations According to the different conditions of the two bridges’ substructures, different strategies were used to model the bridge foundations. For bridge #1, deep pile foundations were used for all piers. PSC piles were driven 20-30 m below the seabed level and eventually founded into rock. The flexibility of the pile shaft and the surrounding soil was modelled as recommended by Priestley (1996). An alternative approach to modelling soil flexibility effects on pile shaft systems is the effective fixity approach (Reese et al., 1974), which also provides good stiffness estimates for seismic bridge analyses (Priestley et al., 1996). For simplicity, the equivalent fixity model was used in this analysis. In this approach the equivalent depth to fixity, df, was determined as the depth that produces, in a fixed-base column with soil removed above the fixed base, the same top-of-pile lateral displacements under the ideal seismic lateral force applied at the pile top (Priestley et al., 1996). The df value was estimated based on the soil subgrade reaction modulus, the pile diameter and the effective pile stiffness. It was found to be approximately 2.4 m and 3.7 m beneath the top face of the sands and clayey sands layer for 610 mm and 457 mm piles, respectively. For bridge #2, where shallow spread footing were used, considering the soil-structure-interaction was necessary as fixed column bases may lead to inaccurate results that are often unconservative (McDaniel, 2006). In order to study the effect of the soil-structure-interaction on the bridge response, both fixed column bases and foundation soil springs were used. The foundation was then modelled by a series of displacement spring elements in three directions. The soil modulus of subgrade reaction was used for vertical soil spring stiffness. The horizontal soil spring stiffness

EIe EIe

1 1


was estimated by Ff/∆f, where Ff represents the total footing-soil base friction resistance and ∆f is the displacement at the maximum frictional resistance. The soil spring stiffness in each direction was then assigned to individual spring elements. SEISMIC RESPONSES OF THE BRIDGES The purpose of the nonlinear time history analyses was to assess the seismic vulnerability of the bridge. Seismic evaluation is typically based on a quantitative assessment of the “capacity” of, and “demands” on, individual components, where capacities include member resistances and displacement capabilities, and demands include force and displacement effects.

Bridge #1 and Bridge #2 were modelled by two and five different models, respectively, summarised in Table 1. Model M1 (bridge #1), M1-NW (bridge #1 without earthquake travelling wave effect) and M2-St (bridge#2 after pier strengthening) were used for the final assessment of the two bridges, while the Models M2-Ex (existing bridge with multi-column piers), M2-NS (no skew), M2-FB (fixed base piers), M2-NW (no earthquake travelling wave effect) were used to study the modelling effects. Each model was run under three different artificial earthquake ground motions, which were representative of a possible severe earthquake, and also run in three different earthquake directions: longitudinal, transverse and vertical directions.

B r id g e s M o d e ls S k e w P ie r F o u n d a tio nE a r th q u a k e w a ve

tra ve llin g e f fe c tP ie r T yp e

M 1 N o E q u iva le n t f ix ity Y e s -

M 1 -N W N o E q u iva le n t f ix ity N o -

M 2 -S t Y e s S o il s p r in g Y e s W a ll M 2 -E x Y e s S o il s p r in g Y e s C o lu m n s

M 2 -N S N o S o il s p r in g Y e s C o lu m n s M 2 -F B Y e s F ix e d b a s e Y e s C o lu m n s M 2 -N W Y e s S o il s p r in g N o C o lu m n s

B r id g e # 1

B r id g e # 2

Table 1 Summary of Bridge Models

The seismic responses of the two bridges are summarised in Table 2 and the critical vulnerabilities are also highlighted. For Bridge #1, only the response in girder-girder (G-G) and girder-abutment (G-A) joints are summarised as no deficiencies were found in the girder-pier joints. Minimal difference was found between the results of M2-Ex and M2-NW, therefore, the results for M2-NW was not listed separately. The capacities of each element are also listed at the bottom of the tables. It should be noted that the estimated actions or displacements were picked up from worst cases from the separate results in three earthquake directions; therefore, the maximum force does not necessarily occur concurrently with the maximum displacement.

Results of the analyses identified the following deficiencies: Bridge #1 (Model M1 and M1-NW)

o the demand of bending moments in pier wall 8 and 14 was found to be greater than the pier wall flexural strength,

o the girder on pier 11 may drop off the adjacent girder since the “drifting away” displacement was greater than the available girder seat length,


Vtr Vtr

(kNm/m) (kNm) (mm) (mm) (kN) (kN) (kN) (kNm/m) (kNm) (mm) (mm) (kN) (kN) (kN)

Mw Mc - + - + Mw Mc - + - +

Ade Abut Exp - - 39 25 0 3020 741 - - 28 25 0 6420 741

1 Exp 86 - 27 26 0 2130 954 87 - 20 26 0 1760 954

2 Exp 99 - 36 26 0 2560 905 111 - 22 26 0 2010 905

3 Exp 96 - 49 26 0 4220 708 104 - 32 26 0 2360 708

4 Exp 85 - 43 26 0 4150 824 102 - 37 26 0 2730 8245 Exp 87 - 33 26 0 3820 846 109 - 23 26 0 3340 846

6 Exp 93 - 71 26 0 4740 845 103 - 34 26 0 3010 845

7 Exp 90 - 71 26 0 3070 738 98 - 58 26 0 4210 738

8 Fix 115 - - - 109 951 218 143 - - - 114 1630 218

9 Fix - 53 - - 219 955 186 - 54 - - 77 1100 186

10 Fix - 67 - - 468 1400 217 - 69 - - 39 1570 217

11 Exp - 88 102 26 1070 3500 656 - 89 32 26 0 4040 656

12 Fix - 85 - - 604 1550 107 - 77 - - 512 1820 107

13 Fix - 54 - - 465 1260 117 - 58 - - 102 1550 11714 Fix 145 - - - 607 1040 205 172 - - - 277 1170 205

15 Fix - 56 - - 331 1250 149 - 57 - - 38 1830 149

16 Fix - 76 - - 168 1310 83 - 77 - - 19 2230 83

17 Exp - 70 77 26 0 1900 610 - 72 62 26 0 3720 610

18 Fix - 48 - - 137 1430 73 - 49 - - 137 2700 73

19 Fix - 43 - - 271 1480 34 - 43 - - 273 2770 34Why Abut Fix - - - - 384 1500 99 - - - - 386 2790 99

Exp 100 25 cable varies 410Fix - - 769 769 769

∆lo Vlo

Pier wall

Pier Column

G-G(A) bearing

type

Abutment/Pier No.

Capacity 123 743

M1

Response in bearings

M1-NW

Response in bearings

∆lo Vlo

Pier wall

Pier Column

(a) Seismic Response of Models for Bridge #1

Vtr

(kNm/m) (mm) (mm) (kN) (kN) (kN) (kN) (kN) (kPa)

Mmax - + - + Flo Ftr p

Exp - 10 10 0 0 1306 - - -east Fix - - 637 640 1112west Exp 5 8 0 0 1071east Fix - - 854 832 730west Fix - - 821 803 1263

Exp - 13 16 0 0 1584 - - -Exp - 20 22 0 0 870 - - -

east Fix - - 409 399 649west Exp 31 42 0 0 890east Fix - - 554 552 857west Fix - - 504 508 460

Exp - 25 33 0 0 788 - - -Exp - 44 45 0 0 895 - - -


Exp - 37 37 0 0 875 - - -Exp - 18 20 0 0 576 - - -


Exp - 17 27 0 0 599 - - -Exp wall-1133 45 157 - - 900Fix column-220 - - 900 900 900

Bearing

type

Abutment &

Pier No./Side

Models

Elements capacity

2

Ade Abut

1

Response in bearings Base reactions

∆lo Vlo

2240 400

726 1270 46

- - -

- -

409 809 1688 58

300

M2-St

Why Abut

M2-Ex

(M2-NW)

M2-NS

M2-FB

2 170

104

Why Abut

-

Why AbutAde Abut

1 110

512 85

2 122 294 564 82

83

Why AbutAde Abut

1 127 285

2 164 540 460

476 435 75

Ade Abut

1

Pier Wall/ Column moment

(b) Seismic Response of Models for Bridge #2

Note: Exp = Expansion bearings, Fix = Fixed bearings, ∆lo = Longitudinal displacement in expansion bearing,

Vlo = total longitudinal shear in fixed bearings or force at expansion joints (concrete pounding or tension in restrainer cables) across the section,

'+' = girder drifts towards, '-' = girder drifts away, Vtr = total transverse shear across the section,

Flo = Total longitudinal friction, Ftr = Total transverse friction, p = Average vertical soil pressure.

Table 2 Summary of Seismic Responses


o the colliding force at the Adelaide abutment was found to be greater than the capacity of the abutment backwall,

o the longitudinal shear force in the fixed bearings was much greater than the dowel capacity,

o the shear capacity of expansion bearings in the transverse direction was found to be inadequate.

Bridge #2 after pier wall strengthening (Model M2-St) o the shear capacity of both fixed and expansion bearings in the transverse

direction was found to be inadequate. The capacities of the remaining bridge elements were found to be adequate. The ductility of each member was also checked. STUDY OF THE MODELLING EFFECTS The effects were quantitatively studied for the sensitivities of bridge seismic modelling by comparing the results derived from the different models. The sensitivity of modelling the bridge skew was investigated between the results of Model M2-Ex and M2-NS. It was found in the results of M2-NS that in the longitudinal direction, the displacements were bigger in expansion bearings and the shear forces were smaller in fixed bearings; in the transverse direction the bearing shear forces were slightly bigger than that of the M2-Ex. Hence, ignoring the skew in bridge modelling might significantly change the capability (‘stiffness’) of the structure components participating in resisting the earthquake in each direction and consequently cause errors in the seismic responses. Comparing the results of model M2-Ex and M2-FB for the effect of soil-structure-interaction, modelling the bridge with fixed base piers may slightly overestimate the bending moment in pier columns and shear forces in bearings, while underestimate the displacement in expansion bearings. Therefore, it is recommended that soil-structure-interaction should be considered in bridge seismic modelling, especially when the fixity of the foundation is uncertain. The earthquake wave travelling effect was studied on both bridges. A significant difference was found in bearing responses in the longitudinal direction for bridge #1. The longitudinal displacement in bearings of Model M1 was found to be much higher than that of M1-NW. However, there was nearly no difference between M2-Ex and M2-NW. It is reported that the earthquake wave travelling effect is significant for long bridges (Priestley et al., 1996). In terms of the analysis results, modelling with earthquake wave travelling effect may give an upper-bound displacement in expansion joints, while the upper-bound forces or bending moments in the elements are normally derived without considering the wave travelling effect. As the earthquake wave travelling speed is uncertain, the adoption of upper-bound displacements and forces is considered to be an acceptable approach. After comparing Model M2-St (Wall pier) and M2-Ex (Multi-column pier), it was found that the bearing shear forces and pier base bending moments predicted by M2-St were much higher and displacements were lower than M2-Ex because the wall piers


are considered to be much stiffer than column piers, and more force can be generated in an earthquake for a stiffer structure. In conclusion, sensitivity studies are recommended to be carried out in all seismic modelling assessments. SEISMIC RETROFITTING SCHEMES The objective of the retrofit schemes was to increase the capacity of the bridge members to satisfy the seismic demands. Several retrofit schemes were developed to address the bridge response vulnerabilities. The bridge model was then modified according to the changes in members and the response of the retrofitted bridge was reanalysed to verify the effectiveness of the retrofitting schemes. The following retrofitting schemes have been applied on bridge #1 and some of them will also be applied on bridge #2. Concrete jacketing was used to improve the ultimate strength of pier walls. Longitudinal restrainer cables were used to prevent excessive displacements at bearings as shown in figure 7a. Tight cables were used at fixed joints and flexible cables (with slackness) were used at pier 11 expansion joint to prevent the girder “dropping off” after movement beyond the available seat length. To address the transverse shear capacity deficiency in bearings, shear walls were constructed between the third and fourth girders (figure 7b). The shear wall will carry the ultimate transverse shear forces. To prevent pounding between girders in the longitudinal direction, rubber blocks were inserted between adjacent girder ends, allowing small girder rotation in the longitudinal plane. Restrainer brackets (figure 7c) at the bottom of girders in the end spans will transfer loads from girders to the abutment main wall, preventing any failure of abutment backwalls or the fixed bearings.

(a) Restrainer Cables (b) Transverse Shear Wall (c) Restrainer Bracket

Figure 7 Seismic Retrofitting Schemes for Bridge #1 CONCLUSIONS AND RECOMMENDATIONS The selection of the best retrofitting schemes for the two bridges required the potential deficiencies of the bridge to be evaluated by the nonlinear time-history analysis program Ruaumoko (Carr, 2007). Retrofitting techniques that could overcome these deficiencies were identified and assessed for their feasibility and effectiveness. If the proposed retrofitting scheme altered element stiffness and


consequently structural response, the model should be reassessed using the modified properties of the retrofitted elements. It is recommended that the soil-structure-interaction is considered and investigated in the seismic modelling of bridges. The earthquake travelling wave effect is recommended to be considered as an additional case, because it gives an upper-bound displacement estimate. Sensitivity analysis of models should be undertaken for all seismic assessments.

An alternative approach to seismic retrofitting is to decrease the bridge seismic actions or displacement demand, rather than increase capacity. Investigations into the use of friction and viscous dampers to improve bridge seismic performance may be justified.

The risk of not installing seismic retrofits on identified high-risk structures is that bridges that are strategically significant to the freight industry, the transport system performance and post disaster relief may be damaged during a severe earthquake that can be reasonably expected to occur at some time. Using an assessment regime where bridges are assessed in terms of importance and vulnerability to damage will help to mitigate the risk and ensure that limited funds are used effectively. REFERENCES Carr, A.J. (2007).Volume 3: User Manual For the 3-Dimensional Version,

Ruaumoko3D. University of Canterbury. Carr, A.J. (2006). “Dynamic Analysis of Structures”, Department of Civil Engineering,

University of Canterbury, ENCI 423 – Structural Analysis. Mander, J. B., M. J. N. Priestley, R. Park. 1988. “Theoretical Stress-Strain Model For

Confined Concrete.” Journal of Structural Engineering, Vol. 114, No. 8, pp. 1804-1826

McDaniel C. C.. 2006. “Seismic Assessment and Retrofit of Existing Multi-Column Bent Bridges”. Report No. WA-RD 639.1. Washington State Department of Transportation, Olympia, WA, US.

Priestley, M. J. N., F. Seible, and G. M. Calvi. 1996. Seismic Design and Retrofit of Bridges, John Wiley & Sons, Inc.

Reese, L. C., W. R. Cox, and F. D. Koop. 1974. “Analysis of Laterally Loaded Piles in Sand.” Proceedings, 6th Offshore Technology Conference, Houston, Texas, pp. 473-483.

Singh, S. P., and G. L. Fenves. 1994. “Earthquake Response of ‘Structure A’ Using Non linear Dynamic Analysis”, in Seismic Design and Retrofitting of Reinforced Concrete Bridges, Proceedings of the Second International Workshop, Queenstown, New Zealand.

Standards Australia/New Zealand. 2002. AS/NZS1700:0: Structural design actions Part 0: General principles, Standards Australia, Sydney & Standards New Zealand, and Wellington.

Standards Australia. 2007. AS1700:4: Minimum design loads on structures Part 4: Earthquake loads, Standards Australia, Sydney.

Standards Australia. 2004. AS5100:2: Bridge design Part 2: Design loads, Standards Australia, Sydney.

seismic assessment of two high risk multi-span bridges · pdf fileseismic assessment of two...

Documents