Fulltext01 1

Wind Loads on Bridges
Analysis of a three span bridge based on theoretical methods and Eurocode 1
M. Sajad Mohammadi Rishiraj Mukherjee

June 2013 TRITA-BKN. Master Thesis 385, 2013 ISSN 1103-4297 ISRN KTH/BKN/EX-385-SE
Examiner and supervisor: Prof. Raid Karoumi @M. Sajad Mohammadi ([email protected]), 2013 @Rishiraj Mukherjee ([email protected]), 2013 Royal Institute of Technology (KTH) Department of Civil and Architectural Engineering Division of Structural Design and Bridges Stockholm, Sweden, 2013
Abstract The limitations lying behind the applications of EN-1991-1-4, Eurocode1, actions on structures-general actions-wind load-part 1-4, lead the structural designers to a great confusion. This may be due to the fact that EC1 only provides the guidance for bridges whose fundamental modes of vibration have a constant sign (e.g. simply supported structures) or a simple linear sign (e.g. cantilever structures) and these modes are the governing modes of vibration of the structure. EC1 analyzes only the along-wind response of the structure and does not deal with the cross wind response. The simplied methods that are recommended in this code can be used to analyze structures with simple geometrical congurations. In this report, the analytical methods which are used to describe the uctuating wind behavior and predict the relative static and dynamic response of the structure are studied and presented. The criteria used to judge the acceptability of the wind load and the corresponding structural responses along with the serviceability considerations are also presented. Then based on the given methods the wind forces acting on a continuous bridge whose main span is larger than the 50 meters (i.e. > 50 meter requires dynamic assessment) is studied and compared with the results which could be obtained from the simplied methods recommended in the EC1. Key words: Wind load, dynamic response of a continuous bridge, theoretical methods, Eurocode 1, vortex shedding, aerodynamic and aeroelastic instabilities, Natural frequencies.
Sammanfattning De begr ansningar som ligger bakom till ampningarna av Eurokod 1: Laster p a b arverk -Del 1-4: Allm anna laster Vindlast leder, byggnadskonstrukt orer till stor f orvirring. Detta beror eventuellt pa grund av det faktum att Euokod 1 bara ger v agledning for broar vars huvudsakliga sv angningsmoder har konstant tecken eller enkelt linj art tecken och att dessa ar de dominerande sv aningsmoder av b arverk. Eurocod 1 analyserar endast hur b arverk r or sig i vindriktningen. De f orenklade metoder som rekommenderas i denna kod till ampas f or att analysera de b arverk som har enkla geometriska kongurationer. I denna rapport diskuteras och redovisas de analytiska metoderna som anv ands f or att beskriva det varierande vindbeteendet och ber akningar av relativ statisk och dynamisk respons av b arverket. De kriterier som anv ands f or att bed omma godtagbarheten av vindlasten och dess motsvarande p averkan p a b arverket, tillsammans med o verv aganden kring funktionsduglighet, presenteras ocks a. Baserat p a de givna metoderna studeras och j amf ors vindlaster som verkar p a en kontinuerlig bro vars st orsta spann ar st orre a n 50 meter (dvs. > 50 meter kr aver dynamisk bed omning), med de resultat som kan erh allas fr an de f orenklade metoderna som rekommenderas i EC1. S okord: Vindlast, dynamisk respons av en kontinuerlig bro, teoretiska metoder, Eurokod 1, virvelavl osning, aerodynamiska och aeroelastisk instabilitet, egenfrekvens..
Acknowledgements We would like to express our sincere gratitude to our project guide Prof. Raid Karoumi for his valuable comments, remarks and suggestions throughout the duration of the master thesis. Furthermore we would like to thank Ali Farhang, Marco Andersson, Jens H aggstr om and Jesper Janzon-Daniel of Ramb oll AB for providing us with the relevant data and information required for the thesis. Also, we would like to thank Johan Jonsson of Trakverket for his guidance and comments. We would like to express our deepest appreciation to all the faculty members of the school of Architecture and Built Environment, KTH who imparted us with the knowledge and skills required for the successful completion of the project.
Contents
1 Introduction 2 Wind load 2.1 Wind load chain . . . . . . . . . . . . . . . . . . . . . . . 2.2 The atmospheric boundary layer . . . . . . . . . . . . . . 2.2.1 The roughness length . . . . . . . . . . . . . . . 2.3 Mean wind velocity- wind prole(homogenous terrain) . . 2.3.1 The logarithmic prole . . . . . . . . . . . . . . 2.3.2 Corrected logarithmic prole . . . . . . . . . . . 2.3.3 Power law prole . . . . . . . . . . . . . . . . . . 2.4 Wind turbulence . . . . . . . . . . . . . . . . . . . . . . 2.4.1 Standard deviation of the turbulence components 2.4.2 Time scales and integral length scales . . . . . . . 2.4.3 Power-spectral density function . . . . . . . . . . 2.4.4 correlation between turbulence at two points . . . 2.5 Static wind load . . . . . . . . . . . . . . . . . . . . . . . 2.5.1 Total wind load on structure- Davenports model 3 Dynamic response of structures to the wind load 3.1 Along-Wind Response . . . . . . . . . . . . . . . . . . 3.1.1 Single degree of freedom . . . . . . . . . . . . . 3.1.2 The Along-wind response of blu bodies . . . . 3.1.3 Design procedure for mode shapes with constant 3.2 Cross-wind vibtrations induced by vortex shedding . . 3.3 Bridge aerodynamics and wind-induced vibrations . . . 3.3.1 Flutter . . . . . . . . . . . . . . . . . . . . . . . 3.3.2 Bueting . . . . . . . . . . . . . . . . . . . . . . 3.3.3 Galloping . . . . . . . . . . . . . . . . . . . . . 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 7 7 8 8 8 9 9 10 10 10 11 13 16 16 17 21 21 21 24 28 30 34 36 39 42 45 45 46 46 47 48 48
. . . . . . . . . sign . . . . . . . . . . . . . . . . . . . . .
Wind Load on Bridges based on Eurocode1 4.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.1 Distinction between principles and application rules 4.1.2 Denitions . . . . . . . . . . . . . . . . . . . . . . . 4.2 Design situation . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 Persistent situations . . . . . . . . . . . . . . . . . 4.2.2 Transient situations . . . . . . . . . . . . . . . . . .
2
Contents
4.3 4.4
4.5 4.6
4.7
4.2.3 Accidental situations . . . . . . . . . . . . 4.2.4 Fatigue situations . . . . . . . . . . . . . . Modeling of wind actions . . . . . . . . . . . . . . Wind Velocity and Wind Pressure . . . . . . . . . 4.4.1 Basic wind velocity . . . . . . . . . . . . . 4.4.2 Mean wind . . . . . . . . . . . . . . . . . 4.4.3 Wind turbulence . . . . . . . . . . . . . . 4.4.4 Wind action . . . . . . . . . . . . . . . . . Structural factor (cs .cd ) . . . . . . . . . . . . . . . Wind actions on bridges . . . . . . . . . . . . . . 4.6.1 Force coecients . . . . . . . . . . . . . . 4.6.2 Force in x-direction- simplied method . . 4.6.3 Wind forces on bridge decks in z-direction 4.6.4 Wind forces on bridge decks in y-direction 4.6.5 Wind eects on piers . . . . . . . . . . . . Annexes . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
48 48 49 49 50 50 50 51 51 52 53 53 54 54 54 54 55 55 55 60 61 61 63 63 66 70 72 72 72 80 80 84 86 86 86 86 87 87 89 90 90 90 90
5 Finite Element Modelling 5.1 LUSAS FEM . . . . . . . . . . . . . . . . . . . 5.1.1 Geometry . . . . . . . . . . . . . . . . . 5.1.2 Cross-sections . . . . . . . . . . . . . . . 5.1.3 Foundations . . . . . . . . . . . . . . . . 5.1.4 Materials . . . . . . . . . . . . . . . . . 5.1.5 Element interactions . . . . . . . . . . . 5.1.6 Finite element modeling in LUSAS . . . 5.1.7 Analysis of dierent boundary conditions 5.2 MATLAB FEM . . . . . . . . . . . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
6 Analysis and Results 6.1 Natural frequencies and Mode shapes . . . . . . . . . . . . . . . 6.1.1 Results from LUSAS . . . . . . . . . . . . . . . . . . . . 6.1.2 Results from Matlab . . . . . . . . . . . . . . . . . . . . 6.2 Power spectral density function . . . . . . . . . . . . . . . . . . 6.2.1 Deck . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2 Piers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3 Response-inuence function . . . . . . . . . . . . . . . . . . . . 6.3.1 Deck . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.2 Piers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4 Wind Load Analysis . . . . . . . . . . . . . . . . . . . . . . . . 6.4.1 Determination of Gust factor . . . . . . . . . . . . . . . 6.4.2 Determination of Maximum Wind Load on the Structure 6.5 Vortex shedding and Aeroelastic instabilities . . . . . . . . . . . 6.5.1 Vortex shedding . . . . . . . . . . . . . . . . . . . . . . . 6.5.2 Galloping . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5.3 Divergence and Flutter Response of the Bridge Deck . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
Contents
7 Conclusion References Appendices A Mass Participation Factor Appendices B MATLAB Code
97 99 102 103 105 106
Chapter 1
Introduction
Wind load is one of the structural actions which has a great deal of inuence on bridge design. The signicant role of wind loads is more highlighted after it caused numbers of bridge structures to either collapse completely, e.g., Tacoma Narrows Bridge (1940) or experience serviceability discomforts e.g. Volgograd bridge (2010). Massive researches and studies have been carried out all over the world in order to analyze and model the uctuating wind behavior and its relative static and dynamic interactions with the bridge components and the corresponding structural responses to the turbulent wind load. The nature of the wind load is dynamic. This means that its magnitude varies with respect to time and space. As a result, analysis and modeling of such a load and its relative eects on structure may be quite complex and require substantial knowledge in mathematics, computational uid dynamics and structural analysis. The EN-1991-1-4 or Eurocode 1 and SS-EN-1991-1-4(2005) are the standard codes for the designers to evaluate both the wind forces acting on the surface of the structure and the corresponding static and dynamic response of the structures. SSEN-1991-1-4(2005) contains the Swedish annex which points out which clauses in the EN-1991-1-4 may or may not be used in Sweden. The mentioned standard code has simplied the complex nature of the wind load and its corresponding eects on the bridge structures by suggesting some simplied methods to model the wind phenomena and also recommends some simplied methods to determine the static and dynamic response of structures. However, great attention must be paid while using the mentioned guidance as one may require knowledge of the background and logics applied behind the given simplications and the corresponding assumptions and limitations to assure that the results represent the actual situations in the eld. The limitations behind the applications of the EN-1991-1-4, Eurocode1, actions on structures-general actions-wind load-part 1-4, lead the structural designers to a great confusion. This may be due to the fact that, EC1 provides only the guidance for the bridges whose fundamental mode of vibrations have constant sign (e.g. simply supported structures) or a simple linear sign (e.g. cantilever structures) and these modes are the governing mode of vibrations of the structure; it analyzes only the along-wind response of the structure and not the cross wind response and the
5
Introduction
simplied methods recommended in this code are covering only the structures with simple geometrical congurations. In this report, the analytical methods which are used to describe the uctuating wind behavior and predict the relative static and dynamic response of the structures along with the eect of serviceability criteria which inuence the performance, are studied and presented in the following chapters. Then based on the given methods the wind forces acting on a continuous bridge whose main span is larger than the 50 meters (i.e. > 50 meter requires dynamic assessment) is studied and compared with the results which could be obtained from the simplied methods recommended in the EC1. In chapter 1, the wind characteristics, wind phenomena, basic wind velocity, mean wind velocity, turbulence, methods to determine the corresponding wind spectral density functions and static wind loads are described. In chapter 2 the corresponding dynamic response of the structure against the uctuating wind load are discussed. The vortex shedding and aerodynamic instabilities are also described in this chapter. Chapter 3 is based on the EN-1991-1-4 and SS-EN-1991-1-4(2005) specications and how the Eurocode 1 deals with the wind actions on bridge structures. Chapter 4 describes the nite element analysis of the given continuous bridge and modeling of the bridge structure in order to determine its relevant natural frequencies and carry out a dynamic assessment of the bridge. Chapter 5 represents the analytical calculation of the wind loads acting on the bridge structure based on the theoretical methods and the nal conclusion is given in the chapter 6. The corresponding tables of mass participation factors, sum mass participations obtained from LUSAS and MATLAB code are given in appendix A and B respectively.
Chapter 2
Wind load
2.1 Wind load chain
Wind load and the wind response of the structure relation is illustrated in the form of a chain by A.G. Daveport, which draws the attentions towards the signicant role of each and every factor in the given chain while designing a safe and stable structure against wind load which is shown below.
Figure 2.1: Wind load chain, suggested by A.G. Daveport
Davenports approach describes that the wind loading on the structures is determined by a combined eect of the wind climate which needs to be calculated statistically; the local wind exposure which depends on the terrain roughness and topography; aerodynamic characteristics of the structures which depend on the shape of the structure; dynamic eect i.e. the wind load magnitude (potential) increases due to the wind-induced resonant vibrations. Sti structures may vibrate in dierent ways when subjected to wind loading. e.g. along-wind vibrations called bueting may occur with the turbulence[15]. Slender structures are especially susceptible to cross-wind vibrations caused by vortex shedding, and within certain ranges of wind velocities, wind load perpendicular to the wind direction may be in resonance with the structure. Cable supported bridges and some other structures may vibrate when vertical and torsional movements are coupled. This phenomenon, known as classical utter, occurs only at high wind velocities. However, bridges where utter is likely to occur must be studied in wind-tunnel experiments, as utter can cause the structure to collapse completely. And nally the clear criteria needs to be established to judge acceptability of the predicted wind load and the corresponding responses. This includes the eect of the wind on the entire structure, each component, exterior envelope and various serviceability considerations which inuence the performance
7
Wind load
and which determine the habitability [9].
2.2
The atmospheric boundary layer
The wind velocity and its corresponding direction near the ground surface changes with respect to the variation of height. When wind is approaching the oshore its velocity reduces close to the ground as the ground surface tends to reduce the wind speed and this eect is minimized as the height of the wind increases from the ground surface. This eect exists up to a height of 1000 meters which is known as height of geostrophic wind above the atmospheric layer [7]. 2.2.1 The roughness length
The roughness length z0 can be interpreted as the size of a characteristic vortex, which is formed as a result of friction between the air and the ground surface. Therefore, z0 is the height above the ground at which the mean velocity is zero.
Figure 2.2: Roughness length z0
2.3
Mean wind velocity- wind prole(homogenous terrain)
The wind velocity increases with the height above the terrain and this variation of the wind velocity is known as wind prole. The variation of the wind load is determined with a logarithmic prole which is discussed below. There are two characteristic lengths to be considered in the boundary layer. In the higher part of the boundary layer close to the free wind ow, the boundary layer height is an important factor where in the lower part of the boundary layer the dominant length scale is a measure of surface roughness. Thus, in the logarithmic prole the surface roughness is taken into account which is valid only up to a height of 50-100 meter above the terrain and in corrected logarithmic prole the length of boundary layer is taken into account which is suitable for the high velocities and it is valid up to a height of 300 meters. Notice that the above logarithmic proles are only used to determine the variation of the mean wind velocity.
Mean wind velocity- wind prole(homogenous terrain)
2.3.1
The logarithmic prole
The friction velocity u is dened by the following formula: u = (2.1)
Where is the shear stress at the ground surface and is the air density. Close to the ground, the velocity gradient dU (z )/dz depends upon and and the height z above the ground. Based upon a dimensional analysis, a dierential equation for the mean wind velocities can be formulated and if there is a long , at terrain upstream, its solutions lead to the following expression for the logarithmic prole. U (z ) = z u . ln z0 (2.2)
Where is von Karamans constant ( 0.4) and z0 is called the roughness length. Eurocode 1 uses the logarithmic prole for the mean wind velocity up to 200 meter above the ground level. The corresponding value of the z0 is given by the following table 4.1 in EC1. 2.3.2 Corrected logarithmic prole
The expression (2.2) for logarithmic prole is not valid at very high altitude above ground. Harris and Deaves (1980) have suggested the following formula.
U (z ) =
u d .[ln zz 0
+ 5.75a 1.88a2 1.33a3 + 0.25a4 ]
Where the actual, eective height, z d, is normalized by the gradient height zg when calculating the non-dimensional argument a. a= The gradient height zg is given by : zg = u 6.fc (2.4) zd zg (2.3)
Where fc is the coriolis parameter and given as follows: fc = 2.. sin (2.5)
Where is the angular velocity of the earth (2./24hours = 7.27.105 rad/sec) and is the latitude.
10
Wind load
2.3.3
Power law prole
There is another empirical formula which is used in Canadian code NBC 1990, which is expressed as: U (z ) = U (zref ).( z zref ) (2.6)
Where, zref is the reference height in m, is given in tabular form in the code which depends on the terrain category.
2.4
Wind turbulence
The wind in the boundary layer is naturally turbulent i.e. the ow varies randomly from interval of a second to several minutes. The statistical methods are used to describe the turbulent ow. A Cartesian coordinate system is applied, with the x-axis in the direction of the mean wind velocity (along wind direction), the y-axis horizontal (cross wind direction) and the z axis vertical , positive upwards. The velocities at a given time t are formulated as: Longitudinal direction : U (z ) + u(x, y, z, t) Lateral direction : v (x, y, z, t) V ertical direction : w(x, y, z, t) (2.7) (2.8) (2.9)
Where the U(z) is the mean velocity and depends only on the height above the ground . u, v, w describe the uctuating part of the wind eld, and can be treated mathematically as stationary, stochastic processes with a zero mean value. A stochastic process is referred to a phenomenon if the phenomenon has been measured/recorded on many occasions during suciently long interval of time, then certain statistical properties of the phenomenon can be deduced. Statistical properties may also be based on mathematical modeling of the phenomenon. The phenomenon itself is then described as a stochastic process, and any measured sample during a time period is called a realization of the stochastic process. Expected values of the process itself, or combinations of the process at dierent times or positions, can be derived from the measurements or mathematical modeling. If theses expected values are time independent, and if the correlation between values at dierent times only depend on time dierences, then the process is called stationary. The gure below illustrates the variation of the mean wind velocity with respect to the height above the ground level and the time along with the turbulence component u(z,t). 2.4.1 Standard deviation of the turbulence components
The standard deviations of the turbulence components in the wind direction u, in horizontal v , and in vertical direction w, up to a height of 100-200m above homogeneous terrain are approximately
Wind turbulence
11
Figure 2.3: Simultaneous wind velocities in the wind direction at dierent heights above the ground (left) and time(right).
u = A.u v 0, 75.u w 0, 5.u
(2.10) (2.11) (2.12)
Where the constant A 2.5if z0 = 0.05m and A 1.8if z0 = 0.3m The turbulence intensity Iu (z ) for the long-wind turbulence component u at height z is dened as: Iu (z ) = u (z ) U (z ) (2.13)
Where u (z ) is the standard deviation of the turbulence component u and U (z ) is mean vind velocity, both at height z . For at terrain, the turbulence intensity is approximately given by: Iu (z ) = 1 ln zz0 (2.14)
Where z0 is the roughness length and u /u is assumed to be 2.5 . Up to 100-200 m above the ground, it is usually reasonable to assume that the turbulence components are distributed normally with a zero mean value and standard deviations as given above. However this does not hold for the tails of the distribution, i.e. when the turbulence components are outside a range of 3 standard deviations. In this case the assumption of normal distribution may lead to signicant errors. 2.4.2 Time scales and integral length scales
Here in this section two correlation functions are introduced. The autocorrelation functions T u (z, ) which is dened as the normalized mean value of the product of the turbulence component u at the time t and u at the time t + ,
12
Wind load
2 T u (z, ) = Eu(x, y, z, t).u(x, y, z, t + )/u (z )
(2.15)
The function indicates that how much information a measurement of the turbulence component u(x, y, z, t) in the mean wind direction will provide about the value of u(x, y, z, t + ) measured at a time later at the same place [9]. The autocorrelation function depends only on height z above ground and on time dierence due to the assumption of the horizontal homogeneous ow. u may be said to have a characteristic time of memory, the so called time scale T (z ). In other words, time scale represents how long the turbulence is being measured i.e. the total time period and is the time segments inside the interval of time scale. One of the important measurements of u taken at time t give a great deal of information about u at the time later if T (z ) , but only little information, if T (z ). the formal denition of time scale T (z ) is
T (z ) =
0
T u (z, ) d ) T (z )
(2.16)
T u (z, ) = exp(
(2.17)
Lx u =
0
u (z, rx )drx
(2.18)
Integral length scale is a measure of the sizes of the vortices in the wind, or in other words the average size of a gust in a given direction. According to Taylors hypothesis, u (z, rx ) = T u (z, )f orrx = U (z ). indicating that the longitudinal integral length scale is equal to the time scale multiplied by the mean velocity, Lx u (z ) = U (z )T (z ). Full scale measurements are used to estimate integral length scales. However results show extensive scatter originating mainly from the variability of the length and degree of stationary of the records being analyzed. The integral length scales depend upon the height z above ground and on the roughness of the terrain, i.e. roughness length z0. The wind velocity may also inuence the integral length scales at site. Counihan(1975), has suggested the following purely empirical expression for the longitudinal integral length at height z in the range of 10-240m.
m Lx u = Cz
(2.19)
Where C and m depend on roughness length z0 which can be determined graphically by referring to the Counihan(1975).
x Ly u 0.3Lu x Lz u 0.2Lu
(2.20) (2.21)
Wind turbulence
13
2.4.3
Power-spectral density function
Power spectral density function is a dimensionless function which describes the frequency in a distributed form for the turbulent along-wind velocity component, u. There are dierent suggestions to determine these functions. The most common and frequently used power spectral density functions are discussed here in this section. The frequency distribution of the turbulent along-wind velocity component u is described by the non-dimensional power spectra density function Rn (z, n) dened as:
2 (z ) RN (z, n) = nSu (z, n)/u
(2.22)
Where n is the frequency in Hertz and Su (z, n) is the power spectrum for the along-wind turbulence component. Turbulent energy is generated in large eddies (low frequencies) and dissipated in small eddies (high frequencies). In the intermediate region, called the inertial sub-range the turbulent energy production is balanced by turbulent energy dissipation, and the turbulent energy spectrum is independent of the specic mechanisms of generation and dissipation. For most of the structures except exible oshore structures (as these structures have very low frequencies), the spectral values for frequencies within this range (inertial sub-range) are the most important. Based on Tylors hypothesis frozen turbulence and considering the frequencies in the inertial sub-range, the non-dimensional power spectrum function RN is given by: RN (z, n) = A.fL Lx u (z )
2/3
(2.23)
Where A is a constant depending slightly on height and fL = nLx u (z )/U (z ) and is a height dependent length scale of turbulence. The constant A should be obtained based on full-scale spectral density functions measured at dierent height, preferably using the integral length scale in the high-frequency calculated by equation above i.e. L(z ) = Lx u (z ). According to ESDU 85020, a function decreases with increasing height and for a structure up to a height of 200-300 m, spectral functions are obtained within accuracy 5% accuracy using A = 0, 14 for all heights assuming L(z ) = Lx u (z ). Dierent suggestions or methods to determine the power spectra density functions in the literatures are Von Karman, Harris, Davenport and Kaimal.
Kaimal spectra density function for longitudinal turbulence component
Kaimal el al. (1972) [17] suggests the following spectral density function which is commonly used: Rn (z, n) = 2.3fz /(1 + fz )5/3 (2.24)
Where, = 50, the non-dimensional parameter used to locate the maximum value of the spectral density obtained for fz = f( z, max) = 3/2. The integral length scale Lx u obtained using the spectral density function is equal to:
14
Wind load
Lx u (z ) = U (z )T (z ) =
U (z )Su (z, 0) = z/6 2 (z ) 4u
(2.25)
The above equation is appropriate for the height higher than 50 meters, for structures lower than this height instead of fz in the spectral density function the following equation needs to be used: fL = nLx u (z )/U (z ) Therefore we obtain the following expression which is used in Eurocode1: Rn (z, n) = 6.8fL /(1 + 10.2fL )5/3 (2.27) (2.26)
Where, fL is calculated from expression (2.25). The spectral density function may be used for the structures whose fundamental frequency of vibration is higher than the lower end of the inertial sub-range. It gives an accurate representation of the turbulent uctuations in the frequency range of interest for most structures.
Von Karman spectra density function for longitudinal turbulence component
Von Karman (1948) [36] has suggested the following expression for the the spectral density function:
2 5/6 Rn (z, n) = 4fL /(1 + 70.8fL )
(2.28)
The above expression is used by the Swedish annex.

Davenport (1967) spectra density function for longitudinal turbulence component
2 2 4/3 Rn (z, n) = 2fL /3(1 + fL )
(2.29)
The only dierence in this expression is that the function is based on the fL = nL/U (z ), where and L 1200m [6].
Harris (1970) spectra density function for longitudinal turbulence component
2 2 5/6 Rn (z, n) = 2fL /3(2 + fL )
(2.30)
With non-dimensional frequency fL = nL/U (z ), where and L 1800m [14]. A comparison is made between the suggested functions which is shown in Figure 2.4. The fuctions are based on the integral length scale of Lx u (z ) = 180m. It can be observed that Davenport gives the highest power spectra and Von Karman gives higher power spectral than the Kaimal( Eurocode1) but after frequency of 1 it is vice versa. The Swedish annex suggests the value of integral length Lx u (z ) = 150m. Therefore, corresponding spectral power density function is shown in Figure 2.5. It can be observed that Von Karman gives a higher power spectra than the Kaimal spectra function but for frequencies higher than 1 the power spectra obtained by Kaimal(Eurocode1) is higher.
Wind turbulence
15
Figure 2.4: Power-spectral density functions for longitudinal turbulence component for Lx u (z ) = 180m.
Figure 2.5: Power-spectral density functions for longitudinal turbulence component for Lx u (z ) = 150m.
16
Wind load
Power spectra of lateral and vertical turbulence components are approximately given by [28]. nSv (z, n) 15fz = 2 u (1 + 9.5fz )5/3 nSw (z, n) 3.36fz = 2 u (1 + 10fz )5/3 2.4.4 correlation between turbulence at two points (2.31)
(2.32)
Davenport [8] suggested on the emperical basis the relaions between two turbulence point having distances ry and rz in horizontal and vertical dimensions (horizontal structures) the following expression: n (Cy ry )2 + (Cz rz )2 ) (2.33) U Where Cy and Cz are the decay constants and they are determined using full scale measurements. u (ry , rz , n) = exp(
2.5
Static wind load
In most of the structures the wind-induced resonant vibrations may be negligible and the wind responses can be determined using the procedures applicable for static loads. The wind load calculation is performed using the probabilistic methods and stochastic process. This means that the wind has a mean wind value and a standard deviation. The peak factor kp is used to include the eect of the highest mean wind load taking place within the measured period of time (i.e. 10 min.). Turbulence gives a uctuating contribution to the wind load which depends on structural geometry and other parameters. Therefore wind always uctuates when acting on structure or the structural components. The characteristic wind load is the wind load with a mean wind value of mux or Fm ax and standard deviation of F obtained within a particular time i.e. 10 minutes which is shown in the equation 2.34. The response of the structure to the characteristic wind load is also expressed as the characteristic response of the structure having mean response value and standard deviation. The static load that generates a characteristic response on the structure or structural components due to the actual uctuating wind load is usually known as equivalent static load. Fmax = Fq + kp F (2.34)
The wind uctuation is proportional to the twice of the turbulence intensity therefore the standard deviation is given by: F = Fq 2Iu kb (2.35)
Static wind load
17
The gust factor is the ratio of the characteristic wind load and the corresponding mean wind load. = Fmax = 1 + kp 2Iu Fq kb (2.36)
The background turbulence factor kb is an integral measure of the load reduction due to the lack of pressure correlation over the surface of the structure. The nonsimultaneous action of wind gust over the structure causes reduction of maximum instantaneous pressure averaged over the surface of the structure. Benjamin Backer 1884 discovered that the wind load acting over the smaller plates experience higher loads proportional to its size than the larger plates and correctly attributed the eect of the size of the wind gust relative to the plates. The concept of the equivalent static gust is commonly used in the codes which is based on the ltering of the time series of either the uctuating wind velocity pressure in the undisturbed wind or of the surface pressure measured at a point by running averaging to remove high-frequency uctuations lasting for periods larger than 5 to 15 seconds. The cut-o frequency is chosen based on the size of the structural area [9]. 2.5.1 Total wind load on structure- Davenports model
Davenport (1962) has developed a method to convert the wind ow with its uctuating nature into the wind load acting on structure. The method is described below [5].
Wind load on small structure
The wind load on small structure or point like structure is calculated by assuming that the size of the wind gusts are smaller than the size of structure or the size of the structural components. Therefore, the eect of the reduced wind load due to the lack of pressure correlation on the surface is negligible and hence the value of kb is taken as 1. Therefore, the gust factor is given by [6]: = Fmax = 1 + kp 2Iu Fq (2.37)
The total static wind load is obtained by equation 2.36. Ftot = Fq + Ft 1 Fq = CA AU 2 2 Ft = CA AU u (2.38) (2.39) (2.40)
Ftot is the total wind load which is determined by adding the mean wind load, Fq and the uctuating wind load, Ft with a mean of zero. The power spectrum of the uctuating wind load Ft is given by the following equation.
18
Wind load
4Fq2 SF (n) = (CA AU ) Su (n) = 2 Su (n) (2.41) U The variance of the wind uctuations is given by integrating the power spectrum SF (n) over all frequencies n [9]:
2 2 = F 0
4Fq2 4Fq2 2 S ( n ) dn = u U2 U2 u
(2.42)
Wind load on large structures
In the structures with considerably large size, the spatial pressure correlation over the surface of the structure must be taken into account. This can be done using aerodynamic admittance function 2 (nl/U ) which argues the ratio between the length l of the structure and the characteristic eddy size of natural wind U/l for line like structures and for structures with rectangular area the ratio between the lengths l1 and l2 and the characteristic eddy size of the natural wind U/l given the aerodynamic admittance function 2 ((nl1 )/U, (nl2 )/U ) [9].
The variance of the uctuation wind load is given by:

2 F
=
0
4Fq2 2 nl1 nl2 4Fq2 2 ( , )Su (n)dn = 2 u U2 U U U
(2.43)
Where the aerodynamic admittance function is equal to: 2 ( 1 nl1 )= U l

l 0
r 2(1 )p (r, n, U )dr l
(2.44)
And for the rectangular area the admittance function is equal to: Where the aerodynamic admittance function is equal to: nl1 nl2 1 ( , )= U U l1 l2
2 l1 0 0 l2
4(1
r1 r2 )(1 )p (r1 , r2 , n, U )dr1 dr2 l1 l2
(2.45)
And the gust factor is given by: = Fmax = 1 + kp 2Iu Fq kb (2.46)
The aerodynamic admittance function value is less than or equal to 1, therefore the corresponding value of kb is also less than or equal to 1. The size factor cs is expressed as the ratio of the gust factor corresponding to large structure and that of small structures. L 1 + kp 2Iu kb cs = (2.47) s 2I 1 + kp u
Static wind load
19
Determination of aerodynamic admittance functions for line like structures
In order to determine the admittance function for line like structures the Davenports model can be used. Extreme wind responses such as bending moment, stresses and deection are estimated based on a statistical description of the uctuating load on the structure. The normalized co-spectrum data described below are used as an input when calculating wind responses on line-like structures. The equivalent static gust which is dened as the shortest-duration, hence smallest, gust which fully loads the structure or structural components is used to determine the extreme wind responses based on the dynamic admittance function[28]. The basic idea is to estimate the extreme wind load on the basis of air turbulence measured only at one point. The spatial distribution of the load is taken into account by time averaging the air turbulence measured. The load on large structures corresponds to long averaging times, where short averaging times are used for small structures.
The normalized co-spectrum of the surface pressure could be described using an exponential decay function [9]: nr ) (2.48) U Where Cr is decay constant and is equal to Cr = 8 in Sweden, n is the structural fundamental frequency in Hertz, r is the distance between two points and U is the mean wind velocity. The response of the structure to the wind load is obtained by the summation of surface pressures multiplied by the response-inuence functions. The response can be bending moment or deection of the structure. These response inuence function should be incorporated into the aerodynamic admittance function that corresponds to the response in question. p (r, n, U ) = exp (Cr
l
R(t) =
0
IR (z )F (z, t)dz
(2.49)
Where R(t) is the response of the structure such as bending moment or deection. IR (z ) is the response-inuence function of the point specied by the coordinate of z and F (z, t) is the wind load at location z and time t. The corresponding admittance function is given by the following expression: 2 () =
1 l l k (r)p (r, n, U )dr 0 l |I (z )|dz )2 (1 l 0 R
(2.50)
Where the non-dimensional parameter is = Cr nr/U . The absolute value of the response inuence function in the denominator of equation 2.48 provides the facility of a normalization that is valid for IR with constant sign as well as response inuence functions with changing sign. The normalized co-inuence function k (r) is obtained by k (r) = 2 l
lr
IR (z )IR (z + r)dz
0
(2.51)
20
Wind load
Rectangular area
The wind load response for only rectangular area is obtained using the following expression [9]:
l1 l2
R(t) =
0 0
IR (z1 , z2 )F (z1 , z2 , t)dz2 dz1
(2.52)
Where IR (z1 , z2 ) is the response-inuence function and F (z1 , z2 , t) is the wind load at the point (z1 , z2 ). The aerodynamic admittance function for rectangular area is calculated as: (1 , 2 ) =
2 1 l1 l2 l1 0 l2 0
kr (r1 , r2 )p (r1 , r2 , n, U )dr2 dr1

l1 0 l2 0
( l11l2
|IR (z1 , z2 )|dz2 dz1 )2
(2.53)
Where 1 = Cp (nl1 )/U and 2 = Cp (nl2 )/U and the normalized co-inuence function is obtained by: k (r1 , r2 ) = Where, IR (z1 , z2 , r1 , r2 ) = IR (z1 , z2 )IR (z1 + r1 , z2 + r2 ) + IR (z1 , z2 + r2 )IR (z1 + r1 , z2 ) (2.55) Structures with constant sign response-inuence functions, give an aerodynamic admittance function which is equal to 1 for full pressure correlation occurring at zero frequency in accordance with the exponential decay function in equation(2.17). Therefore, for the structures with a response-inuence function of constant sign, the aerodynamic admittance function can be approximated using the following expression: 2 (1 , 2 ) = 1+ (G1 1 )2 1 + (G2 2 )2 +
2 ( G1 1 G2 2 )2
2 l1 l2
l1 r1 0 0
l2 r2
|IR (z1 , z2 , r1 , r2 )|dz2 dz1
(2.54)
(2.56)
Where 1 = Cr (nl1 )/U and 2 = Cr (nl2 )/U .
Chapter 3
Dynamic response of structures to the wind load

3.1 Along-Wind Response
There are structures which are sensitive to the uctuating wind load. This means that the uctuating wind load causes the structure to vibrate. Hence the response of the structure must be taken into account while calculating the wind load. For such structures the along-wind load can be calculated with reasonable accuracy by considering the structure to have a single degree of freedom. For such structures the along-wind component is taken into account as the other two components are not of great importance. 3.1.1 Single degree of freedom
Here the analysis is performed for both point-like structure and large structures. As it has already been explained that the lack of pressure correlation on the surface of small structures is not signicant. Therefore, the value of kb is taken as 1. But in large structures the eect of the reduced maximum wind load due to lack of pressure correlation will contribute to the factor kb .
Wind load on point-like structures
The structure is assumed to have a mass of m which is modeled by a spring having stiness k connected parallel with a viscous damper of a damping coecient of cs . Therefore, the simple dynamic equation is as follows: def + cs def + kdef = Ftot m (3.1)
Where is deection and Ftot is the total along-wind force and calculated by the following expression: 1 def )2 Ftot = CD A(U + u 2
21
(3.2)
22
It can be seen that for calculating the wind load the eect of speed of structure against the wind load is taken into account. This is very important as it gives rise to the aerodynamic damping which is often of the same order of magnitude as the structural damping. As a matter of fact the value of mean wind velocity is higher than the along-wind component which leads to the following expression. def )2 = U 2 + 2U u 2U def (U + u (3.3)
Hence, the total wind load is a summation of mean wind load, uctuating wind load and the aerodynamic damping load as below: Ftot = Fq + Ft Fa 1 Fq = CD AU 2 2 Ft = CD AU u def = ca def Fa = CD AU ca = CD AU (3.4) (3.5) (3.6) (3.7) (3.8)
The total damping coecient is given by a summation of structural damping and aerodynamic damping coecients. c = ca + cs
Mean deection
(3.9)
The response of structure to the characteristic wind load has also a mean value and standard deviation. The mean response of the structure is obtained using the following formula: =
Structural vibrations
Fq k
(3.10)
The standard deviation of the response is obtained by the following set of formula: the auto-spectrum S (n) of deection is given by [9]: S (n) = |H (n)|2 SF (n) (3.11)
Where, H (n) is the frequency response function for the structure and SF (n) is the 2 auto-spectrum for load. The variance of deection is obtained by integrating the auto-spectrum S (n) of deection from zero to innity which leads to the following expression:
2 = 0
S (n)dn =
2 4Fq2 u k2 U 2
k 2 |H (n)|2
0
Su (n) dn 2 u
(3.12)
Along-Wind Response
23
As it is already mentioned the structures with constant sign response inuence function give a value of kb = 1 at frequency zero.
kb =
0
k 2 |H (n = 0)|2 k 2 |H (n)|2
Su (n) dn = 1 2 u
(3.13) (3.14)
kr =
0
Su (n) ne Su (ne ) dn = 2 2 u u 4
Therefore, it is a good approximation to calculate the integral as the sum of kb + kr . = 2I And the damping ratio is given by: ca + cs = 2 mk
Wind load on large structures
kb + kr
(3.15)
(3.16)
For a large structure the eect of lack of pressure correlation or the reduced spatial correlation plays a signicant role. The eect of this can be considered by using the aerodynamic admittance function as shown below:
kb =
0
2 (
nl Su (n) ) 2 dn U u
(3.17) (3.18)
kr = 2 (
Gust response factor
ne l Su (ne ) ) 2 U u 4
The maximum response of the structure is obtained as follow: max = + kp = max = 1 + kp 2Iu kb + kr (3.19) (3.20)
The peak factor kp is the ratio of the expected maximum uctuating part of response and standard deviation of response and it is determined as follows [24]: kp = 2 ln (vT ) + 0.577 2 ln (vT ) (3.21)
v=
2 n2 0 kb + ne kr kb + kr
(3.22)
24
Where ne is the resonant frequency (Hz) of the structure for the along-wind vibration of the structure and n0 is the representative frequency (Hz) of the gust loading on rigid structures. n0 is determined as follows: n0 = 3.1.2
2 2 n n ( U )Su (n)dn 0 2 n ( U )Su (n)dn 0
(3.23)
The Along-wind response of blu bodies
In this section the dynamic response of the structure subjected to the along-wind load is calculated. The methods originally presented by A. G. Davenport (1960s) and then developed by Hansen and Krenk (1996) [13,25]. The procedure to estimate the dynamic response of line-like structures subjected to along-wind load is presented. The procedure for plate-like structures is usually used when the width of the structure is of the same order of magnitude with the characteristic eddy size (U/n) [9].The following assumptions are considered: The shape of the structure is simple. The wind load is determined from the undisturbed wind eld. The structure is assumed to be linear-elastic with viscous damping. The along-wind mode considered is uncoupled from other modes. The calculation presented here is not covering the modal coupling. However, for structures with more than one mode contributing to the resonant response, the following calculation can be used to obtain each single mode response r,i . The total response is given by:
2 R = i 2 r,i
(3.24)
There are two frequency functions which are used to describe the dynamic response of the structures. The joint acceptance function and the size reduction function. The joint acceptance function is used to describe the interaction between the mode shape of the structure and the uctuating wind load on the structure. For a structure with a constant sign mode shape the size reduction function is equal to the joint acceptance function normalized to 1 at zero frequency. Therefore, the size reduction function describes the response reduction from the interaction between mode shape and lack of load correlation over the structure as a function of frequency [27,28].
The calculation of joint acceptance function for line-like structures requires the calculation of a double-integral for line-like structures and a four-folded integral for plate-like structures.
Along-Wind Response
25
Extreme structural response
The characteristic response of the structure is expressed in the terms of mean response R , the peak factor kp and standard deviation of structural response R . Rmax = R + kp R (3.25)
The standard deviation of the response of the structure is obtained as follows: R =

2 2 + r b
(3.26)
Where, b is the standard deviation which originates from the background turbulence and r originates from the resonant turbulence which are given by the following expressions: b = R 2Iu,ref b r = R 2Iu,ref r kb kr (3.27) (3.28)
The gust factor is obtained from the following expression: = 1 + kp 2Iu

2 2k b kb + r r
(3.29)
Where, b and r incorporates the eect of dierent inuence functions for the mean and uctuating response.
Response of line-like structures
For line-like structures the total wind load per unit length is given by the following expression: 1 def (z, t))2 d(z )C (z ) F (z, t) = (U (z ) + u(z, t) 2 The wind load is given by the following expressions: F (z, t) = Fq (z ) + Ft (z, t) Fa (z, t) 1 Fq (z ) = U (z )2 d(z )C (z ) 2 Ft (z, t) = U (z )u(z, t)d(z )C (z ) def (z, t)d(z )C (z ) Fa = U (z ) (3.30)
(3.31) (3.32) (3.33) (3.34)
The aerodynamic damping load is taken into account using a logarithmic decrement describing the total damping expressed as: = a + s (3.35)
Where, s is the logarithmic decrement of the structural damping. The aerodynamic damping may be calculated by the following expression:
26
1 Ured a = Cref a 2 Mred Ured = Mred =

h
(3.36) (3.37) (3.38)
Uref ne dred mg /h d2 ref 2 (z ) dz 2 ref
mg =
0
m(z )
(3.39)
a =
1 h
h 0
C (z ) d(z ) U (z ) 2 (z ) dz 2 Cref dref Uref ref
(3.40)
Where, Cref is the reference shape factor, Ured is non-dimensional reduced wind velocity,Mred is a non-dimensional mass ratio, m is the mass per unit length, mg is the normalized, generalized mass of the mode considered and a is a factor that accounts for the actual distribution of shape factor, wind velocity and mode shape along the structure.
Mean response
The mean response originates from the mean wind acting over the structure and is obtained by the multiplication of response-inuence function and the applied wind load.
h
R =
0
Fq (z )IR (z )dz
(3.41)
The mean wind response is given by the following expression: 1 2 R = hdref Cref Uref IR,ref m (3.42) 2 Where, h is the length of the structure, dref is the width of the structure perpendicular to the direction of the wind load, Cref is the reference shape factor, IR,ref is the response-inuence function at the reference point and m gives the integral eect of the function gm . m = 1 h
h
gm (z )dz
0
(3.43)
gm ( z ) =
C (z ) d(z ) U (z )2 IR (z ) 2 Cref dref Uref IR,ref (z )
(3.44)
gm is the non-dimensional function describing the variation of the mean wind load and response-inuence function along the structure.
Along-Wind Response
27
Background turbulence response
The background response is obtained by multiplying the uctuating wind load by the resonance-inuence function and integrating throughout the length of the structure.
h
Rb (t) =
0
Ft (z, t)IR (z )dz
(3.45)
The variance of the background turbulence function is obtained from the following expression:
2 2 2 2 b = (hdref Cref Uref u,ref IR,ref Jb
(3.46)
The non-dimensional response variation is determined by double integral as shown below:

2 Jb
1 = 2 h
h 0 0
u (rz )gb (z1 )gb (z2 )dz1 dz2
(3.47)
The gb (z ) is a non-dimensional function describing the background turbulent wind load variation along the length of the structure and is obtained as follows: gb (z ) = C (z ) d(z ) U (z ) (z ) IR (z ) Cref dref Uref u,ref IR,ref (z ) (3.48)
The gm and gb functions with constant sign are given in Table 3.1 for dierent mode shapes.
2 . and G Table 3.1: Asymptotic behaviour of the non-dimensional response variance Jb b values used to calculate joint acceptance function 2 Asymptote 2 Asymptote Load variation Jb Jb b fungtion g for z 0 for z r G 1 1 2/z 1 1/2 z/h 1/4 2/(3z ) 1/2 3/8 (z/h)2 1/9 2/(5z ) 1/3 5/18 (z/h)3 1/16 2/(7z ) 1/4 7/32 4 (z/h) 1/25 2/(9z ) 1/5 9/50 sin(z/h) 4/ 2 1/z 2/ 4/ 2 2z/h 1 z /15 2/(3z ) 0 -
Resonance turbulence response
The structural response of the dynamic part of the along wind loading may be calculated using modal analysis. The response to gusty wind is usually dominated by the fundamental mode.
h
Q(t) =
0
C (z )U (z )u(z, t)d(z )dz
(3.49)
28
Where, Q(t) is the corresponding generalized uctuating load. The dynamic part of the structural deection may, as an approximation, be written as (z )a(t). Where, a(t) is stochastic amplitude function. The spectral density function of a(t) is given by:
2 Sa (n) = hdref Cref Uref ref |H (n)|2 |Jz (n)|2 Su,ref (n)
(3.50)
The joint acceptance function is obtained by a double-integration given below: 1 |Jz (n)| = 2 h
2 h 0 0 h
gr (z1 , n)gr (z2 , n)F (rz , n, U )dz1 dz2 C (z ) d(z ) U (z ) Cref dref Uref Su (z ) (z ) Su,ref (n) ref
(3.51)
gr (z1 , n) =
(3.52)
Where, gr (z1 , n) is a non-dimensional function describing the resonant wind load variation along the structural length. F (rz , n, U ) is the normalized co-spectrum for the wind load components at two points having a distance rz .H (n)2 is the structural 2 frequency response. The variance of acceleration acc at reference height is given by multiplying r ef 2 (2ne )2 to the integral from zero to innity of the spectral density functions. This is due to the fact that, the inertial force is proportional to acceleration.
2 (2Iu,ref )2 2 2 ( hd C U ) RN (zref , ne )|Jz (ne )|2 (3.53) ref ref ref 2 mg 2
2 2 2 acc = ref (2ne )2 a =
The resonance response such as bending moment and deection in the structure are obtained by multiplying the inertia force to the response-inuence function as shown below:
h
Rr (t) =
0
FI (z, t)IR (z )dz
(3.54) (3.55)
FI (z, t) = m(z )(2ne )2 (z )a(t) 3.1.3 Design procedure for mode shapes with constant sign
The mean wind velocity is obtained using the following expression: z ) (3.56) z0 Where the integral length scale used in the design procedure is obtained as follows: U (z ) = Ubas kT ln( z 0 .3 ) ; 10m z 200m (3.57) z1 0 Where, z1 0 = 10m and L1 0 = 100m. The integral length scale for the other two x z x directions are obtained as follows: Ly u = 1/3Lu and Lu = 1/4Lu . Lx u = L1 0(
Along-Wind Response
29
Gust factor
= 1 + kp 2Iu,ref
2 2k b kb + r r
(3.58)
Where it is a good approximation for most of the structure to consider the b = 1 and r = 1.
Peak factor
kp =
2 ln (vT ) +
0.577 2 ln (vT )
(3.59)
v=
2 n2 0 kb + ne kr kb + kr
(3.60)
n0 =
2 n Ks (n)Su (n, zref )dn 0 Ks (n)Su (n, zref )dn 0
(3.61)
U (zref n0 = 0.3 hb
Background response
hb ; n0 ne Lu
(3.62)
The background turbulence factor for constant sign is approximated by the following expression: kb = 1+
Resonance response
3 2
1
2 ( Lbu
(3.63) +
3 b h 2 ( ) Lu Lu
h 2 (L ) u
The resonance response can be calculated by the following expression: kr = 2 RN (zref , ne )Ks (ne ) 2 (3.64)
Where, Ks (ne ) is given by equation (-). And the total logarithmic decrement of along wind vibrations: = a + s Where the aero-dynamic damping is obtained by: a = CU (zref ) 2ne (3.66) (3.65)
Where, is the mass per unit area of the structure.
30
3.2
Cross-wind vibtrations induced by vortex shedding
When a uid ows over a slender structure, alternative vortices are shed over its sides resulting in the generation of an inconsistent force due to low pressure regions being created in the direction normal to the ow of the uid. This systematic formation pattern of vortices is referred to as the von Karman vortex street. When the shedding frequency of the vortices are in resonance with one of the natural frequencies of the structure, large amplitude vibrations may be expected in a plane normal to the ow [1]. The phenomenon of vortex shedding is generally signicant for the lower natural frequencies of the structure, but for exible structures having a low damping ratio, this might occur at higher frequencies as well. The eect of vortex shedding is generally predominant for slender structures having an aspect ratio of 20 or more (i.e. a width to height ratio of 20). A bridge deck is generally not considered to be a slender structure but vortices will be shed by the ow of wind in the downwind side and large amplitude vibrations may result if the natural frequency of the bridge is in resonance with the shedding frequency. The gure below presented by I. Giosan [12] shows the alternating high and low pressure regions created by wind ow in the downwind direction. The blue and yellow colored vortices represent low pressure and high pressure regions respectively.
Figure 3.1: Vortex shedding phenomena by wind ow over a cylinder (I. Giosan)
For cylindrical cross-sections, the nature of vortex shedding induced depends on the Reynolds number, U.d (3.67) Where, U = wind speed [m/s], d = diameter of the structure [m], = kinematic viscosity [m2 /s]. Re = The shedding frequency of the vortices ns is represented by equation 3.68, U (3.68) d Where: St is the Strouhal number which depends on the wind turbulence, nature of surface roughness and the cross-sectional shape of the structure. Strouhal number is generally considered as 0.15 but for further details one can refer to Simiu and ns = St.
31
Scanlan, 1978. U is the wind speed [m/s], d is the characteristic width or diameter of the structure. The gure below demonstrates the lock-in phenomena for various wind velocities and was given by Simiu and Scanlan, 1986 [28].
Figure 3.2: Vortex shedding trend with velocity (Simiu and Scanlan, 1986)
The ratio between the inertial force and the friction force subjected to the uid is generally represented by the Reynolds number. When the Reynolds number is very low, the ow pattern can be considered to be laminar in nature as the inertia eects can be neglected. At very high Reynolds number, the regularity of the shedding vibrations decreases and is irregular in nature. The vortex shedding phenomenon generally occurs at steady wind ow conditions at a critical velocity. The periodic vibrations of the shed vortices may lock-in with the natural frequency of the structure causing high amplitude vibrations in the transversal plane to the wind ow. Vortex shedding generally does not occur for velocities less than 5 m/s. Vortex shedding generally takes place for steady wind ows with velocities in the range of 5 to 15 m/s. For turbulent wind ow caused due to velocities higher than 15 m/s, vortex shedding will not occur. The oscillations generated by vortex shedding can be quite severe to cause fatigue cracks in structures.
Sinusoidal method
The excitation and vibration caused due to vortex shedding is analyzed as a timedependent load of frequency. The phenomenon of vortex shedding is very complex in nature and as a result load induced is described by a probabilistic method [12]. The load is harmonic and sinusoidal in nature. The load induced per unit length of the structure at a location x may be determined as follows:
32
1 F (x, t) = ..U 2 .D.CL . sin (2..e .t) (3.69) 2 Where U is the mean wind speed [m/s], D is the diameter or width of the crosssection of the structure, S is the Strouhal number, CL is dened as the RMS lift co-ecient and is determined by a stochastic process, is the density of air.
The force to be applied on the structure is calculated by evaluating the maximum force that is caused due to each mode of vibration multiplied with the amplitude of the corresponding modal shape. The calculated force must be applied alternatively on the structure with the natural frequency i of the structure and the corresponding stresses are compared with the limiting values.
Band limited random forcing model
The sinusoidal method is considered to provide conservative results for vortex shedding analysis as it does not take into consideration the wind speed variation with height, turbulent nature of wind and other properties. The band limited random forcing model assumes that the force induced due to vortex shedding tends to behave harmonically only when the motion of the structure is considerably sucient to shed vortices (i.e. the amplitude of the vibrations are of the order of 2-2.5% of the width of the cross-section). In this method, the member is loaded with peak inertia loads which are considered to act statically on the structure and the resulting stresses are computed. The relation to compute the peak inertia load at any location is given as follows [12]: Fi (x) = (2i )2 yi (x)m(x) (3.70)
Where, Fi (x) is the peak inertia member load at any location x on the structure for the ith mode of vibration, [N/m] and m(x) is the mass per unit length at location x of the member, [kg/m] and yi (x) is the peak member displacement caused due to vortex shedding for the ith mode at a location x, [35] yi (x) = i i (x) (3.71)
Where, i is the modal coecient of the oscillatory displacement magnitude for the ith mode of vibration and i (x) is the mode shape amplitude for the ith mode at a location x. The modal coecient can be calculated for non-tapered sections with the procedure described below: i = L D2 0 .25C 3.5C B (4S )2 GMi (3.72)
In case yi (x) is greater than 0.025D, i should be evaluated as, L D3 H |i (x)|dx 2C 0 i = i (4S )2 GMi
(3.73)
33
where, C= (H/D)2 1 + H/2LD

H 0
x3 2 i (x) dx H 1+3
(3.74)
GMi is the generalized modal mass for the ith vibration mode, [kg]:
H
GMi =
0
m(x)2 i (x)dx
(3.75)
i is the critical damping ratio for the ith mode and is the wind velocity prole exponent. Vibrations produced due to vortex shedding may take place in slender structures such as cables, towers, chimneys and bridge decks. The risk of vortex shedding is enhanced if Slender structures are placed in a line and the separation distance between them is less than approximately 10-15 times the width of the structures. Vortices shed by an adjacent solid structure may aect a nearby slender structure. The vortex shedding response can be analyzed using the spectral model or the resonance model. The vortex shedding response analysis of the piers and the bridge deck is based on the vortex resonance model on which Eurocode 1 is based. The vortex resonance model seeks to include the large aero-elastic eects that occur with exible structures. The modal force Q(t) is analyzed as follows:
h
Q(t) =
0
F (z, t) (z )dz
(3.76)
The cross-wind load acting per unit height due to vortex shedding is calculated analytically as F (z, t) = q (z )d(z )cF (z ) sin (2ns t + (z ) ) (3.77)
Where: q (z ) is the velocity pressure, d(z ) is the width of the structure, cF (z ) is the non-dimensional shape factor, ns is the vortex shedding frequency, and is a factor correlating the load and deection direction. The maximum deection amplitude ymax is calculated as, ymax = Fe 2 (2ne ) me s (3.78)
Where, s is the aerodynamic logarithmic decrement of damping, me is the mass per unit length, and Fe is the equivalent load. Fe = max
h 0
q (z )d(z )cF (z ) (z )dz

h 0
2 (z )dz
(3.79)
34
When the vortex shedding load frequency ns is equal to the natural frequency ne of the structure, the maximum deection amplitude is given by ymax = max dref
h q (z ) d(z ) c (z ) (z )dz 0 qref dref F h 4 0 2 (z )dz
1 1 Sc St2
(3.80)
Where: Sc is the Scruton number, St is the Strouhal number. Simplifying the above equations, the following relation can be obtained, 1 1 ymax = K Kw clat dref Sc St2 (3.81)
Where: cl at is the standard deviation of the load and can be obtained from Table E.2 of EC1. K is the mode shape factor, and Kw is the eective correlation length factor. K = max
h (z )dz 0 h 2 (z )dz ) 0
(3.82)
Kw =
L e (z )dz 0 h (z )dz 0
(3.83)
The results obtained for the vortex shedding response for the piers and the bridge deck based on the above analytical method are shown here.
3.3
Bridge aerodynamics and wind-induced vibrations
When a slender structure is subjected to wind ow, forces in three directions may act on the structure, i.e. along the x, y and the z-axis. The three kinds of reactions induced by wind on the bridge deck are shown in Figure 3.3. The wind load acting on the structure is composed of the mean wind load and the uctuating parts (u(t) and w(t)) which vary with time. The force components are the lift force L, the drag force D and the moment generated M. When a slender structure obstructs the path of wind ow, wind circulates around the cross-section and this causes variation in pressure in the wake region of the cross-section due to the turbulent nature of the ow. Vortices may be created in the wake region which are carried forward in the downstream direction and this shedding of vortices cause the structure to vibrate with high amplitudes in a direction perpendicular to the ow of wind. These types of vibrations are known as cross-wind vibrations.
When the structure is not rigidly xed but has a particular stiness in the direction of the wind force, the structure will be subjected to an oscillation of a particular frequency which will be amplied if the vortex shedding frequency is close to the natural frequency of the structure causing resonance. This phenomenon can be prevented by increasing the damping or by stiening the structure.
35
Another type of aerodynamic instability is called galloping. Galloping causes slender structures such as cables to vibrate in the cross-wind directions with suciently high amplitudes which are larger than the cross-sectional dimension of the structure. Galloping is a very common phenomenon in cables and is catalyzed by the formation of ice around the cables. Galloping must be considered for the design of long-span suspension bridges but it is not much of a concern when it is analyzed for simple girder bridges.
Figure 3.3: Reactions induced by wind (Jain, Jones & Scanlan (1995))
The two phenomena mentioned above involve the separation of the wind ow across the cross-section of the structure causing an excitation which is periodic in nature. Another type of aerodynamic instability is known as utter and this phenomenon does not involve the separation of ow. Flutter is generally predominant in streamlined structures and it is a self-excited instability. Flutter can occur at various wind velocities above the critical velocity and the wind forces provide energy to the structure resulting in harmonic oscillations. Flutter can be considered as a case of negative aerodynamic damping for it occurs for a coupled motion in two degrees of freedom. The instability due to utter can be checked and suppressed by increasing the damping and stiness of the structure. Figure 3.4 shows the classication of the various wind induced vibrations and subdivides them into limited-amplitude and divergent-amplitude vibrations. The instability phenomena causing limited-amplitude vibrations do not generally cause structural failure. Instead they cause serviceable discomfort and structural fatigue. On the other hand, the instabilities causing divergent-amplitude vibrations can cause structural catastrophe and failure. Figure 3.5 demonstrates the relation between the resonance amplitudes and the wind velocities inducing these amplitudes for various instability phenomena. It is quite evident from the gure that vortex shedding is generally caused at lower velocities of wind and the maximum amplitude is reached at a resonance value after which the amplitude decreases with further increase in the wind velocity.
36
Figure 3.4: Classication of the wind induced vibrations and Bridge aerodynamics (T. H. Le, 2003)
The resonance amplitude for bueting is lower than that induced by vortex shedding and is caused at higher values of wind velocity. Flutter and galloping instabilities are caused at even higher wind velocities and the resultant resonant amplitude produced are very high compared to the other instabilities and increase with the increase in the velocity. Figure 3.6 demonstrates the possible interactions between the various phenomena causing aerodynamic instabilities. The various methods to perform the analysis are also mentioned in terms of physical and mathematical models. The most important cases to be considered for the design of a bridge are bueting random vibration, utter self-excited vibration and coupled utter with bueting response. The detailed descriptions for the aerodynamic instabilities are mentioned further in this chapter. A brief description of the dierent analytical procedure for each phenomenon is also presented. 3.3.1 Flutter
The phenomenon of utter is an aero-elastic eect on bridges and its occurrence is predominantly due to the aero-dynamic force, inertia force and the elastic force. Flutter is generally considered as an example of negative aerodynamic damping and the deections caused due to it increase to enormous levels until failure of the structure occurs. This phenomenon is known as classical utter and the other types of utter are stall utter and panel utter [28]. The main reason behind the failure of the Tacoma Narrows Bridge is believed to be utter. Classical utter is most common phenomena for bridges and it is treated for a
37
Figure 3.5: Relationship between wind velocity and aerodynamic instabilities (T. H. Le, 2003)
Figure 3.6: Possible interactions between the various phenomena causing aerodynamic instabilities and reduced velocity (T. H. Le, 2003)
Figure 3.7: Denition of the degrees of freedom for utter analysist (G.Morgenthal, 2000)
38
linear elastic system behavior as the structural oscillations are harmonic in nature and the amplitude of the vibration is controlled at the onset of utter. In the case of classical utter, energy is fed by wind into the system during the consecutive cycles counteracting the damping of the bridge. With the increase in the speed of incoming wind, the damping of the structure increases as well but starts decreasing with further increase in wind speed. The velocity at which the damping of the structure tends to approach zero is known as the critical utter velocity and in this case the amplitude of the structure is maintained constant. Any increase in the velocity beyond the critical limit will initiate amplitude of higher oscillations. There are two methods which are adopted to study the critical utter velocity. They are the free oscillation method and the forced oscillation method.
Free oscillation method
In this method [18,31], the structure under analysis is suspended elastically and is given an initial displacement and allowed to oscillate freely. Thus, this method studies the utter stability of the structure during motion. The various coecients such as the drag lift and moment coecients are measured and the translational and rotational motions are governed by the following equations: + ch h + kh h = L mh I + c + k = M (3.84) (3.85)
Where, m Mass of the structure, is the Moment of inertia, h is the displacement of the structure, is the rotation of the structure , L and M are the Lift force and Moment respectively and c and k are the Damping and Stiness coecients respectively. The lift and the moment forces are computed from CFD analysis and are incorporated in the above equations. From these equations, the displacements and rotations at each instant of time are calculated and the results are plotted against time. If the rotational angle due to utter diminishes with the passage of time then it can be concluded that the utter velocity is not reached. On the other hand, if the induced rotational angle keeps on increasing then it signies that the critical utter velocity is reached and is thus calculated from the resulting plots.
Forced Oscillation Method
In this method [18], the structure is subjected to a torsional or a drag force so that it vibrates with a prescribed frequency and amplitude. A lift force and moment is generated due to the applied force and the corresponding aerodynamic derivatives are calculated. These aerodynamic derivatives are used to compute the critical utter velocity. The bridge deck has two degrees of freedom, namely the vertical and the torsional. The moment and the lift force generated can be calculated by the following formulae [28]: 1 h B h L = U 2 (2B )[KH1 + KH2 + K 2 H3 + K 2 H4 ] 2 U U B (3.86)
39
h B 1 2 h M = U 2 (2B 2 )[KA + KA + K 2 A ] (3.87) 1 2 3 + K A4 2 U U B Where, K = B/U , is the reduced non-dimensional frequency, Hi and A i (i = 1, 2, 3, 4) are the aerodynamic derivatives, U is the wind velocity and B is the width of the bridge deck These aerodynamic coecients are generally computed from the results obtained by the wind tunnel experiments. The analytical methods for solving the above equations to obtain the aerodynamic derivatives are quite tedious and cumbersome as the air ow is separated along blu bodies or due to the generation of vortices. Hence this derivation is beyond the scope of this report. After the computation of the aerodynamic derivatives, they are incorporated into the following equations of motion for displacement and rotation h + ch h + kh h = 1 U 2 (2B )[KH h + KH B mh + K 2 H3 + K 2 H4 ] 1 2 2 U U B
(3.88)
h B 1 2 h + KA + K 2 A ] (3.89) I + c + k = U 2 (2B 2 )[KA 1 2 3 + K A4 2 U U B Rewriting the above equations and substituting for angular frequency and damp2 ing ratio, the following equations are derived h = k/m and h = c/2m . Solving the above equations based on plotting the graphs for the roots of the real and imaginary parts against the non-dimensional wind velocity [18] will provide the critical utter velocity as the intersection point of the real and imaginary root curves. U )Xc c B (3.90) B Where, Uc is the Critical utter velocity and Xc is the Intersection point ordinate For classical and stall utter, the criteria for aerodynamic stability is that the critical wind velocity is greater than 1.3 times the value of the reference velocity of the wind at the bridge site. Otherwise, it can be checked that the resulting amplitudes caused by utter are within the allowable levels for the structure. Uc = ( 3.3.2 Bueting
The aero-elastic phenomenon bueting falls in the category of wind-induced vibrations caused due to wind turbulence that are created by the uctuating and inconsistent forces. The velocity of the incoming wind is uctuating in nature and hence results in an inconsistent force on the structure. When the pressure variations in the incoming wind force have a frequency similar to one of the natural frequencies of the bridge, resonance will occur. The response of the bridge to bueting will mainly depend on the turbulence intensity, the natural frequencies and the shape of the structure. Bueting along with utter can cause large aerodynamic instabilities in long span bridges due to the large amplitude vibrations induced by them. Bueting in bridges may cause serviceability discomfort due to high and unpredicted displacements and also cause fatigue failure of structural members of the
40
bridge. Bueting in structures can be a serious threat because it can be caused at variable levels of uctuating velocities and thus had the potential to cause serious damage to a structure. Bueting can also take place in a coupled condition with utter at high velocity ranges. The bueting response analysis can be evaluated by the following two analytical approaches: Frequency-domain approach (Linear behavior) or Time-domain approach (Linear and non-linear behavior)
Figure 3.8: Bueting response prediction classication (T. H. Le, 2003)
Frequency-domain approach
The frequency-domain analysis of bueting response has been used during the recent times due to the fact that the time-domain analysis is time-consuming. In the frequency-domain, a Fourier transform is applied from the time-domain to the frequency-domain with spectral analysis and statistical computation. Also, nDOF systems have been decomposed to single DOF using modal analysis technique. Geometrical and aerodynamic nonlinearity can be taken into account in the time-domain analysis.
41
Assumptions and uncertainties
In the time-domain approach, the wind forces are applied at discrete nodes of the structure. Both quasi-steady sectional forces (aerodynamic admittance) and unsteady forces (indicial functions) can be used in the time-domain models. At higher wind velocities causing coupled utter and bueting, the frequency-domain forces can be transformed into the time-domain forms using rational function approximation with frequency response function (FRF). Newmark and Wilson direct integration methods can also be applied for time-domain analysis for wind-structure interaction and corresponding responses.
Time-domain approach
Both the frequency-domain and the time-domain approaches are based on certain assumptions for simplications specied by Le Thai Hoa [19] as stated below: 1. Gaussian stationary processes assumptions: The uctuating wind loads and velocities are treated as Gaussian stationary random processes. 2. Quasi-steady assumption: The unsteady bueting forces are modeled as quasisteady bueting forces by approximations made in the relative velocity and the unsteady force coecients. The relative velocity is simplied by omitting the unimportant components and force coecients are linearly approximated from the Taylor series expansion. 3. Strip assumption: Line-like structures are divided into span-wise strips and the unsteady forces on a single strip are caused by the forces acting only on that strip. At the same time, the forces acting on a single strip can be used to represent the entire line-like structure. 4. Correlation functions and transfer functions: Certain correlation functions such as aerodynamic admittance, coherence and joint acceptance, and transfer function such as mechanical admittance are added in transformation of statistical computation and single DOF input-output relation. 5. Modal uncoupling assumption: Multi-modal response is validated from generalized response. This assumption can be validated by the fact that the modal frequencies are diverse enough to create modal coupling and due to the complicated mechanism of the dynamic coupling of the modes.
Eects of bueting on bridges
The bueting phenomena can cause serious damages to bridges of which some are listed as follows: Serviceability discomfort. Fatigue damage to structural components.
42
Structural failure when coupled with utter at higher velocities resulting in structural collapse. For the analytical approach of bueting, Davenport [5] proposed the quasi-static method by introducing the aerodynamic admittance function for considering the unsteady nature of the wind eects. The complex nature of various bridge crosssections was taken into account by Scanlan [22] and he suggested the aerodynamic coecients be derived by the wind tunnel tests which are used to compute the selfexcited force. Scanlan [23] has also provided a relation between the aerodynamic admittance functions and the aerodynamic derivatives which forms the foundation of the conventional analysis procedure for bueting. The frequency domain approach for bueting analysis is limited to linear structures without aerodynamic nonlinearities ignoring the aerodynamic coupling of modes. This is a major limitation as this method cannot be performed under ultimate strength conditions but only for serviceability checks. The bueting force can be expressed in the form of matrix as follows:
Where, CL , CD and CM are the coecients for lift, drag and moment respectively, is the angle of attack of wind, [Cb ] is the static coecient matrix, and is the turbulent wind component vector. 3.3.3 Galloping
Galloping is an example of aerodynamic instability and results in cross-wind vibrations due to negative aerodynamic damping and takes place when the resultant wind load is positioned or angled in the same direction as the motion-induced wind load. Galloping is an instability phenomenon which takes place at much lower frequencies than vortex shedding and is generally common in slender structures. The two basic types of galloping phenomena are wake galloping and across wind galloping. The force diagram for galloping depicting the respective force directions are shown in the following gure 3.9 : As depicted in the above gure, if the width of the structure is taken as d and neglecting the turbulent wind components, the various force components can be represented as: 2 FD = 1/2Ur dCD (3.91)
2 FL = 1/2Ur dCL
(3.92)
In the above equation, Ur represents the relative wind velocity;
43
Figure 3.9: Directions of forces in galloing and angle of attack
tan = ( Simultaneously,
def ) U
(3.93)
Fy = 1/2U 2 dCy Cy = (CL + CD tan ) 1 cos
(3.94) (3.95)
Due to the approximation of small angles, a Taylor equation is assumed with respect to , representing the relation with the aerodynamic damping, 1 ( dCL + CD )|=0 Fy = U def (3.96) 2 d The following equation shows the aerodynamic damping on the right hand side and CL and CD are the lift coecent and the drag coecient respectively. The den Hartogs criteria is applied and the equation is simplied. If the den Hartogs criteria is specied then the damping coecient swill be negative. ( dCL + CD )|=0 0 d (3.97)
It should be noted that galloping is not much of a hindrance as utter and bueting as most of the cross-sections do not verify the den Hartogs criteria [20]. The cross-sectional types that are prone to galloping are square and D-shaped crosssections. The mass of the structure should also be taken into account while designing as increasing the mass of the structure (especially steel members) will increase the critical velocity. Hence, galloping is dened as a self-induced vibration in a exible structure in the cross-wind vibration mode. Galloping takes place at a particular onset wind velocity vCG and the amplitude of vibration increases with increasing velocity. EN 1991-1-4 species criteria based on which a structure is analyzed for galloping response. The onset wind velocity for galloping is calculated as
44
2Sc n1,y b (3.98) aG Where: Sc is the Scruton number, aG is the factor of galloping instability, n1,y is the rst fundamental frequency in the cross-wind direction, and b is the cross-wind depth of the bridge deck. vCG =
Chapter 4
Wind Load on Bridges based on Eurocode1

In this chapter the way Eurocode1 treats the wind phenomena and the corresponding bridge responses are discussed in details.
4.1
General
EN-1991-1-4 Eurocode, actions on structures-General Actions (EC1), provides guidance to determine the characteristic wind actions over entire structure, some parts of the structure or a member of the structure. EC1 tends to provide a platform to determine the wind action acting on any land based structures. This in practice is very dicult as a matter of fact that for dierent structures, dierent wind load information is required. For instance, to design bridges the wind load information which is required varies horizontally whereas to design masts the wind load information which is required varies vertically up to the height of the mast. EC1 is used as a guidance for almost all member states. Therefore, it is recommended to use national annex (NA) while following EC1. National annex provides specic data and methods based on the geological, topographical and meteorological characteristics of the state. The current version of the code can only be used for the structures having span length not more than 200 meters or height of 200 meters. The information regarding arch bridges, cable supported bridges, cable state bridges, torsional vibration of the structure, vibrations where more than the fundamental modes need to be considered and response of bridge deck against the transverse wind turbulence are not provided and the user is recommended to refer to the national annexes. EC1 contains a number of informative appendices starting from annex A till annex F. The term informative may represent the fact that an alternative method could be used provided that they are in accordance with the principles of EC1. Annex A provides guidance on dierent terrain categories, eect of orography, geographical conditions of the landscape and the inuence of neighboring structures that need to be considered while determining the mean wind velocity and the cor45
46
responding peak wind pressure. Annex B and C suggests two dierent methods to calculate the structural factor cscd of the entire structure when there is a need for dynamic assessment. Annex D illustrates some gures and diagrams to obtain directly the structural factor. Annex E gives some guidance on vortex shedding and other aeroelastic eects. Annex F propose some methods to determine the dynamic characteristics of the structure with linear behavior. NA can reject any of the above annexes and suggest an alternative method to replace them but it is not permitted to adopt part of the annex or cherry-pick individual clauses except when it is mentioned in the clause [3].
4.1.1
Distinction between principles and application rules
EC1 states that the rules in EN 1990, 2002 clause 1.4 apply in this case. Referring to the given clause it is stated that the principles are the general statements that shall not be replaced by any other statements i.e. there is no alternative for them until it is mentioned in the clause. Principles are represented by capital letter (P) following the paragraph number. On the other hand the application rules are the recognized rules that alternative rules are permitted to be used instead, provided that they comply with the principles and are at least equivalent with regards to structural safety, durability and serviceability with what would be expected when using the Eurocode. Therefore there is no alternative when a statement started by the letter (P) or where the word shall is used in the statement indicating that there is no alternative. Application rules are identied by a letter inside a bracket and also the words should or may in the statements may indicate that the given rule could be replaced by an alternative which complies with EC1 specications. However, by replacing any rule the user cannot claim that the substituted rule is in accordance with the EC1 although the rule is in accordance with the principles given in the EC1. 4.1.2 Denitions
Fundamental basic wind velocity is the basic start-point to estimate the design wind velocity which varies based on the geographical variation of the site. It is dened in EC1 as the 10 minute mean wind velocity with an annual risk of being exceeded of 0.02, irrespective of wind direction, at a height of 10 m above at open country terrain and accounting for altitude eects (if required). This value is usually given in the National annex but it is not the case with the Swedish annex as here the value of the basic wind velocity is directly provide. Basic wind velocity is the modication of the fundamental basic wind velocity including the seasonal and directional factors. In Swedish annex, this value is directly provided for each zone and city which is given in table NA1 and illustrated on a map as well. Notice that the seasonal factor, directional factor and the probability factor which are needed to calculate the basic wind velocity according to the EC1 all are considered to be unity. Therefore the probability factor which accounts for
Design situation
47
the life span of the structure as well as percentage of annual risk of exceeding cannot be modied. Mean wind velocity is modied to include the eects of terrain roughness and orography. The given value is given directly by the Swedish annex in the table NA2a for dierent heights and terrain categories for the basic wind velocities ranging from 21 to 26 m/s. The only disadvantage of this table is that the mean wind velocity cannot be calculated for the basic wind velocities which are lesser than the given basic wind velocities when calculating the eect of wind on the structure during the construction. Force coecients give the overall eect of the wind on a structure, structural element or component as a whole. They may account for the integration of surface pressure distribution. They are specied in wind axes, that is along-wind or drag coecient, cross-wind or lift coecient and rotational or moment coecient. Background response factor is represented by the B and controls the size eect factor, cs which allows for the lack of full correlation of uctuations in the external pressure over the surface of the structure. Resonance response factor is represented by R and controls the dynamic response factor cd which allows for dynamic response in the fundamental mode of the structure. This may represent the fundamental mode of the whole structure, but it would be appropriate for major components of the structure that act independently of the whole structure. For example a tall chimney stack on top of a building.
4.2
Design situation
EC1 recommends that the wind load for dierent design situation shall be determined for each design situations referring to EN 1990 clause 3.2. Therefore referring to the given clause the design situations are as follows: Persistent situations Transient situations Accidental situations Seismic situations The above design situations account for all the situations that can be predicted or may occur during the construction phase or service life. The last situation may not be considered in this report as the possibility of seismic force taking place in Sweden is very small. Additionally, the simultaneous occurrence of seismic force and the maximum wind velocity is very rare. But it is quite possible that accident may take place when there is a storm wind. EC1 denes another extra situation which is known as:
48
Fatigue 4.2.1 Persistent situations
Persistent situations refer to the condition of normal use. This may refer to the wind actions with characteristic risk acting on structures i.e. high wind will occur only within a few hours and the maximum wind load relevant to the static structures will take place for about a second. The structure is required to withstand these persistent loads without any failure or distress. This means that the structure needs to remain intact. The most important parameter in this case specially for building structure is the existing openings in the structure which need to be assumed closed while designing. The eect of persistent wind load acting on structure can be considered using partial coecient gama . The eect of ice, snow and trac must be considered in combined with the wind load wherever relevant. In this report the eect of trac is only taken into account. This was due to the assumption of a regular and immediate removal of snow in Sweden. 4.2.2 Transient situations
Transient situations refer to the condition when the structures are under construction, assembly or repair. The wind load acting on the structure at dierent phases of construction need to be fully considered. The extra components such as frames, walls and temporary structures may change the reference area over which the wind may be applied. The seasonal and directional factors could be modied to minimize the conservatism in wind loads during the construction as the duration of the construction work and the possibility of the occurrence of the peak wind velocity during the work is low. Designers usually specify a limit for wind velocity beyond which the construction will not take place. Transient wind load is considered using a partial coecient gama. In this report the eect of Transient loads are taken into consideration. 4.2.3 Accidental situations
Accidental situations refer to the condition when an accident occurs. Although the possibility of this condition is very low and even may not be happening during the entire life of the structure EC1 recommends this possibility needs to be accounted while designing the structure. Accidental load is referred to an impact load of very short period which may cause to change the reference area over which the wind is applied. The eect of this is accounted by a partial coecient f = 1. This load may even take place in combination with snow and storm. 4.2.4 Fatigue situations
The eect of fatigue needs to be assessed for the susceptible structures. The criteria for such structures are not provided by EC1. Therefore the user is advised to refer to the relevant structural Eurocodes.
Modeling of wind actions
49
4.3
Modeling of wind actions
The term modeling in this chapter refers not to the physical or numerical modeling of the structure which may be required for tunnel testing and the computational uid dynamics (CFD) but it refers to modeling of wind load which is complex in nature by using a set of simplied methods. The wind load is a variable load in both space and time. Therefore this makes the modeling of the wind phenomena very complex. EC1 uses a set of simplied values and formula to represent the wind actions on the surface of the structure. These simplied methods are based on determining the most severe characteristic wind loads i.e. extreme wind actions that can act over the surface of the structure. This hence may lead to conservative results. However EC1 allows for any measurement and testing to be carried out in order to model the wind action. It is found that when the design is assisted by measurement and testing, it results in lower design loads and more ecient structures. In this chapter EC1 states the term characteristic values. These values have a characteristic annual risk of being exceeded by 0.02 each year during its service life. This indicates that the possibility of occurrence of extreme wind velocity may be once in 50 years. Although EC1 introduces an expression for coecient of probability which is a function of the return period and the total life span of the structure but in the Swedish annex the value of basic wind load is directly provided. This means that the expression provided by the EC1 may not be possible to use as this coecient is used to determine the basic wind velocity. EN1 models the wind load by considering the structures size, shape and dynamic properties. It also models aeroelastic structures such as bridges whose motion causes increase in eect of the wind load on the structure.
4.4
Wind Velocity and Wind Pressure
Wind load is dynamic in nature. EC1 uses peak factor model for calculation of wind velocity. The concept is that the wind load that may represent the maximum quasi-steady gust loading or the corresponding dynamic response of the structure may be expressed by addition of a mean, steady part to a turbulent, unsteady part. Therefore, the maximum velocity of the wind can be determined by the following formula[3]. v (m) = vm (z ) + g (t).v (z ) = vm (z ).[1 + g (t).Iv (z )] (4.1) The term g (t) represents the proportionality of the turbulent part which for the gust loads depends on the size of the gusts and for dynamic structures depends on the characteristics of the structure. The term v (z ) is the standard deviation of the turbulence or the root mean square of the turbulent and is calculated as Iv (z ) = v (z )/vm (z ) . Therefore the turbulence intensity of the wind is the ratio of the wind turbulence by the mean wind velocity. Turbulence intensity is inversely proportional to the mean wind velocity which indicates that by increasing the mean wind velocity with height the turbulence will reduce. Based on the square relations
50
of pressure and velocity the given wind pressure may be calculated by the following formula:
2 w = F w .[1 + g (t).Iv (z )]2 = F w .[1 + 2.g (t).Iv (z ) + g 2 (t).Iv (z )] F
(4.2)
w is the total wind pressure where F w F w .[1 + 2.g (t).Iv (z )] F 4.4.1 Basic wind velocity (4.3)
As it is mentioned in the section 1.1.2 the eect of wind direction and seasons and the probability of the extreme wind being exceeded are considered by multiplying their corresponding factors with the characteristic wind speed. For calculation of relevant probability factor, the value of K = 0.2 and n = 0.5, which correspond to Fisher-Tippett Type 1(FT1) distribution of dynamic pressure. The other FT1 is to assume k = 0.1 and n = 1. The dierence in the above two models is not very signicant for the common structures like building and bridges but for the structures like nuclear installation is quite large [3]. 4.4.2 Mean wind
After determining the basic wind velocity, the mean wind velocity is obtained by considering the roughness factor and the orography factor. The value of orography factor is recommended to be equal to 1. In annex A of the EC1 a procedure is suggested to calculate the value of the factor which depends on the slope of the landscape around the structure where the structure is located and the relative length of the clis and escarpments. The roughness factor accounts for the height above the ground and the ground roughness of the terrain upwind of the structure in the wind direction. The expression 4.4 in EC1 denes the given factor which depends on the terrain factor, roughness length and height of the structure. EC1 divides the terrains in ve categories with their respective roughness length and the maximum and minimum heights. The eect of change in roughness length is considered when there is a transition between two terrain categories in annex A of EC1. The expression 4.4 in EC1 is valid when the upstream distance with uniform terrain is long enough to stabilize the prole suciently. The upstream distance is determined based on the direction of the wind with 15 degree variation on either side. The length up to which the distance needs to be considered is left to the national annexes to decide. 4.4.3 Wind turbulence
The expression 4.7 in EC1 is used to determine the turbulence intensity which depends on turbulence factor kI and orography factor c0 . However, setting the recommended value kI = 1 and c0 = 14 then Iv (z ) = 1/ ln(z/z0 ) , we get that
Structural factor (cs .cd )
51
cr (z ).Iv (z ) = kr , so that the turbulence intensity is predicted to be constant for all heights. 4.4.4 Wind action
The mean wind velocity acting on bridges is determined generally by the expression 5.3 given in EC1. The expression can be used for both dynamic and static structures. The values of cs .cd is equal to 1 for the static structures and for dynamic structures should be calculated based on two procedure suggested by the EC1 in corresponding annexes B and C [29]. Fw = cs .cd .cf .qp (ze ).Aref (4.4)
Where Fw is the maximum load acting normal to the surface of structures, cs .cd is the structural factor which is discussed in the next section, qp (ze ) is the peak wind pressure and Aref is the reference area normal to the direction of the wind.
4.5
Structural factor (cs .cd )
In order to be able to use the Eurocodes simplied method given in chapter 8 for determining the wind forces acting on the bridge deck rst a dynamic assessment of the structure is needed. If the structure is found not behaving dynamically against the wind load then the given methods could be used otherwise the structural factor should be calculated. The structural factor accounts for the two following factors: Non-simultaneous occurrence of peak wind pressure over the surface of the structure, generally called size eect. Vibration of the structure in its fundamental mode due to action of turbulence, generally called dynamic factor. The structural factor therefore is consisted of two individual factors. The shape factor which is denoted by cs takes into account the eect of the shape and the size of the structure as the non-simultaneous occurrence of the wind pressure over the surface of the structure will minimize the maximum pressure averaged over the surface. On the other side the dynamic factor which is denoted by cd accounts for the vibration of the structure in its fundamental mode due to action of turbulence. The uctuation of the wind in the same frequency as that of the fundamental mode of vibration of the structure will magnify the response of the structure to the uctuating load. Therefore the eect considered in the structural factor tends to encompass each other [29,30]. In EC1 the structural factor is determined both individually and together. The Swedish annex will not allow the designer to determine the cs .cd individually. However the model used in EC1 to represent the dynamic response component is valid only for the response of building structures in the rst cantilever mode and it is not applicable to bridges or for many of the individual building structures which can be regarded as being static [32,33].
52
The procedure recommended by the EC1 is based on the dynamic response of the structure in the along-wind direction as the root sum-square of a background and resonance component. The background component accounts for the quasi-steady (i.e. not amplied) response of the structure to the atmospheric turbulence and the dynamic component represents the dynamic vibration of the structure in its natural frequencies. This is usually called the Davenport method [4].
4.6
Wind actions on bridges
There exist a large number of limitations in order to use Eurocode1 for determining the eect of wind acting on bridges which some are as follows: The bridge must be rst assessed that it is responding statically against the uctuating wind load. If it is assessed that the bridge is acting dynamically against the uctuating wind load then the procedures given in the annex B and C may not be useful. This is due to the fact that the structural factors which account for the dynamic behavior of the structure are obtained based on the dynamic response component which is valid only for the response of building structures in the rst cantilever mode and it is not applicable to bridges or for many of the individual of building structures which can even be regarded as being static. The methods recommended in chapter 8 of EC1 are valid only for the structures with mode shapes of constant sign or simple linear mode shapes (cantilever structures) and it is not appropriate to use for continuous bridges, guyed masts, cable stayed bridges or arch bridges. The method is covering only the response of the structure in the along-wind direction of the wind. The maximum height of the structure is 200 meter and the same is for the length of main span. The cross section and the depth of the bridge deck should be uniform throughout the span of the bridge. The EN-1991-1-4 recommends a simplied method to use when it is assessed that the structure is not responding dynamically to the uctuating wind load. The general bridge congurations that the given method could be applied are given in the gure (8.1) [10]. The analysis of wind loading is performed both on the deck and the corresponding piers. While, combining these loads the forces need to be considered acting simultaneously to capture the most unfavorable conditions. The eect of the wind load is supposed to be in three directions. The x- axis is the along wind direction, y-axis is parallel to the span in the plane of the deck and the z-axis is perpendicular to the deck. The simultaneous occurrence of wind in x- and y-direction is quite rare as they act usually orthogonal, normal to the span of the deck. But the simultaneous occurrence of the wind in x- and z- axis is very common [10,11].
Wind actions on bridges
53
4.6.1
Force coecients
The wind force in x-direction is usually named as drag force and the coecient of the force in x-direction is called as drag coecient. The value of the drag coecient is cf x,0 = 1.3 or it can be determined from the gure 8.3 in EC1. The drag coecient given is applicable only for the structures without free end ow. This means that the bridge track or road should be continuous over the abutments that are the case for most of the bridges, except during the construction stage. The value of force coecient is reduced by 0.5 % per degree of this inclination for the bridge deck which has vertical inclinations on the wind ward side, which is limited to 30% reductions. The inclined ow takes place when the deck is not horizontal or when the terrain below the structure has slope, in this case the EC1 recommends performing especial studies such as tunnel test for the inclined deck higher than 10 degree. When the bridge deck is transversally inclined or when it is not horizontal to the direction of the wind i.e. when the total resolved height of the deck is greater than the actual depth of the deck then an increase of 0.3% per degree of inclination needs to be considered for the drag coecient, which is limited to a maximum of 25%. The reference area Aref is the total solid area of the structural parts that are projected windward. For bridges with solid barriers the whole area of the solid barriers is taken into account where in the case of open parapets, railing and safety barriers an area of 0.3m2 per meter run is considered. The combination eect of the above is given in the table 8.1 in EC1. The reference area is increased by 2m2 per meter run when the inuence of the trac is included. This may be causing a very conservative result as it considers the whole length of the bridge to be covered by trac. For bridges with multiple main girders the reference area is determined in a way so that the inuence of the girders is taken into account by considering both the girders before the placement of the deck to have exposure to the wind load. In the case of bridges with multiple girders all girders are assumed to have exposure to the wind load since before the placement of the deck the wind can penetrate inside the girders. The reference height of the bridge ze is the distance from the center of the deck structure to the lowest point of the ground under the bridge.
4.6.2
Force in x-direction- simplied method
As it is already mentioned this method can be used when the structure is assessed responds statically to the uctuating wind load. 1 2 Fw = ..C.vb .Aref 2
(4.5)
Where, vb is the basic wind speed, C is the wind load factor, Aref is the reference area and is the density of air.
54
4.6.3
Wind forces on bridge decks in z-direction
The wind force acting in the z-direction is usually called lift force and its corresponding coecient is called lift coecient. This force acts upward or downward normal to the bridge deck. The EC1 recommends the same formula (8.2) in the x-direction to be applied but recommends using the value of the lift coecient as cf,z = 0.9 otherwise the diagram 8.6 in EC1 may be used. Ar ef is the plan area of the deck and no end-eect factor is appropriate. The force is assumed to act with an eccentricity of e = b/4 from the center of the deck unless it is mentioned in the NA [10,11]. 4.6.4 Wind forces on bridge decks in y-direction
Eurocode1 suggests to approximate the value of the wind load in y-direction by taking 25 % of the force in x-direction for plated bridges (bridges with solid plate girders or box-girders) and 50 % for the truss bridges. Eect of this wind force is not very signicant due to the low magnitude of the wind load and the high resistance of the bridge deck in this direction. 4.6.5 Wind eects on piers
The designer is required to determine the most unfavorable direction of the wind load on the structure for each load eect under consideration and make separate calculations for all transient design situations during the construction. There is no provision recommended by EC1 to calculate the wind load acting on the piers of a bridge but the designer is directed to the general provisions of clauses 7.6, 7.8 or 7.9.2 for structural sections and cylinders.
4.7
Annexes
There are ve annexes which are used as informative appendices in EC1. Informative terms may refer to the fact that the suggested methods given in the annexes can be replaced by the methods recommended by the national annexes. The annex A recommends a method to determine the terrain factor. Annex B,C and D are simplied methods to determine the structural factor cs .cd of the structure. In Annex E, a simplied overview of a highly complex set of dynamic instabilities that may aect certain structures is given. In this annex some criteria are suggested that need to be fullled in order to assure that the structure is aerodynamically stable. Annex F provides some methods to determine the dynamic characteristics of the structure. The Swedish Annex suggests that the following annexes should not be used; A.4, A.5, B.1, B.2, B.4, the entire annex B, C, D and E.1.
Chapter 5
Finite Element Modelling

5.1 LUSAS FEM
In this chapter a detailed discription of the LUSAS modeling of the bridge is presented. The goemetric properties of the bridge and the element type and mesh sizes which are used for the dierent elements of the structure are described and reasons for selection of such element types and sizes are also presented. A comparison is then made between dierent boundary conditions or dierent support conditions. 5.1.1 Geometry
Figure 5.1: Full bridge model
The bridge that is under consideration is a simple girder bridge consisting of three spans to be built over a channel at Bergeforsen, Sweden. The total length of the bridge is 166 meters with a mid-span of is 66 meters and two side spans of 50 meter each. The entire length of the bridge is sub-divided by the cross channel bracings into 23 parts at specied distances which are unequal in length at specied distances as mentioned in the drawings. At the supports, the main I-girders are connected by cross I-beams. The deck of the bridge is composite in nature and consists of two main I-girders connected
55
56
Figure 5.2: Plan view of bridge showing the 23 sub-divisionsl
LUSAS FEM
57
by plates and channel sections at intervals. The girders are modeled as two longitudinal I-sections having a ange thickness of 40 mm and 48 mm at the top and bottom respectively. The web thickness is 21 mm. The actual bridge girder has varying geometrical properties at the supports and at the middle of each span. For simplicity, we have assumed the mean value of the thicknesses of each section. The main girder is stiened by four stiening plates placed at the abutments and at intermediate piers. A pathway is made inside the girder section to provide passage for maintenance work which is not included in the nite element modeling due to simplication. The upper deck of the bridge is made up of concrete oor. The cross-sections of the deck at dierent locations are shown below.
Figure 5.3: Deck cross-section at abatments
Figure 5.4: Deck cross-section at sub-divided locations showing the cross channel sections
The total width of the concrete deck is approximately 12 meters and has been assumed to have a uniform cross-sectional thickness of 0.4 meters for simplication as the thickness of the concrete deck varies throughout the width in the actual crosssection. The c/c distance of the main I-girders is 6 meters as shown in the Figure 5.6.
58
Figure 5.5: Deck cross-section at the piers
Figure 5.6: Longitudinal I-girders
The piers and the foundations are described in the gures below. The two intermediate piers are attached to the deck at the top and to the foundation at the bottom. The height of the piers are approximately taken as 25 meters each as per the drawings. The cross-section of the piers are in the shape of a dumbbell and the width increases from 5.2 meters to 7.6 meters at a height of 19.475 meters from the upper surface of the foundation as shown in the following diagrams. The thickness of each pier is 0.6 meters at the stem of the dumbbell cross-section and 1.6 meters at the ends. The c/c distances and other cross-sectional dimensions are shown in Figure 5.8 and Figure 5.9.
The piers are attached to the rectangular foundations at the base. The length, breadth and thickness of the foundations are 11 meters, 6.3 meters and 1.5 meters respectively. The foundations are placed directly on the soil and the cross-sectional
LUSAS FEM
59
Figure 5.7: Elevation of the pier at support 3
Figure 5.8: Plan view of the piers with varying cross-section
60
dimensions are mentioned in the following gures.
Figure 5.9: Plan view of the piers with varying cross-section
The abutments at supports 1 and 4 are assumed to be simply supported in the nite element modeling and have been designed with hinge and roller supports. The actual cross-sectional details of the abutments as per the drawings are shown in the following gure.
Figure 5.10: Side and front view of abutment at support 1
5.1.2
Cross-sections
The cross-sections of the main steel members that compose the deck are shown as per the drawings as follows (note: the drawings are not to scale): The deck of the bridge is orthopedic in nature which signies that the deck has dierent properties in the longitudinal and transverse directions. The deck transfers
LUSAS FEM
61
Figure 5.11: Steel cross sections
the load to the piers via the cross bracings and the main I-girders and the piers transfer the load to the underneath soil via the foundation. 5.1.3 Foundations
The bridge is designed as a simply supported at the end supports and as spring supports at the intermediate piers. The following diagrams illustrate the support conditions applied to the bridge.
Figure 5.12: Simply supported at the abutments (diagram not to scale)
5.1.4
Materials
The quality of steel used for the top and bottom anges of the I-girder is S460 and for the rest of the steel components S355N and S355J2 strength steel has been used. The modulus of elasticity of the steel has been adopted as 210 GPa with a Poissons ratio of 0.3. The density of steel is taken as 7800 kg/m3 . The other detailed mechanical properties of the steel qualities are shown in the table below. The grade of concrete used for the bridge components is C30/37. The modulus of elasticity for concrete is adopted as 30 GPa with a Poissons ratio of 0.2. The density of concrete is taken as 2400 kg/m3 .
62
Figure 5.13: Simply supported and spring stiness at abutment and intermediate pier
Table 5.1: Characteristic strength of steel
LUSAS FEM
63
5.1.5
Element interactions
The interaction that is considered for the bridge is that between the concrete deck and the I-girders. The connection between the deck and the I-girders is assumed to be rigid and is modeled as Constraint Equations or Joint Elements in LUSAS. The deck is modeled as a Master and the corresponding I-girders are modeled as Slaves. With the help of Tied Mesh, lines and areas can be rigidly linked together.
Figure 5.14: Deck and girder joint element constrain
5.1.6
Finite element modeling in LUSAS
The bridge is modeled and analyzed in LUSAS modeler which is based on the Finite Element Method of analysis. Finite Element Modeling and Analysis is based on the discretization technique and is a very dominant tool for analyzing complex structural mechanics problems. The basic idea behind FEM analysis is the discretization of the complex model into simple non-overlapping elements having nite degrees of freedom. The elements are interconnected with each other appropriately and the response of the model is obtained by the superimposition of the individual responses obtained from the discrete elements. A nite element analysis generally requires a thorough understanding of the geometry of the structure, the nodal connecting elements, the mass properties of the structure, boundary conditions and restraints to be applied and the various loads that are to act on the structure.
Beam elements
The beam elements that are used in the LUSAS model are of the type BMS3. The cross bracings are modeled as 3D thick beams BMS3. The properties of the BMS3 beam type are that the shear properties are included for this type of beam and the geometric properties are included throughout the length of the beam. The material properties for this type of beam elements are linear and isotropic. Nonlinear geometric eects cannot be modeled with this type of beam elements.
64
Figure 5.15: Bridge modeled in LUSAS based on FEM
Figure 5.16: Cross bracings modeled as 3D thick beam element BMS3
LUSAS FEM
65
Shell elements
The surfaces are designed as 3D thin shell elements QSI4. These elements take into account both membrane and exural deformations and the transverse shear deformations are not included. The directional material properties are dened in the local axes of the elements. The elements are assigned as isotropic in nature and the mass is modeled as lumped mass. The membrane stresses vary in a linear manner in QSI4 elements and all the loads applied on the membranes are lumped at the nodes. These elements cannot be used for non-linear behavior of materials. The deck surface, the I-girders and the stieners are modeled as QSI4 shell elements. The thicknesses are assigned as per the drawings. The piers are also modeled as thin shell elements and then the thickness is assigned. The following gure provides a description of the thin shell elements in the bridge model.
Figure 5.17: Shell elements represented by the surfaces marked in green
Mesh
A regular mesh of element size 0.5 has been used for both lines and surfaces. The cross bracings are modeled as line elements and the remaining structural members are modeled as surface elements. The mesh size of 0.5 was chosen as it provided the appropriate density of the mesh which generated practically accurate results and also the computer processing time was within limits.
66
Figure 5.18: Mesh of bridge components Loading
The bridge model was assigned its self-weight by the option of Body Force in LUSAS. The acceleration due to gravity value was assigned as 9.81 m/s2 .
Figure 5.19: Self-weight acting on the deck surface
5.1.7
Analysis of dierent boundary conditions
During the modeling of a structure for performing a nite element analysis, the boundary conditions have quite a signicant impact on the results obtained. Suitable assumptions regarding the support conditions and boundary restrains are necessary for the accurate analysis and estimation of results for a structure in a given load case scenario.
For the problem under consideration in this report, various possible outcomes and discrepancies have been observed by editing the boundary conditions. The possible cases that were modeled and analyzed in this report are mentioned as follows:
LUSAS FEM
67
Modeling the foundation of the bridge piers with spring stiness values taking the soil-structure interaction into account and the abutments as being simply supported with hinge and roller supports as shown in the gure 5.20. Modeling the bridge with piers and the foundation slabs and assigning a spring stiness values to the foundation slabs so that they behave as exible structures. The abutments are modeled as simply supported as the previous case shown in the gure 5.21.
Figure 5.20: Bridge modeled with spring supports without considering the footing
Figure 5.21: Bridge modeled with spring supports with considering the footing
Modeling the bridge without the piers and the foundation slabs and assuming the bridge to be simply supported at the abutments and at the intermediate supports. One of the abutments is assigned as a hinge support and the remaining three supports are assigned as roller supports shown in the gure 5.22: The case consisting of the spring stiness boundary condition at the pier supports is further subdivided into two cases. The rst case considers the spring stiness values for the soil-structure interaction computed and provided by Ramb oll AB. The spring stiness values provided for the supports 2 and 3 are shown in the table 5.2 : In the second case, we have computed the spring stiness values for a
68
Figure 5.22: Bridge modeled without considering piers Table 5.2: Spring stiness for supports 2 and 3, case1 (Ramb oll value) and case 2 (Gazetas, 1991) Case 1 Case 2 Support 2 Support 3 Support 2 Support 3 Unit 1 Length of the footing 11 11 11 11 m 2 Breadth of the footing 6.3 6.3 6.3 6.3 m 3 E-value of soil 79 105 79 105 MPa 4 Rotational stiness k,x 6.898.106 9.168.106 15.62.106 20.76.106 kN m/rad 5 Rotational stiness k,x 12.04.106 16.01.106 35.15.106 46.71.106 kN m/rad
rectangular footing based on Literature (Gazetas, 1991) and the values are in table 5.2: The results obtained by the analysis are presented in details in the next chapter. For case 1, the fundamental frequencies were observed to be the lowest when the spring stiness values provided by Ramb oll were used. The rst fundamental frequency was observed to be 0.724 Hz and it had a transversal fundamental mode shape. The second mode of vibration was vertical in shape. When the spring stiness values from Gazetas, 1991 were used, the fundamental frequency increased as expected from 0.724 Hz to 0.92 Hz due to higher spring stiness values. The mode shapes were observed to be identical as the boundary conditions were not altered. For case 2, when the foundation slabs of the bridge under the piers were considered to behave in a exible manner and are assigned with the spring stiness values, the frequencies observed were higher than the previous observed cases. This observation was obvious from the fact that the addition of the foundation slabs to the structure will increase the total mass of the structure resulting in an increase of the stiness of the bridge. In this case the spring stiness values that were provided by Ramb oll were used for the analysis and the rst fundamental frequency of the bridge was observed to be 0.961 Hz. The fundamental mode shapes did not show any variation with the rst case and the rst fundamental mode shape was transversal and the second mode was vertical in shape. For case 3, the bridge deck was modeled as a simply supported beam on one hinge
LUSAS FEM
69
and three roller supports. The model was analyzed for frequencies and mode shapes and the rst fundamental frequency was observed to be 1.578 Hz. The rst fundamental mode shape was vertical in nature and the second fundamental mode shape was transversal in nature. The results for the frequencies and mode shapes obtained were found to be exactly similar to the case if the bridge was modeled with a foundation slab and a xed boundary condition was assigned to the foundation. After a complete understanding of the behavior of the bridge under varying boundary conditions and restrains, we decided to proceed with our calculations based on the most critical case. From the results it is quite evident that the most critical case is the rst case with the spring stiness values provided by Ramb oll. This case had the lowest fundamental frequency of all the cases and had a transversal rst mode shape of vibration. When a structure has a lower fundamental frequency then it is more prone to dynamic behavior being imposed by wind action. Lower frequency indicates a lower stiness and signies that the structure can be easily excited to vibrate in its rst mode by a low frequency harmonic load. This means that a lower wind velocity can induce large amplitude of vibrations on the structure if the frequency of the induced load is in the range of one of the natural frequencies of the structure resulting in resonance. The participation factors and the sum mass participation factors were also generated from LUSAS for all the above discussed cases. A detailed report of the results including the mode shapes, Eigen values and Eigen frequencies, sum mass participation and the participation factors have been presented and discussed in Chapter 6. The sum mass participation factor provides a platform to identify the most signicant modes that need to be analyzed and it generally provides a cut-o boundary of 90
70
5.2
MATLAB FEM
The bridge is also modeled as a 2-dimensional nite element structure and analyzed in MatLab to generate the fundamental frequencies and mode shapes. The bridge elements are designed as Beam Element (Euler-Bernoulli Beam) where the shear deformations are not taken into account. The elemental stiness matrix of the Euler-Bernoulli beam along with the transformation matrix is shown as follows:
A general representation of the degrees of freedom of an Euler-Bernoulli beam is the following gure:
Figure 5.23: Degrees of freedom in 2D beam element
In MatLab, the degrees of freedom of the structure have been specied at each node. The bridge deck is sub-divided into 23 members by various nodes and the piers are sub-divided into three members each.
Figure 5.24: Degrees of freedom of the structural elements
Each node has three degrees of freedom, namely, translation along the x-direction, translation along the y-direction and rotation along the axis perpendicular to the plane of the paper. Hence, it is clearly visible from the diagram that the degrees of
MATLAB FEM
71
freedom of the bridge are numbered from 1 to 90 for all the 30 nodes i.e. 3 degrees of freedom for each node. The boundary conditions that are assigned to the model are similar to those assigned in LUSAS for case 3. The left abutment (node 1) has been modeled as a hinge support with the translations (DOFs 1 and 2) in the x and y-directions as constrained and the rotation (DOF 3) about the perpendicular axis as free. The right abutment (node 24) has been modeled as a roller support with the translation in the y-direction (DOF 71) as constrained and the translation in the x-direction (DOF 70) and the rotation about the perpendicular axis as free (DOF 72). The bases of the piers (node 27 and node 30) are assigned as xed supports with both the translations (DOFs 79, 80, 88 and 89) and the rotations (DOFs 81 and 90) to be xed and restrained. The MatLab program generates the fundamental frequencies and the rst three modes shapes of the 2-dimensional structure. The detailed description of the MatLab program has been shown in Appendix B. The fundamental frequency value of the 2-D bridge obtained from MatLab was found to be 1.476 Hz and the rst three fundamental mode shapes are illustrated as follows:
Figure 5.25: Mode shapes obtained from MATLAB
The fundamental frequency value of 1.476 Hz is quite similar to that obtained from LUSAS modeling the bridge as a simply supported bridge or with xed foundation slabs. This shows that the modeling of the bridge in LUSAS is justied and accurate and hence can be used for further extraction of results.
Chapter 6
Analysis and Results

In this chapter the total wind load acting over the bridge components is calculated as it causes a wind-induced vibration over the bridge structure. The bridge has a length of 166m and three spans which is supported by two intermediate piers with equal distance of 50m away from abatements. The piers have a height of 23m from the footing levels. The dimension of the bridge is described in details in chapter 5. The wind load and the corresponding bridge response is calculated by assuming the deck as a continuous beam of three spans. The procedure is based on the consideration that the deck is a line like structure. This is due to the fact that the vertical dimension of the deck is not in the same order of magnitude with the characteristic eddy sizes or the wave length taken at the natural frequency, (U/n). Also, the lack of correlation along the deck height does not signicantly contribute to the response calculations. If, it is found that the width of the structure is equal or larger than the wave length taken at the natural frequency then the deck has to be analyzed based on the plate-line structures.
6.1
6.1.1
Natural frequencies and Mode shapes

Results from LUSAS
The natural frequencies of the bridge are obtained using nite element model, LUSAS and the results are then veried by MATLAB. The results which are obtained by MATLAB are very close to the values extracted from LUSAS. The bridge is modeled by assuming two dierent boundary conditions; the xed boundary condition and the spring boundary conditions. The values corresponding to the stiness of the spring are supplied by Ramb oll (case 1) and used in analysis. However, the springs stiness may also be determined based on the modulus of the soil beneath the footings for a rectangular footing based on Literature (Gazetas, 1991) (i.e. case 2). The results are presented in table 5.2.
72
73
Spring boundary condition
The rst case is analyzed by modeling the bridge piers with spring support conditions. This case is further subdivided into two corresponding cases; case 1 for the spring stiness values provided by Ramb oll and case 2 for the spring stiness values obtained from literature (Gazetas, 1991). The values of the spring stiness are provided in Table 5.2 for the respective intermediate piers. The spring stiness values obtained from literature (Gazetas, 1991) are higher than those provided by Ramb oll. Hence, the eigen frequencies obtained for case 1 are lower than eigen frequencies obtained for case 2. Both the cases generate fundamental mode shapes which are identical in nature. As case 1 gives lower fundamental frequencies, it is considered to be the more critical case of the two and further analysis is performed for case 1. The fundamental frequency for case 1 is obtained as 0.724 Hz and the corresponding mode shape is transversal in nature (Figure 6.1). The rst fundamental mode is in the along-wind direction and the second mode of vibration is in the cross-wind direction having a frequency of 1.484 Hz. The mass participation along the y-axis is 75.82% for the rst mode which signies a considerable mass contribution of the total mass (Figure 6.4). The mass participations along the x and z-directions are considerably negligible for the rst mode (Figure 6.3 and 6.5).
Fixed boundary condition
In this model, the bridge piers were assigned xed boundary conditions and the translations and rotations about the x, y and z-directions were restricted. The abutments were assigned similar boundary conditions as the previous case as hinge and roller supports. The fundamental frequencies and mode shapes were extracted from LUSAS and were compared and analyzed. The rst fundamental frequency for this case was obtained as 1.491 Hz and the mode shape was a sinusoidal vertical deection mode in nature (Figure 6.6). This mode resembled the second mode shape of the previous case with spring supports. The fundamental frequency was observed to be higher than the previous case as the xed boundary conditions imparted a rigid nature to the bridge. The mass participation along the x-direction was 1.54% which was higher than those along the y and z-directions (Figure 6.8, 6.9 and 6.10). The third mode of vibration showed a signicant mass participation of 66.38% along the y-direction as well (Figure 6.9). The sum mass participation for the rst 50 modes was extracted and it was observed to reach 87% in the x-direction which can be approximated to 90%.
Piers
The piers were also analyzed and the eigen frequencies and mode shapes were extracted from LUSAS. The piers were modeled independently with spring stiness values assigned to the base of the piers. The spring stiness values used for this
74
Mode 1
Mode 2
F1 = 0.7235 Hz
Mode 3 Mode 4
F2 = 1.4848 Hz
F3 = 1.7869 Hz
Mode 5 Mode 6
F4 = 2.2121 Hz
F5 = 2.2765 Hz
Mode 7 Mode 8
F6 = 2.6301 Hz
F7 = 2.6332 Hz
Mode 9 Mode 10
F8 = 3.1753 Hz
Fixed boundary condition The mode shapes and mass participation is shown below.
F9 = 4.1435 Hz
F10 = 4.9304 Hz
Figure 6.1: Mode shapes and fundamental frequencies for the bridge with spring supports
75
Figure 6.2: Modes and frequencies for spring support case
Figure 6.3: Percentage mass participation in x-direction for spring support case
76
Figure 6.4: Percentage mass participation in y-direction for spring support case
Figure 6.5: Percentage mass participation in z-direction for spring support case
77
Mode 1
Mode 2
F1 = 1,491 Hz
Mode 3 Mode 4
F2 = 1,8026 Hz
F3 = 2,064 Hz
Mode 5 Mode 6
F4 = 2,242 Hz
F5 = 2,617 Hz
Mode 7 Mode 8
F6 = 2,647 Hz
F7 = 3,033 Hz
Mode 9 Mode 10
F8 = 3,484 Hz
Mass participation ans natural frequencies:
F9 = 4,396 Hz
F10 = 4,952 Hz
Figure 6.6: Mode shapes and fundamental frequencies for the bridge with xed supports
78
Figure 6.7: Modes and frequencies for xed support case
Figure 6.8: Percentage mass participation in x-direction for xed support case
79
Figure 6.9: Percentage mass participation in y-direction for xed support case
Figure 6.10: Percentage mass participation in z-direction for xed support case
80
case were those provided by Ramb oll. The rst fundamental frequency of the pier was observed to be 1.363 HZ and it had a mode of vibration along the y-direction (Figure 6.11). The second modal frequency was observed to be 3.399 Hz and the mode of vibration was along the x-direction. The rst three modal frequencies and mode shapes for the two corresponding directions x and y were extracted to analyze the pier for dierent directions of wind actions. Mode 1 showed a signicant mass participation of 66.92% along the y-direction which is obvious from the rst mode of vibration (Figure 6.13). The second mode of vibration shows signicant mass participation of 63.97% along the x-direction which is justied by the second mode of vibration (Figure 6.15). The rst two modes of vibration are analyzed further in this chapter for wind actions along the y and x-directions respectively.
Table 6.1: Frequencies, percentage mass participation and percentage sum mass participation for deformation along y-axis Frequency (Hz ) (%) Mass Participation (%) Sum Mass Participation Mode 1 1.362 66.92 66.92 Mode 4 9.287 18.25 85.17 Mode 8 27.137 5.52 90.69
Table 6.2: Frequencies, percentage mass participation and percentage sum mass participation for deformation along x-axis Frequency (Hz ) (%) Mass Participation (%) Sum Mass Participation Mode 2 3.399 63.97 63.97 Mode 7 20.495 19.04 83.01 Mode 12 52.458 7.01 90.02
6.1.2
Results from Matlab
A 2-dimensional model similar to the actual bridge was analyzed in MATLAB and the fundamental frequencies and mode shapes were extracted. The modes of vibrations are along the vertical direction (Figure 6.16). The fundamental frequency was 1.467 Hz and the rst ten fundamental frequencies were extracted (Table 6.3). The results obtained from MATLAB comply with those which are obtained from LUSAS for the xed boundary condition case (Table 6.3). This shows that the modeling in LUSAS is justied and hence will provide satisfactory results for modal frequencies and mode shapes.
6.2
Power spectral density function
There are four dierent Power Spectral Density Functions which can be found in literature for analyzing the wind action on structures (Chapter 2.4.3 Figure 2.4).
81
Mode 1
Mode 2
Mode 3
Mode 4
F1 = 1,363 Hz
Mode 5 Mode 6
F2= 3,399 Hz
Mode 7
F3= 6,679 Hz
Mode 8
F4= 9,287 Hz
F5 = 17,950 Hz
Mode 9 Mode 10
F6= 19,996 Hz
Mode 11
F7= 20,495 Hz
Mode 12
F8= 27,137 Hz
F9=33,339 Hz
F10=35,737 Hz
F11= 46,955 Hz
F12= 52,458 Hz
Figure 6.11: Mode shapes and frequencies of piers
82
Figure 6.12: Modes and frequencies of piers for deformations along y-axis
Figure 6.13: Percentage mas participation of piers for deformations along y-axis
83
Figure 6.14: Modes and frequencies of piers for deformations along x-axis
Figure 6.15: Percentage mas participation of piers for deformations along x-axis
84
Table 6.3: Comparison between the frequencies obtained from MATLAB with the frequencies along z-axis in LUSAS MATLAB LUSAS Mode Frequency(Hz ) Mode Frequency (Hz ) 1 1.467 1 1.491 2 1.742 4 2.242 3 1.867 6 2.647 4 3.611 9 4.397 5 4.685 11 5.077 6 5.683 14 6.939 7 5.905 15 7.232 8 7.973 16 7.429 9 9.100 18 7.912 10 11.060 21 9.118
Among these, the two most commonly used PSDF are von K arm an spectral density and Kaimal spectral density which is shown in the gure 2.5. Eurocode 1 is based on the Kaimal spectral density function (Equation 2.26) and Swedish Annex SSEN-1-4-2005 is based on the von Karman spectral density function (Equation 2.27). Hence, in this report we have analyzed the wind response based on the von K arm an spectral density function. Eurocode 1 suggests the value of integral length scale Lx u to be 180 m and the x Swedish annex suggests a value of Lu to be 150 m. In this report, we have analyzed the spectral response based on an integral length scale of 150 m (Figure 2.5). From the mentioned gure, it is clear that for a non-dimensional frequency fL of 0.04, the von K arm an spectral density gives a higher response value than the Kaimal spectral density and for a non-dimensional frequency fL of 0.6, the Kaimal spectral density gives a higher response value than the von K arm an spectral density.
6.2.1
Deck
In the analysis of the bridge deck against wind action, the two cases for spring support condition and xed support condition are analyzed independently. For the case with spring support, the rst two modes generate a non-dimensional frequency value which seems to be within the range of spectral density function (Figure 6.17). These two cases are rst analyzed separately and then together and the gust factors for both the cases are calculated which is shown further in the chapter. For the case with xed boundary conditions, the only the rst fundamental mode of vibration is analyzed as this generates a non-dimensional frequency fL which is in range of the spectral density function. This value of fL is similar to that generated by the second modal vibration of the previous case (Figure 6.17). The non-dimensional frequency fL for the corresponding vibration modes is calculated according to Equation 2.25.
85
Figure 6.16: Mode shapes obtained from MATLAB
Figure 6.17: Power spectral density function of the deck(von Karman)
86
6.2.2
Piers
For the case of piers, the analysis is performed for two dierent wind directions, namely x and y-direction. The modes and frequencies along these directions are extracted and the corresponding non-dimensional frequencies are calculated. It is found out that only the rst fundamental mode of vibration (f1 = 1.362Hz ) provides a non-dimensional frequency fL which falls within the range of the spectral density function (Figure 6.18). This mode of vibration is along the y-direction and the gust factor and other dynamic analysis parameters are calculated further in this chapter.
Figure 6.18: Power spectral density function of the pier (von Karman)
6.3
6.3.1
Response-inuence function
Deck
In order to calculate the dynamic response of the structure the inuence response function of the structure is required. The response may be in the form of bending moment, deection or stress. For this case i.e. continuous beam with three spans, the normalized bending moment inuence line diagram for the central mid-span is shown in gure 6.19. The bending moment inuence function is normalized with respect to the maximum bending moment value at the central mid-span i.e IR,ref . The IR,ref is the maximum response which here is equal to 11.55. 6.3.2 Piers
The response inuence functions (IR ) of the piers are assumed to have a constant sign. The corresponding response inuence functions for such structures are shown
Wind Load Analysis
87
Figure 6.19: Bending moment inuence function for the three continuous span beam
in Table 3.1. The entire procedure to calculate IR is given in the section 3.1.2.
6.4
Wind Load Analysis
The structure has a main span of the 66m which is higher than the 50m. Therefore a dynamic assessment of the bridge is needed based on the Swedish annex. Additionally the two rst natural freguencies are falling within the range of the power spectral density functions. In this section, rst the corresponding gust factor is obtained and then the maximum wind load acting over the structure is determined. 6.4.1 Determination of Gust factor
The wind analysis of the structure is performed based on three cases mentioned as follows: Bridge with intermediate piers having spring boundary conditions. Bridge with intermediate piers having xed boundary conditions. Piers
Bridge with intermediate piers having spring boundary conditions
This case is subdivided into two cases. In the rst case, only the rst mode of vibration is considered to be the governing mode of vibration and hence the analysis is performed considering only the rst fundamental frequency. For the second case, both the rst and the second modes of vibrations are considered to be the governing
88
modes of vibrations and hence the response of the structure to the wind action is analyzed for both the modes. Table 6.4 provides the input parameters required for the analysis and calculation of gust factor. Table 6.5 shows the calculation of gust factor for the case considering only the rst fundamental mode as the governing mode of vibration. Table 6.6 shows the calculation of the gust factor for the case considering the rst two fundamental modes of vibrations together as the governing modes.
Table 6.4: Input parameters spring xed support support 1 2 3 4 5 6 7 8 9 10 11 Wind climate Reference wind velocity Ubas Air density Site specications Roughness length z0 Terrain factor Structural specications Horizontal dimension b Vertical dimension h Natural frequency ne Logarithmic decrement s Mass per unit of length Mass per unit of area Shape factor C 23 1.25 0.05 0.19 166 3.5 0.742 0.04 10843.37 3098.1 1.56 23 1.25 0.05 0.19 166 3.5 1.48 0.04 10843.37 3098.1 1.56
Unit m/s kg/m3 m
Comment
SS-EN-Table NA1
SS-EN-Table 4.1
m m Hz kg/m kg/m2
FEM (LUSAS) SS-EN-Table F.2
SS-EN-Figure 8.3
Bridge with intermediate piers having xed boundary conditions
For this case, the support condition for the intermediate piers is assigned as xed and the bridge deck is analyzed for the along-wind response. For this case, it has been shown previously from the Power Spectral Density Function that only the rst mode generates a non-dimensional frequency which has a signicant response. Hence, only the rst fundamental mode is considered to be the governing mode and the gust factor is calculated based on the rst fundamental frequency as shown in Table 6.5.
Piers
For the case of the piers, it has been shown previously in this chapter that only the rst fundamental mode of vibration along the y-direction generates a nondimensional frequency which shows a signicant response in the Power Spectral Density (PSDF) diagram. The second mode of vibration generates a response which falls out of the range of the PSDF diagram. Hence, the dynamic wind analysis of the piers is performed for the rst fundamental mode of vibration and the details of the calculation are shown in Table 6.7 and Table 6.8.
Wind Load Analysis
89
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Table 6.5: Design parameters of the bridge deck spring xed support support Unit Reference height zref 25 25 m Mean wind velocity U (zref ) 27.16 27.16 m/s Reference Inuence number 11.55 11.55 N m/N Integral eect of gm i.e. m 0.105 0.105 Mean bending moment at mid-span 506.7 506.7 kN/m Turbulence intensity Iu (zref ) = Iu,ref 0.161 0.161 Integral length scale Lu = Lx 131.64 131.64 m u (zref ) y Lu = Lu /3 43.88 43.88 m 2 Non-dimensional response variance Jb 0.021 0.021 Background bending moment b 32.61 32.61 kN/m Aero-dynamic damping a 0.0113 0.0055 Total damping 0.0513 0.00.0455 Non-dimensional frequency fL 3.51 7.17 Spectral density function RN (zref , ne ) 0.0497 0.0308 y = Cy nb/Uref (Cy = 8) 35.403 72.37 2 Joint acceptance function Jy (ne ) 0.02804 0.0136 Equivalent mass mg 9.105 9.105 kg Variance of acceleration acc 0.0544 0.0318 m Resonant bending moment r 194.57 94.53 kN m Peak factor Kp (v = ne ) 3.651 3.841 Standard deviation of 197.28 99.998 kN m Structural response R Maximum bending moment Rm ax 1226.97 890.83 kN m Gust factor = Rmax /R 2.421 1.758
Comment Eq. 3.56 Figure 3.1 Eq. 3.43 Eq. 3.42 Eq. 2.14 Eq. 3.57 Eq.3.47 Eq. 3.46 Eq. 3.66 Eq. 3.65 Eq. 2.25 Eq. 2.27 Eq. Eq. Eq. Eq. Eq. Eq. 3.51 3.39 3.53 3.53 3.59 3.26
Eq. 3.25 Eq. 3.25
6.4.2
Determination of Maximum Wind Load on the Structure
The maximum wind load calculations for the bridge deck are shown in table 6.9 for the three cases mentioned previously, i.e. spring support condition with rst mode as the governing mode, spring support condition with the rst two modes as the governing modes and xed support condition with the rst fundamental mode as the governing mode of vibration. Table 6.10 also shows three dierent load conditions, i.e. without trac, with trac and under construction. In the case of without trac, the vertical dimension of the deck is equal to 4 m and for the case of with trac, the vertical dimension of the deck is 6 m taking into consideration the height of the vehicles. In the case of under construction, it is assumed that the two girders are placed over the piers before installing the concrete deck. Therefore, the vertical dimensions of the two girders are calculated to be 5.1 m taking into consideration the eect of both the girders. The basic wind velocity vb during the construction is taken as 15 m/s. The corresponding maximum wind loads per unit length of the deck in x-direction (along-wind direction) are shown in table 6.11. The maximum wind in the vertical
90
direction i.e. z-direction is taken as the same as that of in x-direction but the force coecient cf z is taken as 0.9 which is recommended by EC1 (cl. 8.3.3) . The maximum wind force in y-direction is taken as 25% of the force in x-direction for plated bridges based on cl. 8.3.4 of EC1. For the piers, table 6.12 shows the maximum wind load per unit length acting on the pier based on the gust factor calculated in table 6.9.
6.5
6.5.1
Vortex shedding and Aeroelastic instabilities

Vortex shedding
For the bridge deck, the ratio between the critical vortex shedding velocity and the mean wind velocity is calculated to be 1,404. Annex E.1.2(2) species that the vortex shedding response need not be checked if the above ratio exceeds 1,25. Hence, it can be concluded that the bridge deck will be safe against vortex shedding instability.
The maximum deection caused due to the vortex shedding load on the piers is calculated to be 0,036m which is quite negligible when compared to the total height and the cross-sectional dimensions of the pier. Hence, it can be concluded that the piers will not be aected by vortex shedding and the resultant deections are within safe limits. The detailed calculations are shown in the Appendix. 6.5.2 Galloping
Eurocode1 mentions that for a structure to be declared safe against galloping instability, the ratio between the onset galloping velocity and the mean wind velocity should be greater than 1,25. In our case, the above ratio is 2,259 and hence it can be concluded that the bridge deck is safe against galloping instability. 6.5.3 Divergence and Flutter Response of the Bridge Deck
The divergence and utter response of the bridge deck is analyzed with the help of EN1991-1-4 Annex E.4. Divergence and utter are aerodynamic instabilities that can occur for plate-like structures above a certain threshold or critical wind velocity. To avoid divergence and utter, three criteria are given in EC1 to check the susceptibility of a structure to these instabilities. The three criteria are checked in order stated and if any one of the criteria is not met, then the structure will not be prone to either divergence or utter. The three criteria provided by EC1 are as follows: The structure or a substantial part of it has an elongated cross-section similar to a at plate with b/d less than 0.25.
91
Hence, b/d = 3.5/12.01 = 0.291 which is greater than 0.25. Thus, the rst criterion is not met and so the bridge deck will not be prone to either divergence or utter. The remaining two criteria that are mentioned below need not be further checked: The torsional axis is parallel to the plane of the plate and normal to the wind direction. The center of torsion is at least d/4 downwind of the windward edge of the plate. The lowest natural frequency corresponds to a torsional mode, or else the lowest torsional natural frequency is less than 2 times the lowest translational natural frequency.
92
Table 6.6: Determining the gust factor by considering the eect of rst two mode shapes for spring support. Mode 1 Mode 2 Wind climate 1 Reference wind velocity Ubas 23 23 m/s SS-EN-Table NA1 2 Air density 1.25 1.25 kg/m3 Site specications 3 Roughness length z0 0.05 0.05 m SS-EN-Table 4.1 4 Terrain factor 0.19 0.19 Structural specications 5 Horizontal dimension b 166 166 m 6 Vertical dimension h 3.5 3.5 m 7 Natural frequency ne 0.742 1.48 Hz FEM (LUSAS) 8 Logarithmic decrement s 0.04 0.04 SS-EN-Table F.2 9 Mass per unit of length 10843.37 10843.37 kg/m 10 Mass per unit of area 3098.1 3098.1 kg/m2 11 Shape factor C 1.56 1.56 SS-EN-Figure 8.3 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 32 33 34 Reference height zref Mean wind velocity U (zref ) Reference Inuence number Integral eect of gm i.e. m Mean bending moment at mid-span Turbulence intensity Iu (zref ) = Iu,ref Integral length scale Lu = Lx u (zref ) Ly u = Lu /3 2 Non-dimensional response variance Jb Background bending moment b Aero-dynamic damping a Total damping Non-dimensional frequency fL Spectral density function RN (zref , ne ) y = Cy nb/Uref (Cy = 8) Joint acceptance function Jy (ne )2 Equivalent mass mg Variance of acceleration acc Resonant bending moment r Peak factor Kp (v = ne ) Standard deviation of Structural response R Total resonant response R Maximum bending moment Rm ax Gust factor = Rmax /R 25 27.16 11.55 0.105 506.7 0.161 131.64 43.88 0.021 32.61 0.0113 0.0513 3.51 0.0497 35.403 0.02804 9.105 0.0544 194.57 3.651 197.28 221.18 1356.34 2.677 25 27.16 11.55 0.105 506.7 0.161 131.64 43.88 0.021 32.61 0.0055 0.00.0455 7.17 0.0308 72.37 0.0136 9.105 0.0318 94.53 3.841 99.998 m m/s N m/N kN/m m m kN/m
Eq. 3.56 Figure 3.1 Eq. 3.43 Eq. 3.42 Eq. 2.14 Eq. 3.57 Eq.3.47 Eq. 3.46 Eq. 3.66 Eq. 3.65 Eq. 2.25 Eq. 2.27 Eq. Eq. Eq. Eq. Eq. Eq. 3.51 3.39 3.53 3.53 3.59 3.26
kg m kN m kN m kN m kN m
Eq. 3.24 Eq. 3.25 Eq. 3.25
93
Table 6.7: Input parameters of the piers Larger area Smaller area exposed exposed 1 2 3 4 5 6 7 8 9 10 11 Wind climate Reference wind velocity Ubas Air density Site specications Roughness length z0 Terrain factor Structural specications Horizontal dimension b Vertical dimension h Natural frequency ne Logarithmic decrement s Mass per unit of length Mass per unit of area Shape factor C 23 1.25 0.05 0.19 5.2 23.075 1.36 0.03 19903.1 3827.52 1.137 23 1.25 0.05 0.19 1.6 23.075 3.39 0.03 19903.1 12439.44 1.428
Unit m/s kg/m3 m
Comment
SS-EN-Table NA1
SS-EN-Table 4.1
m m Hz kg/m kg/m2
SS-EN-table 7.23
1 2 6 7 11 12 12 12 13 14 15 16 17 18 19 20 21 22 23
Table 6.8: Design parameters of the pier Larger area Smaller area exposed exposed Reference height zref 13.845 13.845 Mean wind velocity U (zref ) 24.57 24.57 Turbulence intensity Iu (zref ) = Iu,ref 0.1778 0.1778 Integral length scale Lu = Lx ( z ) 110.252 110.252 ref u Aero-dynamic damping a 0.00335 0.00052 Total damping 0.03335 0.03052 Background response frequency n0 0.212 0.284 Background response kb 0.7563 0.7607 Non-dimensional frequency fL 6.104 15.208 Spectral density function RN (zref , ne ) 0.0334 0.0182 Mode shape coecient Gy 0.5 0.5 Non-dimensional parameter y 2.302 1.7656 Mode shape coecient Gz 0.2778 0.2778 Non-dimensional parameterz 10.215 25.46 Size reduction factor ks (ne ) 0.212 0.109 Resonant response kr 1.082 0.33 Zero-uppcrossing frequency v 1.0523 1.8805 Peak factor kp 3.752 3.903 Gust factor 2.81 2.45
Unit m m/s m
Comment Eq. 3.56 Eq. 2.14 Eq. 3.57 Eq. 3.66 Eq. 3.65 Eq. 3.62 Eq. 3.63 Eq. 2.25 Eq. 2.27 Table 3.1 Eq. 2.54 Table 3.1 Eq. 2.54 Eq. 2.54 Eq. 3.64 Eq. 3.60 Eq. 3.59 Eq. 3.58
Hz
Hz
94
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Table 6.9: Design parameters of the piers obtained from Eurocode1 Larger area Smaller area exposed exposed Unit Wind climate Reference wind velocity Ubas 23 23 m/s Air density 1.25 1.25 kg/m3 Site specications Roughness length z0 0.05 0.05 m Terrain factor 0.19 0.19 Structural specications Horizontal dimension b 5.2 1.6 m Vertical dimension h 23.075 23.075 m Natural frequency ne 1.36 3.39 Hz Logarithmic decrement s 0.03 0.03 Mass per unit of length 19903.1 19903.1 kg/m Mass per unit of area 3827.52 12439.44 kg/m2 Shape factor C 1.137 1.428 Reference height zref 13.845 13.845 m Mean wind velocity U (zref ) 26.808 26.808 m/s Turbulence intensity Iu (zref ) = Iu,ref 0.161 0.161 Integral length scale Lu = Lx ( z ) 150 150 m ref u Aero-dynamic damping a 0.001 0.001 Total damping 0.031 0.031 Background response factor B 2 0.961 0.97 Non-dimensional frequency fL 7.01 18.968 Spectral density function RN (zref , ne ) 0.03 0.016 Non-dimensional parameter b 0.542 0.607 Non-dimensional parameterh 0.299 0.146 Resonant response Factor R2 0.977 0.291 Zero-uppcrossing frequency v 0.965 1.628 Hz Peak factor kp 3.735 3.872 Structural factor cs cd 1.257 1.128 Gust factor 2.674 2.399
Comment
SS-EN-Table NA1
SS-EN-Table 4.1
SS-EN-table 7.23 EC1,Eq. 4.3 Eq. 4.7 Table F.2 Swedish Annex Swedish Annex
Cl. 6.3.1
Table 6.10: Input parameters to determine maximum wind load acting along the deck length Without With During trac trac construction unit Comment vertical dimension dtot 4 6 5.278 m Reference areaAref 664 996 876.148 m2 Force coecient Cf x 1.56 1.87 1.65 Basic wind velocity vbas 23 23 15 m/s Air Density 1.25 1.25 1.25 kg/m3
95
Table 6.11: Maximum wind load acting along the deck length Spring support Spring support Fixed (1st mode) (1st and 2nd modes) support unit Gust factor 2.422 2.678 1.758 Air Density 1.25 1.25 1.25 kg/m3 Mean load Uz,ref (z ) 27.16 27.16 17.71 m/s Without Trac 6.965 7.699 5.057 kN/m With trac 12.524 13.844 9.093 kN/m Under Construction 4.134 4.57 3.002 kN/m
Comment
Eq. (2.37)(2.39) Eq. (2.37)(2.39) Eq. (2.37)(2.39)
Table 6.12: Maximum wind load acting along the pier length Larger area Smaller area exposed exposed Unit Comment 2 Reference area 120 36.92 m Gust factor 2.81 2.45 Shape factor 1.137 1.428 Air Density 1.25 1.25 kg/m3 Mean load Uz,ref (z ) 24.57 24.57 m/s Maximum wind load 6.23 2.112 kN/m Eq. (2.37)(2.39)
1 2 3 4 5 6 7 8 9 10 11
Table 6.13: Vortex shedding response of bridge deck Values Unit Fundamental frequency in crosswind direction f1 1.48 Hz Width of the deck in the crosswind direction d 3.5 m Width of the deck in the alongwind direction b 12.01 m Strouhal Number St 0.142 Density of air 1.25 kg/m3 Structural logarithmic decrement of damping s 0.04 Mass per unit length me 10843.37 kg/m Critical velocity for vortex shedding Ucrit 36.4788 m/s Mean wind velocity at the center 27.16 m/s of the deck Um Ratio of critical velocity to 1.343 mean wind velocity (Ucrit /Um ) Max. governing ratio for susceptibility 1.25 to vortex shedding (Ucrit /Um )1
Comment
Fig. E.1 EC1
E.1.2(2) EC1
96
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Table 6.14: Vortex shedding response of piers Values Unit Fundamental frequency in crosswind direction f1 1.362 Hz Width of the pier in the crosswind direction d 1.6 m Width of the pier in the alongwind direction b 5.2 m Strouhal Number St 0.107 Density of air 1.25 kg/m3 Structural logarithmic decrement of damping s 0.03 Height of the piers h 23.075 m Mass per unit length me 19903.1 kg/m Scruton Number Sc 373.1831 Assumed correlation length Lj 9.6 m Critical velocity for vortex shedding Ucrit 20.3664 m/s Mean wind velocity at the center 22.53 m/s of the correlation length Um,Lj Basic value of standard deviation of load clat,0 1.1 Ratio of critical velocity to 0.9039 mean wind velocity (Ucrit /Um,Lj ) Standard deviation of load clat 0.9135 Mode shape factor K 0.133 Eective correlation length factor Kw 0.801 Maximum deection amplitude ymax 0.0364 m
Comment
Fig. E.1 EC1
Table E.2 EC1
1 2 3 4 5 6 7 8 9 10 11 12 13
Table 6.15: Galloping response of bridge deck Values Unit Fundamental frequency in crosswind direction f1 1.48 Hz Width of the deck in the crosswind direction d 3.5 m Width of the deck in the alongwind direction b 12.01 m Strouhal Number St 0.142 Density of air 1.25 kg/m3 Structural logarithmic decrement of damping s 0.04 Mass per unit length me 10843.37 kg/m Scruton Number Sc 56.65 Factor of galloping instability aG 10 Onset velocity for galloping Ucg 58.69 m/s Mean wind velocity at the center 27.16 m/s of the deck Um Ratio of onset velocity to 2.161 mean wind velocity (Ucg /Um ) Max. governing ratio for susceptibility 1.25 to galloping (Ucg /Um )1
Comment
Fig. E.1 EC1
E.2.1(2) EC1
Chapter 7
Conclusion
EN-1991-1-4 or Eurocode 1, actions on structures-General Actions part 1-4, provides certain simplied methods that can be used as a guidance to determine the eect of uctuating wind load on various types of structures and structural components. These methods are found to be based on a set of specic assumptions and simplications which restrict the applications of the code to structures having a complicated response functions to the uctuating wind load. If the wind-induced vibration of the structure is negligible i.e. the non-dimensional frequency of the structure is not within the range of the power spectral density function (section 2.4.3), then the wind load is treated as a static wind load case and the simplied method suggested in clause 8.3.2 can be used. The structures whose non dimensional frequency falls within the range, a dynamic assessment needs to be performed. Eurocode 1 presents specic methods to determine the structural factor which are applicable only for particular structural congurations (gure 6.1 EC1) whose responseinuence functions have constant signs. The formula to obtain the aerodynamic admittance functions that are presented in Eurocode 1 (annex B and C) are valid for structures whose response-inuence functions have constant sign. For structures whose response-inuence function have complicated signs like the bending moment inuence function of the given bridge in this report, these simplied methods may not be appropriate to use in order to calculate the aerodynamic admittance function. The other limitations behind the simplied methods to calculate the structural factors are that the along-wind vibration in the fundamental mode needs to be signicant and have a constant sign. This may not be valid for continuous bridges for which more than one mode shape may contribute to the response of the structure against uctuating wind load and one of the contributing fundamental modes may vary in a complicated manner such as a sinusoidal function with changing sign. Hence, the aerodynamic admittance functions are obtained from multiple integrals of the normalized co-spectrum funtions and the mode shapes (section 3.1.2). By comparing the results obtained from Eurocode1 and the analytical method presented in chapter 3 for the piers which are analyzed as structural components having a constant sign response-inuence function and fundamental mode shape, it has been observed that the dierences in the gust factor-values are very minor and both pro97
98
Conclusion
vide similar results. On the other hand, in the case of the three span bridge deck the bending moment inuence function does not vary with a constant sign but rather in a complicated manner, the methods presented in Euocode 1 may not be appropriate to use and hence no comparison could be made between the results obtained from the theoretical methods and Eurocode 1. The bridge is modeled using FEM with two separate boundary conditions, namely spring support and xed support conditions. It has been observed that modeling the intermediate piers of the bridge as xed supports increases the fundamental frequency of the structure and hence results in a lower value of the gust factor. On the other hand, modeling the footing of the intermediate piers as spring supports reduces the fundamental frequency of the structure and changes the fundamental mode shape which results in a higher value of the gust factor. It has also been observed that the resultant wind load acting on the structure increases by 13% when the dynamic response of the structure is considered. From the above observations it may be concluded that one must be aware of the limitations and assumptions behind the simplied methods presented in the Eurocode1 and knowledge of the concepts behind these suggested methods will enhance the capability of a designer towards an ecient use of the given code. Further research is recommended to analyse the deck of the bridge as a plate-like structure to evaluate the dynamic response of the deck in two directions. A wind tunnel test if conducted, may also provide more accurate force coecients for the bridge.
References
[1] Basu, R.I. (1983) Across-wind Response of Slender Structures of Circular Cross-Section to Atmospheric Turbulence PhD Thesis, BLWT-3-1983, University of Western Ontario, London, Ontario, Canada. [2] Bouassida, Y., Bouchon, E., Crespo, P., Croce, P., Davaine, L., Denton, S., Feldmann, M.,Frank, R., Hanswille, G., Hensen, W., Kolias, B., Malakatas, N., Mancini, G., Ortega, M., Raoul, J., Sedlacek, G., Tsionis, G. (2010) Bridge Design to Eurocodes Worked examples. Bridge Design to Eurocodes, Vienna, 4-6 October. [3] Cook, N. (2007) Designers Guide to EN 1991-1-4 Eurocode I: Actions on Structures, General Actions, Part 1-4: Wind Actions. Designers Guide to the Eurocodes. [4] Davenport, A. G. (1961) The application of the statistical concepts to the wind loading of the structures. Proceeding of the Institution of Civil Engineers, 19, 449-471. [5] Davenport, A.G. (1962) The response of slender line-like structures to a gusty wind. Proceedings of the Institution of Civil engineers, 23, 389-408. [6] Davenport, A.G. (1967) Gust loading factors. Journal of the Structural division, ASCE, 93, 11-34. [7] Davenport, A.G. (1977) The prediction of the response of structures to gusty wind. Safety of Structures under Dynamic Loading, Hoalnd, Kavlie, Moe and Sigbj ornsson (eds.), Tapir, Trondheim, pp. 257-284. [8] Davenport, A.G. (1982) The reliability and synthesis of aerodynamic and meteorological data for wind loading. Seminar on Structural Aerodynamics, Insititut Technique du B atiment et des Travaux Publics (ITBTP), St-Remy-LesChevreuse, France, April 20-22. [9] Dyrbye, C. and Hansen, S.O. (1988) Calculation for joint acceptance function for line-like structures. Journal of Wind Engineering and Industrial Aerodynamics, 31, 210-430. [10] Eurocode 1 (1995) Basis of design and actions on structures-Part 2-4: Actions on structures-Wind actions. European Prestandard ENV 1991-2-4.
99
100
References
[11] Eurocode 1 (2005) Basis of design and actions on structures-Part 1-4: Actions on structures-Wind actions. Svensk Standard SS-EN-1991-1-4:2005. [12] Giosan, I.(2000) Vortex shedding induced loads on free standing structures. Structural Vortex Shedding Response Estimation Methodology and Finite Element Simulation, 1-20. [13] Hansen, S.O. and Krenk, S. (1996) Dynamic along-wind response of simple structures. Journal of Engineering, Mechanics ASCE. [14] Harris, R.I.(1971) The nature of wind in the modern design of wind-sensitive structures. London: Construction Industry Research and Information Association. [15] Isyumov, N. (2011) Announcement of the Alan G. Davenport Wind Loading Chain, Boundary Layer Wind Tunnel Laboratory. The University of Western Ontario. [16] Jain, A., Jones, N.P. and Scanlan, R.H. (1995) Fully coupled bueting analysis of long-span bridges. Proceedings of the Ninth International Conference on Wind Engineering, New Delhi, January, pp. 962-971. [17] Kaimal, J.C., Wyngaard, J.C., Izumi, Y. and Cot e, O.R. (1972) Spectral characteristics of surface layer turbulence. Journal of Royal Meteorological society, 98, 563-589. [18] Larsen, A. (1995) Prediction for aeroelastic stability of suspension bridges during erection. Proceedings of the Ninth International Conference on Wind Engineering, New Delhi, January, pp. 917-927. [19] Le, T.H., (2003) Flutter aerodynamic stability analysis and some aerodynamic control approaches of cable-stayed bridges (in Vietnamese). Masters thesis at Vietnam National University of Hanoi. [20] Morgenthal, G. (2000) Fluid-Structure Interaction in Blu-Body Aerodynamics and Long-Span Bridge Design: Phenomena and Methods. PhD Thesis, CUED/D-STRUCT/TR.187, University of Cambridge. [21] Sanpaolesi, L. and Croce, P. (2005) Handbook 4: Design of bridges: Guide to the basis of bridge design related to Eurocodes supplemented by practical examples. Leonardo Da Vinci Pilot Project CZ/02/B/F/PP-134007. [22] Scanlan, R. H. (1978) The action of exible bridges under wind. 2: Bueting theory J. . Sound and Vibration, 60(2), 201211. [23] Scanlan, R. H. (2000) Bridge Deck Aeroelastic Admittance Revisited. ASCE, J. Bridge Engineering, 5(1), pp. 1-7. [24] Simiu, E. (1973) Gust Factor and Alongwind Pressure Correlations. Journal of Structural Division, 99, pp. 773-783.
References
101
[25] Simiu, E. (1974) Wind Spectra and Dynamic Alongwind Response. Journal of Structural Division, 100, pp. 1897-1910. [26] Simiu, E. (1976) Equivalent Static Wind Loads for Tall Building Design. Journal of Structural Division, 102, pp. 719-737. [27] Simiu, E. (1980) Revised Procedure for Estimating Alongwind Response. Journal of Structural Division, 106, pp. 1-10. [28] Simiu, E. and Scanlan, R.H. (1986) Wind Eects on Structures. An Introduction to Wind Engineering. Second edition. Wiley, New York. [29] Solari, G. (1982) Design Wind Loads. Proceedings of the 5th Colloquium on Industrial Aerodynamics, Aachen, Germany. Part 2, pp. 73-86. [30] Solari, G. (1982) Alongwind Response Estimation: Closed Form Solution. Journal of Structural Division, 108, 225-244. [31] Theodorsen, Th. (1934) General theory of aerodynamic instability and the mechanism of utter. NACA Report No. 496, Washington DC. [32] Trakverkets f oreskrifter om a ndring i V agverkets f oreskrifter (VVFS 2004:43) om till ampningen av europeiska ber akningsstandarder (2011) TRVFS 2011:12. Trakverkets f orfattningssamling. [33] TRVK Bro 11 (2011) Trakverkets tekniska krav Bro TRV publ nr 2011:085. Trakverket. [34] Vickery, B.J. (1970) On the reliability of gust loading factors. Proceedings of Technical Meeting concerning Wind Loads on Buildings and Structures, National Bureau of Standards, Washington, DC, 93-10. [35] Vickery, B.J. and Basu, R.I. (1983) Across-wind vibrations of structures of circular cross-section. Part 1. Development of a mathematical model for twodimensional conditions. Journal of Wind Engineering and Industrial Aerodynamics, 12, 49-73. [36] von K arm an, T. (1948) Progress in the statistical theory of turbulence, Journal of Maritime Research, 7.
Appendices
102
Appendix A
Mass Participation Factor
Figure A.1: Modal analysis and mass participation factors of spring case
103
104
Mass Participation Factor
Figure A.2: Modal analysis and mass participation factors of xed case
Appendices
105
Appendix B
MATLAB Code
106
%Determining the Natural frequencies and mode shapes of the given bridge %for fixed boundary conditions clc, clear all % section properties % span properties L_span= 166; Ls = L_span/23; %m A_span= 0.827 ; %m2 I_span =0.687 ; % m4 E_span= 6.868*10^10 ; %Pa EI_span= I_span*E_span ; m_el_span=1907000/23; % column properties A_clm= 1.5*6 ; %m2 I_clm =6*1.5^3/12; % m4 L_clm= 23/3 ; %m density= 2400 %kg/m3 E_clm= 33*10^9; %Pa EI_clm= I_clm*E_clm ; m_el_clm=density*A_clm*L_clm;
% determinin the stiffness matrix and mass matrix element =[1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 11 11 12 12 13 13 14 14 15 15 16 16 17 17 18 18 19 19 20 20 21 21 22 22 23 23 24 8 25 25 26 26 27 17 28 28 29 29 1 3; 4 ; 5; 6; 7; 8; 9; 10; 11; 12; 13 ; 14; 15 ; 16; 17; 18; 19; 20; 21; 22; 23; 24; 25; 26; 27; 28; 29; 30]; 2 ;
edof= [ 1 1 2 3 4 5 6 ; 2 4 5 6 7 8 9; 3 7 8 9 10 11 12 ; 4 10 11 12 13 14 15; 5 13 14 15 16 17 18; 6 16 17 18 19 20 21; 7 19 20 21 22 23 24; 8 22 23 24 25 26 27; 9 25 26 27 28 29 30; 10 28 29 30 31 32 33; 11 31 32 33 34 35 36; 12 34 35 36 37 38 39; 13 37 38 39 40 41 42 ; 14 40 41 42 43 44 45; 15 43 44 45 46 47 48; 16 46 47 48 49 50 51; 17 49 50 51 52 53 54; 18 52 53 54 55 56 57; 19 55 56 57 58 59 60; 20 58 59 60 61 62 63; 21 61 62 63 64 65 66 ; 22 64 65 66 67 68 69; 23 67 68 69 70 71 72; 24 22 23 24 73 74 75; 25 73 74 75 76 77 78 ; 26 76 77 78 79 80 81; 27 49 50 51 82 83 84; 28 82 83 84 85 86 87; 29 85 86 87 88 89 90]; % span stiffness Matrix Kg1 = zeros(90); ne = 23; for i = 1:ne d1=Ls^2/12; k1 = A_span*E_span/Ls; k2= 12*EI_span/Ls^3; k3= 6*EI_span/Ls^2; k4= 4*EI_span/Ls; k5= 2*EI_span/Ls; kel_span = [ k1 0 0 -k1 0 0 0 k2 k3 0 -k2 k3 0 k3 k4 0 -k3 k5 -k1 0 0 k1 0 0 0 -k2 -k3 0 k2 -k3 0 k3 k5 0 -k3 k4]; m_el_s =eye(6,6); m_el_s(3,3)=d1; m_el_s(6,6)=d1; m_span= m_el_s * m_el_span/2; Kg1(edof(i,2:end),edof(i,2:end)) = Kg1(edof(i,2:end),edof(i,2:end)) + kel_span;
end % column stiffness Matrix
Kg2 = zeros(90); ne = 29 for i = 24:ne
Le_clm=3 %m d2=Le_clm^2/12 k1 = A_clm*E_clm/Le_clm; k2= 12*EI_clm/Le_clm^3; k3= 6*EI_clm/Le_clm^2; k4= 4*EI_clm/Le_clm; k5= 2*EI_clm/Le_clm; kel_clm = [ k1 0 0 -k1 0 k2 k3 0 0 k3 k4 0 -k1 0 0 k1 0 -k2 -k3 0 0 k3 k5 0 %transformation matrix T=[0 1 0 0 0 0 -1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 -1 0 0 0 0 0 0 0 1] Kg_clm= T'*kel_clm* T;
0 -k2 -k3 0 k2 -k3
0 k3 k5 0 -k3 k4]
Kg2(edof(i,2:end),edof(i,2:end)) = Kg2(edof(i,2:end),edof(i,2:end)) + Kg_clm; end stiffness_matrix = Kg1+Kg2; K_total=Kg1+Kg2; % boundary conditions for stiffness matrix K_total(90,:)= []; K_total(:,90)= []; K_total(89,:)= []; K_total(:,89)= []; K_total(88,:)= []; K_total(:,88)= []; K_total(81,:)= []; K_total(:,81)= []; K_total(80,:)= []; K_total(:,80)= []; K_total(79,:)= []; K_total(:,79)= []; K_total(71,:)= []; K_total(:,71)= []; K_total(2,:)= []; K_total(:,2)= []; K_total(1,:)= []; K_total(:,1)= [];
stiff_M =K_total;
% Mass Matrix % span mass Matrix Kg_mass_1= zeros(90); ne = 23; for i = 1:ne d1=Ls^2/12; m_el_s =eye(6,6); m_el_s(3,3)=d1; m_el_s(6,6)=d1; m_span= m_el_s * m_el_span/2; Kg_mass_1(edof(i,2:end),edof(i,2:end)) = Kg_mass_1(edof(i,2:end),edof(i,2:end)) + m_span; end % column mass Matrix Kg_mass_2= zeros(90); ne = 29 for i = 24:ne Le_clm=19,5/3 ;%m d2=Le_clm^2/12; %transformation matrix T=[0 1 0 0 0 0 -1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 -1 0 0 0 0 0 0 0 1] Kg_clm= T'*kel_clm* T; %mass matrix column m_el_c =eye(6,6); m_el_c(3,3)=d2; m_el_c(6,6)=d2; m_column= m_el_c * m_el_clm/2; m_clm= T'*m_column* T; Kg2(edof(i,2:end),edof(i,2:end)) = Kg2(edof(i,2:end),edof(i,2:end)) + Kg_clm; Kg_mass_2(edof(i,2:end),edof(i,2:end)) = Kg_mass_2(edof(i,2:end),edof(i,2:end)) + m_clm; end mass_matrix =Kg_mass_1+ Kg_mass_2; M_total= Kg_mass_1+ Kg_mass_2; % boundary conditions for mass matrix M_total(90,:)= []; M_total(:,90)= []; M_total(89,:)= []; M_total(:,89)= []; M_total(88,:)= []; M_total(:,88)= []; M_total(81,:)= []; M_total(:,81)= [];
M_total(80,:)= []; M_total(:,80)= []; M_total(79,:)= []; M_total(:,79)= []; M_total(71,:)= []; M_total(:,71)= []; M_total(2,:)= []; M_total(:,2)= []; M_total(1,:)= []; M_total(:,1)= []; mass_M =M_total;
%obtaining the eigen frequency % AA= eigen vector & BB = eigen values [AA,BB]=eig(stiff_M,mass_M); % the frequecy for i=1:25 frequency(i,1)= sqrt(BB(i,i))/2/pi; end Max= max(abs(AA)); for i=1:81 j=1:81 fi(i,j)=AA(i,j)./Max(1,j); end mode1=[ 0; 0; fi(1:68,1);0;fi(69:75,1);0;0;0; fi(76:81,1);0;0;0] ver_mode1= [mode1(2);mode1(5);mode1(8);mode1(11);mode1(14);mode1(17) mode1(20);mode1(23);mode1(26);mode1(29);mode1(32);mode1(35) mode1(38);mode1(41);mode1(44);mode1(47);mode1(50);mode1(53) mode1(56);mode1(59);mode1(62);mode1(65);mode1(68);mode1(71)]; d=166/23 x=[0 d 2*d 3*d 4*d 5*d 6*d 7*d 8*d 9*d 10*d 11*d 12*d 13*d 14*d 15*d 16*d 17*d 18*d 19*d 20*d 21*d 22*d 23*d]; plot(x,ver_mode1','b'); xlabel('Total length of Bridge(m)') ylabel('Unit displacement') title('Mode shapes') text(98,0.7,'\leftarrow mode 1',... 'HorizontalAlignment','left') hold on
mode2=[ 0; 0; fi(1:68,2);0;fi(69:75,2);0;0;0; fi(76:81,2);0;0;0] ver_mode2= [mode2(2);mode2(5);mode2(8);mode2(11);mode2(14);mode2(17) mode2(20);mode2(23);mode2(26);mode2(29);mode2(32);mode2(35) mode2(38);mode2(41);mode2(44);mode2(47);mode2(50);mode2(53) mode2(56);mode2(59);mode2(62);mode2(65);mode2(68);mode2(71)]; plot(x,ver_mode2','g'); text(38,-0.5,'\leftarrow mode 2',... 'HorizontalAlignment','left') hold on mode3=[ 0; 0; fi(1:68,3);0;fi(69:75,3);0;0;0; fi(76:81,3);0;0;0]
ver_mode3= [mode3(2);mode3(5);mode3(8);mode3(11);mode3(14);mode3(17) mode3(20);mode3(23);mode3(26);mode3(29);mode3(32);mode3(35) mode3(38);mode3(41);mode3(44);mode3(47);mode3(50);mode3(53) mode3(56);mode3(59);mode3(62);mode3(65);mode3(68);mode3(71)]; plot(x,ver_mode3','r'); text(4,0.3,'\leftarrow mode 3',... 'HorizontalAlignment','left'); hold on; y=zeros(24); plot (x,y','k'); hold on;
% m_generalized = generalized mass matrix [e_vec, e_val]=eig(stiffness_matrix, mass_matrix);
m_generalized = e_vec'*mass_matrix*e_vec; % Again, r is the influence vector which represents the displacements of the masses % resulting from static application of a unit ground displacement. For this example, each % mass simply has the same static displacement as the ground displacement. % for i=1:90 % r_dash = eye(90); % % end for i=1:90 j=1:90 r_dash(i,j)= 1; end % the coefficient vector = L_dash
(kg) %(kg)
L_dash = e_vec'*mass_matrix*r_dash ;
% Modal participation factor for each mode =MPF for i=1:90 MPF (i) = L_dash(i,1)/m_generalized (i,i); end %effective modal mass =EMM
for i=1:90 EMM (i) = L_dash(i,1)^2./m_generalized (i,i) end % sum mass participation or sum of effective mass participation

Fulltext01 1

Uploaded by

Copyright:

Available Formats

Fulltext01 1

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fulltext01 1

Uploaded by

Copyright:

Available Formats

Wind Loads on Bridges

Analysis of a three span bridge based on theoretical methods and Eurocode 1

M. Sajad Mohammadi Rishiraj Mukherjee

7 Conclusion References Appendices A Mass Participation Factor Appendices B MATLAB Code

97 99 102 103 105 106

Figure 2.1: Wind load chain, suggested by A.G. Daveport

and which determine the habitability [9].

The atmospheric boundary layer

Figure 2.2: Roughness length z0

Mean wind velocity- wind prole(homogenous terrain)

Mean wind velocity- wind prole(homogenous terrain)

The logarithmic prole

The friction velocity u is dened by the following formula: u = (2.1)

+ 5.75a 1.88a2 1.33a3 + 0.25a4 ]

Power law prole

u = A.u v 0, 75.u w 0, 5.u

(2.10) (2.11) (2.12)

2 T u (z, ) = Eu(x, y, z, t).u(x, y, z, t + )/u (z )

Power-spectral density function

U (z )Su (z, 0) = z/6 2 (z ) 4u

The above expression is used by the Swedish annex.

Static wind load

Static wind load

Wind load on large structures

The variance of the uctuation wind load is given by:

4Fq2 2 nl1 nl2 4Fq2 2 ( , )Su (n)dn = 2 u U2 U U U

Where the aerodynamic admittance function is equal to: 2 ( 1 nl1 )= U l

r 2(1 )p (r, n, U )dr l

r1 r2 )(1 )p (r1 , r2 , n, U )dr1 dr2 l1 l2

And the gust factor is given by: = Fmax = 1 + kp 2Iu Fq kb (2.46)

Static wind load

Determination of aerodynamic admittance functions for line like structures

IR (z1 , z2 )F (z1 , z2 , t)dz2 dz1

kr (r1 , r2 )p (r1 , r2 , n, U )dr2 dr1

|IR (z1 , z2 )|dz2 dz1 )2

|IR (z1 , z2 , r1 , r2 )|dz2 dz1

Where 1 = Cr (nl1 )/U and 2 = Cr (nl2 )/U .

Dynamic response of structures to the wind load

Dynamic response of structures to the wind load

Dynamic response of structures to the wind load

The Along-wind response of blu bodies

Extreme structural response

The standard deviation of the response of the structure is obtained as follows: R =

The gust factor is obtained from the following expression: = 1 + kp 2Iu

(3.31) (3.32) (3.33) (3.34)

Dynamic response of structures to the wind load

1 Ured a = Cref a 2 Mred Ured = Mred =

(3.36) (3.37) (3.38)

Uref ne dred mg /h d2 ref 2 (z ) dz 2 ref

C (z ) d(z ) U (z ) 2 (z ) dz 2 Cref dref Uref ref

C (z ) d(z ) U (z )2 IR (z ) 2 Cref dref Uref IR,ref (z )

Background turbulence response

Ft (z, t)IR (z )dz

The non-dimensional response variation is determined by double integral as shown below:

u (rz )gb (z1 )gb (z2 )dz1 dz2

Resonance turbulence response

C (z )U (z )u(z, t)d(z )dz

Dynamic response of structures to the wind load