Seismic retrofit design of highway-railway combined suspension bridges

Abstract

The seismic retrofit of the Seto-Ohashi Bridges (a group of long span bridges crossing Seto-Inland Sea) was started in 2013 and will be completed by 2020. The seismic performance verification of three suspension bridges (Shimotsui-Seto, Kita Bisan-Seto, and Minami Bisan-Seto Bridges) of the Seto-Ohashi Bridges was started in 2014. These bridges are one of the longest highway-railway combined suspension bridges. Recently, large-scale earthquakes that were stronger than considered in the original design occurred in Japan. Therefore, two types of large-scale earthquake motions coming from plate boundary and inland active faults, expected to occur in the future, were used in the seismic performance verification. All truss members of the stiffening girders were modeled in the analysis to see precise behaviors during earthquakes. As the result of the seismic performance verification, in all three bridges, it is found that main members (such as main towers, stiffening girders, and substructures) will not suffer severe damage and the total number of damaged portions is small. Although some members are damaged, the damage level is found to remain below acceptable level by the supplemental in-depth analysis. As members to be retrofitted are deck bearings only, a retrofit method for road deck bearings was selected from various candidates considering workability and efficiency.

Keywords

Highway-railway combined bridge suspension bridge seismic retrofit seismic verification

1 Introduction

The seismic retrofit of the Seto-Ohashi Bridges (group of long-span bridges crossing Seto-Inland Sea) was started in 2013 and will be completed by 2020. The seismic performance verification of three suspension bridges (Shimotsui-Seto, Kita Bisan-Seto, and Minami Bisan-Seto Bridges) of the Seto-Ohashi Bridges was started in 2014.

The Seto-Ohashi Bridges are highway-railway combined bridges and the three suspension bridges have very long center spans. Such characteristics of the bridges are considered in the seismic retrofit design. Specifications for Highway Bridges [1] were used for verification and retrofit design.

The results of the analysis showed that center stays and road girder bearings will be damaged in all the three bridges. Since it was verified that the girder will not hit the anchorages after the breakage of the center stays, it was decided that the breakage of the center stays was allowed.

Some truss members of the stiffening girder of the Shimotsui-Seto Bridge will be damaged as well. However, by the supplemental in-depth analysis, it was found that expected damage level is small and will not degrade structural safety and serviceability. Therefore, it was decided that the damages were allowed as well.

The seismic retrofit work of the three bridges was started from the spring of 2017.

2 Outline of bridges

The Shimotsui-Seto Bridge is a single-span stiffening truss suspension bridge with the bridge length of 1,400 m. The Kita Bisan-Seto Bridge and the Minami Bisan-Seto Bridge which connect Yoshima Island and Sakaide City are three-span continuous stiffening truss suspension bridges. As highway-railway combined bridges, the three bridges have one of the longest center spans in the world. Figures 1 and 2 show general views and aerial views of the bridges.

Fig.1

General view.

Fig.2

Aerial views of the bridges.

In the original design, large-scale earthquakes coming from plate boundary were considered [2]. While the seismic performance level of the bridges was supposed to be high because they have long period, the reinforcement was expected to be inevitable.

3 Seismic performance verification

3.1 Seismic performance criteria

Since the bridges must undertake a role as an emergency transportation route in case of large-scale earthquakes, it is required not only to remain stable, but also to provide good operations and functions following the large-scale earthquakes. Therefore, target seismic performance criteria were specified in terms of two aspects, seismic safety and serviceability, as follows,

Seismic safety

To prevent the bridge collapse and ensure the safety of human life Seismic serviceability after events

To ensure the serviceability for emergency traffic immediately after events by emergency inspections or temporary repair works

To ensure the serviceability for normal traffic in a short period after events

To be repairable for damaged structural members with normal traffic service

3.2 Earthquake motions

Seismic ground motions of the large-scale earthquakes used for the seismic performance verification are defined in Fig. 3. They are rock outcropping motions on the bedrock whose S-wave velocity is about 1,500 m/s. Two types of scenario earthquakes were considered. One is the Tounankai-Nankai Earthquake which is a plate boundary earthquake with a magnitude of 8.6, and the other is an earthquake with a magnitude of 7.5 coming from the part of an inland active fault, the Japan Median Tectonic Line (hereinafter referred to as MTL). A fault model for the Tounankai-Nankai Earthquake and MTL are shown in Fig. 4. In addition to the two types of scenario earthquakes, two more types of seismic ground motions generated by unknown fault earthquakes with a magnitude of M6.7 occurring just beneath the site were considered. This is a consideration for the possibility that an unknown inland active fault might exist near the site. All the seismic ground motions were estimated by a hybrid method (Green’s function method + 3D finite difference method or discrete wave number method).

Fig.3

Large-scale Earthquake.

Fig.4

Fault model.

Input seismic motions for the analysis were calculated with two-dimensional FEM model considering the kinematic interaction of soil and structure from the large-scale earthquakes defined before, for longitudinal and transverse directions. In order to take into account the three-dimensional behavior of earthquake acting on these long-span bridges, longitudinal, transverse and vertical seismic motions were input simultaneously. For the earthquakes from the unknown faults, longitudinal and transverse seismic motions were input individually and the bridges were verified with average response.

3.3 Train load

In the seismic retrofit design of the Seto-Ohashi Bridges, a train load is considered. The details of the train load are shown in Table 1. The train load is loaded on one side line (east side of the bridge).

Table 1
Train load

Standard train load EA-17 (axle load = 170 kN)

Train load per unit length 35 kN/m

Upper limit of Longitudinal direction: 0.2 g,

inertia force transverse direction: 0.3 g

Load length The whole bridges

The number of loaded lines 1

3.4 Analysis procedure

All truss members of the stiffening girders (maintenance ways were not included) were modeled in the analysis to see precise behaviors during earthquakes. In order to conduct the analysis efficiently, it is divided into several steps. The analysis process is explained in the following.

Step 1: In the analysis, linear elements are used and some members are simplified. The objective of Step 1 is to grasp overall damage level as well as to verify the validity of the model.

Step 2: In the analysis, the members that aren’t modeled in Step 1 (road girders, train girders, etc) are precisely modeled. The objective in Step 2 is to see precise behaviors of the members.

The Kita Bisan-Seto Bridge and the Minami Bisan-Seto Bridge share BB4A anchorage. Effect of one bridge to the other was examined between Step 1 and Step 2. The results of the individual bridge model and combined bridge model were compared. It was found that the difference is negligible. Therefore, in the subsequent analysis, the individual bridge model is used.

3.5 Analytical model

The precise model of the Kita Bisan-Seto Bridge is shown in Fig. 5. The towers and stiffening girders are modeled as fiber elements because they could be plasticized and the large variation of axial force and biaxial bending affect them. The connections of members of the stiffening truss girders are modeled as rigid. As steel road decks are separate for north and south-bounds, they are modeled as two beams. And all the bearings of the girder for road decks are modeled to simulate their actual behavior.

Fig.5

Analytical model (Kita Bisan-Seto Bridge).

Main cables and suspender ropes are modeled with geometric nonlinear elements. Models of the stay ropes can simulate breakage and do not resist to compression force.

The tower links and end links are modeled as truss elements with hinge ends. In order to take into account geometric nonlinearity, relationship between relative displacement in a horizontal direction and reaction force is defined as in Fig. 6.

Fig.6

Force-displacement relation of link.

Wind bearings are modeled as nonlinear spring to restrain transverse movement of the girder. End stoppers are modeled as spring element considering the space between stiffening girder and anchorage.

Since the embedment is shallow compared with the bottom area for all the foundations, embedment is not considered. Nonlinear springs distributed at even intervals in longitudinal and transverse direction are connected to the centroid of the bottom of foundations radially with rigid elements (Winkler model) [3].

Damping ratios for individual members are set based on the Specifications for Highway Bridges [1] as shown in Table 2.

Table 2

Damping constant

Category	Member	Model	Linearity	Damping ratio
Tower	shaft, horizontal strut	fiber element	Non-linear	0.01
Main truss	upper and lower chord, diagonal member,	fiber element	Non-linear	0.01
	vertical member
Floor truss	upper and lower chord, bracket etc	fiber element	Non-linear	0.01
Lateral bracing	upper and lower lateral bracing	fiber element	Non-linear	0.01
Floor system	road girder	linear beam element	Linear	0.02
	road girder bearing	linear spring element	Linear	0
	rail girder bearing
Displacement restriction	wind bearing	spring element	Linear	0
	end stopper	spring element	Non-linear	0
Cable	main cable, suspender rope center stay	non-linear beam element	Linear	0.01
Anchorage	body	rigid	Linear	0
	splay saddle	linear beam element	Linear	0
Foundation-ground system	spread foundation	Winkler	Non-linear	0.1

The natural periods of each bridge are shown in Table 3.

Table 3

Natural period

	Direction
	Longitudinal	Transverse
BB2P	0.95 sec	2.36 sec
BB3P	1.04 sec	2.37 sec
BB5P	1.18 sec	2.49 sec
BB6P	1.03 sec	2.47 sec
Girder of Kita	7.48 sec	11.22 sec
Bisan-Seto Bridge
Girder of Minami	8.07 sec	12.96 sec
Bisan-Seto Bridge

As the Seto-Ohashi Bridges are flexible and affected by geometric nonlinearities, dynamic analyses considering both material and geometric nonlinearities are conducted [4].

It was confirmed that the vibration frequencies in the original design are consistent with the ones measured in actual bridge. And the ones in this model used in the seismic verification were approximately the same as the ones in the original design. Therefore it is thought that this model is valid.

3.6 Results

For the verification of the steel members, two criteria are used. One is to check the strain of the fiber elements and the other is the stability verification formula for members under axial force and bending moment. The support members (tower links, bearings, etc) are assessed by their strength that corresponds to the lowest yield stress of the components and the displacement.

The analysis result is shown in Table 4. Overall, the total number of damaged portions is small in all three bridges. Especially, main members such as main towers and stiffening girders will not suffer severe damage. This is probably because these bridges are the structures with long – period and there is no great difference between earthquake force in the original design and the one in this analysis in the frequency range near the natural frequencies of the fundamental vibration modes of stiffening girders.

Table 4
Summary of seismic performance verification

Structure Structural element Evaluated item Results

Shimotsui-Seto Kita Bisan-Seto Minami Bisan-Seto

Bridge Bridge Bridge

Superstructure Stiffening girder Stability 0.995 0.90 0.99

Yield 1.99 0.80 0.97

Tower Stability 0.93 0.75 0.85

Yield 0.55 0.50 0.56

Shear (horizontal strust) 0.10 0.08 0.10

Base, anchor 0.08 0.09 0.08

Link Tower link 0.48 0.28 0.30

End link 0.41 0.29 0.38

Wind bearing Tower 0.38 0.44 0.59

Anchorage 0.44 0.39 0.45

Saddle Friction 0.41 0.72 0.59

End stopper Gap between end stopper 0.29 0.30 0.37

and anchorage

Road girder bearing Displacement 0.64 0.51 0.82

Yield strength 2.36 3.49 4.88

Center stay Breakage All stay All stay All stay

broken(12/12) broken(12/12) broken(12/12)

Main cable Yield 0.40 0.44 0.40

Suspender rope Yield 0.29 0.33 0.33

Substructure Stability Area of upward separation 44% 47% 49%

of the foundation from the ground

Bearing capacity OK OK OK

Verification of members OK (SB2P,SB3P) OK (BB2P,BB3P) OK (BB5P,BB6P)

(upper slab of caisson foundation)

Structure	Structural element	Evaluated item	Results
Superstructure	Stiffening girder	Stability	0.995	0.90	0.99
		Yield	1.99	0.80	0.97
	Tower	Stability	0.93	0.75	0.85
		Yield	0.55	0.50	0.56
		Shear (horizontal strust)	0.10	0.08	0.10
		Base, anchor	0.08	0.09	0.08
	Link	Tower link	0.48	0.28	0.30
		End link	0.41	0.29	0.38
	Wind bearing	Tower	0.38	0.44	0.59
		Anchorage	0.44	0.39	0.45
	Saddle	Friction	0.41	0.72	0.59
		End stopper	Gap between end stopper	0.29	0.30	0.37
		and anchorage
	Road girder bearing	Displacement	0.64	0.51	0.82
		Yield strength	2.36	3.49	4.88
	Center stay	Breakage	All stay	All stay	All stay
			broken(12/12)	broken(12/12)	broken(12/12)
	Main cable	Yield	0.40	0.44	0.40
	Suspender rope	Yield	0.29	0.33	0.33
Substructure	Stability	Area of upward separation	44%	47%	49%
		of the foundation from the ground
		Bearing capacity	OK	OK	OK
		Verification of members	OK (SB2P,SB3P)	OK (BB2P,BB3P)	OK (BB5P,BB6P)
		(upper slab of caisson foundation)

Numbers: Maximum of response/allowable value. Over 1.0 is NG.

Center stays and road girder bearings will be damaged for all the bridges. Since it was verified that the girder will not hit the anchorages after the breakage of the center stays in the verification, it was decided that the breakage of the center stays was allowed.

The road girder bearings (pot bearing) support steel decks that have 6 spans and are separate for north and south bounds. Each support point has four bearings. Many of the bearings on the center and outer sides (Fig. 7) will be damaged, as shown in Fig. 8. The damage is not caused by that the steel decks cannot keep up the deformation of the truss girder but by the shortage of the strength against inertia force. Especially, the damage concentrates on bearings for G1 girder. This is because they are not directly loaded by live load and therefore designed with lower strength compared with other bearings.

Fig.7

Location of bearings.

Fig.8

Results of seismic performance verification for bearing.

In the truss members of the stiffening girder of the Shimotsui-Seto Bridge, it is found that some upper chords and middle chords of the floor trusses near the end supports and the middle supports yield. The damaged portion of truss members is shown in Fig. 9. This damage is caused by the deformation of the stiffening girders in transverse direction. The damage level is assessed further by the following steps.

Fig.9

Damaged portion of truss members and bearing.

For the upper chords and middle chords of the floor truss which will yield during earthquake, the axial force is small and the exceedance of the strain over yield strain is small and local.

Therefore it is thought that the damage will not lead to the degradation of load bearing capacity in terms of structural safety. Moreover, the fiber model of the stiffening truss does not consider the connection structures such as the gusset plates. In order to assess the force flows transmitted through the gussets of the connection, the floor truss which will be damaged most as shown in Fig. 10 was replaced with the precise FEM shell elements including the gussets of the connection and the dynamic response analysis of the whole bridge was conducted. Dynamic characteristics of this model were confirmed to be consistent with the previous model. As a result, maximum stress is about 0.6 times of the yield stress in the von Mises stress. Moreover, the gusset will suffer no damage.

Fig.10

Deformation when the maximum stress occurs in the upper-chord of floor truss (Shimotsui-Seto Bridge; the Tounankai-Nankai earthquake; magnification of 100 times).

From the above, damage is considered to be within repairable level after earthquake. Thus it was decided that the members would not be retrofitted.

4 Seismic retrofit design

4.1 Evaluation of measure

From the evaluation in the previous chapter, since it was decided that the damage of the stiffening truss of the Shimotsui-Seto Bridge and the breakage of the center stays of all the three bridges were allowed, only members to be retrofitted are the road girder bearings.

It was evaluated whether the damage to the road girder bearings can be allowed. The analysis used two models. In the first model, damage of the bearings is considered. That is, if bearings break, force born by those bearings will be redistributed to the others and damage will propagate. In the second model, bearings are assumed to be reinforced and not to be damaged. That is, earthquake force is born by each bearing (same model used in the verification). Figure 11 shows distribution of damaged bearings of the Kita Bisan-Seto Bridge by the Tounankai-Nankai earthquake. As shown in the figure, many bearings will be damaged by the redistribution of earthquake force in the first model. The results are similar for other bridges though the number of the damaged bearings varies. Retrofit of bearings is judged to be advantageous in terms of serviceability and repairability after earthquake.

Fig.11

Difference of seismic performance results due to analytical condition of bearing.

However, since the number of bearings to be retrofitted is large, additional measures that can reduce the number were studied. Table 5 lists candidate measures to reduce earthquake force acting on the bearings and their evaluations. Regarding measures 1, 2, and 3, workability is low and even some bearings that have enough strength must be modified. Measures 4 and 5 are found to be not effective. Eventually, additional measure is not taken and bearings are retrofitted.

Table 5

Additional seismic measures for bearings

Number	1	2	3	4	5
Measures	Installation of damper for road	Distribution of inertial force	Connection of road decks	Installation of damper for	Installation of damper for
	deck girder			center stays	stiffening girder
Outline	By the dampers for road deck girder, inertial force is distributed to multi bearings. Effective only for longitudinal direction.	By the additional rubber bearings between movable bearings, horizontal force acting on the road deck is distributed. Effective for both directions.	Several road decks are connected and becomes one vibration unit. And by the alteration of supporting condition (fix – >move), horizontal force will be concentrated to arbitrary bearings.	Additional damping and alteration of vibration characteristics are expected by the damper.	Additional damping and alteration of vibration characteristics are expected by the damper between girder and anchorage/tower.
Evaluation	Effect of the dampers is small, because the relative displacement between deck and stiffening girder is small (30 mm).	Distribution effect is small, because the relative displacement between deck and stiffening girder is small (30 mm).	Connection method of road decks has to be studied.	By the installation of damper, center stays become dysfunctional for original purpose.	The number of damaged bearings increases by this measure.

4.2 Seismic retrofit design

For the design of the additional members to reinforce bearings, extent of the excess of earthquake force over the strength of bearings is considered. If the excess is small, additional member is designed for the shortage of the strength. If the excess is large, additional member is designed for the earthquake force acting on the bearing.

The additional retrofit member for the longitudinal direction is shown in Fig. 12. For bearings in which excess is large, the additional member attached to the road girder is designed to receive the reaction force from the upper chord of the floor truss. The additional retrofit member for the transverse direction is shown in Fig. 13. The member is attached on the upper chord of the floor truss receiving reaction force through the lower plate of the bearing regardless of the degree of the excess.

Fig.12

Seismic retrofit for bearing (longitudinal direction).

Fig.13

Seismic retrofit for bearing (transverse direction).

If the bearings under the G1 girders are retrofitted, the inertial forces of road deck will concentrate on fixed bearings under the G1 girders. In this case, there should be no difference between the displacement of the G1 girder and those of the G2-4 girders. Currently, lateral bracings are installed between G1 and G2 near the bearings (Fig. 14). If the bearings under G1 girders are retrofitted, excessive axial force will act on the lateral bracings. Thus the lateral bracings will be retrofitted against this axial force.

Fig.14

Lateral bracings to be retrofitted.

5 Conclusions

This paper presented the seismic retrofit design of the suspension bridges of the Seto-Ohashi Bridges. Even with the earthquakes larger than the ones considered in the original design, damage level of the bridges were relatively small. And with the additional in-depth analysis, the number of members to be retrofitted was minimized. Finally, it was decided that some road bearings and some other members would be retrofitted.

Footnotes

Acknowledgments

The authors would like to acknowledge useful supports from the Honshi Seismic Retrofit Study Committee members (chairman: Dr. Iemura, the professor emeritus of Kyoto University).

References

Japan Road Association. The Specifications for Highway Bridges (in Japanese), 2012.

Honshu-Shinkoku Bridge Authority. Seismic Design Code (in Japanese), 1997.

Harada

, Nonaka

, Magoshi

, Iwamura

, Oho

. A Nonlinear Dynamic Soil-Large Foundation Interaction Model Using Fiber Element”, the 12th Japan Earthquake Engineering Symposium, pp. 646–9 (in Japanese), 2006.

Seismic Analysis Research Center, SeanFEM ver1.2.2.

Standard train load	EA-17 (axle load = 170 kN)
Train load per unit length	35 kN/m
Upper limit of	Longitudinal direction: 0.2 g,
inertia force	transverse direction: 0.3 g
Load length	The whole bridges
The number of loaded lines	1