behaviour of a steel bridge

7/30/2019 Behaviour of a Steel Bridge

1/10

EACS 20125th

European Conference on Structural Control Genoa, Italy18-20 June 2012

Paper No. # 074

1*

Corresponding author

Behaviour of a Steel Bridge Equipped with Seismic Isolation Devices

Ionu Radu RCNEL 1*, Dan Ilie CREU 2Technical University of Civil Engineering, Department of Strength of Materials, Bridges and Tunnels

122-124 Lacul Tei Bvd., Bucharest, 020396, [email protected], [email protected]

Costin Stelian MUTU 3

S.C. I.P.T.A.N.A. S.A., Department of Structures Computation36-38 Dinicu Golescu Bvd., Bucharest, 010873, Romania

[email protected]

ABSTRACT

Bridges are structures subjected to several types of external loads during their service life. One ofthese external actions which have a high order of uncertainty is the seismic action and in order to design a

safe structure many scenarios concerning this type of action were made. Earthquakes are one the most severenatural actions which can occur during the service life of a structure. For this reason, in the last decade, all

the countries placed in active seismic regions have produced specific standards and provisions having theaim to design safer structures.

Romanian territory includes several seismic faults covering a major part of the country, the seismicaction being considered in the design stage for a new structure.

In the past, generally, the desired safety factor for a bridge structure considering the seismic actionwas reached by increasing the dimensions of the structural elements.

By introducing the use of passive control devices and isolating the base of tall buildings a new era

began also in the field of bridges. Using special bearing devices as lead rubber bearings (LRB), highdamping rubber bearing (HDRB), dampers or friction pendulum isolators (FPS) or combinations of these, thebehavior of the bridge under seismic loads can be significantly improved. Thus, really benefits regarding thecost of the whole structure can be obtained.

In this paper, the behavior of a new steel bridge equipped with passive control devices is

investigated. The bridge has some special characteristics: large spans (200m), tall piers with height between

30 and 70m and bad soil conditions for the substructure elements. In the same time, the site is characterizedby a high value of the ground acceleration (ks=0.32g). Several types of passive control devices are used,finally a comparison between their influences of the bridge response being presented.

Keywords:bridge, earthquake, seismic action, lead rubber bearings, friction isolators, stresses.

1 INTRODUCTIONThe seimic action is one of the most severe which cand occur during the lifetime of a

structure and can produce damages, in the worst situations leading even to the collapse of the

structures and also casualties. Bridges are complex structures which are supporting roads or

railways being also exposed to the effects of the earthquakes. The level of redundancy of a bridge ismany times lower than in the case of buildings because of their geometry (mainly constructed in


2/10

2

one direction) and their structure. The bridge substructure consisting in piers and abutments is

organized, almost in all cases, with the structural elements in a row, so that the effects of a severe

damage on one element affect imediatelly all other elements. Last recent seismic events occured all

arround the world shown what could happen if the seimic action is not or is wrong considered in the

estimation of the structural behaviour.

Because of the uncertainity of seimic action in terms of intensity, duration, place of occurence

and mode of action it is very difficult to establish the best method to be use in order to protect astructure and limit the effects of an earthquake. In the past, the concept of increasing stiffness of the

structural elements was first adopted in the design. According to this, the resistance capacity of the

structure was increased and as a consequence, the dynamic response was kept in the domain with

high frequencies, low periods and high acceleration values. Supplementary, the structural elements

are forced to take high levels of loads which can lead to damages on their cross sections.

In the last decade the performance based design, which accounts the needs of a structure in

terms of displacements for, became an adopted solution in the design and induced the fact that

structural flexibility can be an alternative to the structural stiffness to improve the dynamic

response. The correct estimation of the needs of a structure, in terms of displacements, for a certain

level of the seismic action can lead to significant reduction of costs by limitng the seismic action

effects and keeping the structure further on in service.The use of innovative theoretical methods in structural response estimation, together with the

introduction of passive control devices have shown that a real improvement in the structural

behaviour can be achieved. The isolation devices were used first in the case of tall buildings to

isolate the base and limit the values of accelerations induced by the ground motion. The solution

was later implemented also for bridges and by placing such kind of devices between super and

substructure, a certain level of decoupling of these to parts of bridge can be obtained. As a result, an

improvement of the behaviour in terms of reducing the internal forces and limit the displacements

can be observed.

The study presented in this paper shows the benefits of using passive devices (Seismic

Isolators and Tune Mass Dampers) to control and improve the dynamic response of a steel bridge

subjected to seismic action. A well arrangement between the level of displacements and stressesinduced in a structure can have favourable effects both from structural safety and economical point

of view.

2 DESCRIPTION OF THE ANALYZED BRIDGE STRUCTUREThe bridge analyzed in this paper will be errected on the national road DN 2D near the city of

Focani. Because of the rough ground on the site and also because of the high level of the ground

acceleration (0.32g) the adopted solution had some special characteristics. In order to reduce the

inertial effects induced by the seismic action, a light superstructure and tall piers were adopted. For

the lateral spans, the superstructure is a composite one, in concrete and steel and for the main spans,in the central part of the viaduct, a complete steel superstructure was designed. This part of the

bridge was chosen for the analyses in this paper.

The total length of the bridge is 1442 m in the following succesion of spans:

3 70+160+200+160+3 70+3 70+4 70m. In figure 1 an elevation of the bridge main part

including the major spans is shown. Because of the large values of the spans an orthotropic deck

solution was designed for the superstructure. The steel box has a variable height in the range 4.00-

7.50 m and the distance between the vertical webs is 7.00m. The steel plates at the bottom and top

part of the box as well the webs are stiffened using open ribs with flanges. Each 5.50 meters, the

webs are connected in transverse direction through cross girders placed at the bottom as well at the

top of the box, but also through vertical bracings. The thickness of steel sheets forming the boxvaries between 12 and 20 mm. The bridge superstructure sustains a roadway with three lanes (two

upgrade an one downgrade) each 3.50 m width, two shoulders of 0.40 m and two footways of 1.50


3/10

3

m each side of the road. It results a total width of 14.30 m of the superstructure at the top part. The

cross sections through the steel box of the superstructure on the bearing (cross section A-A) and in

the midspan (cross section B-B) are presented in figure 2.

Figure 1General layout of the main part of the viaduct.

Figure 2Cross section through the bridge superstructure.

The bridge substructure consists in 2 abutments and 15 piers which have deep foundations on

piles having the diameter of 1.50 m and a length of 30.00 m. Because of their height, the piers

elevation has special box shape with a depth of 4.50 m in bridge longitudinal direction and a width

of 7.00 m and a variable height between 39.00 m and 55.00 m. The cross section of the piers differson the length: 20.00 m above the piles cap a single reinforced concrete box was designed and on the

remaining height this was replaced two separate boxes also in reinforced concrete (Fig.3). This

solution was chosen in order to increase the piers flexibility regarding the effects of the seismic

action.

The length of the piles which are forming the bridgefoundations is justified by the high values

the piers height, but also by the presence of several soil layers with bad geotctechnical

characteristics.

The connection between super- and substructure is made through neoprene bearing devices

which were installed below the webs of the steel box. The piers shape allows to place on each pier

four bearings, this approach offering advantages regarding the dimensions of the bearing devices.

Figure 3

Cross section through piers elevations.


4/10

4

3 DISCRETE MODELS USED FOR THE ANALYSESIn order to obtain the bridge response following the seismic action, several three dimensional

simple finite element models were considered. The difference between these models consists in the

modeling, characteristics and behaviour of the bearings used to support the bridge superstructure on

the substructure elements. A 3D view of one of the finite element models is presented in figure 4

bellow.

Figure 43D view a finite element model.

The bridge steel box orthotropic deck was modelled using straight frame elements with twojoints. These elements concentrate the geometrical characteristics for the whole deck cross section.

In order to respect the position of the deck neutral axis with respect to the piers supporting saddles,

but also to the bearings, rigid connections (rigid link elements) were used in horizontal and vertical

direction respectivelly, as can be seen in figure 5. For the modelling of the bearing devices, link

elements with linear or nonlinear behaviour were introduced in the models, thus to obtain the

desired response of the structure. The standard elastomeric have into the model linear elastic

characteristics and the LRBs (Lead Rubber Bearings) together with FPSs (Friction Pendulum)

isolators have nonlinear characteristics according to a bilinear response curve, as will be shown

later in this paper. The TMDs (Tune Mass Dampers) attached to the piers top part were modelled

using a viscous damper and a linear spring element which are acting in paralell. In figure 5 the two

nodes straight frame elements modelling the piers elevation can also be seen.

Figure 5Modelling of the connection between bridge super- and substructure.

The soil-structure interaction was considered in the finite element models through six linear

springs (three for translation and three for rotation degrees of freedom of the foundation system)

placed at the bottom of the pile elevation, their stiffness equivaleting the stiffness of the pile group

for translation and rotation respectivelly. The equivalence was made assuming that along each pile

of the group are distributed, two orthogonal in horizontal plane linear springs and one spring in

vertical direction. The springs characteristics were computed according to the soil parameters


5/10

5

obtained from bohrholes on the site. All dynamic analyses were performed using these finite

element models in order to investigate the bridge responses under seismic loads.

4 PERFORMED ANALYSES AND RESULTSIn order to calibrate the models for the dynamic response, linear eigenvector analysesfollowed by response spectrum and linear time-history analyses were performed on the bridge

equipped with standard elastomeric bearing devices. The resulted maximum values for the internal

forces (bending moments and shear forces) on the piers as well the nodal displacements at

superstructure level were compaired for the response spectrum and linear time-history analyses. The

used design response spectrum according to the romanian norm P100-2006 is presented in figure 6

and it describes the seismic input for the bridge location characterized by a corner period T c=1 s and

a peak value of the ground acceleration ag=4.3 m/s2. According to the piers height and cross section

geometry and stiffness, a behaviour factor q=2 was considered to obtain the values in the design

spectrum. For the linear time-history analysis a number of five artificial accelerograms based on the

previously presented response spectrum were generated. The average response spectrum according

to the generated accelerograms was plotted also in figure 6 together with the design responsespectrum. In figure 7 the corresponding displacements response spectra are plotted.

Figure 6

Accelerations response spectra Figure 7

Displacements response spectra

From the modal analysis, the fundamental period of the structure is T1=2.56 s which places

the response of the structure in the range of medium accelerations but large displacements. In Table

1 are presented for comparison the values of the bending moments and shear forces at the base of

the tallest pier P5, but also the values of the horizontal displacement at the superstructure level

following the response spectrum and linear time-history analyses. Only the first generated

accelerogram was used at this stage in the time-history analysis.

Table 1Bending moments, shear forces and displacements.

Analysis type Bending moments M, [kNm] Shear forces V, [kN] Dsiplacements d, [m]

Response spectrum 201539 4677 0.31Linear time-history 190477 4878 0.31

Differences [%] -5.5 +4.1 0

According to the values presented in table above it can be concluded that the internal forces

on the cross section of the pier P5 are high, leading to high values of reinforcement percentage.

Concerning the displacement value it is over the usual limits (0.20, 0.25m) that allow the use of

common expansion joint devices. Thus, the goal of the study is first to decrease the values of the

superstructure displacement at maximum 0.25m and second to adjust the values of the internal

forces on the piers cross section. The tallest pier P5 will be monitored for this purpose.

Improvement of the bridge behaviour under seismic loads claim the use of special bearing

devices. In this study two types of isolators are used separately: lead rubber bearings (LRB) andfriction pendulum systems (FPS). Finally, another aproach based on the use of TMDs (Tune Mass


6/10

6

Dampers) placed inside the piers cross section is also accounted for. Considering the values of the

internal forces on pier P5 cross section and of the superstructure displacement following nonlinear

time-history analyses a comparison is made in order to establish the most effective solution to

obtain the desired response of the structure. The study is made considering all five artifficially

generated accelerograms matching with the acceleration response spectrum presented in figure 6.

The LRBs and FPSs were placed at the top of all piers of the bridge. Both isolation systems

LRBs and FPSs were modeled into the structural analysis program using nonlinear link finiteelements. For the LRBs the rubber isolator element was chosen and for the FPSs the plastic Wen

element was selected. For both isolation devices the force-displacement relationship was described

using a bilinear curve. In figure 7, the theoretical function used to describe the nonlinear behaviour

of the isolators is shown. Figure 8 represents the schemes of LRB and FPS bearings used into the

analyses.

Figure 8The function used for the isolators Figure 9Schemes of seismic isolators

In figure 8 Fmax, Dmax are the maximum force and the maximum displacement of the isolator,

Fy, Dy are the yileding force and the yielding displacement, K1 is the initial stiffness, K2 is the post

yield stiffness and Keffis the effective stiffness.

The fundamental period of the structure is 2.56 s, outside of the amplification domain of the

response spectrum. Starting from this value, the characteristics of the used isolators can be

established. For the LRB isolators the following equations were used to find the basic

charasteristics:

2str

mT

k(1)

1 PIERS streff

PIERS str

k kk

n k k(2)

where T is the fundamental period of the structure, m is the modal mass, keff is the efective

stiffness of one isolator, kPIERS is the bending stiffness of the piers and kstr is the overall stiffness of

the bridge structure. Based on this value ofkeff from the Algasism catalogue the LRN D700 B750Z550 was chosen, with =4% and G=0.9 MPa. The effective damping efffor this type of isolator is

30% and the maximum displacement is 140 mm. In this case, K1 is the lead core and K2 is the

elastomer contribution respectivelly.

For the friction pendulum system the equations used are as follows:

2R

Tg

(3)

VF uV D

R(4)


7/10

7

2;

/eff eff

F uk

D u D R(5)

1 2;y

uV VK K

D R(6)

whereR is the radius of the sliding surface;gis the gravity constant;Fis the horizontal force

taken by the friction pendulum system; u represents the dynamic friction constant in the range 3%-12% according to [4]; Vis the total vertical load on a isolator; D is the isolator displacement. For

this case study the period of the isolator was chosen smaller (T=2.15 s) than the structure

fundamental period in order to take into account also the piers contribution to the general

displacement of the bridge. The resulted value for the radius of the sliding surface wasR=1.15 m.

Based on this equations and following nonlinear time history analyses using all five

artificially generated accelerograms, the complete dynamic response of the bridge in terms of

displacements and internal forces are obtained. In figures 10 and 11 the force-deformation plots are

presented both for a LRB and a FPS on pier P4 using as input the first accelerogram.

Figure 10Force-displacement plot for a LRB Figure 11Force-displacement plot for a FPS

Aa an alternative to the LRB and FPS isolators, in this study also the influence of TMDs

(Tune Mass Dampers) on the dynamic response of the bridge was investigated. This devices are

vibration absorbers attached to the primary oscilating system, in this case the bridge structure. Ingeneral they consist in a mass which is connected to the primary system through a oscilating system

formed by a spring and a viscous damper acting in paralell. The accurate tunning of the frequency

of the TMD with the frequency of the structure lead to introduction into the primary system of

inertial forces, which will counterbalance its movements, the result being a reduction of

displacements and internal forces of the primary system. While the displacements of the TMD are

large, those of the primary system will be reduced.

In this paper, the use of TMDs is proposed as an alternative to other types of isolators,

because in some cases, the resistance capacity of the bearings do not allow to take large values of

vertical reactions produced by static or dynamic actions. The TMDs were placed only on the two

central piers, P4 and P5 respectivelly.The modeling of the TMD in performed numerical analyses is presented in figure 12. They

were placed at the top of piers P4 and P5 and consist in two systems viscous damper-elastic spring

which are connected through rigid elements. The mass is disposed in the middle as presentd in

figure 12.

Figure 12

Scheme of the TMD used in the analyses


8/10

8

According to [5] the otpimum frequency of the TMD should be in the range 95-99% of the

frequency of the primary system. Knowing the fundamental frequency of the bridgefs and the value

of the modal mass ms, the mass mt, the spring stiffness kt and the damping constant ct of the TMD

can be established using the equations bellow:

1

st

t

s

ff

m

m

(7)

1 1;

2 2

t st s

t s

k kf f

m m(8)

3

3

8 1

t

sopt

t

s

m

m

m

m

(9)

2 ; 2t t t t t opt f c m (10)

whereft, t and opt are the frequency, the circular frequency and the optimum damping ratio

of the TMD. Using these equations the characteristics of the absorber were determined and inputed

into the nonlinear time-history analyses. The ground motion was simulated using the five artificially

generated accelerograms. Finally, the results obtained using accelerogram four, which lead to the

largest values of horizontal displacement of the superstructure, were selected for comparison for all

type of analyzed isolation devices and are summarized in the table 2 and in figure 13 bellow.

Table 2Horizontal displacements of the superstructure in meters

Input ground motionStructure with

neoprene bearings

Structure with

LRBs

Structure with

FPSs

Structure with

TMDsGenerated accelerogram

number four0.35 0.19 0.18 0.23

Differences in % with respect to

the values obtained on thestructure with neoprene bearings

- -46 -49 -35

Figure 13Time histories of the horizontal displacement of the superstructure

Figures 14 and 15 show a comparison in terms of internal stresses on the piers P5 cross

section computed at the base of the pier. These are absolute maximum or minimum values obtained

from the envelope of the internal forces, resulted following the nonlinear time-history analyses

performed for each of five accelerograms. The values in the figures 14 and 15 were obtained for the

accelerogram four who leads to the maximum displacements of the bridge superstructure.


9/10

9

Figure 14Bending moments (base of pier P5) Figure 15Shear forces (base of pier P5)

During the motion induced by the seismic action, the mass attached to the TMD exhibits large

displacements which reduce significantly the overall displacements of the structure. Figure 16

shows time-history responses in terms of displacements for the mass of the TMD, for a point

situated at the top of the pier elevation and for a point on the bridge superstructure.

Figure 16Displacements of the TMD mass, pier and superstructure

5 CONCLUSIONIn this paper the behaviour under the seismic action of a steel bridge equipped with standard

bearings, but also with special isolation devices, was investigated. For this purpose, several finite

element models were built and analyzed by the add of linear response spectrum, linear and

nonlinear time-history analyses. The ground motion was simulated using a design response

spectrum function foreseen in the romanian norm P100-2006 for the site where the bridge is placed,

but also five artificially generated accelerograms matched to comply with this response spectrum

function. All analyses were performed on the structure equipped with standard elastomeric bearings

(neoprene), but also on the structure equipped on all piers with lead rubber bearings (LRB) or with

friction isolators (FPS) and with tune mass dampers only on the central piers P4 and P5.

Because the fundamental period of the structure is outside the amplification domain of the

response spectrum, the goal of the study is to establish the appropriate isolation device to be used in

order to keep the superstructure displacement bellow 0.25 m. Thus, the use of common expansion

joint devices is possible.

The structure with standard elastomeric bearings exhibits large displacements at the

superstructure level (0.35 m for accelerogram four) and also large stresses at the base of the piers.

This because of the impossibility of the beearins to dissipate energy. By introducing the isolation

devices a significant improvement in the bridge behaviour can be achieved. Thus, the superstructure

displacements and the bending moments at the base of the piers can be reduced with almost 50%

using FPSs, the reduction in terms of shear force being only arround 18% in the case of using

LRBs. Globally, the FPSs seems to offer the best solution to improve the dynamic response of the

bridge structure, both in terms of displacements and stresses on the piers cross section.Supplementary, the capacity of the FPSs system to restore the strructure near to undeformed

position reccomends it for the future use to solve this kind of problems.


10/10

10

Regarding the use of TMDs it can be concluded that their efficiency in terms of reducing

displacements and stresses is smaller compaired to other two used devices. The tuning of a TMD is

very complexe because of the large composition in frequency of the induced ground motion. To

obtain a better behaviour, the TMD should be tuned for each range of significant frequency

contained in recorded ground motion for real earthquakes. In this paper, the tuning of the TMD was

made to fit on all five artificially generated accelerograms. Even thus, it can be observed that the

reduction with respect to the structure equipped with standard elastomeric devices is, fordisplacements of 35%, for bending moments of 19% and for shear forces of 10%. Of course a

problem can be the installation of such devices on bridges because their wheight and dimensions.

REFEERENCES

[1]Naeim, F. and Kelly, J.M. 1999: Design of Sesimic Isolated structures: from theory to practice,John Wiley&Sons, Inc., N.Y.

[2]Chopra, A.K. 2007:Dynamics of Structures. Theory and Applications to EarthquakeEngineering, Pearson Prentice Hall International.

[3]Zekioglu, A., Darama, H. and Erkus, B. 2009:Performance-Based seismic design of a largeseismically isolated structure: Istanbul Sabiha Gken Intrenational Airport Terminal Building.

SEAOC 2009 Convention Proceedings ,409-427.

[4]EPS, Inc. 2003: Technical characteristics of friction pendulum bearings, Vallejo, California,USA

[5]Bachmann, H. Et al. 1995: Vibration problems in structures: practical guidlines, Birkhuser,Berlin

[6]***P100-2006 2006: Romanian Seismic Design Code-Part I: Design Provision for Buildings[7]*** SR-EN 1998-2. 2006: Design of structures for earthquake resistance. Part 2: Bridges[8]Computers and Structures Incorporated 2010: CSI Analysis Reference Manual for SAP2000,

Etabs and Safe, Berkeley, California, USA.

behaviour of a steel bridge

Documents