Professional Documents
Culture Documents
4 Schedule of classes
Date
Topic 1. Introduction 2. SDoF systems: Equation of motion and modelling 3. Free vibrations Assignment 1 Assignment 1 4. Harmonic excitation 5. Transfer functions 6. Forced vibrations (Part 1) 6. Forced vibrations (Part 2) 7. Seismic excitation (Part 1) 7. Seismic excitation (Part 2) Assignment 2 Assignment 2 8. MDoF systems: Equation of motion 9. Free vibrations 10. Damping 11. Forced vibrations 11. Forced vibrations 12. Seismic excitation (Part 1) 12. Seismic excitation (Part 2) Assignment 3 Assignment 3
11:00 - 12:30 14:30 - 16:00 16:30 - 18:00 9:00 - 10:30 11:00 - 12:30 14:30 - 16:00 16:30 - 18:00 09:00 - 10:30
2 Suggested reading
[Cho11] [CP03] Chopra A., Dynamics of Structures, Prentice Hall, Fourth Edition, 2011. Clough R., Penzien J., Dynamics of Structures, Second Edition (revised), Computer and Structures Inc., 2003.
[Hum12] Humar J.L., Dynamics of Structures. Third Edition. CRC Press, 2012.
3 Software
In the framework of the course the following software will be used by the lecturer to solve selected examples: [Map10] [Mic07] [VN12] Maplesoft: Maple 14. User Manual. 2010 Microsoft: Excel 2007. User Manual. 2007 Visual Numerics: PV Wave. User Manual. 2012
As an alternative to [VN12] and [Map10] it is recommended that students make use of the following software, or a previous version thereof, to deal with coursework: [Mat12] MathWorks: MATLAB 2012. User Manual. 2012
Page 1/2
Page 2/2
An-Najah 2013
An-Najah 2013
Table of Contents
Table of Contents...................................................................... i 1 Introduction
1.1 Goals of the course .............................................................................. 1-1 1.2 Limitations of the course..................................................................... 1-1 1.3 Topics of the course ............................................................................ 1-2 1.4 References ............................................................................................ 1-3
3.3 The logarithmic decrement .............................................................. 3-12 3.4 Friction damping ............................................................................... 3-15
4.2.1
5 Transfer Functions
5.1 Force excitation .................................................................................... 5-1
5.1.1 5.1.2 5.1.3 Comments on the amplification factor V.................................................... 5-4 Steady-state displacement quantities ........................................................ 5-8 Derivating properties of SDoF systems from harmonic vibrations....... 5-10
5.2 Force transmission (vibration isolation) ......................................... 5-12 5.3 Base excitation (vibration isolation)................................................. 5-15
5.3.1 5.3.2 5.3.3 Displacement excitation ........................................................................... 5-15 Acceleration excitation ............................................................................. 5-17 Example transmissibility by base excitation .......................................... 5-20
3 Free Vibrations
3.1 Undamped free vibrations ................................................................... 3-1
3.1.1 3.1.2 3.1.3 Formulation 1: Amplitude and phase angle............................................... 3-1 Formulation 2: Trigonometric functions .................................................... 3-3 Formulation 3: Exponential Functions ....................................................... 3-4
6 Forced Vibrations
6.1 Periodic excitation .............................................................................. 6-1
6.1.1 Steady state response due to periodic excitation..................................... 6-4
Table of Contents
Page i
Table of Contents
Page ii
An-Najah 2013
An-Najah 2013
7 Seismic Excitation
7.1 Introduction .......................................................................................... 7-1 7.2 Time-history analysis of linear SDoF systems ................................. 7-3
7.2.1 7.2.2 7.2.3 Newmarks method (see [New59]) .............................................................. 7-4 Implementation of Newmarks integration scheme within the Excel-Table SDOF_TH.xls.................................................................. 7-8 Alternative formulation of Newmarks Method. ....................................... 7-10
Table of Contents
Page iii
Table of Contents
Page iv
An-Najah 2013
An-Najah 2013
Flexibility formulation ................................................................................. 8-3 Principle of virtual work............................................................................... 8-5 Energie formulation...................................................................................... 8-5 Direct Stiffness Method............................................................................ 8-6 Change of degrees of freedom.................................................................. 8-11 Systems incorporating rigid elements with distributed mass ............... 8-14
9.3 Modal matrix and Spectral matrix ..................................................... 9-12 9.4 Properties of the eigenvectors.......................................................... 9-13
9.4.1 9.4.2 Orthogonality of eigenvectors .................................................................. 9-13 Linear independence of the eigenvectors................................................ 9-16
12 Seismic Excitation
12.1Equation of motion............................................................................. 12-1
12.1.1 Introduction................................................................................................. 12-1 12.1.2 Synchronous Ground motion.................................................................... 12-3 12.1.3 Multiple support ground motion ............................................................... 12-8
9.5 Decoupling of the equation of motion.............................................. 9-17 9.6 Free vibration response..................................................................... 9-22
9.6.1 9.6.2 Systems without damping ......................................................................... 9-22 Classically damped systems..................................................................... 9-24
12.2Time-history of the response of elastic systems .......................... 12-18 12.3Response spectrum method ........................................................... 12-23
10 Damping
10.1Free vibrations with damping ........................................................... 10-1 10.2Example .............................................................................................. 10-2
10.2.1 Non-classical damping ............................................................................ 10-3 10.2.2 Classical damping .................................................................................... 10-4
12.3.1 Definition and characteristics ................................................................. 12-23 12.3.2 Step-by-step procedure ......................................................................... 12-27
12.4Practical application of the response spectrum method to a 2-DoF system ............................................................................ 12-29
12.4.1 Dynamic properties ................................................................................. 12-29 12.4.2 Free vibrations.......................................................................................... 12-31
Table of Contents
Page v
Table of Contents
Page vi
An-Najah 2013
An-Najah 2013
12.4.3 Equation of motion in modal coordinates.............................................. 12-38 12.4.4 Response spectrum method ................................................................... 12-41 12.4.5 Response spectrum method vs. time-history analysis ........................ 12-50
14.2.1 Footbridge (Computed, without TMD) ...................................................... 14-4 14.2.2 Tuned Mass Damper (Computed) ............................................................. 14-4
14.3Test programme ................................................................................. 14-5 14.4Free decay test with locked TMD ...................................................... 14-6 14.5Sandbag test ....................................................................................... 14-8
14.5.1 Locked TMD, Excitation at midspan ....................................................... 14-9 14.5.2 Locked TMD, Excitation at quarter-point of the span .......................... 14-12 14.5.3 Free TMD: Excitation at midspan .......................................................... 14-15
14.6One person walking with 3 Hz......................................................... 14-17 14.7One person walking with 2 Hz......................................................... 14-20
14.7.1 Locked TMD (Measured) .......................................................................... 14-20 14.7.2 Locked TMD (ABAQUS-Simulation) ...................................................... 14-22 14.7.3 Free TMD .................................................................................................. 14-24 14.7.4 Remarks about One person walking with 2 Hz .................................. 14-25
Table of Contents
Page vii
Table of Contents
Page viii
An-Najah 2013
An-Najah 2013
1 Introduction
1.1 Goals of the course
Presentation of the theoretical basis and of the relevant tools; General understanding of phenomena related to structural dynamics; Focus on earthquake engineering; Development of a Dynamic Feeling; Detection of frequent dynamic problems and application of appropriate solutions.
1 Introduction
Page 1-1
1 Introduction
Page 1-2
An-Najah 2013
An-Najah 2013
1.4 References
Blank page Theory
[Bat96] [CF06] Bathe KJ: Finite Element Procedures. Prentice Hall, Upper Saddle River, 1996. Christopoulos C, Filiatrault A: "Principles of Passive Supplemental Damping and Seismic Isolation". ISBN 88-7358-0378. IUSSPress, 2006. Chopra AK: Dynamics of Structures. Fourth Edition. Prentice Hall, 2011. Clough R, Penzien J: Dynamics of Structures. Second Edition (Revised). Computer and Structures, 2003. (http://www.csiberkeley.com) Den Hartog JP: Mechanical Vibrations. Reprint of the fourth edition (1956). Dover Publications, 1985. Humar JL: Dynamics of Structures. Third Edition. CRC Press, 2012. Inman D: Engineering Vibration. Prentice Hall, 2001. Przemieniecki JS: Theory of Matrix Structural Analysis. Dover Publications, New York 1985. Weawer W, Timoshenko SP, Young DH: Vibration problems in Engineering. Fifth Edition. John Wiley & Sons, 1990.
[Cho11] [CP03]
1 Introduction
Page 1-3
1 Introduction
Page 1-4
An-Najah 2013
An-Najah 2013
The principle is based on the idea of a fictitious inertia force that is equal to the product of the mass times its acceleration, and acts in the opposite direction as the acceleration The mass is at all times in equilibrium under the resultant . force F and the inertia force T = mu y = x ( t ) + l + us + u ( t ) = x + u y (2.5) (2.6) (2.7)
= m ( x + u ) T = my
Introducing the spring force f k ( t ) = ku ( t ) and the damping ( t ) Equation (2.2) becomes: force f c ( t ) = cu ( t ) + cu ( t ) + ku ( t ) = F ( t ) mu (2.3)
To derive the equation of motion, the dynamic equilibrium for each force component is formulated. To this purpose, forces, and possibly also moments shall be decomposed into their components according to the coordinate directions.
2 Single Degree of Freedom Systems Page 2-2
Page 2-1
An-Najah 2013
An-Najah 2013
Virtual displacement = imaginary infinitesimal displacement Should best be kinematically permissible, so that unknown reaction forces do not produce work
Ai = Aa
(2.13)
k
(2.14)
Kinetic energy T (Work, that an external force needs to provide to move a mass) Deformation energy U (is determined from the work that an external force has to provide in order to generate a deformation) Potential energy of the external forces V (is determined with respect to the potential energy at the position of equilibrium) Conservation of energy theorem (Conservative systems) E = T + U + V = T o + U o + V o = cons tan t dE = 0 dt (2.15) (2.16)
O
F k = a sin ( 1 ) k a 1 k Fm = 1 l m Fp = m g
F k a cos ( 1 ) + F m l F p l sin ( 1 ) = 0
(2.20)
Page 2-3
Page 2-4
An-Najah 2013
An-Najah 2013
(2.21)
F k cos ( 1 ) a 1 k Fm = 1 l m F p sin ( 1 ) m g 1
a k ------------g -2 l ml
(2.22)
u m = 1 l
(2.27)
(2.23)
( a 1 k ) 1 a + ( 1 l m m g 1 ) 1 l = 0 (2.29)
Fpsin(1) um
(2.30)
l a 1 1
uk
sin(1) ~ 1 O cos(1) ~ 1
Page 2-5
Page 2-6
An-Najah 2013
An-Najah 2013
Energy Formulation
m (1-cos(1)) l ~ 0.5 l 12
for small angles 1 we have: 1 1 - and ----- = 1 cos ( 1 ) cos ( 1 ) = 1 ----2 2 and Equation (2.33) becomes: E pot,p = ( m g 0.5 l 1 ) Energy conservation:
1 sin(1) ~ 1 cos(1) ~ 1
2 2 2
(2.35)
(2.36)
l a
(2.37) (2.38)
Derivative of the energy with respect to time: dE = 0 dt Derivation rule: ( g f )' = ( g' f ) f' (2.39) (2.40)
(2.32) (2.33)
2 2 ( m l ) 1 1 + ( k a m g l ) 1 1 = 0
E pot,p = ( m g ) ( 1 cos ( 1 ) ) l
(2.41)
The equation of motion given by Equation (2.41) corresponds to Equations (2.21) and (2.30).
Page 2-8
An-Najah 2013
An-Najah 2013
2.3 Modelling
(2.42) (2.43) 2.3.1 Structures with concentrated mass Water tank
Tank: Mass=1000t
F(t)
F(t)
By equating KE and PE we obtain: 2 k m g l 1,max = a ------------------------------------ 1 2 ml 1,max = 1 is independent of the initial angle 1 the greater the deflection, the greater the maximum velocity.
RC Walls in the longitudinal direction
(2.44) (2.45)
al din itu n ng ctio o L ire d
Ground
k =
3EI w k = 2 ----------3 H
F(t)
12EI s k = 2 ------------3 H
+ ku = F ( t ) mu
2 Single Degree of Freedom Systems Page 2-9 2 Single Degree of Freedom Systems
(2.46)
Page 2-10
An-Najah 2013
An-Najah 2013
Ai =
0 ( EIu'' [ u'' ] ) dx
and = U u [ u'' ] = '' U
(2.53)
Transformations:
u'' = ''U u = U
(2.54)
and
(2.55)
Using Equations (2.54) and (2.55), the work A a produced by the external forces is:
u ( x, t ) = ( x ) U ( t ) ( x, t ) t ( x, t ) = mu f ( x, t )
(2.47) (2.48)
L U ) dx + L ( f U ) dx Aa = ( m U 0 0
(2.56)
L m 2 dx + L f dx = U U
0 0
(2.49)
L
Using Equations (2.54) and (2.55) the work A i produced by the internal forces is:
0
L
( t u ) dx + ( f u ) dx
0 L 0
(2.50)
Ai =
(2.57)
u ) dx + ( f u ) dx = ( mu Ai =
(2.58) (2.59)
0 ( M ) dx where:
and
= [ u'' ]
(2.51) (2.52)
Page 2-11
* * * m U +k U = F
M = EIu''
Page 2-12
An-Najah 2013
An-Najah 2013
Circular frequency
* 0 ( EI ( '' ) 2 ) dx k - = -----------------------------------= -----* L 2 m m dx 0 L
2 n
(2.60)
-> Rayleigh-Quotient Choosing the deformation figure - The accuracy of the modelling depends on the assumed deformation figure; - The best results are obtained when the deformation figure fulfills all boundary conditions; - The boundary conditions are automatically satisfied if the deformation figure corresponds to the deformed shape due to an external force; - A possible external force is the weight of the structure acting in the considered direction. Properties of the Rayleigh-Quotient - The estimated natural frequency is always larger than the exact one (Minimization of the quotient!); - Useful results can be obtained even if the assumed deformation figure is not very realistic. x - , = 1 cos ----- 2L x - 2 - cos -----'' = ----- 2L 2L (2.61)
m =
0 m 1 cos -----2L
x dx + 2 ( x = L ) M
2 L
(2.62)
+M
0
Page 2-13
Page 2-14
An-Najah 2013
An-Najah 2013
(2.63)
L
x x x + 2 cos ------ sin ------ L 4 2L 2L 1 - -------------------------------------------------------------- -- = EI ----- 2L 2 EI 3EI EI - = 3.04 ----- -------- ----= ----3 3 32 L 3 L L 3EI ----------------------------------------3 ( 0.23mL + M ) L
4
(2.65)
x ------ : ( 0 ) = 0 ? -> ( x ) = 1 cos 2L x - sin ------ : ' ( 0 ) = 0 ? -> ' ( x ) = -----2L 2L 2 x ------ cos ------ : '' ( L ) = 0 ? -> '' ( x ) = 2L 2L
0 m 1 cos -----2L
2 x dx + 2 x = L -- M + 2 ( x = L ) M 2 2 1
2 (3 8) * - mL + 1 cos -- M 1 + 1 2 M 2 m = ------------------ 4 2
Page 2-15
Page 2-16
An-Najah 2013
An-Najah 2013 2
13
(2.75) (2.76)
2 * x - 4 L ----- dx k = EI -----cos 2L 2L 0
By means of Equation (2.73) we obtain: (2.71) EI = 1.673 -----------------------8.638 10 - = 4.9170 = 1.673 ----------3 3 10 10 ML
14.9170 - = 0.783Hz f = ---- = --------------2 2
4
(2.77)
(2.78)
From Equation (2.74): 1.652 EI 1.652 8.638 104 - ----------- = ------------ -----------------------f = ------------ = 0.773Hz 3 2 ML 3 2 10 10 (2.79)
The exact first natural circular frequency of a two-mass oscillator with constant stiffness and mass is: = 3.007EI - = 1.652 ----------EI ----------------------------3 3 ( 1.102M ) L ML (2.74)
The first natural frequency of such a dynamic system can be calculated using a finite element program (e.g. SAP 2000), and it is equal to: T = 1.2946s , f = 0.772Hz (2.80)
As a numerical example, the first natural frequency of a L = 10m tall steel shape HEB360 (bending about the strong axis) featuring two masses M 1 = M 2 = 10t is calculated.
Equations (2.78), (2.79) and (2.80) are in very good accordance. The representation of the first mode shape and corresponding natural frequency obtained by means of a finite element program is shown in the next figure.
Page 2-17
Page 2-18
An-Najah 2013
An-Najah 2013
2.3.3 Damping
M = 10t
Types of damping
Damping
Internal Contact areas within the structure Relative movements between parts of the structure (Bearings, Joints, etc.)
External
Material
M = 10t
Typical values of damping in structures Material Reinforced concrete (uncraked) Reinforced concrete (craked Reinforced concrete (PT) Reinforced concrete (partially PT) Composite components Steel
Table C.1 from [Bac+97]
SAP2000 v8 - File:HEB_360 - Mode 1 Period 1.2946 seconds - KN-m Units
Damping 0.007 - 0.010 0.010 - 0.040 0.004 - 0.007 0.008 - 0.012 0.002 - 0.003 0.001 - 0.002
HEB 360
Page 2-19
Page 2-20
An-Najah 2013
An-Najah 2013
Bearings
Dissipators
Page 2-21
An-Najah 2013
An-Najah 2013
3 Free Vibrations
A structure undergoes free vibrations when it is brought out of its static equilibrium, and can then oscillate without any external dynamic excitation
n - [1/s], [Hz]: Number of revolutions per time f n = ----2 2 - [s]: T n = ----n Time required per revolution
(3.1)
Transformation of the equation of motion ( t ) + 2 u ( t ) = 0 u n (3.2) (3.3) Determination of the unknowns A and : The static equilibrium is disturbed by the initial displacement (0) = v : u ( 0 ) = u 0 and the initial velocity u 0 A = v0 2 v0 2 - , tan = ----------u 0 + ---- n u0 n (3.11) (3.10)
Visualization of the solution by means of the Excel file given on the web page of the course (SD_FV_viscous.xlsx)
3 Free Vibrations
Page 3-1
3 Free Vibrations
Page 3-2
An-Najah 2013
An-Najah 2013
3.1.2 Formulation 2: Trigonometric functions ( t ) + ku ( t ) = 0 mu Ansatz: u ( t ) = A 1 cos ( n t ) + A 2 sin ( n t ) ( t ) = A 2 cos ( t ) A 2 sin ( t ) u 1 n n 2 n n By substituting Equations (3.13) and (3.14) in (3.12): A 1 ( n m + k ) cos ( n t ) + A 2 ( n m + k ) sin ( n t ) = 0 (3.15) n m + k = 0 n = k m Natural circular frequency
2 2 2
(3.19)
(3.20) (3.21)
(3.16) (3.17)
Determination of the unknowns A 1 and A 2 : The static equilibrium is disturbed by the initial displacement (0) = v : u ( 0 ) = u 0 and the initial velocity u 0 v0 A 1 = u 0 , A 2 = ----n (3.18)
+ C2 e
i n t
(3.25)
(3.26) (3.27)
= cos ( ) + i sin ( ) , e
= cos ( ) i sin ( )
3 Free Vibrations
Page 3-3
3 Free Vibrations
Page 3-4
An-Najah 2013
An-Najah 2013
Equation (3.25) can be transformed as follows: u ( t ) = ( C 1 + C 2 ) cos ( n t ) + i ( C 1 C 2 ) sin ( n t ) u ( t ) = A 1 cos ( n t ) + A 2 sin ( n t ) Equation (3.29) corresponds to (3.13)! (3.28) (3.29)
(3.30)
(3.31)
(3.32)
( t ) = et , u ( t ) = 2 e t ,u
(3.33)
= 0
m + c + k = 0
2 c - ------1 = ------ - c 4km 2m 2m
3 Free Vibrations
Page 3-5
3 Free Vibrations
Page 3-6
An-Najah 2013
An-Najah 2013
Damping ratio
c c c= -------------= ------------- = ----2 c cr 2 km nm
(3.38)
u(t)/u0 [-]
0.5
-0.5
-1
Types of vibrations:
c<1 : = ----c cr c= 1 : = ----c cr c ->1 : = ----c cr
0.5
1.5
2.5
3.5
t/Tn [-]
Underdamped free vibrations Critically damped free vibrations Overdamped free vibrations
3 Free Vibrations
Page 3-7
3 Free Vibrations
Page 3-8
An-Najah 2013
An-Najah 2013
The determination of the unknowns A 1 and A 2 is carried out as usual by means of the initial conditions for displacement ( 0 ) = v ) obtaining: ( u ( 0 ) = u 0 ) and velocity ( u 0 (3.42) v 0 + n u 0 A 1 = u 0 , A 2 = --------------------------d 3.2.2 Formulation 1: Amplitude and phase angle (3.43) Equation (3.50) can be rewritten as the amplitude and phase angle: u ( t ) = Ae (3.44) with (3.45) A = (3.46) (3.47) v 0 + n u 0 v 0 + n u 0 2 2 u 0 + --------------------------- , tan = -------------------------- d d u0 (3.53)
n t
(3.51)
cos ( d t )
(3.52)
d = n 1 damped circular frequency = n i d The complete solution of the ODE is: u ( t ) = C1 e u(t) = e u(t) = e
( n + i d ) t i d t
The motion is a sinusoidal vibration with circular frequency d and decreasing amplitude Ae
n t
+ C2 e
( n i d ) t
n t n t
( C1 e
+ C2 e
i d t
( A 1 cos ( d t ) + A 2 sin ( d t ) )
3 Free Vibrations
Page 3-9
3 Free Vibrations
Page 3-10
An-Najah 2013
An-Najah 2013
Notes - The period of the damped vibration is longer, i.e. the vibration is slower
1 0.9
Td u0
Free vibration
u1
Displacement
0.8 0.7
Tn/Td
0.6 0.5 0.4 0.3 03 0.2 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
d = n 1 Tn T d = ----------------2 1
Time (s)
Damping ratio
(3.55)
with A =
v 0 + n u 0 2 2 u 0 + --------------------------- d
(3.54)
with
e
n ( t + T d )
- Visualization of the solution by means of the Excel file given on the web page of the course (SD_FV_viscous.xlsx)
= e
n t n T d
(3.56) (3.57)
3 Free Vibrations
Page 3-11
3 Free Vibrations
Page 3-12
An-Najah 2013
An-Najah 2013
we obtain:
u0 T 1 ---- = ---------------- = e n d T n d u1 e
Evaluation over several cycles (3.58) u0 u1 uN 1 u0 T N N n T d ----- = ---- ---- ------------ = (e n d) = e uN u1 u2 uN u0 1 - ln ---- = --N u N Halving of the amplitude 1 u0 --- ln ----N u N = ---------------------- = 2 1 --- ln ( 2 ) N 1 - --------1----------------- = ----- 2 9N 10N (3.61)
Logarithmic decrement u 0 2 - 2 (if small) - = n T d = ---------------- = ln --- u 1 2 1 The damping ratio becomes: - ---- = -----------------------(if small) 2 2 2 4 +
10 9
(3.62)
(3.59)
(3.60)
(3.63)
Logarithmic Decrement
m 1, k 1, c 1
Damping ratio
c1 1 = ------------------2 k1 m1
3 Free Vibrations
Page 3-13
3 Free Vibrations
Page 3-14
An-Najah 2013
An-Najah 2013
Displacement
5 0 -5
a)
( t ) f k ( t ) f = mu ( t ) + ku ( t ) = f mu
b)
( t ) f k ( t ) + f = mu ( t ) + ku ( t ) = f mu
-10 -15
-20 0 1 2 3 4 5 6 7 8 9 10
Time (s)
- Step 1:
(0) = 0 Initial conditions u ( 0 ) = u 0 , u A1 = u0 u , A2 = 0 u ( t ) = [ u 0 u ] cos ( n t ) + u 0 t < ----n
(3.66) (3.67)
An-Najah 2013
An-Najah 2013
- Step 2:
(0) = 0 Initial conditions u ( 0 ) = u 0 + 2u , u A 1 = u ( 0 ) + u = u 0 + 2u + u = u 0 + 3u , A 2 = 0 (3.68) u ( t ) = [ u 0 + 3u ] cos ( n t ) u 0 t < ----n
Comparison Viscous damping vs. Friction damping Free vibration: f=0.5 Hz , u0=10 , v0 = 50, uf = 1 Logarithmic decrement: (3.69)
1 2 3 4 Average U0 18.35 18.35 18.35 18.35 UN 14.35 10.35 6.35 2.35 N 1 2 3 4 0.245 0.286 0.354 0.514 [%] 3.91 4.56 5.63 8.18 5.57
- Step 3:
Initial conditions ....
Important note: The change between case a) and case b) occurs at velocity reversals. In order to avoid the build-up of inaccuracies, the displacement at velocity reversal should be identified with adequate precision (iterate!)
Displacement
Comparison:
20 15 10 5 0 -5 -10 -15 -20 0 1 2 3 4 5 6 7 8 9 10
Friction damping Viscous damping
Visualization of the solution by means of the Excel file given on the web page of the course (SD_FV_friction.xlsx) Characteristics of friction damping - Linear decrease in amplitude by 4u at each cycle - The period of the damped and of the undamped oscillator is the same: 2 T n = ----n
Time (s)
3 Free Vibrations Page 3-17 3 Free Vibrations Page 3-18
An-Najah 2013
An-Najah 2013
(4.4)
(4.5)
Complete solution: u = u p + Cu h An harmonic excitation can be described either by means of a sine function (Equation 4.1) or by means of a cosine function (Equation 4.2): + cu + ku = F sin ( t ) mu o + cu + ku = F cos ( t ) mu o (4.1) (4.2)
2 + 2 u u n + n u = f o cos ( t )
(4.6)
(4.3)
where: n : Circular frequency of the SDoF system : Circular frequency of the excitation
2 fo = Fo m = ( Fo k ) n
Page 4-1
Page 4-2
An-Najah 2013
An-Najah 2013
Complete solution of the ODE: (4.8) fo - cos ( t ) u = A 1 cos ( n t ) + A 2 sin ( n t ) + -----------------2 2 n (4.16)
(4.9) (4.10)
By means of the initial conditions given in Equation (4.7), the constants A 1 and A 2 can be calculated as follows: fo A 1 = u 0 -----------------2 2 n Denominations: - Homogeneous part of the solution: transient steady-state , v0 A 2 = ----n (4.17)
= fo
Fo fo 1 - = ---- -----------------------------A o = -----------------2 2 k 1 ( n )2 n fo - cos ( t ) u p = -----------------2 2 n Ansatz for the solution of the homogeneous ODE (see section on free vibrations) u h = B 1 cos ( n t ) + B 2 sin ( n t )
Visualization of the solution by means of the Excel file given on the web page of the course (SD_HE_cosine_viscous.xlsx)
(4.14)
(4.15)
By means of the initial conditions given in Equation (3.7), the constants A 1 and A 2 can be calculated as follows: A1 = u0
v0 fo ( n ) - ----------------------A 2 = ----n 2 2
n
Page 4-3
Page 4-4
An-Najah 2013
An-Najah 2013
4.1.1 Interpretation as a beat + 2 u = f cos ( t ) with u n o The solution is: fo - [ cos ( t ) cos ( n t ) ] u ( t ) = -----------------2 2 n and using the trigonometric identity + - t sin ------------t cos ( ) cos ( ) = 2 sin ----------- 2 2 one gets the equation n + n 2f o - sin ----------------t u ( t ) = -----------------t sin ---------------2 2 2 2 n that describes a beat with: Fundamental vibration: f G Envelope: (4.22) (4.21) (4.20) (0) = 0 u(0) = u (4.18)
Displacement
(4.19)
Time [s]
Displacement
f + fn = -----------2
f fn f U = ----------2
(4.23)
A beat is always present, but is only evident when the natural frequency of the SDoF system and the excitation frequency are close (see figures on the next page)
Time [s]
Page 4-5
Page 4-6
An-Najah 2013
An-Najah 2013
Transition to f = fn
4.1.2 Resonant excitation ( = n) + 2 u = f cos ( t ) u n o n f 2.0500 --- = --------------fn 2.0000 Ansatz for the particular solution u p = A o t sin ( n t ) = A sin ( t ) + A t cos ( t ) u p o n o n n = 2A cos ( t ) A 2 t sin ( t ) u p o n n o n n By substituting Equations (4.25) and (4.27) in (4.24):
2A o n cos ( n t ) A o n t sin ( n t ) + A o n t sin ( n t ) = f o cos ( n t )
2 2
(4.24)
(4.28) 2A o n = f o 2.0125 f--= --------------fn 2.0000 fo Fo n - = ----A o = -------- ----2 n k 2 fo - t sin ( n t ) u p = --------2 n f 2.0000 --- = --------------fn 2.0000 Resonance! Ansatz for the solution of the homogeneous ODE (see section on free vibrations) u h = B 1 cos ( n t ) + B 2 sin ( n t ) (4.32) (4.29) (4.30)
(4.31)
Page 4-7
Page 4-8
An-Najah 2013
An-Najah 2013
(4.37)
By means of the initial conditions given in Equation (4.7), the constants A 1 and A 2 can be calculated as follows: A1 = u0 , v0 A 2 = ----n (4.34)
Ansatz for particular solution u p = A 3 cos ( t ) + A 4 sin ( t ) = A sin ( t ) + A cos ( t ) u p 3 4 = A 2 cos ( t ) A 2 sin ( t ) u p 3 4 By substitution Equations (4.38) to (4.40) in (4.37): (4.38) (4.39) (4.40)
Special case u 0 = v 0 = 0 (The homogeneous part of the solution falls away) fo - t sin ( n t ) u = --------2 n Is a sinusoidal vibration with amplitude: fo -t A = --------2 n (4.36) (4.35)
(4.41) Equation (4.41) shall be true for all times t and for all constants A 3 and A 4 , therefore Equations (4.42) and (4.43) can be written as follows: ( n ) A 3 + 2 n A 4 = f o 2 n A 3 + ( n ) A 4 = 0 The solution of the system [(4.42), (4.43)] allows the calculations of the constants A 3 and A 4 as:
n 2 n - , A 4 = f o --------------------------------------------------------A 3 = f o --------------------------------------------------------2 2 2 2 2 2 2 2 ( n ) + ( 2 n ) ( n ) + ( 2 n )
2 2
(4.42) (4.43)
- The amplitude grows linearly with time (see last picture of interpretation beat); - We have A when t , i.e. after infinite time the amplitude of the vibration is infinite as well. Visualization of the solution by means of the Excel file given on the web page of the course (SD_HE_cosine_viscous.xlsx)
(4.44)
4 Response to Harmonic Excitation Page 4-9 4 Response to Harmonic Excitation Page 4-10
An-Najah 2013
An-Najah 2013
Ansatz for the solution of the homogeneous ODE (see Section 3.2 on damped free vibrations) uh = e with: d = n 1 damped circular frequency Complete solution of the ODE: u = e
n t 2 n t
( B 1 cos ( d t ) + B 2 sin ( d t ) )
(4.45)
Displacement
Total response
(4.46)
14
16
18
20
Time [s]
By means of the initial conditions of Equation (4.7), the constants A 1 and A 2 can be calculated. The calculation is laborious and should be best carried out with a mathematics program (e.g. Maple). Denominations: - Homogeneous part of the solution: - Particular part of the solution: transient steady-state
Displacement
Visualization of the solution by means of the Excel file given on the web page of the course (SD_HE_cosine_viscous.xlsx)
Time [s]
Page 4-11
Page 4-12
An-Najah 2013
An-Najah 2013
4.2.1 Resonant excitation ( = n) By substituting = n in Equation (4.44) the constants A 3 and A 4 becomes:
fo A 3 = 0 , A 4 = -----------2 2 n
- The amplitude is limited, i.e. the maximum displacement of the SDoF system is:
fo Fo u st - = -------- = ----u max = -----------2 2k 2 2 n
(4.53)
(4.48)
where u st = F o k is the static displacement. - For small damping ratios ( 0.2 ) d n and hence Equations (4.51) becomes: 1 1
2
i.e. if damping is present, the resonant excitation is not a special case any more, and the complete solution of the differential equation is: u = e
n t
Special case u 0 = v 0 = 0
A1 = 0
fo fo - = ------------------A 2 = ------------------------------2 n d 2 2 2 n 1
(4.55)
(4.56)
- After a certain time, the homogeneous part of the solution subsides and what remains is a sinusoidal oscillation of the amplitude:
fo A = -----------2 2 n
(4.57)
(4.52)
(4.58)
Page 4-13
Page 4-14
An-Najah 2013
An-Najah 2013
Dynamic amplification
50
40
= 0.01
40
30
abs(uj) / ust
umax / ust
30
0.02
20
20
10
10
0.05 0.10
0 0 5 10 15 20 25 30 35 40 45 0.20 50
Cycle
0.2
0.8
0.1
abs(uj) / umax
0.6
0.4
0.2
0 0 5 10 15 20 25 30 35 40 45 50
Cycle
Page 4-15
Page 4-16
An-Najah 2013
An-Najah 2013
5 Transfer Functions
5.1 Force excitation
The steady-state displacement of a system due to harmonic excitation is (see Section 4.2 on harmonic excitation): u p = a 1 cos ( t ) + a 2 sin ( t ) with
n 2 n - , a 2 = f o --------------------------------------------------------a 1 = f o --------------------------------------------------------2 2 2 2 2 2 2 2 ( n ) + ( 2 n ) ( n ) + ( 2 n )
2 2
The maximum dynamic amplitude u max given by Equation (5.5) can be transformed to:
u max = n 2 n 2 - + f o --------------------------------------------------------- f o -------------------------------------------------------- ( 2 2 ) 2 + ( 2 ) 2 ( 2 2 ) 2 + ( 2 ) 2
n n n n 2 2 2
(5.7)
u max = f o ( n ) + ( 2 n ) ---------------------------------------------------------------2 2 2 2 2 [ ( n ) + ( 2 n ) ]
2 2 2 2
(5.1)
(5.8)
(5.2)
(5.9)
(5.10)
(5.3)
Introducing the maximum static amplitude u o = F o k = f o 2 n the dynamic amplification factor V ( ) can be defined as: u max 1 V ( ) = ----------- = ---------------------------------------------------------------------------------uo 2 2 2 [ 1 ( n ) ] + [ 2 ( n ) ] (5.11)
(5.5)
The maximum amplification factor V ( ) occurs when its derivative, given by Equation (5.12), is equal to zero.
(5.6)
5 Transfer Functions
Page 5-1
5 Transfer Functions
Page 5-2
An-Najah 2013
An-Najah 2013
2 n [ n ( 1 2 ) ] dV = -------------------------------------------------------------------------------4 2 2 2 4 (3 2) d [ 2 ( 1 2 ) n + n ] dV 2 = 0 when: = 0 , = n 1 2 d
(5.13)
The maximum amplification factor V ( ) occurs when: 1 2 - 0.71 = n 1 2 for < -----2 and we have:
= n : = n 1 2 :
2
1 V = ----2 1 V = -----------------------2 2 1
(5.15) (5.16)
(5.17)
at n = 1 we have:
(5.19)
5 Transfer Functions Page 5-4
5 Transfer Functions
Page 5-3
An-Najah 2013
An-Najah 2013
Amplification factor
10 9
Example:
0.00 0.01 = 0.05
An excitation produces the static displacement F o cos ( t ) u st = ------------------------k and its maximum is: (5.21)
8 7 6 5 4 3 2 1 0 0 0.5
0.50
Fo u o = ----k The steady-state dynamic response of the system is: u p = u max cos ( t )
(5.22)
/n
Phase angle
180
(5.23)
therefore:
0.10 0.20 0.50 0.70
u st ----- = cos ( t ) uo
up ---- = V cos ( t ) uo
(5.24)
90
In the next plots the time histories of u st u o and u p u o are represented and compared. The phase angle is always positive and because of the minus sign in Equation (5.24) it shows how much the response to the excitation lags behind.
0 0 0.5 1 1.5 2
Z/Zn
5 Transfer Functions Page 5-5 5 Transfer Functions Page 5-6
An-Najah 2013
An-Najah 2013
5.1.2 Steady-state displacement quantities Displacement: Corresponds to Equation (5.4) up -----------= V ( ) cos ( t ) Fo k Velocity: Obtained by derivating Equation (5.25) u p -----------= V ( ) sin ( t ) Fo k u p ------------------------ sin ( t ) = V ( ) ----n ( F o k ) n (5.26) (5.25)
V t
n = 0.9
= 0.9 6.28 = 5.65 rad/s V = 3.82 = 43.45 = 0.76 rad 0.76 - = -------- t = -- 5.65 = 0.14 s
u / uo
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Time [s]
5 4 3 2 1 0 -1 -2 -3 -4 -5 0.00
n = 1.0
t
= 1.0 6.28 = 6.28 rad/s V = 5.00 = 90.00 = 1.57 rad --------- = 1.57 t = -- 6.28 = 0.25 s
(5.27)
u / uo
u p --------------------- = V v ( ) sin ( t ) with V v ( ) = ----- V ( ) (5.28) F o km n Acceleration: Obtained by derivating Equation (5.26) u 2 p -----------= V ( ) cos ( t ) Fo k
2 u p ---------------------------- cos ( t ) = V ( ) 2 2 ( F o k ) n n 2 u p - V() ------------- = V a ( ) cos ( t ) with V a ( ) = ----2 Fo m n
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Time [s]
5 4 3 2 1 0 -1 -2 -3 -4 -5 0.00
n = 1.1
V t
= 1.1 6.28 = 6.91 rad/s V = 3.28 = 133.66 = 2.33 rad
Excitation Steady-state response
(5.29)
u / uo
(5.30)
(5.31)
0.25
0.50
Time [s]
= 0.34 s
5 Transfer Functions
Page 5-7
5 Transfer Functions
Page 5-8
An-Najah 2013
An-Najah 2013
Amplification factors
10
0.00
9
0.01
8 7 6 5 4 3 2 1
Resonant displacement
2
5.1.3 Derivating properties of SDoF systems from harmonic vibrations Half-power bandwidth
10
= 0.05
V(Resonance)
= n 1 2 1 V = ---------------------2 1
1.5 2
9 8 7 6 5 4 3 2 1 0 0 0.5
V(Resonance)
0.50
0 0 10 0.5 1
/n
0.00
9
0.01
8 7 6 5 4 3 2 1 0 0 10 0.5
Resonant velocity
= n
= 0.05
= 0.05
2 a
1
0.50
1 V = ----2
1.5
/n
/n
0.00
9
(5.32) (5.33)
= 0.05
(5.34) (5.35)
1 V = ---------------------2 1
/n
5 Transfer Functions
Page 5-9
Page 5-10
An-Najah 2013
An-Najah 2013
(5.36) The mass-spring-damper system, shown here on the right, is excited by the harmonic force F ( t ) = F o cos ( t ) What is the reaction force F T ( t ) , which is introduced in the foundation?
This yield the solution for the half-power bandwidth: b a 2 = -----------------n Remarks on the frequency response curve
The natural frequency of the system can be derived from the resonant response. However, it is sometimes problematic to build the whole frequency response curve because at resonance the system could be damaged. For this reason it is often better to determine the properties of a system based on vibration decay tests (see section on free vibration) The natural frequency n can be estimated by varying the Excitation until a 90 phase shift in the response occurs. Damping can be calculated by means of Equation (5.15) as: uo 1 - --------- = -2 u max However, it is sometimes difficult to determine the static deflection u o , therefore, the definition of half-power bandwidth is used to estimate the damping. Damping can be determined from the slope of the phase angle curve using Equation (5.19).
(5.37)
The reaction force F T ( t ) results from the sum of the spring force F s and the dampers force F c (t) F T ( t ) = F s ( t ) + F c ( t ) = ku ( t ) + cu (5.38)
The steady-state deformation of the system due to harmonic excitation F ( t ) is according to Equation (5.4):
Fo u p = u max cos ( t ) with u max = u o V ( ) = ----- V ( ) k
(5.39)
By substituting Equation (5.39) and its derivative into Equation (5.38) we obtain:
Fo F T ( t ) = ----- V ( ) [ k cos ( t ) c sin ( t ) ] k
(5.40)
Page 5-12
5 Transfer Functions
Page 5-11
5 Transfer Functions
An-Najah 2013
An-Najah 2013
- cos ( t ) F T ( t ) = F o V ( ) 1 + 2 ---- n
2
7 6 5 4 3 2 1 0 0 0.5 1
(5.42)
(5.43)
where the quantity TR ( ) is called Transmissibility and it is equal to: - TR ( ) = V ( ) 1 + 2 ---- n = Special case:
1 + 4 - = 1 = --------------------TR ---- n 2
2
2
1.5 2 2.5 3 3.5 4
(5.44)
2
/n
- When n > 2 then TR < 1 : Vibration isolation - When n > 2 damping has a stiffening effect - High tuning (sub-critical excitation) - Low tuning (super-critical excitation): Pay attention to the starting phase! (5.45)
1 + [ 2 ( n ) ] -----------------------------------------------------------------------------2 2 2 [ 1 ( n ) ] + [ 2 ( n ) ]
5 Transfer Functions
Page 5-13
5 Transfer Functions
Page 5-14
An-Najah 2013
An-Najah 2013
The right hand side of the ODE (5.48) can be interpreted as an : external excitation force F ( t ) = ky + cy
F ( t ) = ky go cos ( t ) cy go sin ( t ) - sin ( t ) = ky go cos ( t ) 2 ----n - 2 = ky go 1 + 2 ----cos ( t + ) n
(5.49)
(5.50)
The mass-spring-damper system, shown here above is excited by the harmonic vertical ground displacement y g ( t ) = y go cos ( t ) (5.46)
According to Equations (5.10) and (5.11) the maximum displacement of the system due to such a force is equal to:
Fo - 2 u max = ----- V ( ) = y go 1 + 2 ----V() k n
What is the absolute vertical displacement u ( t ) of the system? The differential equation of the system is: + c ( u y ) + k( u y) = 0 mu after rearrangement: + cu + ku = ky + cy mu (5.48) (5.47)
(5.51)
5 Transfer Functions
Page 5-15
5 Transfer Functions
Page 5-16
An-Najah 2013
An-Najah 2013
The steady-state relative deformation u rel of the system due to is given by Equation (5.1): the harmonic ground acceleration y g u rel = a 1 cos ( t ) + a 2 sin ( t ) (5.59)
(5.60)
The mass-spring-damper system, shown above here, is excited by the harmonic vertical ground acceleration.
( t ) = y cos ( t ) y g go
By double derivation of Equation (5.59), the relative acceleration can be calculated as: u rel
= a 2 cos ( t ) a 2 sin ( t ) u rel 1 2
(5.53)
(5.61)
( t ) of the system? What is the absolute vertical acceleration u The differential equation of the system is:
+ c ( u y ) + k(u y) = 0 mu
(5.62)
(5.54)
after rearrangement:
+ c ( u y ) + k ( u y ) my = my mu y ) + c ( u y ) + k ( u y ) = my m(u + cu + ku = my mu rel rel rel g + cu + ku = my cos ( t ) mu rel rel rel go
By substituting the constants a 1 , a 2 and f o given by Equations (5.2) and (5.60), and after a long but simple rearrangement, the equations for the maximum absolute vertical acceleration of the system is obtained as: u max ----------= TR ( ) y go (5.63)
5 Transfer Functions
Page 5-17
5 Transfer Functions
Page 5-18
An-Najah 2013
An-Najah 2013
Additional derivation: The maximum relative displacement given by Equation (5.58) can be easily determined by means of Equations (5.10), (5.11) and (5.60) as:
u rel, max fo y go - V ( ) = -------------V() = ----2 2 n n
(5.64)
( t ) = A cos ( t ) y g 0 0
(5.65)
Natural frequency SDoF system: Excitation frequency: Excitation amplitude: f n = 0.5Hz f 0 = 2.0Hz A 0 = 10m s 2
Sought is the maximum absolute acceleration u max of the SDoF system for = 2% and for = 20% . The steady-state maximum absolute acceleration is: = 2% , 0 n = 4 :
2 TR = 0.068 and u max = 0.68m s
Is the steady-state maximum absolute acceleration really the maximum absolute acceleration or at start even larger absolute accelerations may result?
Assumptions: starting time t a = 80s , sinusoidal start function for excitation frequency and excitation amplitude. Numerical computation using Newmarks Method (see Section 7)
5 Transfer Functions Page 5-19 5 Transfer Functions Page 5-20
An-Najah 2013
An-Najah 2013
f=fn=0.5Hz
0 0 10 20 30 40 50 60 70 80 90 100
f=fn=0.5Hz
0 0 10 20 30 40 50 60 70 80 90 100
10
10
0 0 10 5 0 -5 -10 0 20
10
20
30
40
50
60
70
80
90
100
0 0 10 5 0 -5 -10
10
20
30
40
50
60
70
80
90
100
Excitation [m/s2]
10
20
30
40
50
60
70
80
90
100
Excitation [m/s2]
10
20
30
40
50
60
70
80
90
100
10
20
30
40
50 Time [s]
60
70
80
90
100
5 Transfer Functions
Page 5-21
5 Transfer Functions
Page 5-22
An-Najah 2013
An-Najah 2013
f=fn=0.5Hz
0 0 10 20 30 40 50 60 70 80 90 100
f=fn=0.5Hz
0 0 10 20 30 40 50 60 70 80 90 100
10
10
0 0 10 5 0 -5 -10 0 20
10
20
30
40
50
60
70
80
90
100
0 0 10 5 0 -5 -10
10
20
30
40
50
60
70
80
90
100
Excitation [m/s2]
10
20
30
40
50
60
70
80
90
100
Excitation [m/s2]
10
20
30
40
50
60
70
80
90
100
10
20
30
40
50 Time [s]
60
70
80
90
100
5 Transfer Functions
Page 5-23
5 Transfer Functions
Page 5-24
An-Najah 2013
An-Najah 2013
Notes
The excitation function in the starting phase has the form: ( t ) = A ( t ) cos ( ( t ) t ) y g The excitation angular frequency varies with time, and is:
(5.72)
(t) = d ((t) t ) dt
Linear variation of the excitation circular frequency 0 -t : ( t ) = -----2t a t ( t ) = 0 --- ( 0 t t a ) ta
(5.67)
(5.73)
(5.68)
Force excitation:
(5.69)
Force transmission:
Sinusoidal variation of the excitation circular frequency 2 0 ta t t - cos -- ---- -- ( t ) = ------------: ( t ) = 0 sin ( 0 t ta ) 2 t a 2 t a t
Displacement excitation:
(5.70)
Double-sinusoidal variation of the excitation circular frequency t sin --- t a t a 0 t - : ( t ) = 0 1 cos --- - 1 --------------------------- ( t ) = ---- t 2 t a
Acceleration excitation
(5.71)
Visualization of the solution by means of the Excel file given on the web page of the course (SD_HE_Starting_Phase.xlsx)
5 Transfer Functions Page 5-25 5 Transfer Functions Page 5-26
An-Najah 2013
An-Najah 2013
6 Forced Vibrations
6.1 Periodic excitation
4.0 3.5 3.0
with the fundamental frequency 2 0 = ----T0 Taking into account the orthogonality relations:
Half-sine excitation
(6.3)
T0
0 0 0
T0
for n j 0 sin ( n 0 t ) sin ( j 0 t ) dt = T 0 2 for n = j for n j 0 cos ( n 0 t ) cos ( j 0 t ) dt = T 0 2 for n = j cos ( n 0 t ) sin ( j 0 t ) dt = 0
(6.4)
2.5
T0
2.0 1.5 1.0 0.5 0.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
(6.5)
T0
(6.6)
the Fourier coefficients a n can be computed by multiplying Equation (6.2) by cos ( j 0 t ) first, and then integrating it over the period T 0 . j = 0 0 n = , , 1, 0, 1, , (6.1) 0
T0 T0
Time (s)
F ( t ) cos ( j 0 t ) dt =
T0
a 0 cos ( j 0 t ) dt
T0 T0 0
(6.7)
The function F ( t ) can be represented as a sum of several harmonic functions in the form of a Fourier series, namely: F ( t ) = a0 +
n=1 0
F ( t ) dt =
T0
a 0 dt = a 0 T 0
(6.8) (6.9)
n=1
[ an cos ( n 0 t ) + bn sin ( n 0 t ) ]
(6.2)
1 T0 - a 0 = ----F ( t ) dt T 0 0
6 Forced Vibrations
6 Forced Vibrations
Page 6-1
Page 6-2
An-Najah 2013
An-Najah 2013
j = n 0
T0
F ( t ) cos ( j 0 t ) dt =
a 0 cos ( j 0 t ) dt
(6.10)
T0 0
n=1
T0
+ cu + ku = F ( t ) mu F(t) 2 + 2 u u n + n u = --------m F ( t ) = a0 +
T0
T0 F ( t ) cos ( n 0 t ) dt = a n ----2
n=1
[ a n cos ( n 0 t ) + b n sin ( n 0 t ) ]
(6.11) (6.12)
Static Part ( a 0 )
a0 u 0 ( t ) = ---k
2- T0 F ( t ) cos ( n 0 t ) dt a n = ---- T 0 0
(6.17)
Similarly, the Fourier coefficients b n can be computed by first multiplying Equation (6.2) by sin ( j 0 t ) and then integrating it over the period T 0 . 2 T0 - b n = ----F ( t ) sin ( n 0 t ) dt T 0 0 Notes - a 0 is the mean value of the function F ( t ) - The integrals can also be calculated over the interval [ T 0 2,T 0 2] - For j = 0 no b-coefficient exists (6.13)
(6.19) The steady-state response u ( t ) of a damped SDoF system under the periodic excitation force F ( t ) is equal to the sum of the terms of the Fourier series. u ( t ) = u0 ( t ) +
n=1
un
Co sin e
(t) +
n=1
un
Sine
(t)
(6.20)
6 Forced Vibrations
Page 6-3
6 Forced Vibrations
Page 6-4
An-Najah 2013
An-Najah 2013 2
6.1.2 Half-sine A series of half-sine functions is a good model for the force that is generated by a person jumping.
4.0 3.5 3.0
(6.23)
T0 tp
Half-sine excitation
2.5 2.0
The approximation of the half-sine model for T 0 = 0.5s and t p = 0.16s by means of 6 Fourier terms is as follows:
4.0
1.5 1.0
3.0
T0
Static term (n=0) First harmonic (n=1) Second harmonic (n=2) Third harmonic (n=3) Total (6 harmonics)
0.5 0.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)
(6.21)
-2.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)
The Fourier coefficients can be calculated at the best using a mathematics program: A t p t 2A - sin ---- dt = --------a 0 = ---- tp T 0 0 with tp = ----T0 (6.22)
Note The static term a 0 = 2A = G corresponds to the weight of the person jumping.
6 Forced Vibrations
Page 6-5
6 Forced Vibrations
Page 6-6
An-Najah 2013
An-Najah 2013
Investigated cases Length [m] 26.80 19.00 15.50 13.42 12.01 10.96 Notes - When the excitation frequency f 0 is twice as large as the natural frequency f n of the beam, the magnification factor V is small. - Taking into account the higher harmonics can be important! Frequency fn [Hz] 1 2 3 4 5 6 umax [m] 0.003 0.044 0.002 0.012 0.001 0.004 V [-] 1.37 55.94 3.62 41.61 4.20 25.02
3.5 3.0
Jumping frequency: f 0 = 2Hz Period: T 0 = 0.5s Contact time: t p = 0.16s Persons weight: G = 0.70kN
2.5 2.0 1.5 1.0 0.5 0.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)
Amplitude: A = 3.44kN
6 Forced Vibrations
Page 6-7
6 Forced Vibrations
Page 6-8
An-Najah 2013
An-Najah 2013
0.0015
Static term (n=0) First harmonic (n=1)
Displacement [m] D
0.0020 0.0015 0.0010 0.0005 0 0000 0.0000 -0.0005 -0.0010 -0.0015 0 0.1 0.2 0.3 0.4 0.5 0.6
Displacement [m] D
Total (6 harmonics)
0.7
0.8
0.9
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Time (s)
Time (s)
0.0100
Displacement [m] D
Displacement [m] D
First harmonic (n=1) Second harmonic (n=2) Third harmonic (n=3) Total (6 harmonics)
0.7
0.8
0.9
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Time (s)
Time (s)
6 Forced Vibrations
Page 6-9
6 Forced Vibrations
Page 6-10
An-Najah 2013
An-Najah 2013
0.0008 0.0006
6.2.1 Step force The differential equation of an undamped SDoF System loaded with a force F 0 which is applied suddenly at the time t = 0 is: + ku = F mu 0 There is a homogeneous and a particular solution
u h = A 1 cos ( n t ) + A 2 sin ( n t ) (see free vibrations) up = F0 k
Total (6 harmonics)
Displacement [m] D
0.0004 0.0002 0.0000 -0.0002 -0.0004 -0.0006 -0.0008 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
(6.25)
(6.26) (6.27)
Time (s)
The overall solution u ( t ) = u h + u p is completely defined by the in ( 0 ) = 0 and it is: itial conditions u ( 0 ) = u
F0 u ( t ) = ----- [ 1 cos ( n t ) ] k
(6.28)
0.0040 0.0030
First harmonic (n=1) Second harmonic (n=2) Third harmonic (n=3) Total (6 harmonics)
Notes
The damped case can be solved in the exact same way. On the web page of the course there is an Excel file to illustrate this excitation. The maximum displacement of an undamped SDoF System under a step force is twice the static deflection u st = F 0 k . The deflection at the time t = of a damped SDoF System under a step force is equal to the static deflection u st = F 0 k .
Displacement [m] D
0.0020 0.0010 0.0000 -0.0010 -0.0020 -0.0030 -0.0040 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)
6 Forced Vibrations
Page 6-11
6 Forced Vibrations
Page 6-12
An-Najah 2013
An-Najah 2013
Displacement
The differential equation of an undamped SDoF system under a rectangular pulse force excitation is: mu + ku = F 0 mu + ku = 0 for t t 1 for t > t 1 (6.29)
Time (s)
Up to time t = t 1 the solution of the ODE corresponds to Equation (6.28). From time t = t 1 onwards, it is a free vibration with the following initial conditions:
F0 u ( t 1 ) = ----- [ 1 cos ( n t 1 ) ] k F0 ( t ) = ---- sin ( n t 1 ) u 1 k n
Displacement
(6.30) (6.31)
(6.32)
Time (s)
and through the initial conditions (6.30) and (6.31), the constants A 1 and A 2 can be determined.
6 Forced Vibrations Page 6-14
6 Forced Vibrations
Page 6-13
An-Najah 2013
An-Najah 2013
Short duration of excitation ( t 1 T n is small) The series expansion of sine and cosine is:
( n t1 )2 2 -+ - t 1 = 1 ----------------cos ( n t 1 ) = cos ---- Tn 2 ( n t1 )3 2 - t 1 = n t 1 + -----------------+ sin ( n t 1 ) = sin ---- Tn 6
(6.33) therefore, the maximum amplitude of a short excitation is: (6.34) v0 A = ----n Notes
(6.38)
(6.39)
sin ( n t 1 ) n t 1
(6.35)
The damped case can be solved in the exact same way. On the web page of the course there is an Excel file to illustrate this excitation. Rectangular pulse force excitation: When t 1 > T n 2 , the maximum response of the SDoF system is equal to two times the static deflection u st = F 0 k . Rectangular pulse force excitation: When t 1 > T n 2 , for some t 1 T n ratios (z.B.: 0.5, 1.5, ...) the maximum response of the SDoF system can even be as large as 4F 0 k . Rectangular pulse force excitation: try yourself using the provided Excel spreadsheet. Short excitation: The shape of the excitation has virtually no effect on the maximum response of the SDoF system. Important is the impulse. Short excitation: Equation (6.59) is exact only for t 1 T n 0 and = 0 . For all other cases, it is only an approximation, which overestimates the actual maximum deflection.
F0 2 F0 t1 ( t ) = ---- n t 1 = --------u 1 k m
(6.36)
Equation (6.36) shows, that a short excitation can be interpreted as a free vibration with initial velocity v0 = I m (6.37)
0 F ( t ) dt
t1
6 Forced Vibrations
Page 6-15
6 Forced Vibrations
Page 6-16
An-Najah 2013
An-Najah 2013
Displacement
1 0 -1 -2
Displacement
-3 -4 0 1 2 3 4 5 6 7 8 9 10
0 -0.5 0 1 2 3 4 5 6 7 8 9 10
Time (s)
Time (s)
Displacement
Displacement
2 1 0 -1 -2 -3 -4 -5 0 1 2 3 4 5 6 7 8 9 10
2 1 0 -1 -2 -3 -4 -5 0 1 2 3 4 5 6 7 8 9 10
Time (s)
Time (s)
6 Forced Vibrations
Page 6-17
6 Forced Vibrations
Page 6-18
An-Najah 2013
An-Najah 2013
Displacement
1 0 -1 -2
Displacement
1.5 1 0.5 0
-3 -4 0 1 2 3 4 5 6 7 8 9 10
-0.5 0 1 2 3 4 5 6 7 8 9 10
Time (s)
Time (s)
Displacement
Displacement
Time (s)
Time (s)
6 Forced Vibrations
Page 6-19
6 Forced Vibrations
Page 6-20
An-Najah 2013
An-Najah 2013
- Action
t 1 0.3ms is by sure much
Modelling option 1 Within a simplified modelling approach, it is assumed that the slab remains elastic during loading. Sought is the maximum deflection of the slab due to the explosion. - Simplified system
Cross-Section
shorter than the period T n = 64ms of the slab (see Equation (6.51)). Therefore, the excitation can be considered as short. - Equivalent modal SDoF system (see Section Modelling)
(6.40)
6 Forced Vibrations Page 6-21 6 Forced Vibrations Page 6-22
An-Najah 2013
An-Najah 2013
Boundary conditions:
( 0 ) = 0 , ( L ) = 0 , ' ( 0 ) = 0 , '' ( L ) = 0
k* =
(6.41)
P* =
0 ( EI ( '' )
L 2 =3.45m
1
(6.45) (6.46)
By means of the mathematics program Maple Equation (6.40) can be solved for the boundary conditions (6.41) and we get:
[ sin ( L ) + sin h ( L ) ] [ cos ( x ) cos h ( x ) ] 1.508 = sin ( x ) sin h ( x ) + ---------------------------------------------------------------------------------------------------------------- cos ( L ) cos h ( L )
L =1.55m ( p dx )
For this example, the modal properties characterizing the equivalent modal SDoF system are:
m * = 0.439 2.06 5 = 4.52t 52184 - = 43571kN/m k * = 104.37 -------------53 P * = 0.888 192000 = 170496kN = k* m* = 43571 4.52 = 98.18rad/s
(6.42) with
L = 3.927
(6.43)
\ [-]
0.6 0.4 0.2 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
T n = 2 = 0.064s
The maximum elastic deformation of the SDoF system can be calculated using the modal pulse as follows:
I * = 0.5 P * t 0 = 0.5 170496 0.3 10
3
= 25.6kNs
(6.52)
x/L [-]
And with the equations given in Section Modelling, the modal properties of the equivalent SDoF system are determined:
m* =
(6.53)
0 m
dx = 0.439mL
(6.44)
(6.54)
6 Forced Vibrations
Page 6-23
6 Forced Vibrations
Page 6-24
An-Najah 2013
An-Najah 2013
Modelling option 2 Within a simplified modelling approach, it is assumed that the slab remains elastic during loading. Sought is the maximum deflection of the slab due to the explosion. - Simplified system
[-]
0.8
0.9
x/L [-]
And with the equations given in Section Modelling, the modal properties of the equivalent SDoF system are determined:
m* =
0 m
L
dx = 0.5mL
2
0 ( EI ( '' )
L 2 =8.45m
1
P* =
L =6.55m ( p dx )
For this example, the modal properties characterizing the equivalent modal SDoF system are: (6.55)
m * = 0.5 2.06 10 = 10.3t 52184 - = 40666kN/m k * = 779.27 -------------10 3
Boundary conditions:
( 0 ) = 0 , ( L ) = 0 , '' ( 0 ) = 0 , '' ( L ) = 0
(6.56)
6 Forced Vibrations
An-Najah 2013
An-Najah 2013
T n = 2 = 0.10s
(6.64)
Modelling option 3 As a third option, the slab is modelled using the commercial finite element software SAP 2000. - Numerical Model
The maximum elastic deformation of the SDoF system can be calculated using the modal pulse as follows:
I * = 0.5 P * t 0 = 0.5 180672 0.3 10
3
= 27.1kNs
(6.65)
(6.66)
(6.67)
The distributed load q is replaced by n = 19 concentrated forces F i : 192000 - = 10105kN F i = ----------------19 The first period of the system is: T 1 = 0.100s which corresponds to Equation (6.64). (6.69) (6.68)
6 Forced Vibrations
Page 6-27
6 Forced Vibrations
Page 6-28
An-Najah 2013
An-Najah 2013
Blank Page
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
Time [s]
The effect of the higher modes can be clearly seen! Comparison System m* [t] 4.52 10.30 k* [kN/m] 43571 40666 P* [P] 0.888 0.941 T [s] 0.064 0.100 0.100 m, e [m] 0.058 0.042 0.064
6 Forced Vibrations
Page 6-29
6 Forced Vibrations
Page 6-30
An-Najah 2013
An-Najah 2013
7 Seismic Excitation
7.1 Introduction
The equation of motion for a base point excitation through an ac ( t ) can be derived from the equilibrium celeration time-history u g of forces (see Section 2.1.1) as: + cu + f (u,t) = mu mu s g (7.1)
From the previous figure it can be clearly seen that the time-history of an earthquake ground acceleration can not be described by a simple mathematical formula. Time-histories are therefore usually expressed as sequence of discrete sample values and hence Equation (7.1) must therefore be solved numerically.
2 xg [m/s2] 1 0 -1 -2
2 x [m/s2] 1 0 -1 -2 9.5
t t+t g
, u and u are motion quantities relative to the base point where u of the SDoF system, while f s (u,t) is the spring force of the system that can be linear or nonlinear in function of time and space. The , u and u for a given SDoF time-history of the motion quantities u system are calculated by solving Equation (7.1).
4
9.6
Ground acceleration
2 xg [m/s ]
9.6
4
2 x [m/s ]
2 0 -2 -4 0 10
30
40
( t ) are known The sample values of the ground acceleration u g from beginning to end of the earthquake at each increment of time t (time step). The solution strategy assumes that the motion quantities of the SDoF system at the time t are known, and that those at the time t + t can be computed. Calculations start at the time t = 0 (at which the SDoF system is subjected to known initial conditions) and are carried out time step after time step until the entire time-history of the motion quantities is computed, like e.g. the acceleration shown in the figure on page 7-1.
7 Seismic Excitation Page 7-2
7 Seismic Excitation
Page 7-1
An-Najah 2013
An-Najah 2013
7.2.1 Newmarks method (see [New59]) Incremental formulation of the equation of motion + c u + k u = m u mu g
t + t
(7.4)
t + t t u = u + u
u = u + u ,
t + t
t , u = u + u
(7.5)
In the case of a linear SDoF system Equation (7.1) becomes: + cu + ku = mu mu g (7.2) t t + t t u ( ) = 1 --( u u + u) = u + -----2 2 u () = u + u ( ) d = tu + tu + ------ ( t) u 2
t t
and by introducing the definitions of natural circular frequency n = k m and of damping ratio = c ( 2m n ) , Equation (7.1) can be rearranged as: + 2 u + u n
2 n u
(7.6)
= u g
(7.3)
(7.7)
u ( ) d = u + tu + tu + ------ ( t ) d u() = u + u 2
t t t t
(7.8)
7 Seismic Excitation
Page 7-4
An-Najah 2013
An-Najah 2013
( t ) 2 t t t u - ---------------u() = u + u ( t) + u + ----- 2 2
(7.9)
Linear Acceleration: 1 1 - , = - = -6 2
The increments of acceleration, velocity and displacement during the time step are: = u = u
t + t t + t t u u = u
t --- 0.551 T
(7.10) (7.11)
t t u - t = u + -----u u 2
It is important to note that the average acceleration-method is unconditionally stable, while the linear acceleration-method is only stable if the condition t T 0.551 is fulfilled. However, the linear acceleration-method is typically more accurate and should be preferred if there are no stability concerns. For a discussion on stability and accuracy of the Newmarks methods see e.g. [Cho11] and [Bat96]. Solution of the differential equation: Option 1 Substituting Equations (7.13) and (7.14) into Equation (7.4) gives Equation (7.15), which can be solved for the only remain : ing variable u
t - 2 = m u c tu t k tu t + tu ------( m + c t + k t ) u g 2
2
t 2 t t u - ------u = u t + u + ----- 2 2
(7.12)
Introducing the parameters and into Equations (7.11) and and u , respectively, can be generalized as follows: (7.12) for u = ( u + u ) t u t ) ------ u = u t + ( u + 2 u 2
t t 2 t
(7.13) (7.14)
(7.15)
where different values of the parameters and correspond to different assumptions regarding the variation of the acceleration within the time step: Average Acceleration: 1 1 - , = - = -4 2
t ---- T
into Equations (7.11) and (7.12) gives the increSubstituting u and of the displacement u . In conjuncments of the velocity u tion with Equation (7.5), these increments yield the dynamic response of the SDoF system at the end of the time step t + t .
7 Seismic Excitation
Page 7-5
7 Seismic Excitation
Page 7-6
An-Najah 2013
An-Najah 2013
Solution of the differential equation: Option 2 Equation (7.14) can be transformed to:
t t uu - ---- u ------- = ---------- u 2 t t 2
7.2.2 Implementation of Newmarks integration scheme within the Excel-Table SDOF_TH.xls Equation (7.15) is implemented in the Excel-Table as follows: (7.17)
( m + c t + k t ) meq
2
u da
and substituting Equation (7.17) into (7.13) we obtain: t u u = --------- + t 1 ----- u u ----- t 2
t
F(t)
dv
(7.18)
In the columns C to E the so-called predictors dd, dv and da are computed first: dd = u t + u ------2
t dv = u t t t t 2
Substituting Equations (7.17) and (7.18) into Equation (7.4) gives Equation (7.19), which can be solved for the only remaining variable u :
m m c t m c- - t k + ---------- + -------- t 1 ----- u + ----+ -------u = mu c u + ---g 2 t 2 t 2 t
c dv k dd m u g da = ------------------------------------------------------ = u meq
Substituting u into Equations (7.18) and (7.17) gives the incre and of the acceleration x . In conjuncments of the velocity u tion with Equation (7.5), these increments yield the dynamic response of the SDoF system at the end of the time step t + t . For linear systems we have: m , c and k are constant throughout the whole time-history. m in Equation (7.15), as well as k in Equation (7.20), are also constant and have to be computed only once.
7 Seismic Excitation Page 7-7
Afterwards, in the columns F to H the ground motion quantities at the time step t + t are computed by means of so-called correctors:
t + t t + t t u = u + da t u = u + dv + ( da t )
t + t
7 Seismic Excitation
u = u + dd + ( da t ) u
u
2
dd
Page 7-8
m u g
ut
t tu t + tu ------ 2 k
An-Najah 2013
An-Najah 2013
Finally, in column I the absolute acceleration u abs at the time step t + t is computed as follows:
t + t uabs
7.2.3 Alternative formulation of Newmarks Method. The formulation of the Newmarks Method presented in Section 7.2.1 corresponds to an incremental formulation. It is possible to rearrange the methodology to obtain a total formulation. The equation of motion at the time t + t can be written as: m
t + t
t + t
u+
t + t ug
Observations about the use of the Excel-Table Only the yellow cells should be modified:
The columns A and B contain the time vector and the ground ac ( t ) at intervals t ; for this ground motion the receleration u g sponse of a Single-Degree-of-Freedom (SDoF) system will be computed. To compute the response of the SDoF system for a dif ( t ) , the time and acceleration vector of the ferent ground motion u g new ground motion have to be pasted into columns A and B. ( t ) , the response of a linear SDoF For a given ground motion u system is only dependent on its period T = 2 n and its damping . For this reason, the period T and the damping can be chosen freely in the Excel-Table.
g
u+c
t + t
u+k
t + t
u = m
t + t ug
(7.21)
where
t + t t + t t u = u + u t u = u + u
(7.22) (7.23)
and u given by Equations (7.17) Using the expressions for u and (7.18), the acceleration and the velocity at the time t + t can be written as:
t + t t t 1 t + t 1 t 1 - u ----- 1 u u = ----------( u u ) -------2 t 2 t t t + t t t -( - u + t 1 ----- u u = -------u u ) + 1 + - t 2
The mass m is only used to define the actual stiffness of the SDoF 2 system k = m n and to compute from it the correct spring force f s = k u . However, f s is not needed in any of the presented plots, hence the defaults value m = 1 can be kept for all computations. In the field Number of periods (cell V19) one can enter the number of periods T i for which the response of the SDoF is to be computed in order to draw the corresponding response spectra.
(7.24)
t + t
(7.25)
The response spectra are computed if the button compute response spectra is pressed. The macro pastes the different periods T i into cell S3, computes the response of the SDoF system, reads the maximum response quantities from the cells F6, G6, H6 and I6 and writes these value into the columns L to P.
7 Seismic Excitation Page 7-9
Introducing Equations (7.24) and (7.25) into Equation (7.21) and t + t solving for the only unknown u we obtain: k
t + t
u =
t + t
(7.26)
where:
7 Seismic Excitation
Page 7-10
An-Najah 2013
An-Najah 2013
k = k + a1
t + t
(7.27)
t + t ug t t t + a1 u + a2 u + a3 u
p = m
(7.28) (7.29)
(7.30)
(7.31)
This formulation corresponds to the implementation of Newmarks method presented in [Cho11]. Strength f y of the nonlinear SDoF system f el k el u el - = ----------------f y = ----Ry Ry
R y = force reduction factor f el = maximum spring force f s that a linear SDoF system of the same period T and damping would experience if submitted to the same ground motion u g
(7.32)
An-Najah 2013
An-Najah 2013
7.3.1 Equation of motion of nonlinear SDoF systems In the equation of motion for a base point excitation through an ( t ) acceleration time-history u g + cu + f (u,t) = mu mu s g (7.34)
7.3.2 Hysteretic rules The next figure shows typical hysteretic rules (or models) for nonlinear SDoF systems.
of a nonlinear SDoF system, the spring force f s (u,t) is no longer constant and varies in function of time and location. Most structural components are characterised by a continuously curved force-deformation relationship like the one shown by means of a thin line on the right of the figure on page 7-12, which however is often approximated by a bilinear curve (thick line in the same figure). When the loading of the nonlinear SDoF system is cyclic, then f s is no longer an unambiguous function of the location u and also for this reason Equation (7.34) shall be solved incrementally. For this reason f s (u,t) must be described in such a way, that starting from a known spring force f s (u,t) at the time t , the still unknown spring force f s (x + u,t + t) at the time t + t can easily be computed. This description of the cyclic force-deformation relationship is known as hysteretic rule. In the literature many different hysteretic rules for nonlinear SDoF systems are available (See e.g. [Saa91]). In the following section a few hysteretic rules are presented and discussed.
7 Seismic Excitation
Page 7-13
7 Seismic Excitation
Page 7-14
An-Najah 2013
An-Najah 2013
In the following the often used Takeda hysteretic model is discussed in some detail. The Takeda model was first described in [TSN70] and later modified by various authors. The assumed force-deformation relationship shown in the following figure was derived from the moment-curvature relationship described in [AP88].
Large amplitude cycles
+ p
Unloading occurs along a straight line with stiffness k u . This unloading stiffness is computed by means of Equation (7.35) as a function of the elastic stiffness k el and taking into softening effects in proportion of the previously reached maximum displacement ductility . The parameter controls the unloading stiffness reduction and varies from structural element to structural element.
+ = k ( max { + } ) ku el
fs
k pl =r o kel >0 u rev k el uy ku fy A+
+ p
fs
A+ C+
+ p B+
- = k ( max { - } ) ku el
(7.35)
<0
+ ku
+ ku
u rev
u
ku <0 Xxp
u rec
u rev + u rec
Reloading follows a straight line which is defined by the force reversal point (u rev,0) and the point A. The location of point A is determined according to the figure on the previous page as a function of the last reversal point B, the plastic strain p and the damage influence parameter . The parameter allows taking into account softening effects occurring during the reloading phase. Again, in the case of RC walls meaningful parameters and lay in the following ranges: = 0.2 0.6 and = 0.0 0.3 . These rules, which are valid for cycles with large amplitude, are typically based on observations of physical phenomena made during experiments. On the other hand, rules for small amplitude cycles are based on engineering considerations rather than on exact observations. They are designed to provide reasonable hysteresis lops during an earthquake time-history, thus avoiding clearly incorrect behaviours like e.g. negative stiffnesses. The rules for small amplitude cycles are shown on the right of the figure on the previous page.
7 Seismic Excitation Page 7-16
C>0 B- Xxp
The initial loading follows the bilinear force-deformation relationship for monotonic loading mentioned in the previous section. The exact definition of this so-called skeleton curve depends on the structural element at hand. For example, in the case of Reinforced Concrete (RC) structural walls the elastic stiffness k el corresponds to 20 to 30% of the uncracked stiffness, while the plastic stiffness k pl = r o k el is approximated assuming an hardening factor r o = 0.01 0.05 .
7 Seismic Excitation
Page 7-15
An-Najah 2013
An-Najah 2013
If a reloading phase starts from a force reversal point (u rev,0) lay+ ing between the two extreme force reversal point (u rec,0) and (u rec,0) , then reloading does no longer occurs towards point A, but towards a newly calculated point X , which lies between points A and B. The position of point X is calculated using the + auxiliary variables x and x defined in Equation (7.36). u rev u rec + - p , x = -----------------------+ u rec u rec
+ +
7.3.3 Newmarks method for inelastic systems The Newmarks numerical method discussed in Sections 7.2.1 to 7.2.3 can be easily modified for application to nonlinear systems. The following modifications are required: The mass m and the damping c are typically constant throughout the whole time-history. , re The stiffness k changes during the time-history, hence m spectively k , are no longer constant. If the stiffness changes within the time step iterations are needed (e.g. Newton-Raphson). For nonlinear systems the second solution strategy presented in Section 7.2.1 (Option 2) has the advantage that the factors m- c m- - and b = ---- t 1 ----c a = -------+ --- t 2 2 on the RHS of Equation (7.19) are constant throughout the whole time-history and do not need to be recomputed at every time step. Remark For the actual implementation of the nonlinear version of Newmarks time stepping method, it is suggested to use the formulation presented in Section 7.4.4 in conjunction with the NewtonRaphson algorithm described in Section 7.4.3.
(7.36)
When a load reversal occurs before point X is reached, a new point C is defined as a temporary maximum and minimum. The reloading in the subsequent cycles, which are smaller than the temporary maximum and minimum is then always in the direction of point C. These rules for cycles with a small amplitude are a simplification of those described in [AP88], however they lead to very satisfactory results and can be programmed very easily and efficiently.
7 Seismic Excitation
Page 7-17
7 Seismic Excitation
Page 7-18
An-Najah 2013
An-Najah 2013
=28.1cm
40 20 0 20 40 20 0 Vel=39.5kN 20
20 0
=25.9cm
10
20 0
=21.6cm
10
20 0 10
=21.2cm
20 Time [s]
7 Seismic Excitation
7 Seismic Excitation
Page 7-20
An-Najah 2013
An-Najah 2013
=11.3cm
10 20 30
200 10 0
Vel=246.7kN 10
=1.5cm
=7.7cm
200
10 0 10 20 30
Vy=123.4kN 10
=1.2cm
Displacement [cm]
10
=1.5cm
10 0
=9.4cm
10 20 30
Vy=61.7kN 10
Displacement [cm]
10
=2.2cm
10 0
=10.4cm
10 20 Time [s] 30
200 10
7 Seismic Excitation
Page 7-21
7 Seismic Excitation
Page 7-22
An-Najah 2013
An-Najah 2013
7.3.5 Example 2: A 3-storey RC wall As a second example, the behaviour of the RC wall WDH4 presented in Section 7.6.3 is simulated. Wall WDH4 is actually a 3DoF system and its behaviour is simulated by means of an equivalent SDoF system. For this reason the relative displacement of the SDoF system shall be multiplied by the participation factor = 1.291 to obtain an estimation of top displacement of WDH4. To simulate the behaviour of WDH4 a nonlinear SDoF system with Takeda hysteretic model is used. The parameter used to characterise the SDoF system are: = 5% , f n = 1.2Hz , R y = 2.5 , r o = 0.03 , = 0.2 , = 0.15 (7.38) where is the damping rate, f n is the natural frequency of the SDoF system for elastic deformations (i.e. with k = k el ), and R y is the force reduction factor. The parameters given in (7.38) were subsequently adjusted to obtain the best possible match between the simulation and the experiment. In a first phase, a calculation using a linear SDoF system is performed. The latter has the same damping rate and natural frequency as the Takeda SDoF system and allows an estimation of the maximum elastic spring force f el and of the maximum elastic deformation u el . The yield force f y and the yield displacement u y of the Takeda SDoF system are then estimated using the force reduction factor R y as follows: f el u el R y = ----- = -----fy uy
7 Seismic Excitation
Time-history of the top displacement The time-history of the top displacement (see below) shows that: (i) plastic phenomena affect the behaviour of the wall WDH4 significantly, and that (ii) the Takeda-SDoF System is able can describe the global behaviour of the wall WDH4 quite accurately.
80
13
14
15
16
17
18
Time [s]
80
(7.39)
13
14
15
16
17
18
Time [s]
Page 7-23
7 Seismic Excitation
Page 7-24
An-Najah 2013
An-Najah 2013
Hysteretic behaviour of the nonlinear SDoF System Force-deformation relationship of the Takeda-SDoF System subjected to the same ground motion as Wall WDH4. In both diagrams the same curve is plotted: On the left in absolute units and on the right in normalised units.
Normalised spring force fs/fy [kN]
80 60
1.0 0.5 0.0 -0.5 -1.0
In this equation F ( t ) is the time-dependent vector of the internal forces of all DoFs of the structure and R ( t ) the time-dependent vector of the external forces. The vector R ( t ) depends on the problem analysed and is known. 7.4.2 Nonlinear static analysis For linear-elastic systems the internal forces can be computed by means of Equation (7.41): F = KU (7.41)
where U is the vector of the displacements of the DoF and K is the stiffness matrix of the structure. Equation (7.40) can therefore be rewritten as KU = R (7.42)
7 Seismic Excitation
Page 7-25
7 Seismic Excitation
Page 7-26
An-Najah 2013
An-Najah 2013
In Equation (7.42) R is known and K is also known, therefore the unknown vector U can be computed by means of Equation (7.43): U = K R
1
The actual solution of the equilibrium conditions of Equation (7.40) is often obtained by the numerical method for the iterative solution of nonlinear equations that was originally developed by Newton. In the next section the so-called Newton-Raphson Algorithm for the solution of Equation (7.40) will be discussed. 7.4.3 The Newton-Raphson Algorithm The Newton-Raphson Algorithm allows the solution of the loading of nonlinear springs with the following equilibrium condition: F(U(t) ) = R(t) (7.47)
(7.43)
The equilibrium condition of Equation (7.40) can only be solved for linear-elastic systems by means of Equation (7.43). For inelastic systems, due to successive yielding of the structure, the stiffness matrix K is not constant over the course of the loading. For this reason Equation (7.40) must be solved in increments (=small load steps) and iteratively. The approach is as follows: The nodal displacements U at the time t are known from the previous load step; The nodal displacements U at the end of the load step t are determined by means of n iterations of Equations (7.44) and (7.45).
t + t t + t t
F ( U ( t ) ) = R ( t ) represents the internal spring force, which is a given nonlinear function of U ( t ) . The external force R ( t ) is a function of the time t . For a system with 1 DOF the solution method of the NewtonRaphson Algorithm can be illustrated by the figure on page 7-29. The algorithm consists of the following steps:
KT U = U = U
i1
t + t
i1
(7.44) (7.45)
t + t i
t + t i 1
+ U
0) Up to time step t a solution was obtained and at time step t t t the system is in equilibrium with ( U, R) ; 1) The initial conditions at the beginning of the iteration are determined. The iteration commences with the nodal displacement, the tangent stiffness and the internal force that have resulted at the end of the previous time step t . The external t + t 0 initial loading increment R within the time step is determined by Equation (7.51).
7 Seismic Excitation Page 7-28
where:
t + t
i1
t + t
t + t i 1
(7.46)
7 Seismic Excitation
Page 7-27
R
t+tR
t+tR2
7 Seismic Excitation
t+tU1 t+tU3 t+tU2 t+tR1 t+tK 1 T t+tK 2 T t+tR0 tR t+tK 0 T tR=t+tF0 t+tF1 t+tF2 t+tFn ~ t+tF = t+tR tU=t+tU0 t+tU1 t+tU2 t+tU=t+tUn
An-Najah 2013
Page 7-29
R
t+tR
t+tU1
t+tU2
t+tU3
t+tR2
7 Seismic Excitation
tR
t+tR1
t+tK 0 T
tR=t+tF0
t+tF1
t+tF2
tU=t+tU0
t+tU1 t+tU2
t+tU=t+tUn
U
Page 7-30 Modified Newton-Raphson algorithm to solve systems of nonlinear equations (according to [AEM86])
An-Najah 2013
An-Najah 2013
An-Najah 2013
t + t 0
U = U = KT
t t
Comments on the Newton-Raphson Algorithm The Newton-Raphson Algorithm for systems with several or many DoFs follows exactly the same procedure as the algorithm for SDoF systems. Only difference: Scalar values are replaced by the corresponding vectorial quantities. Apart from the Newton-Raphson Approach (Full NewtonRaphson iteration) the Modified Newton-Raphson iteration is often applied. This algorithm is illustrated on page 7-30.
Unlike in the Newton-Raphson Algorithm, in the Modified NewtonRaphson Algorithm the tangent stiffness matrix K T is updated only at the beginning of the time step and is kept constant over all the iterations within this time step. U more iterations are re To reach the target displacement quired for the Modified Newton-Raphson Algorithm than for the Full Newton-Raphson Algorithm. However, these can be computed more quickly since assembling the tangent stiffness matrix K T T (Step 6) and in particular its LDL -decomposition (Step 2) are only required at the beginning of a time step and not at each iteration within the time step. This is particularly advantageous for systems with many DoFs.
t + t
t + t 0 KT t + t 0 t + t
F = F R =
0 t + t
t + t 0
(7.51)
i
2) Computation of the -displacement increment U by means of Equation (7.52) (i starts from 1). t + t i 1 For systems with more DoFs, KT is a matrix (tangent stiffness matrix) and Equation (7.52) is best solved by means t + t i 1 of a LDL T -decomposition of the matrix KT .
t + t i 1 i KT U
ith
t + t
i1 t + t i
(7.52)
U at the end of the ith-
U =
t + t i 1
+ U
i t + t i
(7.53)
F and the new external
R =
t + t
t + t i
(7.54)
5) If U i and/or R are so small that they can be neglected: Continue with Step 7; 6) Determine the new tangent stiffness Step 2;
t + t i KT
t + t
In most FE-analysis programs both Newton-Raphson Algorithms as well as other algorithms are typically combined in a general solver in order to obtain a successful convergence of the iteration process for many structural analysis problems. Other algorithms for the solution of the equilibrium conditions can be found in Chapters 8 (static analysis) and 9 (dynamic analysis) of [Bat96].
7) If the analysis time has not yet ended, start a new time step and start again the procedure at Step 1.
7 Seismic Excitation Page 7-31 7 Seismic Excitation Page 7-32
An-Najah 2013
An-Najah 2013
Convergence criteria A in-depth discussion of the convergence criteria can be found in [Bat96] and [AEM86]. This section provides only a short overview. In Step 5) criteria are required in order to decide whether convergence of the iteration was obtained. Possible convergence criteria can be based on displacements, force or energy considerations. Since within the time step the unknown target displacement t + t U needs to be determined, it makes sense to prescribe that the target displacement is reached within a certain tolerance interval. For this reason a possible displacement criterion for the convergence is:
U ----------------- D t + t U
i
For this reason the displacement criterion is typically used in conjunction with other convergence criteria.
A force criterion, which checks the residual forces, is given in Equation (7.56).
R F-------------------------------------- F t + t t R F
t + t t + t i
(7.56)
nitude of the residual force after the ith-iteration against the first load increment of the time step.
As for the displacement criterion, this force criterion should not be applied on its own because in some cases the target displacement t + t U may not have been reached. This may happen for systems with small post-yield stiffness.
(7.55)
The energy criterion in Equation (7.57) has the advantage that it checks the convergence of the displacements and the forces simultaneously.
(U ) ( R F ) ---------------------------------------------------------------- E t + t T t 1 (U ) ( R F)
i T t + t t + t i 1
(7.57)
work of the residual forces of the ith-iteration against the work of the residual forces of the first load increment of the time step. Choosing the tolerances D , F or E too large, can yield wrong results, which can lead to the divergence of the solution in the following load steps.
7 Seismic Excitation Page 7-34
An-Najah 2013
An-Najah 2013
Choosing the tolerances D , F or E too small, results in an unnecessary increase of the required iteration steps. The increased accuracy is typically not useful. For numerical reasons it can also happen that too small convergence tolerances do not allow to reach convergence at all. For strongly inelastic systems it is recommended to check the sensitivity of the results to the chosen convergence criteria and the chosen tolerances. 7.4.4 Nonlinear dynamic analyses Similar to Equation (7.44) the equilibrium condition for nonlinear dynamic analyses is:
M
t + t i
The numerical time integration of Equation (7.58) is often performed by means of the Newmarks Algorithms ([New 59]). According to these algorithms the displacement vector at the time t + t is estimated as follows:
t + t t t t + t t t U = U+ U t + [ ( 1 2 ) U + 2 U ] ------2 2
(7.61) (7.62)
t + t
= tU + [ ( 1 ) tU + t + tU ] t U
(7.63)
U +C
t + t i
U +(
t + t i 1
t + t
i1 i KT U )
t + t
(7.58) (7.59)
(7.64)
t + t i
U =
t + t i 1
+ U
For base excitation by means of ground accelerations the vector of the external forces is computed according to Equation (7.60).
t + t
R = M 1
t + t
ag
(7.60)
The expressions for the displacement, the velocity and the acceleration at the time t + t from Equations (7.59), (7.64) and (7.63) can be substituted into the differential equation of motion (7.58) which can then be solved for the only remaining unknown U i :
1 t + tKi 1 + ---------- M + -------- C Ui = T 2 t t
t + t
where M is the mass matrix of the structure, 1 the norm vector with entries of unity for all DoFs in the direction of the excitation t + t and ag the ground acceleration at the time t + t . For this type of excitation the differential equation of motion (7.58) has to be integrated numerically and due to the inelastic behaviour of the system the equation must be solved iteratively and incrementally.
7 Seismic Excitation Page 7-35
t + t i 1
(7.65)
7 Seismic Excitation Page 7-36
An-Najah 2013
An-Najah 2013
K T U =
i1
t + t
i1
7.4.5 Comments on the solution algorithms for nonlinear analysis problems (7.66) Contrary to the analysis of elastic systems, the analysis of inelastic systems is often interrupted before the targeted load or deformation state is reached.
This situation arises if in one of the time steps convergence cannot be reached.
Equation (7.66) corresponds exactly to Equation (7.44) and is also solved iteratively by the Newton-Raphson Algorithm. When dynamic analyses are carried out, typical convergence criteria also consider the inertia forces and, if present, damping forces. Possible, often used convergence criteria are:
t + t i 1 t + t i 1 t + t t + t i 1 R F M U C U ------------------------------------------------------------------------------------------------------------------ F RNORM
(7.67) (7.68)
(U ) ( R F M U C U ) ------------------------------------------------------------------------------------------------------------------------------------ E t t t 1 T t + t (U ) ( R F M U C U)
i T t + t
t + t i 1
t + t i 1
t + t i 1
with
RNORM =
m ij g
(7.69)
As alternative, depending at which point during the iteration process convergence is checked, both criteria can be rewritten as:
t + t i t + t i t + t t + t i R F M U C U--------------------------------------------------------------------------------------------- F RNORM
(7.70) (7.71)
Apart from the algorithms by Newmark, many other algorithms have been developed for solving the differential equation of motion (7.58) (e.g. Houbolt Method, Wilson Method, Method). These methods are described in detail in [Bat96]. The -Method allows to introduce numerical damping, which can be useful. The -Method is described in detail in [HHT77]. Choice of the time step t for static analyses:
For static analyses the time t has no physical meaning. For this reason the size of the time step t can be chosen almost arbitrarily. As long as the algorithm converges and the variation of the external loads is captured correctly, the size of the time step t has only a minor influence on the results.
7 Seismic Excitation Page 7-38
C ) (U ) ( R F M U U ---------------------------------------------------------------------------------------------------------------- E t t t 1 T t + t (U ) ( R FMU CU )
i T t + t
t + t i
t + t i
t + t i
As for static analyses, different convergence criteria exist also for dynamic analyses and a discussion of these can also be found in [Bat96].
7 Seismic Excitation Page 7-37
An-Najah 2013
An-Najah 2013
However, when geometric nonlinearities are considered or when the constitutive laws are a function of the strain history, care should be taken also for static analyses when choosing the size of the time step. The time step size influences the convergence of the algorithm: For small t the algorithm converges more quickly; however, more steps are required. Time steps of a variable size can be advantageous. If the system is elastic or almost elastic, large times steps can be chosen. If the system is close to its capacity, small time steps should be chosen. Certain analysis programs (see for example [HKS03]) determine the time step size within chosen limits as a function of the convergence and the number of required iterations.
An in-depth discussion on the accuracy of time-integration methods and on so-called conditionally stable integration methods can be found in [Bat96]. Example: The time-integration method by Newmark with = 1 2 and = 1 6 (linear variation of the acceleration over the length of the time step, see Section 7.2.1) is only stable if the criterion given in Equation (7.73) is met for all natural periods T n of the system. For systems with many DoFs, higher modes can be especially problematic and a very short time step is generally required when this time-integration method is used. t ---- 0.551 Tn
(7.73)
For this reason the unconditionally stable time-integration algorithm by Newmark with = 1 2 and = 1 4 is often used in seismic engineering.
(7.72)
Certain time-integration algorithms can become instable if the time step size t is too large.
7 Seismic Excitation
Page 7-39
7 Seismic Excitation
Page 7-40
An-Najah 2013
An-Najah 2013
7.4.6 Simplified iteration procedure for SDoF systems with idealised rule-based force-deformation relationships In the case that the hysteretic behaviour is described by a sequence of straight lines, a so called idealized rule-based forcedeformation relationships like the Takeda model presented in Section 7.3.2, it is possible to avoid implementing a NewtonRaphson Iteration strategy. In this case adjustments are needed in the case of:
Stiffness change during loading Velocity reversal Transition between unloading an reloading.
In first case a secant stiffness can be iteratively computed until the target point lays on the backbone curve, while in the second and third case it is often enough to reduce the size of the time step to limit error. These strategies are shown in the following figure:
Response spectra are used to represent the seismic demand on structures due to a ground motion record and design spectra are used for the seismic design of structures. Response spectra shall be computed for all periods and damping rates likely to be found in structures. Unless specified otherwise, the response spectra presented in the following belongs to the north-south component of the May 18, 1940 El Centro Earthquake (see [Cho11]). Additional ground motion records can be downloaded for free from: 1) http://db.cosmos-eq.org/scripts/default.plx 2) http://peer.berkeley.edu/nga/
7 Seismic Excitation
Page 7-41
7 Seismic Excitation
Page 7-42
An-Najah 2013
An-Najah 2013
0.10
1.00
10.00
100.0
+ 2 u + 2u = 0 u a ( t ) = u ( t ) + u ( t ) = u ( t ) u = 0u a g g
2.0 0.0 2.0 0 0.4 vg [m/s]
+ 2 u + 2u = 0 u a ( t ) = u ( t ) u ( t ) = u ( t ) u g g
ag [m/s2]
= = = = =
0% 2% 5% 10% 20%
0 0.01 0.6 Relative displacement [m] 0.5 0.4 0.3 0.2 0.1 0.01
vg,max = 36.1cm/s
0.10
1.00
10.00
100.0
dg [m]
= = = = =
0% 2% 5% 10% 20%
dg,max = 21.1cm
0.1 0.0
Baseline correction!
0.10
10.00
100.0
0.1 0
10 Time [s]
20
30
7 Seismic Excitation
Page 7-43
7 Seismic Excitation
Page 7-44
An-Najah 2013
An-Najah 2013
7.5.2 Pseudo response quantities Pseudo-velocity S pv S pv = S d - S pv has units of a velocity - S pv is related to the peak value of the strain energy E s kS d k ( S pv ) mS pv - = -------------------------- = -----------E s = -------2 2 2 Pseudo-acceleration S pa = 2 S d - S pa has units of an acceleration - S pa is related to the peak value of the base shear V V = kS d = k ( S pa 2 ) = mS pa (7.77) (7.76)
2 2 2
(7.74)
Acceleration Pseudo-acceleration = 5%
= 50%
(7.75)
0.10
1.00
10.00
100.00
Velocity Pseudo-velocity = 5%
0.5
= 50%
0.0 0.01
0.10
10.00
100.00
For = 0 are acceleration and pseudo-acceleration identical. For T the pseudo-velocity tends to zero. Pseudo-velocity and pseudo-acceleration match well the true motion of a SDOF system with < 20% and T < 1s
7 Seismic Excitation
Page 7-45
7 Seismic Excitation
Page 7-46
An-Najah 2013
An-Najah 2013
Pseudo-acceleration Pseudo-velocity
=0
S pa S pv = ------
log ( S pv ) = log ( ) + log ( S d ) log ( S pv ) = log ( f ) + log ( 2 ) + log ( S d ) log ( S pv ) = log ( T ) + log ( 2 ) + log ( S d ) log ( S pv ) = log ( ) + log ( S pa ) log ( S pv ) = log ( f ) log ( 2 ) + log ( S pa ) log ( S pv ) = log ( T ) log ( 2 ) + log ( S pa )
0.0
-1 x Displacement
1.0 0 1.0 5 10 15
Pseudo-acceleration Pseudo-velocity
>0
Pseudovelocity Spv [cm/s]
100
Spv = 71 cm/s
10 00 0
0.0
44 7
cm /s 2
10
pa
10
-1 x Displacement
1.0 0 5 Time [s] 10 15
u(t) = A(t) At u max : u a = A however A < A max Shift of the location of the maxima through damping
T=1s
0.1 1.0 Period [s] 10.0
7 Seismic Excitation
Page 7-47
7 Seismic Excitation
0.
is
pl
ac
( t ) = 2 u ( t ) 2 u (t) u a
(7.78)
em
en
tS
[c
10
10
Sd
00 Ps eu do
= . 11 2 cm
10 0 a cc .S
pa
10 /s
2
[c m ]
Page 7-48
An-Najah 2013
An-Najah 2013
100
vg = 36.1 cm/s
10
Spa / ag [-]
T [s]
10 00
10
10 0 Ps eu
/s 2
10
dg = .1 21 cm
cm
31
tS
em
en
pl
ac
0.1
10.0
Response spectra typically show spectral regions where the response is sensitive to different motion quantities, i.e. they show an acceleration sensitive region (small periods), a displacement sensitive region (large periods) and a velocity sensitive region laying in between.
7 Seismic Excitation Page 7-49 7 Seismic Excitation
Spv / vg [-]
0.
is
T [s]
Page 7-50
/d /a
g
[c
[-] Sp
.S
pa
do a cc 10 /s
2
[c m ]
[-]
An-Najah 2013
An-Najah 2013
Zone 3b, = 5%
5 Pseudoacceleration Spa [m/s2]
100
aa g
vvg
D
dg d
10 00 0
C
vg
10
10
B
Ps eu
10
[c
10 00 10 0 do a cc .S
dg
Ground Class A Ground Class B Ground Class C Ground Class D Ground Class E
0.10 Period [s] 1.00 10.00
is
pl
ac
em
en
0 0.01
tS
0.
pa
10 /s
2
[c m ]
TA=1/33s
0.1
TB=1/8s
1.0 Period [s]
Ground class B: deposit of extensive cemented gravel and sand with a thickness >30m. Ground class C: deposits of normally consolidated and uncemented gravel and sand with a thickness >30m. Ground class D: deposits of unconsolidated fine sand, silt and clay with a thickness >30m. Ground class E: alluvial surface layer of GC C or D, with a thickness of 5 to 30m above a layer of GC A or B. Ground class F: deposits of structurally-sensitive and organic deposits with a thickness >10m.
7 Seismic Excitation Page 7-52
10.0
TF=33s
Median(50%) a v d 2.74 2.03 1.63 2.12 1.65 1.39 1.64 1.37 1.20 1.17 1.08 1.01
One sigma (84%) a v d 3.66 2.92 2.42 2.71 2.30 2.01 1.99 1.84 1.69 1.26 1.37 1.38
Page 7-51
An-Najah 2013
An-Najah 2013
Zone 3b, = 5%
20 Displacement Sd [cm]
15
Ground Class A Ground Class B Ground Class C Ground Class D Ground Class E
Ground Class A Ground Class B Ground Class C Ground Class D Ground Class E
10
Zone 3b, = 5%
0 0.00 0.50 1.00 1.50 Period [s] 2.00 2.50 3.00
0 0.01
1.00
10.00
25
Displacement Sd [cm]
The displacement spectra are computed from the acceleration spectra using equation (7.79)
20
S pa S d = ------2
15
Ground Class A Ground Class B Ground Class C Ground Class D Ground Class E
(7.79)
10
Displacement spectra are an important design tool (even within force-based design procedures) because they allow a quick estimate of the expected deformations, hence of the expected damage.
Zone 3b, = 5%
0 0.00 0.50 1.00 1.50 Period [s] 2.00 2.50 3.00
7 Seismic Excitation
Page 7-53
7 Seismic Excitation
Page 7-54
An-Najah 2013
An-Najah 2013
3.0
B
2.5
Spa / Ag []
Spa [g]
2.0
Sd [cm]
1.5
5 0 0
1.0
A
Elastic design spectrum according to SIA 261, Ground Type B Newmarks elastic design spectrum
0.0 0
2 3 Period [s]
2 3 Period [s]
D
0.5
1.00
10.00
Spa [g]
T = const.
The SIA 261 spectra, like the spectra of the majority of the standards worldwide, were defined using the same principles as Newmarks spectra. However, different ground motion were used: - SIA 261 takes into account different ground classes; - Different seismic sources were considered; - A larger number of ground motions was considered. Note: in SIA 261 the corner period T A is not defined.
10 Sd [cm]
15
Periods T correspond to lines running through the origin of the axes, because: S pa = 2 S d and after reorganizing: T = 2 S d S pa
Page 7-56
7 Seismic Excitation
Page 7-55
7 Seismic Excitation
An-Najah 2013
An-Najah 2013
0.4
T=0.77s: Overestimation
0.3
Spa [g]
T=0.87s: Underestimation
0.2
= 5%
0.1
Where:
0.0 0 5 Sd [cm] 10 15
(7.80)
Design spectra are defined based on averaged response spectra. For this reason, the spectral values of single response spectra may differ significantly from the design spectra. This is a crucial property of design spectra and should be kept in mind during design!
Maximum restoring force that the elastic SDOF system ( t ) (7.81) reaches over the course of the seismic excitation u g Yield force of the inelastic SDOF system
(7.82) (7.83)
Maximum displacement that the inelastic SDOF system ( t ) (7.84) reaches over the course of the seismic excitation u g Yield displacement of the inelastic SDOF system
(7.85)
7 Seismic Excitation
Page 7-57
7 Seismic Excitation
Page 7-58
An-Najah 2013
An-Najah 2013
Results
0.15
Quantity Elastic SDOF system Inelastic SDOF system (Ry=2) Inelastic SDOF system (Ry=6) T [s] Fmax [kN] Ry [-] uy [m] um [m] [-]
Inela. SDOF Ry=2 Inela. SDOF Ry=6 2.0 67.35 2.0 0.068 0.147 2.16 2.0 22.45 6.0 0.023 0.126 5.54
0.10
Displacement [m]
0.05
0.00
-0.05
Comments
Both inelastic SDOF systems show a stable seismic response.
-0.10
-0.15 0 150
10
15
20 Time [s]
25
30
35
40
100
50 Force [kN]
-50
-100
Elastic SDOF system Inelastic SDOF system (Ry=2) Inelastic SDOF system (Ry=6)
-0.10 -0.05 0.00 Displacement [m] 0.05 0.10 0.15
-150 -0.15
7 Seismic Excitation
Page 7-59
7 Seismic Excitation
Page 7-60
An-Najah 2013
An-Najah 2013
7.6.2 Seismic behaviour equation For seismic collapse prevention, the following approximate relationship applies quality of seismic behaviour strength ductility (7.86)
If the strength of the structure reduces, the stiffness typically reduces too. If the masses do not change significantly (which is typically the case), the fundamental period T of the softer structure is longer. Structures with a longer fundamental period T are typically subjected to larger deformations, i.e., the deformation demand is larger.
7 Seismic Excitation
Page 7-61
7 Seismic Excitation
Page 7-62
An-Najah 2013
An-Najah 2013
Moment-curvature-relationship at the base of the plastic hinge zone. Despite reaching and exceeding its elastic limit the wall did not collapse. The plastic deformation capacity of structures can really be taken into account for seismic design purposes.
7 Seismic Excitation
Wall WDH4
20
10 0 10 Curvature [1/km]
20
30
Wall WSH6 [DWB99]
Page 7-63
7 Seismic Excitation
Page 7-64
An-Najah 2013
An-Najah 2013
Wall WSH3
Fy 0.75 Fy
Comments
The ductility capacity is a property of the structural member. The ductility demand is a result of the seismic excitation and also a function of the dynamic properties of the structure. A structural member survives the earthquake if:
Plastic region of test unit WSH6 (left) and close-up of the left boundary region (right). Both photos were taken at displacement ductility 6.
7 Seismic Excitation
Page 7-65
7 Seismic Excitation
Page 7-66
An-Najah 2013
An-Najah 2013
strain ductility
u = ---y
10
= 5%
curvature ductility
u = ---y
=1 =2 =4 =6
0.10 1.00 10.00 100.0
Sa [m/s2]
0 0.01 1.0
rotation ductility
u = ---y
0.5
=1 =2 =4 =6
0.10 1.00 10.00
= 5%
Sv [m/s]
100.0
displacement ductility
u = ----y
=1 =2 =4 =6
0.10 1.00 Period [s] 10.00
= 5%
Sd [m]
100.0
7 Seismic Excitation
Page 7-67
7 Seismic Excitation
Page 7-68
An-Najah 2013
An-Najah 2013
Displacement ductility TA TB TC TD TE = 5% TF
Displacement ductility []
10
=6
10
Ry=6 Ry=4
Ry []
=4
=2 Ry=(2-1)0.5 Ry=
Ry=2
Ry =
Ry =
2 1
In the small period range, already small reductions of the elastic strength of the SDOF system yield very large ductility demands. If the ductility demand is very large, it can be difficult to provide the structure with a sufficiently large ductility capacity. This problem will be further discussed during the design classes. Also in the large period range where the equal displacement principle applies large discrepancies between real and estimated ductility demand can occur. The equal displacement principle and the equal energy principle are historical Ry--Tn relationships. In recent years a lot of research has been done to come up with more accurate formulations (see e.g. works by Krawinkler [KN92], Fajfar [VFF94], Miranda [Mir01], ...)
7 Seismic Excitation Page 7-70
7 Seismic Excitation
Page 7-69
An-Najah 2013
An-Najah 2013
7.7.1 Inelastic design spectra Inelastic design spectra in combined D-V-A format
g
= 5% =1 =2 =4 =6
Elastic design spectrum
V=vvg
A=
= 5%
100
100
D
D dg d =
C
vg
a
g
aa g
Vy [cm/s]
Vy [cm/s]
10
10
B
10 00
C
1) 0
.5 -
V/
D
10
10
10
A/
(2
B A A
A= a
1
g
[c
E
1
10
[c
0.
TA=1/33s
0.1
TB=1/8s
TE=10s
10.0
TF=33s
0.1
10.0
um = Dy
Yield strength of the SDOF system:
fy = m Ay
Page 7-72
10 00 0 10 00 10 0 m /s
2
dg D /
F
D
g =d
Ay [c
10 0
] 10 1
Ay [c ] 1
F
D /
An-Najah 2013
An-Najah 2013
Construction of the spectra using Ry--Tn relationships The inelastic design spectra are computed by means of Ry-Tn relationships:
A y = S pa, inelastic 1= ----S R y pa, elastic
TB
TC
TD
TE
TF =1 =2 =4 =6
(7.87) (7.88)
Ay / ag []
TA
(7.90)
Sd / dg [] 2
TB =1 =2 =4 =6
TC
TD
TE
TF
(7.91) (7.92)
T c = Corner period between the constant Spa and the constant Spv regions T c' = Corner period between the constant Spa and the constant Spv regions of the inelastic spectrum
7 Seismic Excitation Page 7-73
7 Seismic Excitation
Page 7-74
An-Najah 2013
An-Najah 2013
Ry--Tn relationships according to [VFF94] In [VFF94] Ry--Tn relationships are defined as follows:
cR T n -+1 c 1 ( 1 ) ----T0 Ry = c c1 ( 1 ) R + 1 Where:
cT T0 = c2 Tc Tc
TB
TC
TD
= 5% =1 =2 =4 =6
4.0
Tn T0 Tn > T0
(7.93)
Ay [m/s2]
3.0
2.0
(7.94)
1.0
T c = Corner period between the constant Spa and the constant Spv regions
0.5
1.0
1.5
2.0
2.5
3.0
TB
TC
TD
= 5%
c1
1.0 0.75 1.35 1.10
cR
1.0 1.0 0.95 0.95
c2
0.65 0.65 0.75 075
cT
0.30 0.30 0.20 0.20
0.10 D [m]
Q Bilinear
0.05
=1 =2 =4 =6
Bilinear
and where the Q-hysteretic rule is a stiffness degrading rule similar to the Takeda-hysteretic rule presented in Section 7.3.2. The table shows the dependency of the Ry--Tn relationships both on damping and hysteretic model.
0.00 0.0
0.5
1.0
2.0
2.5
3.0
7 Seismic Excitation
Page 7-75
7 Seismic Excitation
Page 7-76
An-Najah 2013
An-Najah 2013
For the Q-hysteretic model and mass-proportional damping, the Ry--Tn relationships by [VFF94] specialise as:
Tn -+1 ( 1 ) ----Ry = T0 Where: Tn T0 Tn > T0 (ED principle) (7.96) (7.95)
TB
TC
TD
= 5% =1 =2 =4 =6
4.0
T c = Corner period between the constant Spa and the constant Spv regions
Ay [m/s2]
0.3 T T T 0 = 0.65 c c
3.0
2.0
1.0
0.5
1.0
1.5
2.0
2.5
3.0
TB
TC
TD
= 5%
0.10 D [m]
0.05
=1 =2 =4 =6
0.5 1.0 1.5 Period [s] 2.0 2.5 3.0
0.00 0.0
7 Seismic Excitation
Page 7-77
7 Seismic Excitation
Page 7-78
An-Najah 2013
An-Najah 2013
TB
TC
= 5% [NH82] =1 =2 =4 =6 TD
7.7.2 Determining the response of an inelastic SDOF system by means of inelastic design spectra in ADRSformat In this section the response of two example inelastic SDOF systems is determined by means of inelastic design spectra in ADRS-format.
SDOF system 1 with Tn = 0.9 s SDOF system 2 with Tn = 0.3 s The spectra according to [VFF94] will be used (see Section 7.7.1)
4.0
Spa [m/s2]
3.0
2.0
1.0
0.02
0.04
0.06 Sd [m]
0.08
0.10
0.12
TB
TC
= 5% [VFF94] =1 =2 =4 =6 TD
fy = 80 kN uy = 0.016 m
kpl
4.0
Spa [m/s2]
3.0
uy
um
2.0
Response of the elastic SDOF system 1: 100- = 0.9s Tn = 2 m --- = 2 ----------k 4874 S pa = 2.62m/s 2 S d = 0.054m f el = 261.7kN
1.0
0.0 0.00
0.02
0.04
0.06 Sd [m]
0.08
0.10
0.12
7 Seismic Excitation
Page 7-79
7 Seismic Excitation
Page 7-80
An-Najah 2013
An-Najah 2013
Response of the inelastic SDOF system 1: f el - = 261.7 ------------ = 3.27 R y = ---80 fy = R y = 3.27 (From Equation (7.95) since T n > T c = 0.5s ) u m = u y = 0.016 3.27 = 0.054m = S d
Representation of the inelastic SDOF system 1 in the inelastic design spectrum in ADRS-format:
5.0
= 1
4.0
uy
um
= 3.27
2.0
1.0
um = 0.054m
0.0 0.00 0.02 0.04 0.06 Sd [m]
Capacity curve
In this second example two different inelastic SDOF systems will be considered: (a) A SDOF system with a rather low fy and (b) a SDOF system with a rather high fy.
0.12
0.08
0.10
Response of the second inelastic SDOF system 2a f el - = 471 -------- = 3.93 R y = ---120 fy In this case the resulting displacement ductility is so large, that Equation (7.96) T 0 = T c = 0.5s results. After rearranging Equation (7.95), the displacement ductility can be computed as:
7 Seismic Excitation Page 7-82
If the force-deformation relationship of the inelastic SDOF system is divided by its mass m, the capacity curve is obtained, which can be plotted on top of the spectrum in ADRS-format. The capacity curve and the inelastic spectrum intersect in the performance point.
7 Seismic Excitation Page 7-81
An-Najah 2013
An-Najah 2013
T c = 0.65 5.88
Representation of the inelastic SDOF system 2a in the inelastic design spectrum in ADRS-format:
5.0
(7.97)
Tn = 0.3s
4.0
(*)
Spa [m/s2] 3.0
Performance point
Tn 0.3 - + 1 = ( 1.73 1 ) -----------+ 1 = 1.57 = R y ( 1 ) ----T0 0.383 The maximum displacement response is therefore: u m = u y = 0.0068 1.73 = 0.012m > S d
2.0
1.0
= 1
0.0 0.00
0.02
0.04
0.06 Sd [m]
0.08
0.10
0.12
Note that the line (*) is no longer vertical as in Example 1, but inclined according to the equation = ( R y 1 ) ( T c T n ) + 1 .
7 Seismic Excitation
Page 7-83
7 Seismic Excitation
Page 7-84
An-Najah 2013
An-Najah 2013
Representation of the inelastic SDOF system 2b in the inelastic design spectrum in ADRS-format:
5.0
Tn = 0.3s
Tn -+1 ( 1 ) ----Ry = Tc
(7.98)
4.0
This approximation is particularly satisfactory, if the large uncertainties associated with smoothed spectra are considered.
Comments
A discussion of similar examples can be found in [Faj99]. For computing the response of inelastic SDOF systems by means of inelastic design spectra, the Ry--Tn relationships in Section 7.7.1 are sufficient. The spectra in ADRS-format are not absolutely necessary, but they illustrate the maximum response of inelastic SDOF systems very well. Ry--Tn relationships should only be used in conjunction with smoothed spectra. They should not be used to derive the inelastic response spectra of a single ground motion Remember:
- Design spectra are very useful tools to design structures for the expected seismic demand. Design spectra represent the average effect of an earthquake with design intensity. - If a single earthquake is considered, the spectra may underestimate the seismic demand for a certain period range (... overestimate ...). - This characteristic of design spectra should be considered when designing structures: The seismic design should aim at structures that are as robust as possible.
Spa [m/s2]
3.0
um = 0.012m
2.0
1.0
(*)
= 1 = 1.73
0.0 0.00
0.02
0.04
0.06 Sd [m]
0.08
0.10
0.12
Note that the curve (*) is no longer a straight line as in Examples 1 and 2a. In Example 2b the curve (*) needs to be computed numerically. In Example 2a the curve (*) is only an approximation of the curve (*) in Example 2b. As soon as T 0 = T c both curves are identical. In Example 2 this is the case if S pa < 1.6m/s 2 . When T 0 < T c (i.e. when S pa > 1.6m/s 2 ) the curve (*)2a predicts larger maximum displacements u m than curve (*)2b. The difference is, however, small. For this reason, in most cases Equation (7.95) can be approximated as:
7 Seismic Excitation
Page 7-85
7 Seismic Excitation
Page 7-86
An-Najah 2013
An-Najah 2013
The equal displacement and the equal energy principles represent a strong simplification of the real inelastic behaviour of SDOF systems. q = o Ry
Design spectra are a powerful tool to design structures to resist the expected seismic action. On average, design spectra are a good representation of the expected peak behaviour of structures.
0 = Overstrength
However, if single ground motions are considered, then it can easily be the case that design spectra significantly underestimate the expected peak behaviour of structures.
Ground Class B
TB
TC
TD
3 Sd / agd [-]
This characteristic of the design spectra shall be taken into account during design by aiming at robust structures.
q=1.5
2
q=2
1
0=1.5
q=3 q=4
0 0.01
1.00
10.00
7 Seismic Excitation
Page 7-87
7 Seismic Excitation
Page 7-88
An-Najah 2013
An-Najah 2013
ug(t) .. ug(t)
kel
Elastic
(with effective stiffness or secant stiffness keff)
= 5% (constant, proportional to kel) m = 100t k el = 4874kN/m f y = 80kN Takeda-hysteresis with r o = 0.05 , = 0.5 , = 0.0
fm fy
kel
k u = k el ( max { } ) k u = k el ( max { } )
-
kpl keff
Inelastic
uy
um
It is postulated that the maximum response u m of an inelastic SDOF system can be estimated by means of a linear equivalent SDOF system (SDOFe). The properties of the SDOFe are: Stiffness: Damping: k eff = f m u m e (7.99) (7.100)
The maximum response of the SDOF system when subjected to the NScomponent of the 1940 El Centro Earthquake is: x m = 0.073m , f m = 93.0kN The properties of the corresponding SDOFe are: fm 93.0 kN m- = 2 ----------100- = 1.76s - = 1274 ------ , T e = 2 ------- = -----------k eff = ----0.073 m um k eff 1274 e = 22.89% , the viscous damping e was determined iteratively!
7 Seismic Excitation Page 7-89 7 Seismic Excitation Page 7-90
The question is how the viscous damping e of the SDOFe can be determined so that max ( u ( t ) ) = u m .
An-Najah 2013
An-Najah 2013
Comments regarding the example: The damping e is in general larger than the damping , since e of the SDOFe needs to compensate for the hysteretic energy absorption of the inelastic SDOF system. However, in rare cases it happens that e < . This shows again the difficulties that are associated with the prediction of the seismic response of inelastic SDOF systems. In the example, the viscous damping e was determined iteratively until a value for e was found for which the response of the SDOFe system was equal to the maximum response of the inelastic SDOF system. Hence, if a method was available for estimating the viscous damping e , then the maximum response of the inelastic SDOF system could indeed be estimated by means of the linear equivalent SDOF system.
0.08
= 5% e = 22.89%
50
Force [kN]
-50
fm
-100 -0.08 0.08 0.06 0.04 Deformation [m] 0.02 0.00 -0.02 -0.04 -0.06 -0.08 0
= 5% e = 22.89%
The stiffness k eff and the period T e of the SDOFe system are only known once the maximum response of the inelastic SDOF system are known. Section 7.8.2 shows how the equivalent viscous damping e can be estimated without knowing the stiffness k eff and the period T e of the SDOFe system a priori.
um
5 10 15 20 Time [s] 25
The inelastic SDOF system dissipates energy due to and due to the inelastic deformations, which are a function of its inelastic force-deformation relationship. The equivalent SDOF system, however, dissipates energy solely due to its viscous damping. For this reason the following relationship applies:
7 Seismic Excitation Page 7-92
7 Seismic Excitation
Page 7-91
An-Najah 2013
An-Najah 2013
e = + eq
(7.102)
fm = [ 1 + ro ( 1 ) ] fy fo = ( 1 ro ) fy u1 -f f 1 = ---------------u1 + u3 2 f2 = [ 1 + ro ( 1 ) ( 1 ) ] fy u1 = uy u2 1 + ro ( 1 ) - uy u 2 = -------------------------------- u3 = [ ( 1 ) ] uy
where eq is the viscous damping equivalent to the hysteretic energy absorption of the inelastic system. The simplest method for estimating the equivalent viscous damping is to assume that the inelastic system and the linear equivalent system dissipate the same energy within one displacement cycle. According to this assumption [Cho11] defines the equivalent elastic damping as: 1- A h ----- eq = ----4 Ae Where: A h : Energy dissipated by the inelastic SDOF system due to the inelastic deformation of the system. The dissipated energy corresponds to the area of the force-displacement hysteresis of the considered displacement cycle; A e : Potential energy of the equivalent SDOF system at maximum displacement: k eff u m A e = ------------------2 The inelastic force-deformation relationship of many structural RC elements can be described by the Takeda-hysteresis rule. According to Equation (7.103) the equivalent viscous damping of this hysteresis rule is: ( f m + f o ) u y + f 1 u 1 f m u 2 ( f m + f o ) u 3 1- --------------------------------------------------------------------------------------------------- eq,Tak = ----( fm uy ) 2 4 Where: (7.104) Rule according to Clough (Clo) [CP75]: 2 3 1 - ----- ------------- eq, Clo = - 2 (7.107) Bilinear (BL) rule: 2 ( 1 ) ( 1 ro ) - ----------------------------------------- eq, BL = - ( 1 + ro ro ) (7.106)
2
(7.103)
The equivalent viscous damping of other important hysteresis rules is: Elasto-plastic (EP) rule: 1 2 - ------------- eq, EP = - (7.105)
7 Seismic Excitation
Page 7-93
7 Seismic Excitation
Page 7-94
An-Najah 2013
An-Najah 2013
Comments regarding the comparison of the theoretical value with the computed value of e for the Takeda-SDOF system when excited by the El Centro earthquake: The computed value of e was determined iteratively. Eight different inelastic SDOF systems with different periods T n were considered. The strength of each inelastic SDOF system was varied in such a way that seven different displacement ductilities resulted ( =2 to 8). The results show that e is not only dependent on but also on the period T n of the SDOF system. This effect is not considered by Equations (7.103) and (7.104), respectively.
9 10
eq [-]
40 30 20 10 0 0 1 2 3
[-]
The next figure compares the theoretical value for e for the Takeda-SDOF (Equ. (7.104), ro=0.05, =0.5, =0) with the computed value (for El Centro):
50 11 10 40 9 Theory (Eq. (3.46)) Tn=4.00s Tn=2.00s Tn=1.33s Tn=1.00s Tn=0.67s Tn=0.50s Tn=0.33s Tn=0.25s
In some cases the difference between the theoretical value and the computed vale for e is considerable. For this reason there are also considerable differences between , t arg et (target ductility) and , actual (actual ductility obtained from the time-history analysis of the SDOFe system with the viscous damping e according to Equation (7.104)). Typically these differences increase as the target ductility increases. Similar observations were made when the computation of the inelastic spectra was discussed. This shows again the difficulties associated with the prediction of the seismic response of inelastic SDOF system.
,actual [-]
8 7 6 5 4 3
30
e [-]
20
10
2 1
0 0
,target [-]
0 0
,target [-]
7 Seismic Excitation
Page 7-95
7 Seismic Excitation
Page 7-96
An-Najah 2013
An-Najah 2013
Takeda Thin (TT): Hysteretic rule that characterises RC structures which lateral stiffness is provided by walls and columns. Takeda Fat (TF): Hysteretic rule that characterises RC structures which lateral stiffness is provided by frames. Ramberg-Osgood (RO): Hysteretic rule that characterises ductile steel structures. Flag-Shaped, =0.35 (FS): Hysteretic rule that characterises prestressed structures with unbonded tendons. eq - -relationships for the most important hysteresis rules according to [GBP05]:
20 1) EPP 2) BL (ro=0.2) 3) Takeda Thin 4) Takeda Fat 5) FS (beta=0.35) 6) RO
Hysteresis rule 1) Elasto-Plastic (EPP) 2) Bilinear, ro=0.2 (BI) 3) Takeda Thin (TT) 4) Takeda Fat (TF) 5) Ramberg-Osgood (RO) 6) Flag-Shaped, =0.35 (FS)
15
eq [-]
10
Te=2.0s
0 0 1 2
The hysteresis rules 1) to 6) were chosen because they can be used to represent the hysteretic behaviour of typical structural types: Elasto-Plastic (EPP): Hysteretic rule that characterises systems for the seismic isolation of structures (sliding systems that are based on friction). Bilinear, ro=0.2 (BI): Hysteretic rule that also characterises systems for the seismic isolation of structures. The value of the postyield stiffness rokpl may vary significantly between different systems.
7 Seismic Excitation Page 7-97
[-]
Important comments: With these relationships an in a statistical sense improved estimate of the damping e is obtained. For single systems subjected to a specific ground motion differences between the maximum response of the inelastic system and the maximum response of the equivalent SDOF with e according to these improved eq - -relationships can still be significant!
7 Seismic Excitation Page 7-98
An-Najah 2013
An-Najah 2013
7.8.1 Elastic design spectra for high damping values To compute the response of the equivalent SDOF systems, elastic design spectra can be used. The damping values of equivalent SDOFe systems are in general larger than the typical 5%. For this reason the design spectra needs to be computed for higher damping values. The design spectra for higher values of damping are often obtained by multiplying the design spectra for 5% damping with a correction factor : S pa (T n,) = S pa (T n, = 5%) (7.109) Correction factor [-]
[BE99] [TF99]
0.05 0.10 0.15 0.20
Damping [-]
0.25
0.30
0.35
0.40
0.45
0.50
The literature provides different estimates for this correction factor . Two of these are: [TF99]: [BE99]: = = 1.5 ----------------- where 0.05 0.5 1 + 10 1 - where 0.05 0.3 ---------------------0.5 + 10 (7.110) (7.111)
Comments:
A discussion of the different approaches for computing the design spectra for high values of damping can be found in [PCK07]. Equations (7.110) and (7.111) were derived for ground motions without near-field effects. Equations (7.110) and (7.111) were derived from the statistical analysis of several response spectra for different ground motions. For this reason Equ.s (7.110) and (7.111) should only be used in conjunction with smoothed response or design spectra. As for all statistical analyses the resulting design spectra correspond only in average with the true highly damped spectral ordinates. For single periods and ground motions the differences between the highly damped spectral ordinates obtained by Equ.s (7.110)/(7.111) and by time-history analyses of SDOF systems can be significant.
Equation (7.111) corresponds to Equation (29) in the Swiss Code SIA 261 [SIA03]. The correction factors obtained with Equ.s (7.110) and (7.111) are plotted for different damping values in the next figure:
7 Seismic Excitation
Page 7-99
7 Seismic Excitation
Page 7-100
An-Najah 2013
An-Najah 2013
TC = 5%
TD
TB
TC
[BE99]
4.0
4.0
Spa [m/s2]
Ay [m/s2]
3.0
3.0
= 5%
2.0
2.0
TD
1.0
1.0
0.5
1.0
1.5
2.0
2.5
3.0
0.02
0.04
0.06 Sd [m]
0.08
0.10
0.12
TB
TC
TD = 5%
TB
TC
[TF99]
10%
Spa [m/s2]
20% 30%
0.05
3.0
2.0
1.0
40 50%
30
= 5% 10% 20%
TD
0.00 0.0
0.5
1.0
2.0
2.5
3.0
0.0 0.00
0.02
0.04
0.06 Sd [m]
0.08
0.10
0.12
7 Seismic Excitation
Page 7-101
7 Seismic Excitation
Page 7-102
An-Najah 2013
An-Najah 2013
7.8.2 Determining the response of inelastic SDOF systems by means of a linear equivalent SDOF system and elastic design spectra with high damping The computation of the seismic response of inelastic systems by means of linear equivalent systems was studied by Sozen and his co-workers in the seventies (see for example [GS74], [SS76] and [SS81]). Today this approach gains new attention since the Direct Displacement-Based Design (DDBD) approach, which was developed by Priestley and his co-workers, is based on the idea of the linear equivalent system ([PCK07]). This sections outlines the procedure for computing the response of an inelastic SDOF system by means of an linear equivalent SDOF system and elastic design spectra with high damping. Example: SDOF system with Tn= 0.9 s
m Properties: C k m = 100 t kel = 4874 kN/m kpl = 244 kN/m = 5% fy
kel
Response of the elastic SDOF system 100- = 0.9s Tn = 2 m --- = 2 ----------4874 k S pa = 2.62m/s 2 S d = 0.054m Response of the inelastic SDOF system The maximum response of the inelastic SDOF system will be computed by means of the ADRS-spectra (page 7-106). Step 1: The capacity curve of the SDOF system is plotted on top of the ADRS spectra. Step 2: By means of Equ.s (7.102) and (7.104) the nonlinear scale, which represents the damping e as a function of the maximum response of the SDOF system, is plotted along the capacity curve. Step 3: Several spectra for different values of damping are plotted. Step 4: The Performance Point is the point where the spectrum with damping e intersects the capacity curve at the same value of e . For the considered example the maxmimum response of the inelastic SDOF system is: S d = 0.065m Comments regarding the example: To determine the Performance Point exactly, an iterativ approach is typically required.
uy
um
The linear equivalent SDOF system is fully defined by the period T e and the damping e . The period T e results from the slope of the line that connects the origin with the Performance Point.
For the example the spectra according to [BE99] will be used (Section 7.8.1).
7 Seismic Excitation Page 7-103 7 Seismic Excitation Page 7-104
An-Najah 2013
An-Najah 2013
The damping values eq used in the figures on page 7-106 were determined according to Equation (7.104). In the figures on page 7-107 the damping eq was determined using Equation (7.108). The difference is, however, rather small. It should be noted that in both cases the computed maximum response of the inelastic SDOF system does not comply with the equal displacement principle. The linear equivalent SDOF system leads often to results that do not agree with the equal displacement principle. This applies in particular to SDOF systems with long periods or systems with large ductility demands. A second example is presented on page 7-107. It is a SDOF system with a shorter period and a smaller ductility demand than in Example 1. In this second example the equal displacement principle is approximately confirmed.
4.0
Spa [m/s2]
3.0
20%
25%
2.0
1.0
e = 5% 10% 15%
0.04
20% e = 18.5%
0.06 Sd [m] 0.08 0.10 0.12
0.0 0.00
0.02
Alternative representation:
5.0
= 5%
4.0
Spa [m/s2]
3.0
= 18.5% Tn = 0.9s
Performance point
2.0
Te = 1.67s
1.0
= 3.98
0.0 0.00
e= 18.5%
0.10 0.12
uy = 0.016m
0.02
0.04
0.06 Sd [m]
um = 0.065m
0.08
7 Seismic Excitation
Page 7-105
7 Seismic Excitation
Page 7-106
An-Najah 2013
An-Najah 2013
7.9 References
[AEM 86] Anderheggen E., Elmer H., Maag H.: Nichtlineare Finite Element Methoden: Eine Einfhrung fr Ingenieure. IBK Vorlesungsskript. Institut fr Baustatik und Konstruktion (IBK), ETH Zrich. 1986.
= 5%
4.0
= 17.8%
Spa [m/s2] 3.0
Tn = 0.90s
Performance point
[AP88]
Allahabadi R., Powell G.: Drain-2DX, User Guide. Report UBC/EERC-88/06. Earthquake Engineering Research Center, University of California, Berkeley 1988. Bathe K-J.: Finite Element Procedures. ISBN 0-13-301458-4. Prentice-Hall, Upper Saddle River, New Jersey 1996. Bommer J.J., Elnashai A.S.: Displacement spectra for seismic design. Journal of Earthquake Engineering vol. 3, No. 1, 1999. Chopra A.K.: Dynamics of Structures. Fourth Edition. Prentice Hall, 2011.
2.0
Te = 1.69s
1.0
[Bat96] [BE99]
0.0 0.00
= 4.07
0.02 0.04
e= 17.8%
0.12
uy = 0.016m = 5%
[Cho11]
Performance point
4.0
= 12.8%
Tn = 0.63s Te = 0.88s
[DWB99] Dazio A., Wenk T., Bachmann H.: Versuche an Stahlbetontragwnden unter zyklisch-statischer Einwirkung. (Tests on RC Structural Walls under Cyclic-Static Action). IBK-Report No. 239, IBK-ETH Zurich. Birkhuser Verlag, Basel 1999. (http://e-collection.ethbib.ethz.ch/view/eth:23296) [Faj99] Fajfar P.: Capacity Spectrum Method Based on Inelastic Demand Spectra. Earthquake Engineering and Structural Dynamics, 28, 979-993, 1999. Hilber H.M., Hughes T.J.R, Taylor R.L.: Improved Numerical Dissipation for Time Integration Algorithms in Structural Dynamics. Journal of Earthquake Engineering and Structural Dynamics, Vol. 5, 1977. Hibbit Karlsson & Sorensen: ABAQUS Version 6.4 - Users Manual. Pawtucket, RI, 2003.
Page 7-108
Spa [m/s2]
3.0
2.0
[HHT77]
1.0
= 1.97
0.0 0.00
e= 12.8%
0.08 0.10 0.12
uy = 0.02m
0.02
um = 0.039m
0.04
0.06 Sd [m]
[HKS03]
7 Seismic Excitation
Page 7-107
7 Seismic Excitation
An-Najah 2013
An-Najah 2013
[KN92]
Krawinkler H., Nassar A.A.: Seismic design based on ductility and cumulative damage demands and capacities. Nonlinear Seismic Analysis and Design of Reinforced Concrete Buildings, eds P. Fajfar and H. Krawinkler. New York: Elsevier Applied Science. 1992. Grant D.N., Blandon C.A., Priestley M.J.N.: Modelling Inelastic Response in Direct Displacement-Based Design. Rose School, Report 2005/03, IUSS Press, Pavia, 2005. Gulkan P., Sozen M.: Inelastic Responses of Reinforced Concrete Structures to Earthquake Motions. ACI Journal, Title No. 71-41. December 1974. Iwan W.D., Gates N.C.: Estimating Earthquake Response of Simple Hysteretic Structures. ASCE, Journal of the Engineering Mechanics Division, Vol. 105, EM3, June 1979. Jacobsen L.S.: Damping in composite structures. Proceedings of the 2nd World Conference in Earthquake Engineering. Vol. 2. Tokio and Kyoto, Japan, 1960. Jennings P.C.: Equivalent Viscous Damping for Yielding Structures. ASCE, Journal of the Engineering Mechanics Division, Vol. 94, No. EM1, February 1968. Miranda E.: Estimation of inelastic defromation demands of SDOF Systems. ASCE, Journal of structural engineering, Vol. 127, No. 9, 2001. Newmark N.M.: A Method of Computation for Structural Dynamics. ASCE, Journal of the Engineering Mechanics Division, Vol. 85, No. 3, July 1959. Newmark N.M., Hall W.J.: Earthquake Spectra and Design. Earthquake Engineering Research Institute (www.eeri.org), 1982.
[PCK07] [Saa91]
Priestley M.J.N., Calvi G.M., Kowalsky M.J.: DisplacementBased Seismic Design of Structures. IUSS Press, 2007. Saatcioglu M.: Modeling Hysteretic Force-Deformation Relationships for Reinforced Concrete Elements. In Publication SP 127 Earthquake-Resistant Concrete Structures, Inelastic Response and Design, ACI, Detroit 1991. Swiss standard SIA 261: Actions on structures. SIA, Zrich 2003. Saiidi A., Sozen M.: Simple Nonlinear Seismic Analysis of R/C Structures. ASCE, Journal of the Structural Division, Vol. 107, No. 5, May 1981. Shibata A., Sozen M.: Substitute Structure Method for Seismic Design in R/C. ASCE, Journal of the Structural Division, Vol. 102, No. 1, January 1976. Tolis S.V., Faccioli E.: Displacement design spectra. Journal of Earthquake Engineering vol. 3, No. 1, 1999. Takeda T., Sozen M.A., Nielsen N.N.: Reinforced Concrete Response to Simulated Earthquakes. ASCE, Journal of Structural Engineering, Vol. 96, No. 12, December 1970. Vidic T., Fajfar P., Fischinger M.: Consistent inelastic design spectra: strength and displacement. Earthquake Engineering and Structural Dynamics 23(5) 507-521. 1994.
[GBP05]
[SIA03] [SS81]
[GS74]
[IG79]
[SS76]
[Jac60]
[TF99] [TSN70]
[Jen68]
[Mir01]
[VFF94]
[New59]
[NH82]
7 Seismic Excitation
Page 7-109
7 Seismic Excitation
Page 7-110
An-Najah 2013
An-Najah 2013
The degrees of freedoms are the horizontal displacements u 1 and u 2 at the level of the masses m 1 and m 2
Stiffness matrix
K = k 11 k 12 k 21 k 22 = ( k1 + k2 ) k2 k2 k2
Unit displacement u 1 = 1
m1 u1 + c1 u1 + k1 u1 = f1 ( t ) + c2 ( u2 u1 ) + k2 ( u2 u1 ) + c ( u m2 u 2 2 2 u1 ) + k2 ( u2 u1 ) = f2 ( t ) m1 u1 + ( c1 + c2 ) u1 c2 u2 + ( k1 + k2 ) u1 k2 u2 = f1 ( t ) c u m2 u 2 2 1 + c2 u2 k2 u1 + k2 u2 = f2 ( t ) m1 0 0 m2 u ( c1 + c2 ) c2 1 + u2 c2 c2
(8.1)
(8.2)
u ( k1 + k2 ) k2 u1 f (t) 1 = 1 + (8.3) u2 k2 k2 u2 f2 ( t )
+ Cu + Ku = f ( t ) Mu
8 Multi Degree of Freedom Systems
(8.4)
Page 8-1 8 Multi Degree of Freedom Systems Page 8-2
An-Najah 2013
An-Najah 2013
Mass matrix M
M = m1 0 0 m2
(8.6)
By means of the principle of virtual forces the vertical displacement at location d due to a unit force F = 1 acting at location a can be readily determined.
FL 3 a d - and = - (,) = ( 2 + 2 1 ) --------- with = -L L 6EI
(8.9)
Equation of motion
m1 0 0 m2 u ( k1 + k2 ) k2 u1 1 + = 0 u2 k2 k2 u2 0
(8.10) (8.11)
+ Ku = 0 Mu
Flexibility matrix D
(8.16)
Page 8-3
Page 8-4
An-Najah 2013
An-Najah 2013
Stiffness matrix K
162 EI - 8 7 - ----K = D 1 = -------5 L3 7 8
Mass matrix M
M = m1 0 0 m2
(8.18) The stiffness matrix K of a beam element with constant flexural and axial stiffness is well known: (8.19) (8.20)
F1 F2 F3 F4 F5 F6 = EA ------L 0 0 F = Ku EA ------L 0 0 EA ------L 0 0 EA ------L 0 0
Equation of motion
m1 0 0 m2 u u 162 EI 1 - 8 7 1 = 0 - ----+ -------5 L3 7 8 u u 0 2 2
(8.21)
0 12EI ----------L3 6EI -------L2 0 0 6EI -------L2 4EI -------L 0 0 12EI ----------L3 6EI -------L2 0 0 6EI -------L2 2EI -------L 0 u1 u2 u3 u4 u5 u6
+ Ku = 0 Mu
8.1.4 Principle of virtual work See e.g. [Hum12] 8.1.5 Energie formulation See e.g. [Hum12]
(8.22)
If the axial elongation of the beam is not considered, the matrix can be further simplified as follows:
Page 8-5
Page 8-6
An-Najah 2013
An-Najah 2013
Equation of motion:
mL 0 0 0 0 12 6L 12 6L 0 0 0 0 0 u 1 u
2
24
0 8L 2
12 6L 2L 2
u1 u2
mL - 0 0 ------2 0 0 0
u 3 u
4
EI 6L - 0 + ---- L 3 12 6L 12 6L u3 2 2 u4 6L 2L 6L 4L
0 = 0 (8.25) 0 0
2 2 EI - 6L 4L 6L 2L K = ----3 L 12 6L 12 6L
(8.23)
mL 0 0 0 0 0 0 mL - 0 0 ------2 0 0 0 0 0 0 0 u EI 12 12 6L 6L - 3 + ---- 3 = 0 (8.26) 3 u2 L u2 0 0 6L 8L 2 2L 2 0 u u4 6L 6L 2L 2 4L 2 4 u 1 u 24 12 0 6L u1
6L 2L 2 6L 4L 2
Example: Cantilever
Static condensation:
mL 0 0 0 0 0 0 u 1 u 24 12 0 6L u1
mL - 0 0 ------2 0 0 0 0 0 0
0 u EI - 12 12 6L 6L 3 = 0 (8.27) 3 + ----3 L u u2 0 0 6L 8L 2 2L 2 2 0 u u4 6L 6L 2L 2 4L 2 4
2 2 EI - 6 L + 6L 4L + 4L = ----L3 12 6L
(8.24)
m tt 0 0 0
u t + k tt k t0 u t = 0 0 k 0t k 00 u0 u0
(8.28)
6L
2L 2
with L = L 2
Page 8-7
Page 8-8
An-Najah 2013
An-Najah 2013
m tt u t + k tt u t + k t0 u 0 = 0 k 0t u t + k 00 u 0 = 0
(8.29)
EI 6 16 5 = ----- -- k tt L3 7 5 2
(8.36)
From the second row of Equation (8.29) the following expression can be derived:
u0 =
1 k 00 k 0t u t
after substituting L = L 2 :
EI = ----- 48 ----- 16 5 k tt 3 L 7 5 2
(8.30)
(8.37)
The final equation of motion of the cantilever is therefore: mL ------- 0 u u EI 48 16 5 2 - ----- 1 + ---- 1 = 0 3 L 7 5 2 0 u u3 mL 3 0 ------4 Notes
(8.31) (8.32)
(8.38)
= 0
and with k t0 = k 0t :
+ ( k k k k ) u = 0 m tt u t tt 0t 00 0t t
T 1
(8.33) (8.34)
The Direct Stiffness Method is often used in the Finite Element Method. The derivation of the stiffness matrix K for a beam element and instructions for assembling the stiffness matrix of entire structures can be found e.g. in the following references: [Prz85] Przemieniecki J.S.: Theory of Matrix Structural Analysis. Dover Publications, New York 1985. Bathe K-J.: Finite Element Procedures. Prentice Hall, Upper Saddle River, 1996.
+ k = k kT k1 k tt u t = 0 with k m tt u t tt 0t 00 0t tt
is the condensed stiffness matrix, and in our case it is Where k tt equal to:
1 ------- 2 EI 24 12 0 6L 7L - = ----L 3 12 12 1 6 L 6 L ---------- 14L 2 1 ---------- 14L 2 0 6 L 26L 6 L ------- 7L 2
k tt
[Bat96]
(8.35)
8 Multi Degree of Freedom Systems Page 8-9 8 Multi Degree of Freedom Systems Page 8-10
An-Najah 2013
An-Najah 2013
8.1.7 Change of degrees of freedom The equation of motion for free vibration of the 2-DoF system depicted in the following can be immediately set up if the DoFs u 1 and 1 are considered.
or: M u u 1 +K 1 = 0 0 1 1 (8.40)
As an alternative, the motion of the system can be also expressed in terms of the DoFs u 1 and u 2 . To this purpose, the relationship between the two sets of DoFs can be immediately written as: u1 = u1 L - u 2 = -2 1
(8.41)
1 0 0 2L A
or u = Au
(8.42)
Using Equation (8.23), the equation of motion for free vibrations of the system becomes 3m ------2 0
2
The matrix A is called coordinate transformation matrix and can be used to transform the mass matrix, the stiffness matrix and the load vector from one set of DoF to the other, i.e. K = A KA
T
mL 0 ---------8
u u1 12 6L EI - 1 + ---- = 0 3 2 L 1 0 1 6L 4L
(8.39) M = A MA F = A F
T T
Page 8-11
An-Najah 2013
An-Najah 2013
For the example at hand, the stiffness matrix K expressed in the set of DoFs u 1 and u 2 becomes: 12 6L EI T - 1 0 K = A KA = 1 0 ----3 2 0 2L L 0 2L 6L 4L EI - 12 12 K = ----3 L 12 16 while the mass matrix M becomes: 3m ------2 0
2
8.1.8 Systems incorporating rigid elements with distributed mass The 2-DoF system depicted in the following incorporates a rigid element with distributed mass .
(8.46)
(8.47)
M = A MA = 1 0 0 2L
T
mL 0 ---------8
1 0 0 2L
(8.48)
3m ------- 0 M = 2 m 0 --2
(8.49) The elements of the 2x2 mass matrix can be determined by imparting a unit acceleration a=1 to one degree of freedom while keeping the acceleration of the other degree of freedom equal to zero (b=0). The resulting inertia forces are then applied as static forces acting onto the system, and the elements of the mass matrix are computed as the reactions to these static forces.
which yields the equation of motion of the 2-DoF systems expressed in terms of the DoFs u 1 and u 2 3m ------- 0 2 m 0 --2 u u EI 1 - 12 12 1 = 0 + ----3 L 12 16 0 u u2 2
(8.50)
Page 8-13
Page 8-14
An-Najah 2013
An-Najah 2013
In the example at hand, if the DoFs u 1 and u 2 are considered, the elements of the mass matrix can be easily computed as follows: 2 - L m 11 = -3 1 - L m 21 = -3 1 - L m 12 = -3 2 - L m 22 = -3 Hence the mass matrix becomes: L - 2 1 M = -----6 1 2 (8.55) (8.51) (8.52) (8.53) (8.54)
u u L EI - 28 10 1 = 0 ------ 2 1 1 + ----3 6 10 4 1 2 u u2 0 L 2
(8.57)
Due to the fact that the mass is distributed, off-diagonal terms are present and therefore the mass matrix is coupled. The stiffness matrix of the 2-DoF system can be easily computed by means of the methods discussed so far as: EI - 28 10 K = ----3 10 4 L (8.56)
and the equation of motion of the system for free vibration becomes:
8 Multi Degree of Freedom Systems Page 8-15 8 Multi Degree of Freedom Systems Page 8-16
An-Najah 2013
An-Najah 2013
9 Free Vibrations
9.1 Natural vibrations
+ Ku = 0 Mu
Also in the case of Equation (9.7), there is always the trivial solution n = 0 , which corresponds to an absence of movement. If the matrix A has an inverse A 1 , then Equation (9.7) can be rearranged as follows: (9.1)
A A n = A 0
1 1
Ansatz:
u ( t ) = q n ( t ) n where q n ( t ) = A n cos ( n t ) + B n sin ( n t )
(9.8)
(9.2)
and therefore
n = 0
(9.9)
(9.3) (9.4)
This means that if matrix A has an inverse A 1 , then Equations (9.6) and (9.7) have only the trivial solution given by Equation (9.9). The inverse of Matrix A has the form:
A
1
+ K n ] qn ( t ) = 0
(9.5)
1 -A = -----A
(9.10)
Equation (9.5) is satisfied if q n ( t ) = 0 , which is a trivial solution n = 0 . meaning that there is no movement, because u ( t ) = q n ( t ) To obtain a nontrivial solution the term in brackets in Equation (9.5) must be equal to zero, i.e.:
[
2 n M
If the determinant A is equal to zero, then the matrix is singular and has no inverse. Therefore, Equation (9.6) has a nontrivial solution only if:
n M + K = 0
2 2
+ K ] n = 0
(9.6)
(9.11)
or:
A n = 0 with A = n M + K
2
(9.7)
The determinant yields a polynomial of order N in n which is called characteristic equation. The N roots of the characteristic equation are called eigenvalues and allow the calculation of the N natural circular frequencies n of the system.
9 Free Vibrations
Page 9-1
9 Free Vibrations
Page 9-2
An-Najah 2013
An-Najah 2013
As soon as the natural circular frequencies n are computed, also the vectors n can be computed within a multiplicative constant by means of Equation (9.6). There are N independent Vectors which are called eigenvectors or natural modes of vibration of the system. Summary A MDoF system with N degrees of freedom has N circular frequencies n ( n = 1, 2, 3, , N ) and N eigenvectors. Each eigenvector has N elements. The circular frequencies are arranged in ascending order, i.e.: 1 < 2 < < n . Natural circular frequencies and eigenvectors are properties of the MDoF system and depends only from its mass and stiffness properties. The index n refers to the numbering of the eigenvectors and the first mode of vibration ( n = 1 ) is commonly referred to as the fundamental mode of vibration.
(9.12)
2k n m k
k k n m
2
= 0
(9.13)
9 Free Vibrations
Page 9-3
9 Free Vibrations
Page 9-4
An-Najah 2013
An-Najah 2013
(9.19)
(9.15)
k 11 + k 11 = 0 11 = 11
For each eigenvalue 2 n we can now compute an eigenvector and a natural circular frequency. 9.2.2 Fundamental mode of vibration
3 5 k- --With the smallest eigenvalue 2 we obtain the 1 = --------------2 m
As expected, the eigenvector is determined within a multiplicative constant, and can therefore be arbitrarily normalized as follows:
so that the largest element of the eigenvector is equal to 1 so that one particular element of the eigenvector is equal to 1 so that the norm of the eigenvector is equal to 1 ...
1. circular frequency 1 =
(9.16)
[ K 1 M ] 1 =
3 5 k- - --2k --------------m 2 m k
k 3 5 k - --- m k -------------- 2 m
Fundamental mode:
11 21 = 0 0 1 = 3 5 kk --------------- --= 0.618 --2 m m 11 21 2 --------------= 1 + 5 = 0.618 1 1
(9.17) we obtain two independent equations that can be used to determine the elements of the first eigenvector 1 . The first row of the system yield the equation:
( 1 + 5)k ----------------------- 11 k 21 = 0 2
1 =
and
(1 + 5) - 11 21 = -------------------2
(9.18)
An-Najah 2013
An-Najah 2013
9.2.3 Higher modes of vibration Additionally to the fundamental mode of vibration, the considered 2-DoF system has a second mode of vibration. The properties of this second mode of vibration can be computed in analogy to the fundamental mode and the following results are obtained: Second mode
2 = 3 + 5 kk--------------- --= 1.618 --2 m m 1
9.2.4 Free vibrations of the 2-DoF system According to Equation (9.2), the free vibration of the 2-DoF system is:
u = [ C 1 cos ( 1 t ) + C 2 sin ( 1 t ) ] 1 + [ C 3 cos ( 2 t ) + C 4 sin ( 2 t ) ] 2
(9.20)
u1 u2 = [ C 1 cos ( 1 t ) + C 2 sin ( 1 t ) ] 11 21 12 22
(9.21)
+ [ C 3 cos ( 2 t ) + C 4 sin ( 2 t ) ]
2 =
12 22
1 = 1 5 = 0.618 --------------2
The still unknown constants C 1 to C 4 can be computed using the initial conditions given by Equation (9.24) and become:
22 u1 12 u2 C 1 = ------------------------------------ 11 22 21 12 11 u2 21 u1 C 3 = ------------------------------------ 11 22 21 12
22 v1 12 v2 C 2 = ------------------------------------------------( 11 22 21 12 ) 1 11 v2 21 v1 C 4 = ------------------------------------------------( 11 22 21 12 ) 2 = u1 = u2 = v1 = v2
(9.22)
(9.23)
(9.24)
An-Najah 2013
An-Najah 2013
0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 -0.8 -1.0 -1.2 0 5 10 15 20
0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 -0.8 -1.0 -1.2 0 5 10 15 20
Time [s]
1.2 1.0 0.8 1.2 1.0 0.8
Time [s]
0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 -0.8 -1.0 -1.2 0 5 10 15 20
0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 -0.8 -1.0 -1.2 0 5 10 15 20
Time [s]
Time [s]
9 Free Vibrations
Page 9-9
9 Free Vibrations
Page 9-10
An-Najah 2013
An-Najah 2013
(9.25)
Spectral matrix
Time [s]
2 0 0 1
0.8 0.6
=
First mode Second mode Total displacement
2 0 0 2
0
2 0 N
(9.26)
(9.27)
and it is immediately apparent that the equation for all eigenvalues and all eigenvectors can be expressed in terms of modal and the spectral matrices, as follows:
K = M
Time [s]
2
(9.28)
9 Free Vibrations
Page 9-11
9 Free Vibrations
Page 9-12
An-Najah 2013
An-Najah 2013
vectors are linearly independent and can be chosen so that they are orthogonal (proof complicated). So far we have shown that T n M r = 0 for n r . By means of Equation (9.32) we can prove also the T n K r = 0 for n r . We have already seen that for n r the right hand side of Equation (9.32) is equal to zero. For this reason also the left hand side of Equation (9.32) must be equal to zero, which conclude the verification. Example: 2-DoF system In the following the orthogonality of the eigenvectors of the 2DoF system presented in Section 9.2 is checked: Relative to the mass matrix
2 2m ( 5 + 5 ) T 2 m 0 -------------- 1.382m 1 M 1 = --------------1 1 + 5 = ----------------------------2 0 m 1+ 5 (1 + 5) 1
(9.29)
and can be proven by means of Equation (9.27). Equation (9.27) is first to be set up for the eigenvector vector n , and then preT multiplied with r on both sides:
2 M r K n = n r n T T
(9.30)
Afterwards, Equation (9.30) shall be transposed making use of T T the symmetry properties of the matrices K = K and M = M :
2 M n K r = n n r T T
(9.31)
Now, Equation (9.27) shall be set up for the eigenvector vector T r , and then pre-multiplied with n on both sides:
T n K r
T r2 n M r
(9.34)
1 T 2 m 0 1 M 2 = --------------1 1 5 = 0 0 m --------------1+ 5 2 2 T m 0 -------------- 5 2 M 1 = 1 1 --------------1+ 5 = 0 0 m 2 1
(9.32) (9.35)
Equation (9.32) can now be subtracted from Equation (9.31) yielding the following equation:
2 2 ) ( n r n M r = 0 T
(9.33) (9.36)
In the case that the eigenvalues are different, then for n r we 2 2 ) 0 and the expression T M must be zero. In have ( n r n r the case that an eigenvalue occurs more than once, the eigen-
9 Free Vibrations
Page 9-13
9 Free Vibrations
Page 9-14
An-Najah 2013
An-Najah 2013
9.4.2 Linear independence of the eigenvectors The eigenvectors are linearly independent. To prove this, it needs to be shown that if (9.37)
1 1 + 2 2 + + n n = 0
(9.42)
T
then all scalars i must be equal to zero. To this purpose, we left-multiply Equation (9.42) by i M and we obtain: (9.38)
i M ( 1 1 + 2 2 + + n n ) = i M i i = 0
T
T T
(9.43)
T 1 K 2
1 2 2k k = -------------- 1 5 = 0 - 1 k k --------------1+ 5 2
(9.39)
In Section 9.4.1 we have shown that i M i 0 , therefore i = 0 meaning that the eigenvectors are linearly independent. The property that the eigenvectors are linearly independent, is very important because it allows to represent any displacement vector as a linear combination of the eigenvectors.
(9.40)
(9.41)
9 Free Vibrations
Page 9-15
9 Free Vibrations
Page 9-16
An-Najah 2013
An-Najah 2013
* = M mn n n * = K kn n n T
(9.49) (9.50)
(9.51)
(9.45)
For each n we can set up such an equation, which yields to N decoupled Single Degree of Freedom systems. The total displacement of the system can then be computes as the sum of the contribution of all decoupled SDoF systems, i.e.:
N
qi :
u(t) =
i=1
q i ( t ) i
(9.52)
(9.46)
in matricial form:
q1 ( t ) qN ( t )
can be substituted into Equation (9.44), and the latter can be leftT multiplied by n yielding the following equation:
T T ( t ) n M q i + n K q i ( t ) i = 0 i i i
u ( t ) = q ( t ) with q =
(9.53)
(9.47)
+ K * q = 0 M*q
(9.54)
Because of the orthogonality properties of the eigenvectors only one term of the summations remains, i.e.:
T ( t ) n M n q n
with
* 0 m1 * 0 k1
T n K n qn ( t )
= 0
(9.48)
M* = M =
and
* 0 mn
K* = K =
* 0 kn
where:
(9.55)
9 Free Vibrations Page 9-17 9 Free Vibrations Page 9-18
An-Najah 2013
An-Najah 2013
Example: 2-DoF system (9.56) The modal masses and modal stiffness of the 2-DoF system of Section 9.2 were already checked during the verification of the orthogonality of the eigenvectors (See Equations (9.34), (9.37), (9.38) and (9.41)). They are:
(5 + 5) * = T M = 2m ----------------------------- 1.382m m1 1 1 2 (1 5)
* = T M = m --- ( 5 5 ) 1.382m m2 2 2 2
u(t) =
i=1
q i ( t ) i = q ( t )
the LHS and the RHS of the resulting equation can be pre-mulT tiplied by n M and we obtain:
N
n Mu ( t ) =
n M i qi ( t )
i=1
(9.57)
Because of the orthogonality of the eigenvectors (see Section 9.4.1), Equation can be further simplified to: n Mu ( t ) = n M n q n ( t )
T T
(5 5) * = T K = 2k ------------------------- 0.528k k1 1 1 2 (1 5)
* = T K = k -- ( 5 + 5 ) 3.618k k2 2 2 2
(9.58)
(9.59)
or introducing the definition of the modal mass given by Equation (9.49) we obtain the equivalent expression qn ( t ) =
T n Mu ( t ) ---------------------* mn
( t ) + 0.528kq ( t ) = 0 1.382mq 1 1 1 =
* k1 ------ = * m1
(9.60)
Equations (9.59) and (9.60) will be later used to compute the response of MDoF systems.
9 Free Vibrations Page 9-19 9 Free Vibrations Page 9-20
An-Najah 2013
An-Najah 2013
( t ) + 3.618kq ( t ) = 0 1.382mq 2 2 2 =
* k2 ------ = * m2
and making use of the possibility of decoupling of the equation of motion, the total deformation u ( t ) under free vibration can be computed as the sum of the contribution of all modes. The equation of motion of the nth decoupled SDoF system is:
*q ( t ) + k * q ( t ) = 0 mn n n n
(9.72)
and its solution can be computed as discussed in Chapter 3 for SDoF systems. If we make use of the second formulation with trigonometric functions (see Section 3.1.2), the solution is: q n ( t ) = A n cos ( n t ) + B n sin ( n t ) (9.73)
u(t) =
i=1
i qi ( t ) =
i=1
i [ A i cos ( i t ) + B i sin ( i t ) ]
(9.74)
The 2 N constants A i and B i can be computed by means of the (0) = v . initial conditions u ( 0 ) = u 0 and u 0 To this purpose, the vector of the velocity is needed and can easily be computed by deriving Equation (9.74), i.e.:
9 Free Vibrations
Page 9-21
9 Free Vibrations
Page 9-22
An-Najah 2013
An-Najah 2013
(t) = u
i=1
(t) = i q i
i=1
(9.75)
The equation of motion for free vibration of a system with damping is + Cu + Ku = 0 Mu (9.80)
u(0) =
i=1
(0) = i q i ( 0 ) and u
i=1
(0) i q i
(9.76)
Making use of Equation (9.59) we can now write the equations to compute the initial conditions of the nth decoupled SDoF system as: qn ( 0 ) =
T n Mu ( 0 ) ----------------------T n M n T (0) n Mu ----------------------T n M n
As it will be shown in Chapter 10, if the MDoF system is classically damped, the equation of motion can be decoupled analogously to system without damping, and the total deformation u ( t ) under free vibration can be computed again as the sum of the contribution of all modes. The equation of motion of the nth decoupled SDoF system is:
*q * ( t ) + c * q mn n n n ( t ) + kn qn ( t ) = 0
(9.81)
(9.77)
or ( t ) + 2 q ( t ) + 2 q ( t ) = 0 q n n n n n n (9.82)
(0) = q n
(9.78)
where
* cn
In Section 3.1.2 (see Equation 3.18) it as been shown that the ( 0 ) , respecconstants A n and B n are equal to q n ( 0 ) and q n n tively, hence Equation (9.74) can be rewritten as:
N
T n C n
(9.83)
u(t) =
i=1
(9.79)
The solution of Equation (9.82) can be computed as discussed in Chapter 3 for SDoF systems. According to Equation (3.50) we have: qn ( t ) = e where:
n t
[ A n cos ( nd t ) + B n sin ( nd t ) ]
(9.84)
9 Free Vibrations
Page 9-23
9 Free Vibrations
Page 9-24
An-Najah 2013
An-Najah 2013
nd = n 1 damped circular frequency of the nth mode (9.85) The the total deformation u ( t ) under free vibration is hence
N N
Blank Page
u(t) =
i=1
i qi ( t ) =
i=1
i e
n t
[ A i cos ( id t ) + B i sin ( id t ) ]
(9.86)
As in the case of the undamped systems, the 2 N constants A i and B i can be computed by means of the initial conditions (0) = v . u ( 0 ) = u 0 and u 0 For the nth decoupled SDoF system according to Equation (3.51) the constants A n and B n can be expressed in function of the initial conditions of the modal coordinate q as follows: An = qn ( 0 ) ( 0 ) + q ( 0 ) q n n n B n = ------------------------------------------ nd (9.87) (9.88)
(0) where the initial displacement q n ( 0 ) and the initial velocity q n can be computed by means of Equations (9.77) and (9.78). Hence, the total displacement of a classically damped MDoF system under free vibration can be computed as:
N
u(t) =
i=1
i e
n t
(9.89)
An-Najah 2013
An-Najah 2013
10 Damping
10.1 Free vibrations with damping
The differential equation to compute the free vibrations of a MDoF system is:
+ Cu + Ku = 0 Mu
10.2 Example
and
(0) = v u 0
(10.2)
The displacement vector u ( t ) may be expressed as linear combination of the eigenvectors, i.e. u ( t ) = q ( t ) , and Equation (10.1) becomes:
+ C q + Kq = 0 Mq
, ,
m2 = m k2 = k
(10.6) (10.7)
while the damping characteristics will be defined later. Natural frequencies and eigenvectors The natural frequencies and the eigenvectors of the 2-DoF system can be easily computed as: Natural frequencies: Eigenvectors:
1 = k- , ------2m 2 = 2k ----m
(10.3)
+ C * q + K*q = 0 M*q
Definition: A system is classically damped if the matrix C * is diagonal A system is non-classically damped if the matrix C * is not diagonal
1 = 1 2 , 1
2 = 1 1
10 Damping
Page 10-1
10 Damping
Page 10-2
An-Najah 2013
An-Najah 2013
10.2.1 Non-classical damping The damping characteristics of the 2-DoF system are chosen as:
c1 = c
10.2.2 Classical damping The damping characteristics of the 2-DoF system are chosen as:
c 1 = 4c
c 2 = 4c
(10.10)
c 2 = 2c
(10.15)
The equation of motion of the system can be easily assembled by means of the equilibrium formulation:
u u u m 2 0 1 + c 5 4 1 + 3 1 1 = 0 1 1 u2 0 1 u 4 4 u 0 2 2
The equation of motion of the system can be easily assembled by means of the equilibrium formulation:
u m 2 0 1 + c 6 2 0 1 u 2 2 2 u u 1 + 3 1 1 = 0 u2 1 1 u2 0
(10.11)
(10.16)
It is now attempted to decouple the equations by computing the modal properties of the 2-DoF system:
1 1 3 -- 1 -- 1 --m 0 2 2m 0 2 = M = = 2 0 m 1 1 0 3m 1 1
T
It is now attempted to decouple the equations by computing the modal properties of the 2-DoF system:
3 --m 0 M* = 2 0 3m 3k ----- 0 K* = 4 0 6k
M*
(10.12)
(10.17)
3 1 1 --k 0 -- 1 -- 1 = 4 3k k 2 K* = K = 2 k k 0 6k 1 1 1 1
T
(10.13)
1 1 -- 1 -- 1 = 6c 2c 2 C* = C = 2 2c 2c 1 1 1 1
T
3 --c 2
(10.18)
0 12c
1 1 -- 1 -- 1 2 5c 4c = C* = C = 2 4c 4c 1 1 1 1
T
5 --c 4 7 --c 2
7 --c 2 17c
(10.14)
The Matrix C * is not diagonal, hence it is not possible to decouple the equations!
10 Damping Page 10-3 10 Damping Page 10-4
An-Najah 2013
An-Najah 2013
(10.25)
(10.19)
(10.26)
(10.20)
and using the results for MpD and SpD damping ratio n becomes:
a0 a1 - n = --------- + ---2 n 2 n
(10.27)
(10.21)
The coefficients a 0 and a 1 may be computed for vibration modes i and j by means of equation (10.28): (10.22)
a0 1 a1 - ---- i = i - + --- --- 2 i 2 a0 1 a1 ---- j = j ---- + --- 2- 2 j
(10.28)
(10.23)
(10.29)
Remark Both MpD and SpD, taken alone, are not a good approximation of the behaviour of real structures. Studies have shown that different modes of vibration exhibit similar damping ratios.
10 Damping Page 10-5
10 Damping
Page 10-6
An-Najah 2013
An-Najah 2013
10.3.4 Example A damping matrix shall be assembled so that in the case of the 2-DoF system shown on page 10-2 both modes of vibration are characterised by the same damping ratio . The natural frequencies are:
1 = k -------, 2m 2 = 2k ----m
(10.34)
(10.35)
(10.30)
and the representation of the damping ratio in function of the natural circular frequency is:
10 9
Mass proportional damping Stiffness proportional damping
(10.31)
8 7 6 5 4 3 2 1 0 0 1 2
Rayleigh Damping
(10.32) Check:
C* = C =
T
10
Circular frequency
Remarks - If there are more than 2 modes of vibrations, then not all of them will have the same damping ratio. - If more than 2 modes of vibrations should have the same damping, then a different damping modal shall be used. To this purpose see e.g. Caughey-Damping in [Cho11].
10 Damping Page 10-8
10 Damping
Page 10-7
An-Najah 2013
An-Najah 2013
11 Forced Vibrations
11.1 Forced vibrations without damping
11.1.1 Introduction
(11.4) (11.5)
+ K * q = F * ( t ) M*q
Where
* - M * : Diagonal matrix of the modal masses m n * - K * : Diagonal matrix of the modal stiffnesses k n * - F * : Vector of the modal forces F n
For the considered 2-DoF system, Equation (11.5) can be rearranged as:
* * * m1 q1 + k1 q1 = F1 * *q = F * m2 q2 + k2 2 2
(11.6)
Sought is the response of the 2-DoF system as a result of the external excitation force F ( t ) given by Equation (11.1)
F(t) = F1 ( t ) F2 ( t )
or as alternative:
* F1 2 q 1 + 1 q 1 = -----* m1 * F2 2 -----q + q = 2 2 2 * m2
(11.1)
(11.7)
(11.2)
The displacement vector u ( t ) can be represented as a linear combination of the eigenvectors of the 2-DoF system, u ( t ) = q ( t ) , and Equation (11.2) becomes:
+ K q = F ( t ) Mq
(11.3)
T
The two equations of the system (11.7) are decoupled and can be solved independently. The constants resulting from the solution of the system can be determined by means of the initial con (0) = v . ditions u ( 0 ) = u 0 and u 0
11 Forced Vibrations
Page 11-1
11 Forced Vibrations
Page 11-2
An-Najah 2013
An-Najah 2013
F 0 sin ( t ) F 0 sin ( t ) + 2 q = ------------------------- q - = ------------------------ = f 1 sin ( t ) 1 1 1 2 (3 2)m 3m F 0 sin ( t ) F 0 sin ( t ) 2 - = --------------------------- = f 2 sin ( t ) q 2 + 2 q 2 = --------------------------3m 3m
(11.13)
with
F0 f 1 = ------3m
and
F0 f 2 = ------3m
(11.14)
, , ,
m2 = m k2 = k c2 = 0
Each equation of the system (11.13) corresponds to the equation of motion of an undamped SDoF system under an harmonic sine excitation. The complete solution of these differential equations has been discussed in Chapter 4 and it is:
fn - sin ( t ) q n = A 1 cos ( n t ) + A 2 sin ( n t ) + -----------------2 2 n
(11.15)
(11.11)
and the modal excitation force is calculated using the modal matrix:
1 -- 1 2 1 1 F 0 sin ( t ) 0 F 0 sin ( t ) -----------------------2 F 0 sin ( t )
(11.16)
F*( t ) = F( t ) =
(11.12)
The 4 constants A 1 to A 4 can be easily computed for the initial ( 0 ) = 0 by means of the mathematical softconditions u ( 0 ) = u ware Maple. They are:
11 Forced Vibrations
Page 11-3
11 Forced Vibrations
Page 11-4
An-Najah 2013
An-Najah 2013
u(t) = q(t) =
n qn ( t )
n
= 1 q1 ( t ) + 2 q2 ( t )
(11.23)
1 -u = 2 1
(11.24)
where:
F0 F0 - , f = ------- , 1 = f 1 = ------3m 2 3m k ------- , 2 = 2m 2k ----m
(11.25)
(11.21)
or
q = f 1 1 q2 = f2 sin ( t ) ( 1 ) sin ( 1 t ) ----------------------------------------------------------------2 2 1 sin ( t ) ( 2 ) sin ( 2 t ) ----------------------------------------------------------------2 2 2
(11.22)
11 Forced Vibrations
Page 11-5
11 Forced Vibrations
Page 11-6
An-Najah 2013
An-Najah 2013
11.1.3 Example 2: RC beam with Tuned Mass Damper (TMD) without damping RC Beam
Damping ratio n = 0.0 Modal mass M n = 5.626t Modal stiffness K n = 886kN m Natural frequency f n = 2Hz
Excitation As excitation a vertical harmonic sine force acting only on the beam is assumed.
F ( t ) = F o sin ( t )
(11.26)
with:
Solution Both the transient and the steady-state part of the solution are considered.
11 Forced Vibrations Page 11-8
11 Forced Vibrations
Page 11-7
An-Najah 2013
An-Najah 2013
0.03
0.04
0 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
0.04
Displacement of the TMD
0.03
A TMD = u 2 u st
A Beam = u 1 u st
where u st = F o K n - The solution was computed by means of the Excel file given on the web page of the course (SD_MDOF_TMD.xlsx) - The Tuned Mass Damper (TMD) is blocked - The natural frequency of the beam with TMD is: f n = 1.94Hz - At f = f n resonance occurs. In the diagram above the amplification factor is limited, because the response of the system was only calculated during 60 seconds.
11 Forced Vibrations Page 11-10
Time [s]
11 Forced Vibrations
Page 11-9
An-Najah 2013
An-Najah 2013
0.006
0.008
0 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
0.04
Displacement of the TMD
0.03
A TMD = u 2 u st
A Beam = u 1 u st
where u st = F o K n - The solution was computed by means of the Excel file given on the web page of the course (SD_MDOF_TMD.xlsx) - The Tuned Mass Damper (TMD) is free to move - No resonance at f = f n occurs. Resonance occurs in correspondence of the first and of the second natural frequencies of the 2-DoF system. In the diagram above the factor A is limited, because the response of the system was only calculated during 60s.
11 Forced Vibrations Page 11-12
Time [s]
11 Forced Vibrations
Page 11-11
An-Najah 2013
An-Najah 2013
Where:
* - M * : Diagonal matrix of the modal masses m n * - K * : Diagonal matrix of the modal stiffnesses k n * - F * : Vector of the modal forces F n
- C * : Matrix of the modal damping constants. It is diagonal only if the system is classically damped (see Chapter 10). For the considered classically damped 2-DoF system, Equation (11.31) can be rearranged as: Sought is the response of the 2-DoF system as a result of the external excitation force F ( t ) given by Equation (11.27)
F(t) = F1 ( t ) F2 ( t )
* * * * m1 q1 + c1 q1 + k1 q1 = F1 * *q + k *q = F * m2 q2 + c2 2 2 2 2
(11.32)
(11.27)
or as alternative:
* F1 2 -----q + 2 q + q = 1 1 1 1 1 1 * m1 * F2 2 -----q + 2 q + q = 2 2 2 2 2 2 * m2
(11.28)
(11.33)
The displacement vector u ( t ) can be represented as a linear combination of the eigenvectors of the 2-DoF system, u ( t ) = q ( t ) , and Equation (11.28) becomes:
+ C q + Kq = F(t) Mq
T We can now multiply Equation (11.29) by obtaining: T + T C q + T K q = T F ( t ) Mq
(11.29)
The two equations of the system (11.33) are decoupled and can be solved independently. The constants resulting from the solution of the system can be determined by means of the initial con (0) = v . ditions u ( 0 ) = u 0 and u 0
(11.30) (11.31)
Page 11-13 11 Forced Vibrations Page 11-14
An-Najah 2013
An-Najah 2013
2) Compute the natural circular frequencies n and the eigenvectors n - Compute the modal properties of the MDoF system ( M * , K* ) 3) Compute the response of every mode of vibration - Set up the equation of motion of the modal SDoF systems
* Fn + 2 * q + 2 q = ------ and solve it for q n q n n n n n n * mn
- Compute the modal displacements u n ( t ) = n q n - Compute the sectional forces by means of the static equiv2M q (t) alent forces F n ( t ) = Ku n ( t ) = K n q n = n n n 4) Sum up (respectively combine) the contribution from all modes of vibration to obtain the total response of the system.
11 Forced Vibrations
Page 11-15
11 Forced Vibrations
Page 11-16
An-Najah 2013
An-Najah 2013
12 Seismic Excitation
12.1 Equation of motion
12.1.1 Introduction In analogy to Section 2.1.1, the equation of motion of the system depicted here can be formulated by means of the dAlembert principle F + T = 0 applied to each one of the masses.
y 1 = x ( t ) + l 1 + u s1 + u 1 ( t ) = x ( t ) + u ( t ) y 1 1 = m ( x + u ) T1 = m1 y 1 1 1 +m g F 1 = k 1 ( u s1 + u 1 ) c 1 u 1 1 u ) + k 2 ( u s2 + u 2 u 1 ) + c 2 ( u 2 1 +c u F1 = ( k1 + k2 ) u1 + k2 u2 ( c1 + c2 ) u 1 2 2 y 2 = x ( t ) + l 1 + u s1 + l 2 + u s2 + u 2 ( t ) = x ( t ) + u ( t ) y 2 2 = m ( x + u ) T2 = m2 y 2 2 2 u )+m g F 2 = k 2 ( u s2 + u 2 u 1 ) c 2 ( u 2 1 1 c u F2 = k2 u1 k2 u2 + c2 u 1 2 2
(12.1)
(12.2)
(12.3)
or:
m1 0 0 m2 + u x c + c2 c2 u k + k2 k2 u1 1 1 + 1 + 1 = 0 x + u2 c2 c2 u2 k2 k2 u2 0
(12.4)
or:
m1 0 0 m2 m 0 u c + c2 c2 u k + k2 k2 u1 1 1 + 1 + 1 = 1 0 m2 u c2 c2 u k2 k2 u2 2 2 x x
(12.5) which is similar to Equation (8.3) meaning that the base point excitation x ( t ) can be considered equivalent to two external forces ( t ) and f ( t ) = m x f1 ( t ) = m1 x 2 2 ( t ) acting on the masses m 1 and m 2 . This is the same interpretation given in Section 2.1.1 for SDoF systems.
12 Seismic Excitation Page 12-2
12 Seismic Excitation
Page 12-1
An-Najah 2013
An-Najah 2013
12.1.2 Synchronous Ground motion As shown in the previous section, the equation of motion of a system subjected to a base excitation is: + Cu + Ku = 0 Mu a (12.6)
Influence vector for some typical cases Planar system with translational ground motion (Case 1) In this case all DoFs of the system undergo static displacements u s ( t ) which are equal to the ground displacement u g ( t ) , hence: 1 = 1 = 1 1
is vector of the absolute accelerations of the DoFs of where u a and u are the vectors of the relative velocithe system while u ties and of the relative displacements of the DoFs of the system, respectively. The absolute displacement u a of the system can be expressed as: ua = us + u (12.7)
(12.11)
where 1 is a vector of order N , i.e. the number of DoFs, with all elements equal to 1. Planar system with translational ground motion (Case 2) The axial flexibility of the elements of the depicted system can be neglected, hence 3 DoFs are defined. In this case DoFs 1 and 2 undergo static displacements which are equal to the ground displacement, while the static displacement of DoF 3 is equal to 0, i.e.: 1 = 1 0
12 Seismic Excitation
where u s is displacement of the DoFs due to the static application (i.e. very slow so that no inertia and damping forces are generated) of the ground motion, and u is again the vector of the relative displacements of the DoFs of the system. The static displacements u s ( t ) can now be expressed in function of the ground displacement u g ( t ) as follows: us ( t ) = ug ( t ) (12.8)
where is the so-called influence vector. Equation (12.6) can now be rewritten as: ) + Cu + Ku = 0 + u M(u g + Cu + Ku = M u ( t ) Mu g
12 Seismic Excitation
(12.9) (12.10)
(12.12)
Page 12-3
Page 12-4
An-Najah 2013
An-Najah 2013
Spatial system with multiple translational ground motion Consider the spatial frame depicted here:
U6
U5
U4
U3 U2
The depicted system is subjected to a rotational ground motion g which generates the following static displacements of the DoFs: h1 L Remark If the planar system with rotational ground motion has more than one support and every support is subjected only to the base rotation g , then the static application of the base rotations typically create stresses within the system. Such a case must be considered like a multiple support excitation (see Section 12.1.3).
Picture from: Chaudat T., Pilakoutas K., Papastergiou P., Ciupala M. A. (2006) Shaking Table Tests on Reinforced Concrete Retrofitted Frame With Carbon Fibre Reinforced Polymers (CFRP), Proceedings of the First European Conference on Earthquake Engineering and Seismology, Geneva, Switzerland, 3-8 September 2006
U1
u s ( t ) = h g ( t ) hence = h 2 2 L
h1 (12.13)
The equation of motion of the frame structure for the ground motions u gx ( t ) and u gy ( t ) neglecting damping is:
12 Seismic Excitation Page 12-6
12 Seismic Excitation
Page 12-5
An-Najah 2013
An-Najah 2013
u 1 u
2
u1 u2
12.1.3 Multiple support ground motion Structures with a significative spatial extension may be subjected to ground motion time-histories that are different from support to support. A typical example for such structures is the bridge shown in the following figure.
u u 3 ( t ) M u + K 3 = M x u y gy ( t ) gx u4 u4 u u
5 5
(12.14)
u 6 u 1 u
2
u6 u1 u2 = M 1 0 0 u ( t ) gy 1 0 0
u u M 3 +K 3 u u4 4 u u
5 5
u 6
u6
0 1 0 u ( t ) + gx 0 1 0
(12.15)
Example of structure where often multiple support excitation is applied: Plan view of the dynamic model for the seismic analysis of a bridge in the transverse direction. The springs represent the piers.
and with
m1 m2 M = 0 I3 m4 m5 I6 0 u 1 u
2
In this case it is distinguished between the DoFs of the structure u a , which are free to move and whose displacements are expressed in absolute coordinates, and those of the ground u g , which undergo the displacements imposed to the support. The vector containing the displacements of all DoFs is hence:
u1 u2 m1 u gy m u
2 gx
u u 0 we obtain M 3 + K 3 = u4 u4 m4 u gy u u u m 5 5 5 gx 0 u u
6 6
u =
ua ug
(12.17)
The equation of motion of the system can hence be expressed as (see [Cho11]): m mg m g m gg
T
(12.16) Remarks For other cases see [Cho11] Sections 9.4 to 9.6.
12 Seismic Excitation Page 12-7
c cg u k kg u u 0 a a a (12.18) = + + T T p ( t ) u ug u g c g c gg g k g k gg g
12 Seismic Excitation
Page 12-8
An-Najah 2013
An-Najah 2013
Where p g ( t ) are the forces resulting at the supports when the supports undergo the displacements u g ( t ) . In Equation (12.18) the different matrices g , gg are not computed separately, they just result form the partition of the overall system of equations when the DoFs of the structure and of the ground are collected as it is shown in the example of page 12-12. The vector of Equation (12.17) can be rewritten as: u = ua ug = us ug + u 0 (12.19)
m mg m g m gg
T
+ u +u c cg u k kg us + u u 0 s s + + = T T p ug ug ug g(t) c g c gg k g k gg
According to the first line of Equation (12.20) ku s + k g u g = 0 and hence Equation (12.22) becomes: + cu + ku = ( mu + m u mu s g g ) ( cu s + c g u g ) (12.23)
where u s is the vector of the displacements of the DoFs of the structure when the ground displacements u g ( t ) are applied statically, and u is the vector of the relative displacements of the DoFs of the structure. The relationship between u s and u g ( t ) is given by the following system of equation: k kg us 0 p g, s
the so-called influence matrix can be computed, again making use of the first line of Equation (12.20), i.e.: k g u g = ku s = k u g and after rearranging we obtain: (12.25)
T k g k gg u g
(12.20) = k kg
1
(12.26)
where p g, s is the vector of the support forces needed to impose the displacements u g statically. If the system is statically determinated, p g, s is equal to zero (See example of page 12-12). By introducing Equation (12.19) into Equation (12.18) we obtain the new system of equations:
The influence matrix is a N N g matrix where N is the number of DoFs of the structure and N g is the number of DoFs of the supports. By introducing Equation (12.26) into Equation (12.23) the final equation of motion of the system is obtained:
12 Seismic Excitation
Page 12-9
12 Seismic Excitation
Page 12-10
An-Najah 2013
An-Najah 2013
+ cu + ku = ( m + m ) u mu g g ( c + cg ) ug
(12.27)
Example: 2-DoF system The following 2-DoF system is subjected to multiple support ground motion. Two different ground motions are applied to the degrees of freedom u 1 and u 5 . Sought is the equation of motion of the system:
Analogously, with the second line of Equation (12.21), an equation for the computation of the forces at the supports p g ( t ) can be setup and solved. Remarks In Equation (12.27), the masses associated with the support are often equal to zero, i.e. m g = 0 . If this is the case, Equation (12.27) simplifies to: + cu + ku = m u ( c + c ) u mu g g g (12.28)
And considering that in most cases the damping forces on the LHS of the equation are small (and they are zero if no damping is present) compared to the inertia forces (see [Cho11]), Equation (12.28) can be further simplified to: + cu + ku = m u mu g (12.29)
The stiffness matrix of the system is assembled by means of the Direct Stiffness Method and the following degrees of freedom are considered:
u = u3 u4 u1 u5
(12.32)
In the case that the movement of the supports is the same at all supports, u g becomes: ug = 1 ug and with = 1 Equation (12.29) becomes + cu + ku = m u mu g which is the same as Equation (12.10). (12.31) (12.30)
ug =
u0 =
u2 u6
12 Seismic Excitation
Page 12-11
12 Seismic Excitation
Page 12-12
An-Najah 2013
An-Najah 2013
12 6L 6L 4L
2
12 6L
6L 2L
2
0 0
0 0
u1 u2
12 6L 12 6L 6L 2L 6L 2L 0 0 0 0
2 2 2
u3 u4 u1 u2 u5 u6
u 12 6L 12 + 12 6L + 6L 12 6L EI - K = ----,U = 3 3 2 2 2 2 u4 L 6L 2L 6L + 6L 4L + 4L 6L 2L u5 0 0 12 6L 12 6L 0 0 6L 2L
2
EI 12 6L 12 6L - K = ----3 2 2 L 6L 2L 6L 4L 12 6L 0 0 0
,U =
(12.38)
6L 4L
u6
12 6L 6L 4L
(12.35)
12 6L 12 6L 6L 4L 6L 2L
2 2
6L 2L
0 0
0 0
u1 u2
u 12 6L 24 0 12 6L EI - ,U = 3 K = ----3 2 2 2 u4 L 6L 2L 0 8L 6L 2L u5 0 0 12 6L 12 6L 0 0 6L 2L 6L 4L
2 2
12 12 6L 6L 6L 6L 2L 2L 0 12 0 6L 0 4L 0
2 2 2
u3 u4 u1 u5 u2 u6
(12.36)
u6
12 6L 12 EI - K = ----3 12 6L 0 L 6L 2L 6L 2L
2 2
0 6L 0 4L
2
,U =
(12.39)
6L
0 6L
0
2
12 6L 0 0
u3 u2
By means of static condensation we can now eliminate DoF 2 and 6. We racall that:
EI k tt k t0 EI 1 - - ( k tt k T , ktt = ----K = ----0t k 00 k 0t ) 3 3 k 0t k 00 L L
6L 2L
u 0 EI 12 6L 12 6L 0 - ,U = 1 K = ----3 2 2 2 u4 L 0 2L 6L 8L 6L 2L u5 12 0 0 6L 12 6L 6L 0 0 2L 6L 4L
2 2
(12.40)
(12.37)
u6
12 Seismic Excitation
Page 12-13
12 Seismic Excitation
Page 12-14
An-Najah 2013
An-Najah 2013
3 3
u3
1 k 00
(12.41)
0 6L 3L 3L u 4 u1 3 3L 3 0 u5 3 3L 0 3
0 0 = p1 ( t ) p5 ( t )
(12.46)
6L 6L k 0t k 00 k 0t =
T 1
1 -------- 0 2 2 2 2 2L 2L 4L 6L 2L 6L 0 2 1 6L 0 6L 2L 0 6L 0 -------2 4L 0 6L 9 9
2
(12.47)
0 2L 3L 3L 9 3L 9 0 9 3L 0 9 12 12 9 9 6L 6L 0 2L 2 3L 3L 12 0 9 3L 9 0 9 3L 0 9 0 12 18 0
(12.43)
k
1
(12.48)
24 0 EI 0 8L 2 - k tt = ----3 L 12 6L 12 6L 6 0
2
(12.44)
1 -2 1 -----2L
(12.49)
3 3
EI - 0 6L 3L 3L k tt = ----3 L 3 3L 3 0 3 3L 0 3
(12.45)
12 Seismic Excitation
Page 12-15
12 Seismic Excitation
Page 12-16
An-Najah 2013
An-Najah 2013
(12.50)
(12.52)
(12.51)
The following drawings show the interpretation of the elements of the influence matrix : if u g1 = 1 then:
1 1 - and u 4 = -----u 3 = -2 2L
if u g5 = 1 then:
1 1- and u 4 = -----u 3 = -2 2L
(12.53)
Remarks: See Section 9.7 of [Cho11] for an example with a statically indeterminated system. In Finite Element analysis, when applying multiple support excitation, support displacements instead of support acceleration are often used. For more details see [Bat96].
12 Seismic Excitation Page 12-17
or: n M + 2 * q + 2 q = ---------------- -u q n n n n n n g T n M n
T
(12.54)
12 Seismic Excitation
Page 12-18
An-Najah 2013
An-Najah 2013
u(t) =
n=1
n qn ( t )
(12.55)
* and the modal participation factor , Unlike the modal mass m n n * the effective modal mass m n is independent of the normaliza, eff tion of the eigenvectors. The following equation holds: N n=1 N * mn , eff = n=1
nth eigenvector of the MDoF system n : q n ( t ) : nth modal coordinate of the MDoF system
* Further variables in Equation (12.53) are the modal mass m n th * and the modal stiffness k n of the n mode. These parameters are defined as follows: * = M mn n n * = kn T n T
mn
= m tot
(12.60)
where m tot is the total mass of the dynamic system. The effective modal height h n of the nth mode is:
* hn *
(12.56)
2 n * mn
K n =
(12.57)
Ln - with L n = = ----Ln
N j=1
hj mj jn and Ln
= n M (12.61)
n :
* Significance of the effective modal mass m n , eff * The effective modal mass m n , eff is the lumped mass of a singlestorey substitute system which is subjected to a base shear force V bn equal to the nth modal base shear force of a multi-storey system.
The modal participation factor n is a measure for the contribution of the n-th mode to the total response of the system. It is defined as follows: n M n = ----------------T n M n
T
(12.58)
In addition the so-called effective modal mass of the nth mode is defined as:
* * mn , eff = n m n 2
If in addition the height of the single storey substitute system with * * the lumped mass m n , eff equals the modal height h n , the singlestorey system is subjected to a base moment M bn which is equal to the nth modal base moment of the multi-storey system. The following holds:
(12.59)
N
V bn =
* mn , eff
S pa, n =
j=1
fjn
(12.62)
12 Seismic Excitation
Page 12-19
12 Seismic Excitation
Page 12-20
An-Najah 2013
An-Najah 2013
* M bn = m n , eff
* S pa, n h n
j=1
fjn hj
where S pa, n is the pseudo-acceleration of the nth mode. Distribution of the internal forces If the internal forces of the entire system are to be determined, the modal equivalent static forces f jn should be computed first: f n = s n S pa, n where f n = f 1n f 2n f nn (12.65) (12.64)
h
hs
The excitation vector s n is defined according to equation (12.66) and specifies the distribution of the inertia forces due to excitation of the nth mode: sn = n M n (12.66)
hs
MDOF System
m3 m1*Sa1 m2 h hs m1 hs Vb Mb EI h1*=2.50hs
12 Seismic Excitation
Page 12-21
12 Seismic Excitation
h2*=0.72hs
sn
= M
(12.67)
Page 12-22
An-Najah 2013
An-Najah 2013
u i, max
n=1
in q n, max
(12.70)
The assumption that all maxima occur at the same instant and in the same direction yields an upper bound value for the response quantity. This assumption is commonly too conservative. Square-Root-of Sum-of-Squares (SRSS) Combination Rule
N
1*) = n ---- S pa ( n, n 2 n
(12.68)
u i, max
n=1
( in qn, max )
(12.71)
This rule is often used as the standard combination method and yields very good estimates of the total maximum response if the modes of the system are well separated. If the system has several modes with similar frequencies the SRSS rule might yield estimates which are significantly lower than the actual total maximum response. Complete Quadratic Combination (CQC) Combination Rule u i, max where u i, max and u i, max are the max. modal responses of modes j and k jk is the correlation coefficient between nodes j and k :
12 Seismic Excitation Page 12-24
The maxima of different modes do not occur at the same instant. An exact computation of the total maximum response on the basis of the maximum modal responses is hence impossible. Different methods have been developed to estimate the total maximum response from the maximum modal responses.
12 Seismic Excitation Page 12-23
j = 1k = 1
(j)
(k)
(12.72)
(j)
An-Najah 2013
An-Najah 2013
jk
Number of modes to be considered. All modes which contribute to the dynamic response of the system should be considered. In practical applications, however, only those modes are considered which contribution to the total response is above a certain threshold. It should be noted that in order to achieve the same accuracy for different response measures (e.g. displacements, shear forces, bending moments, etc.) different numbers of modes might need to be considered in the computation. For a regular building the top displacement can be estimated fairly well on the basis of the fundamental mode only. To estimate the internal forces, however, higher modes need to be considered too. According to Eurocode 8 Design of Structures for Earthquake Resistance [CEN04] all modes should be considered (starting from the lowest) until the sum of the effective modal masses m n, eff of all considered modes corresponds to at least 90% of the total mass m tot . As an alternative, Eurocode 8 allows the de* signer to show that all modes with m n , eff > 0.05m tot were considered in the computation.
This method based on random vibration theory gives exact results if the excitation is represented by a white noise. If the frequencies of the modes are well spaced apart, the result converge to those of the SRSS rule. More detailed information on this and other combination rules can be found in [Cho11] Chapter 13.7. Internal forces The aforementioned combination rules cannot only be applied on displacements but also on internal forces. The maximum modal internal forces can be determined from equivalent static forces F n, max = K u n, max , (12.74)
which, as a first option, are computed from the equivalent static displacements. Alternatively, the equivalent static forces can be determined from the inertia forces:
*) = M S ( , *) F n, max = s n S pa ( n, n n n pa n n
(12.75)
with s n being the excitation vector which represents the distribution of the inertia forces of the nth mode (see Equation 12.67). Attention: It is wrong to compute the maximum internal forces from the maximum displacement of the total response u max .
12 Seismic Excitation Page 12-25 12 Seismic Excitation Page 12-26
An-Najah 2013
An-Najah 2013
12.3.2 Step-by-step procedure The maximum response of a N-storey building can be estimated according to the following procedure: 1) Determine the properties of the MDOF system
Choose DOFs Determine mass matrix M and stiffness matrix K .
* Estimate modal damping ratios n
response spectrum for pseudo-accelerations. (The spectral dis*) should be determined in the same manner) placement S d ( n, n Computation of the maximum displacements
*) u n, max = n n S d ( n, n
F n, max = s n S pa ( n, n ) = n M n S pa ( n, n )
Computation of the maximum internal forces on the basis of the forces F n, max
( K n M ) n = 0
Compute the modal properties of the MDOF system ( M * , K * )
* = M , k * = K mn n n n n n T T
4) Estimate the total response in terms of displacements and internal forces by means of suitable combination rules. Different combination rules might be applied (ABSSUM, SRSS, CQC). Comment In order to consider the non-linear behaviour of the structure the equivalent lateral static forces F n, max can be determined from the spectral ordinate S pa ( n, n, q ) of the design spectrum for pseudo-accelerations: F n, max = s n S pa ( n, n, q ) = n M n S pa ( n, n, q ) (12.76)
n M n = ----------------T n M n 3) The maximum response of the n-th mode should be determined as described in the following. This should be done for all modes n = 1, 2, , N which require consideration.
*, For all periods T n and for the corresponding damping ratios n the spectral response S a ( n, n ) should be determined from the
12 Seismic Excitation
Page 12-27
12 Seismic Excitation
Page 12-28
An-Najah 2013
An-Najah 2013
Stiffness The horizontal stiffness of each story is k = 1 , hence the stiffness matrix K is: K = k 11 k 12 k 21 k 22 = 1 1 1 2 (12.78)
1. unit displacement u 1 = 1
This 2-DoF system corresponds to the system presented in Chapter 9 with the only difference that the 2 DoFs are swapped. Degrees of freedom (DoF) Horizontal displacements u 1 and u 2 in correspondence of the masses m 1 and m 2 Masses Both story masses have unit value, i.e. m 1 = m 2 = 1 , hence the mass matrix M is: M = m1 0 0 m2
12 Seismic Excitation
2. unit displacement u 2 = 1
Damping Damping is small and is neglected, hence the damping matrix C is: C = 0 0 0 0
12 Seismic Excitation
= 1 0 0 1
(12.77)
(12.79)
Page 12-29
Page 12-30
An-Najah 2013
An-Najah 2013
2) Natural modes of vibration and natural frequencies For each eigenvalue a natural mode of vibration and a natural frequency can be computed. (12.80)
1
2
1 2
=0
The nontrivial solution for the eigenvector = 0 exists if 2 the determinant is equal to zero: det ( K M ) = 0
2 1 = 0 det ( K M ) = det 1 2 1 2 2 2
(12.81) (12.82)
3 5 --------------- = 0.62 2
(12.86)
1 2
=0
(12.87)
can be solved for the corresponding vector 11 (fundamental 21 mode) to within a multiplicative constant: 3 5 1 --------------2 1 1 3 5 2 --------------2
(12.84)
11 21
= 0 0
(12.88)
12 Seismic Excitation
Page 12-31
12 Seismic Excitation
Page 12-32
An-Najah 2013
An-Najah 2013
Normalizing the largest coordinate of the eigenvector to unity ( 11 = 1 ), 21 becomes: 2 ( 3 5 ) -------------------------- 21 = 0 2 or 51 - = 0.62 21 = --------------2 Hence the first natural mode of vibration is: (12.91) (12.90)
Higher mode of vibration 3+ 5 2 - leads to the The largest eigenvalue 2 = --------------2 2. circular natural frequency 2 = 3+ 5 --------------- = 1.62 2 (12.93)
In analogy to the fundamental mode, the second mode of vibra2 tion can be computed introducing the second eigenvalue 2 into the system of equations: 3+ 5 1 --------------2 1 1 3+ 5 2 --------------2
12 22
= 0 0
(12.94)
The first row yields following equation: 11 21 = 1 5 1 = 0.62 --------------2 1 (12.92) 2 ( 3 + 5 ) --------------------------- 12 1 22 = 0 2 (12.95)
Normalizing the largest coordinate of the eigenvector to unity ( 22 = 1 ), 12 becomes: 2 ( 3 + 5 ) --------------------------- 12 1 = 0 2 or: 2 1 5 - = 0.62 12 = --------------= --------------2 1+ 5 (12.97) (12.96)
12 Seismic Excitation
Page 12-33
12 Seismic Excitation
Page 12-34
An-Najah 2013
An-Najah 2013
Orthogonality with respect to the mass matrix The modal mass matrix M * is:
51 51 --------------1 --------------1 0 2 2 0 1 51 51 --------------1 --------------1 2 2 1 1 51 --------------2 0 51 - 1 + -------------- 2
2
=
51 --------------2
51 --------------2 1
2
51 --------------2 1
3) Orthogonality of modes In the following the orthogonality of the modes of vibration should be checked. Hence, following matrix of the eigenvectors is needed: 1 = 51 --------------2 51 --------------2 1
51 - 1 + -------------- 2 0
11 12 21 22
(12.99)
12 Seismic Excitation
Page 12-35
12 Seismic Excitation
Page 12-36
An-Najah 2013
An-Najah 2013
Orthogonality with respect to the stiffness matrix The modal stiffness matrix K * is:
51 51 --------------1 --------------1 1 2 2 1 2 51 51 --------------1 --------------1 2 2 1
51 2 2
K* = K =
12.4.3 Equation of motion in modal coordinates The equation of motion in modal coordinates of a system without damping ( C * = 0 ) is: + K * q = L u ( t ) M* q g where: (12.101) L = L1 L2 , and q = q1 q2 is the vector of the modal coordinates (12.107)
51 - 2 + ( 5 1 ) + -------------- 2
Computation of the elements of the vector L : 51 --------------2 L = M = 1 0 1 = 1.62 0 1 1 0.382 51 --------------1 2 (12.108)
T
(12.102)
2 2 52+52 5+14 51 51 - + --------------- 1 = --------------------------------------------------------------- = 0 k* 12 = -------------- 4 2 2
(12.103) k* 21 = k* 12 = 0 (12.104)
12 Seismic Excitation
Page 12-37
12 Seismic Excitation
Page 12-38
An-Najah 2013
An-Najah 2013
The influence vector represents the displacement of the masses resulting from the static application of a unit ground displacement u g = 1 : = 1 1 (12.109)
Additional important modal quantities The modal participation factor n is defined as: Ln n = --------mn * (12.114)
The substitution of L into the equation of motion in modal coordinates + K * q = L u ( t ) M* q g leads to: 1.38 0 q 1 + 0.528 0 q 1 = 1.62 u ( t ) g 0 1.38 0 3.618 0.382 q2 q2 (12.111) Checking the circular natural frequency computed using modal coordinates against the results of Section 12.4.2: 1 = k 11 * -----------= m 11 * k 22 * -----------= m 22 * 0.528 ------------ = 0.62 OK! 1.38 3.618 ------------ = 1.62 OK! 1.38 (12.112) (12.110)
and substituting L n and m n * into this definition gives following values for 1 and 2 : L1 1.62 - = --------- = 1.17 1 = --------m1 * 1.38 L2 0.382 - = ------------ = 0.28 2 = --------m2 * 1.38 The effective modal mass is defined as:
* * mn , eff = n m n 2
(12.115)
(12.116)
(12.117)
and substituting n and m n * into this definition gives following * * values for m 1 , eff and m 2, eff :
* * 2 m1 , eff = 1 m 1 = 1.17 1.38 = 1.894 * * 2 m2 , eff = 2 m 2 = 0.28 1.38 = 0.106 2 2
2 =
(12.113)
12 Seismic Excitation
Page 12-39
12 Seismic Excitation
Page 12-40
An-Najah 2013
An-Najah 2013
12.4.4 Response spectrum method The 2-DoF system analysed in the previous Sections shall be used to illustrate the response spectrum method. For this reason real masses and stiffnesses shall be assumed. The seismic action on the 2-DoF system is represented by the elastic response spectrum of the El Centro earthquake. 1) Model Similar to Section 12.4.1, however with a new definition of masses and stiffnesses: m 1 = m 2 = 1kg Hence, the mass matrix is: M = m1 0 0 m2 = 1 0 kg 0 1
2) Natural modes of vibration and natural frequencies The results of the previous Sections computed using unit masses and unit stiffnesses shall be multiplied by the factor: k --- = m 100kg s -----------------------= 1kg
2
100s
(12.123)
Fundamental mode Natural frequency: Natural period: Higher vibration mode Natural frequency: Natural period: 2 = 1.62 100s
1
1 = 0.62 100s
= 6.2 Hz
= 16.2 Hz
The stiffness chosen for each story is k 1 = k 2 = k = 100 N/m and an appropriate units transformation leads to: k = 100 N/m = 100 kgm/s 2 m 1 = 100 kg/s 2 Hence, the stiffness matrix is: K = k 11 k 12 k 21 k 22
12 Seismic Excitation
The eigenvectors are dimensionless quantities and remain unchanged: 1 = 51 --------------2 51 --------------2 1
(12.121) = (12.122) 11 12 21 22
(12.124)
Page 12-41
12 Seismic Excitation
Page 12-42
An-Najah 2013
An-Najah 2013
4) Peak modal response The peak modal response of both vibration modes can be computed like in the case of SDoF systems using the spectral value given by the relevant response spectrum. The peak value of the modal coordinate q 1 is: L1 - S ( , ) = 1 S d ( 1, 1 ) q 1, max = --------m1 * d 1 1 where: S d ( 1, 1 ) : spectral displacement for a natural frequency 1 and a damping 1 (here 1 = 5% ) (12.130)
1 0 kg u 1 + 100 100 kg s 2 u 1 = 1 0 kg 1 u ( t ) g 100 200 0 1 01 1 u u 2 2 (12.126) Variables transformation in modal coordinates q 1 and q 2 : u = q where: : Modal Matrix, i.e. the matrix of the eigenvectors The equation of motion in modal coordinates q 1 and q 2 (without damping) is: + K * q = L u ( t ) M* q g (12.128) (12.127)
If an acceleration instead of a displacement response spectrum is used, then the peak value of the modal coordinate q 1 is: L1 1 1 - S pa ( 1, 1 ) = ----- S pa ( 1, 1 ) - ----q 1, max = --------2 m1 * 2 1 1 where: (12.131)
1.38 0 kg q 1 + 52.8 0 kg s 2 q 1 = 1.62 kg u ( t ) g q 0 1.38 0 362 0.382 q 2 2 (12.129) yielding the equation of motion in modal coordinates of two independent SDoF systems.
12 Seismic Excitation
Page 12-43
12 Seismic Excitation
Page 12-44
An-Najah 2013
An-Najah 2013
The spectral values given by the elastic acceleration response spectrum of the El Centro earthquake for the periods T 1 and T 2 are: S pa1 = 4.25 m/s 2 S pa2 = 7.34 10 Pseudo acceleration [m/s2] Spa2 = 7.34 m/s2 m/s 2 and (12.132) (12.133)
5) Inverse transformation
The peak deformations and internal forces belonging to each mode of vibration in the original reference system are obtained by multiplying the relevant eigenvector with the corresponding peak value of the modal coordinate.
Fundamental mode
u max = q 1, max 1 = 0.130m
(1)
= 5%
100 200
1 = 130 mm 0.62 81
(12.136)
0.081
(12.137)
2 5 S pa1 = 4.25 m/s
(12.138)
(12.139)
0 0.01
0.725
3.2
1.62kg 1 - 4.25 m/s 2 = 0.130m - --------------------q 1, max = ---------------1.38kg ( 6.2Hz ) 2 0.38kg 1 - 7.34 m/s 2 = 0.008m - -----------------------q 2, max = ---------------1.38kg ( 16.2Hz ) 2
(12.134) (12.135)
12 Seismic Excitation
Page 12-45
12 Seismic Excitation
Page 12-46
An-Najah 2013
An-Najah 2013
6) Combination (12.140) The total peak response is obtained from the peak response of the single vibration modes using e.g. the SRSS combination rule (SRSS = Square Root of the Sum of Squares). Peak displacements u 1, max =
0.005 =
0.008
(12.141)
Alternatively (allow an approximate consideration of nonlinearities): s 2 = 2 M 2 = 0.28 1 0 0.62 = 0.173 01 1 0.28
(1) f max = s 2 S pa2 = 7.34 0.173 = 1.3
k=1
( u1
2
(n) 2
) =
(12.142) u 2, max =
0.28
2.1
(12.143)
k=1
( u2
(n) 2
) =
In this case the total peak displacements are almost identical to the peak displacements of the fundamental mode. The relatively small contributions due to the second vibration mode basically disappear because of the SRSS combination rule.
12 Seismic Excitation
Page 12-47
12 Seismic Excitation
Page 12-48
An-Najah 2013
An-Najah 2013
Peak sectional forces (Shear force V) Upper shear force: V 1, max = ( 4.9N ) + ( 1.3N ) = 5.1N (12.146) Lower shear force: V 2, max = ( 8.1N ) + ( 0.8N ) = 8.1N (12.147) Compared to the peak sectional forces due to the fundamental mode, the total peak sectional forces show a slight increase in the upper story of the 2-DoF system.
2 2 2 2
Case study 1 Pay attention to following pitfall! It is wrong to compute the total peak sectional forces using the total peak displacements: V 1, max 100N m ( 0.130m 0.081m ) = 4.9N V 2, max 100N m 0.081m = 8.1N The sectional forces would be underestimated. (12.148) (12.149) Masses: m 1 = 1.0kg m 2 = 1.0kg Stiffnesses: k 1 = 100N/m k 2 = 100N/m
Case study 1 corresponds to the model analysed in Section 12.4.4. Case study 2 represents a dynamic system where the second vibration mode is important.
12 Seismic Excitation
Page 12-49
12 Seismic Excitation
Page 12-50
An-Najah 2013
An-Najah 2013
2) Results Dynamic properties Case study 1 Periods: T 1 = 1.02s T 2 = 0.39s Eigenvectors: 1: 11 = 1 , 2: 12 = 1 , Part. factors: 21 = 0.62 22 = 1.62 1 = 1.17 2 = 0.17 Eigenvectors: 1: 11 = 1 , 2: 12 = 1 , Part. factors: 21 = 0.27 22 = 0.37 1 = 2.14 2 = 1.14 Case study 2 Periods: T 1 = 0.74s T 2 = 0.54s
Demand Case study 1 Displacements: 1: 2: Sum: SRSS: Time-history: = 0.129m = 0.005m = 0.134m = 0.130m = 0.130m Case study 2 Displacements: 1: 2: Sum: SRSS: Time-history: = 0.130m = 0.072m = 0.202m = 0.148m = 0.165m
Note that in this case the eigenvectors are normalized to yield unit displacement at the top of the second story. Therefore, the eigenvectors and the participation factors of case study 1 differ from the values obtained in previous sections.
12 Seismic Excitation
Page 12-51
12 Seismic Excitation
Page 12-52
An-Najah 2013
An-Najah 2013
First mode
Displacement [cm] 10 Displacement [cm] 10
First mode
10 20 0 20
10 20 0 20
10
15
20
10
15
20
Second mode
Displacement [cm] 10 Displacement [cm] 10
Second mode
10 20 0 20
10 20 0 20
10
15
20
10
15
20
Sum
Displacement [cm] 10 Displacement [cm] 10
Sum
10 20 0
10 20 0
10 Time [s]
15
20
12 Seismic Excitation
Page 12-53
12 Seismic Excitation
Page 12-54
An-Najah 2013
An-Najah 2013
Time-histories: Summary
20
Blank page
Case study 1
10 Displacement [cm]
10
20 0 20
Case study 2
10 Displacement [cm]
10
20 0
12 Seismic Excitation
Page 12-55
12 Seismic Excitation
Page 12-56
An-Najah 2013
An-Najah 2013
13.1.1 Dynamic action a) People-induced vibrations - Pedestrian bridges - Floors with walking people - Floors for sport or dance activities - Floor with fixed seating and spectator galleries - High-diving platforms b) Machinery-induced vibrations - Machine foundations and supports - Bell towers - Structure-borne sound - Ground-transmitted vibrations c) Wind-induced vibrations - Buildings - Towers, chimneys and masts - Bridges - Cantilevered roofs d) Vibrations induced by traffic and construction activity - Roads and bridges - Railways - Construction works
Nevertheless vibration sensitive structures are often designed for static loads only
Goal of this chapter Give an overview of possible causes of vibration problems in buildings and of potential countermeasures Description of practical cases with vibration rehabilitation
13 Vibration Problems
Page 13-1
13 Vibration Problems
Page 13-2
An-Najah 2013
An-Najah 2013
13.1.2 References
[Bac+97] Bachmann et al.:Vibration Problems in Structures. ISBN 3-7643-5148-9. Birkhuser Verlag, Basel 1997. [BB88] Baumann K., Bachmann H.: Durch Menschen verursachte dynamische Lasten und deren Auswirkungen auf Balkentragwerke. IBK Bericht Nr. 7501-3, 1988. Bachmann H., Weber B.: Tuned Vibration Absorbers for Damping of Lively Structures. Structural Engineering International, No. 1, 1995. SIA: Wind Kommentar zum Kapitel 6 der Normen SIA 261 und 261/1 (2003) Einwirkungen auf Tragwerke. SIA Dokumentation D0188. Zrich 2006. SIA: Aktuelle Probleme der Brckendynamik. SIA Dokumentation D0198. Zrich 2003. SIA: Dynamische Probleme bei Brcken- und Hochbauten. SIA Dokumentation D0138. Zrich 1996. Vogt R., Bachmann H.: Dynamische Krfte beim Klatschen, Fussstampfen und Wippen. IBK Bericht Nr. 7501-4, 1987.
[BW95]
Tr rasmissibility TR
7 6 5 4 3 2 1 0 0 0.5 1
[SIA06]
2
1.5 2 2.5 3 3.5 4
/n
- High tuning (subcritical excitation) - Low tuning (supercritical excitation) Amplitude limitation
13 Vibration Problems
Page 13-3
13 Vibration Problems
Page 13-4
An-Najah 2013
An-Najah 2013
13.2.3 Calculation methods Computation of the natural frequencies The natural frequencies of structures have to be determined by means of realistic models. Approximate formulas that are often found in design codes and literature shall be checked carefully. Computation of the Amplitude If the frequency of a harmonic of the excitation coincides with a natural frequency of the structure (resonance), the maximum deflection of the structure can be estimated as follows (See Chapter 5):
Fo u p = ----- V ( ) cos ( t ) k
f a, 1
(13.1)
(13.2)
The maximum velocity and the maximum acceleration can be determined from Equation (13.2) as follows:
Fo 1 u ----max = u max = ----k 2
2 2 Fo 1 ----u max = u max = ----k 2
(13.3) (13.4)
Increase of the damping - Installation of dampers or absorbers - Plastic energy dissipation Tuned Mass Dampers
13 Vibration Problems Page 13-5
13 Vibration Problems
Page 13-6
An-Najah 2013
An-Najah 2013
The amplitude of the nth harmonic component of a force generated by people excitation is proportional to the mass of the person ( F o = G n = g M n , see Equation (13.8)).
M n 1 g n k- g 2 g M n 1 - ----- = --- ----- = M ---- ----------- ---------------------u max = ---------------------k 2 m k 2 m 2 g n M - -----------u max = ---m 2
(13.5) (13.6)
Remarks - A soft structure is more prone to vibration than a rigid one. See Equations (13.2) to (13.4). - The acceleration amplitude is directly proportional on the ratio of the mass of the people to the building mass.
F ( t ) = a0 +
n=1
[ a n cos ( n 0 t ) + b n sin ( n 0 t ) ]
(13.7)
F(t) = G +
n=1
G n sin ( n 2 f 0 t n )
(13.8)
Where: - G = Weight of the person - n = Fourier coefficient for the nth harmonic - G n = Amplitude of the nth harmonic of the excitation force - f 0 = Step frequency of the excitation force - n = Phase shift of the nth harmonic ( 1 = 0 ) - n = Number of the nth harmonic - N = Number of considered harmonics
13 Vibration Problems
Page 13-7
13 Vibration Problems
Page 13-8
An-Najah 2013
An-Najah 2013
The steady-state response of a SDoF system under periodic excitation can be computed in analogy to Chapter 6 as:
N
u ( t ) = u0 ( t ) +
n=1
un ( t )
(13.9)
Where
G - (Static displacement) u 0 ( t ) = --k G n ( 1 n ) sin ( n 0 t n ) 2 n cos ( n 0 t n ) - --------------------------------------------------------------------------------------------------------------u n ( t ) = -------------2 2 2 k ( 1 n ) + ( 2 n ) 0 = 2 f0
2
(13.10)
n 0 n = --------n
13 Vibration Problems
Page 13-9
13 Vibration Problems
Page 13-10
An-Najah 2013
An-Najah 2013
13 Vibration Problems
Page 13-11
13 Vibration Problems
Page 13-12
An-Najah 2013
An-Najah 2013
13 Vibration Problems
Page 13-13
13 Vibration Problems
Page 13-14
An-Najah 2013
An-Najah 2013
Remarks regarding Table G.2 - Coefficients and phase angles represent averages. - Phase angles have strong scattering and therefore, in many cases, it is difficult to provide reasonable values. In such cases (e.g,. running and dancing) in Table G.2 no values are specified. - Decisive are cases in which resonance occurs. In such cases the phase angle no longer plays a role. - Coefficients and phase were checked and discussed internationally. 13.3.2 Example: Jumping on an RC beam Here the same example as in Section 6.1.3 is considered again:
RC Beam The RC beam has a length of 19 meters. The natural frequency is thus:
f n = 2Hz .
(13.13)
Excitation Here jumping is described by means of the Fourier-series given in Table G.2. In Section 6.1 periodic excitation, jumping was described by means of a half-sine function. Jumping frequency: Contact time:
f 0 = 2Hz
(13.14)
(13.15) Weight of the person: G = 0.70kN Results Excel Table: Equation (13.2):
u max = 0.043m
(13.16)
(13.17)
(13.18) Remarks - Shape of the excitation similar as half-sine - Maximum deflection very close to the solution obtained by means of the half-sine function
13 Vibration Problems
Page 13-15
13 Vibration Problems
Page 13-16
An-Najah 2013
An-Najah 2013
Excitation
3.5
Static component (n=0)
13.3.3 Footbridges Frequency tuning - Vertical: Avoid natural frequencies between 1.6 and 2.4Hz. In the case of structures with low damping (Steel), avoid also natural frequencies from 3.5 to 4.5Hz. - Horizontal transverse: Avoid natural frequencies between 0.7 and 1.3Hz (absolutely safe: fht,1 > 3.4Hz). - Horizontal longitudinal: Avoid natural frequencies between 1.6 and 2.4Hz. Amplitude limitation - Calculation of the acceleration maximum amplitude.
a max < ca. 0.5m s = 5% g
2
3.0 2.5
First harmonics (n=1) Second harmonics (n=2) Third harmonics (n=3) Total (3 harmonics)
2.0 1.5 1.0 0.5 0.0 00 -0.5 -1.0 -1.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0.8
0.9
Time (s)
(13.19)
Response
0.0500
Static component (n=0)
- When walking or running, the effectiveness of people is limited, because the forces are not always applied at midspan;
System
Displacement D [m]
0.0200 0.0100 0.0000 -0.0100 -0.0200 0 0200 -0.0300 -0.0400 -0.0500 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)
13 Vibration Problems
Page 13-17
13 Vibration Problems
Page 13-18
An-Najah 2013
An-Najah 2013
- People need a finite number of steps in order to cross the bridge (This limited excitation time may be too short to reach the maximum amplitude)
1
0.2
0.8
0.1
abs(uj) / umax
0.6
0.4
0.2
0 0 5 10 15 20 25 30 35 40 45 50
Cycle
- Not all people walk in the step (Exception: Lateral vibrations Synchronisation effect) To take into account the specificities of the amplitude limitation, sophisticated methods are available. From [Bac+97] the following one is adopted:
a max = 4 f y [ m s ]
2 2 2
The acceleration a max given in Equation (13.20) is the acceleration generated by one person crossing the footbridge. If n people are on the bridge at the same time, the maximum acceleration is typically less than n a max because not all people walk in step across the bridge. . The square root of the number of people is often chosen as the multiplication factor, i.e. n a max
(13.20)
Where: - y : Static deflection at half the span - : Fourier coefficient - : dynamic amplification factor
13 Vibration Problems Page 13-19
13 Vibration Problems
Page 13-20
An-Najah 2013
An-Najah 2013
Rough estimate of maximum displacement and acceleration The maximum displacement, and the maximum acceleration can be estimated by means of Equations (13.2) and (13.4):
1G = 0.001128m = 0.11cm u max,st = ------ = -------886 Kn G 1 1 0.4 1 - ----- = 1 -------------- -------------------= 0.0133m = 1.33cm u max,1 = -------------886 2 0.017 Kn 2
(13.21)
(13.23) (13.24)
Estimate of the maximum displacement and acceleration using the improved method The maximum acceleration is computed by means of Equation (13.20) as follows: - Stiffness at mid span: K n = 886kN m - Natural frequency: f n = 2Hz - Damping: = 0.017 Excitation - Walking with f 0 = 2Hz according to Table G.2. - Step length: S = 0.70m - Weight of the Person: G = 1kN Amplification factor: From Equation (13.20):
2 2 1.00 - 0.40 23 = 1.64m s 2 a max = 4 2 --------886
(13.29)
13 Vibration Problems
Page 13-21
13 Vibration Problems
Page 13-22
An-Najah 2013
An-Najah 2013
Computation of displacements and accelerations by means of the FE Programme ABAQUS Displacements and accelerations are computed by means of time-history analysis: - Excitation
Time [s]
1.4
F1 F2 F3
F15 F28
10
12
14
16
Time [s]
Time [s]
13 Vibration Problems
Page 13-23
13 Vibration Problems
Page 13-24
An-Najah 2013
An-Najah 2013
Remarks - The refined method and the time history calculations show lower values compared to the rough method; - The refined method and the time history calculations are in good agreement; - The time history of the displacement is not symmetric compared to the time axis, because of the static component of the displacement caused by the weigh of the crossing person; - The time history of the acceleration is symmetric compared to the time axis, because there is no static component of the acceleration.
13.3.4 Floors in residential and office buildings Frequency tuning - If the excitation is generated by walking ( f max 2.4Hz ), the following natural frequencies shall be exceeded:
Damping > 5% < 5% Natural frequency [Hz] >5 > 7.5 Remark Avoid resonance due to the second harmonic Avoid resonance due to the third harmonic
Amplitude limitation Swinging footbridge on the Internet http://www.londonmillenniumbridge.com/ http://www.youtube.com - Calculation of the acceleration maximum amplitude
a max < ca. 0.05m s = 0.5% g
2
(13.30)
- Because of the many non-structural components (wallpaper, furniture, suspended ceilings, technical floors, partitions, ....) it is difficult to estimate the dynamic properties of the floors. - Where possible measure the dynamic properties. Response of people to vibrations The sensitivity of people to vibration depends on many parameters:
13 Vibration Problems
Page 13-25
13 Vibration Problems
Page 13-26
An-Najah 2013
An-Najah 2013
- Position (standing, sitting, lying) - Direction of the action compared to the spinal column - Activity (resting, walking, running, ...) - Type of vibration - ....
Description Barely noticeable Clearly noticeable Disturbing Not tolerable Frequency 1 to 10 Hz amax [m/s2] 0.034 0.1 0.55 1.8 Frequency 10 to 100 Hz vmax [m/s] 0.0005 0.0013 0.0068 0.0138
Vertical harmonic vibration action on a standing person. Accepted averages; scatters up to a factor of 2 is possible (from [Bac+97])
G2 limit for vibrations transverse to the spinal column ISO 2631 standard a eff =
T 2 1 -- a ( t ) dt T 0
(13.31)
Where T is the period of time over which the effective acceleration was measured. 3 limits are defined: - G1: Reduced comfort boundary - G2: Fatigue-decreased proficiency boundary ~ 3 x G1 - G3: Exposure limit ~ 6 x G1
13 Vibration Problems
Page 13-27
13 Vibration Problems
Page 13-28
An-Najah 2013
An-Najah 2013
13.3.5 Gyms and dance halls Due to gymnastics or dancing, very large dynamic forces are generated. This is readily understandable when the Fourier coefficients in Table G2 are considered: - Walking: 1 = 0.4 , 2 = 0.1 , 3 = 0.1 - Running: 1 = 1.6 , 2 = 0.7 , 3 = 0.2 - Jumping: 1 = 1.9 , 2 = 1.6 , 3 = 1.1 - Dancing: 1 = 0.5 , 2 = 0.15 , 3 = 0.1 (however: a - many people moving rhythmically. b - certain dances are very similar to jumping) Frequency tuning - If the excitation is generated through jumping ( f max 3.4Hz ) or dancing ( f max 3.0Hz ), then the following natural frequencies shall be exceeded: Construction
Reinforced concrete Prestressed concrete Composite structures Steel Gyms Natural frequency [Hz] > 7.5 > 8.0 > 8.5 > 9.0 Dance halls Natural frequency [Hz] > 6.5 > 7.0 > 7.5 > 8.0
(13.32)
- Limits depend on the activity, if e.g. people are sitting in the dance hall, as well, this limit shall be reduced. - Because of the large forces that can be generated through these activities, the dynamic characteristics of the structure shall be estimated as precisely as possible. 13.3.6 Concert halls, stands and diving platforms See [Bac+97].
13 Vibration Problems
Page 13-29
13 Vibration Problems
Page 13-30
An-Najah 2013
An-Najah 2013
Vortex shedding: Periodic action transversely to the wind direction - Vortex are not shedded left and right at the same time. If the time-interval of the vortex shedding is equal to the oscillation period of the structure, resonance excitation occurs.
fe d u crit = ---------S
(13.33)
Where
u crit : fe : d: S:
Critical wind velocity Natural frequency of the structure transverse to the wind direction Diameter of the structure Strouhal number (about 0.2 for circular cross sections)
Buffeting: Periodic action in wind direction - Vortex detached from an obstacle hit the structure
13 Vibration Problems
Page 13-31
13 Vibration Problems
Page 13-32
An-Najah 2013
An-Najah 2013
Gallopping and Flutter: Unstable interaction between wind flow and structural motion - Gallopping: Motion of the structure transversely to the flow direction. - Flutter: Combined flexural-torsional motion of the structure. Work done by wind forces during flutter
[Den85]
13 Vibration Problems
Page 13-33
13 Vibration Problems
Page 13-34
An-Najah 2013
An-Najah 2013
2 [ mH + i ( cH + cT ) + ( kH + kT ) ] UH + [ i cT kT ] UT = FH 2 [ i cT kT ] UH + [ mT + i cT + kT ] UT = 0
(13.36) To facilitate the solution of the system, some dimensionless parameters are now introduced: = mT mH :
T = H = kT mT : kH mH :
Mass ratio (TMD Mass/Mass of the structure) Natural frequency of the TMD Natural frequency of the structure without TMD Ratio of the natural frequencies Ratio of the excitation frequency to the natural frequency of the structure Damping ratio of the TMD Damping ratio of the structure Static deformation of the structure
T : H :
U H0 = F H k H :
(13.34)
For an harmonic excitation of the type F ( t ) = F H cos ( t ) , a possible ansatz for the steady-state part of the solution is:
uH = UH e
it
uT = UT e
it
F ( t ) = FH e
it
(13.35)
Using the complex numbers formulation allows a particularly elegant solution to the problem. The equations of motion become:
(13.37)
13 Vibration Problems
Page 13-35
13 Vibration Problems
Page 13-36
An-Najah 2013
An-Najah 2013
The system of equation can be easily solved using Maple, and we obtain the following expression for the amplification function U H U H0 :
UH ( ) + 2i T -------- = -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------2 2 2 2 2 2 2 2 2 2 U H0 [ ( ) ( 1 ) + ( 4 H T ) ] + 2i [ ( ) H + ( 1 ) T ]
2 2
Thus, the norm of the dynamic amplification function U H U H0 can be easily calculated:
UH = -------U H0 ( ) + ( 2 T ) -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------2 2 2 2 2 2 2 2 2 2 2 2 [ ( ) ( 1 ) + ( 4 H T ) ] + 4 [ ( ) H + ( 1 ) T ]
2 2 2 2
(13.38)
(13.44) A similar procedure can be followed to compute the dynamic amplification function U T U H0 . Next figure show a representation of Equation (13.44) in function of for an undamped structure H = 0 . Curves for different values of the parameters , and T are provided.
16
The complex expression given in Equation (13.38) shall now be converted into the form:
z = x + iy
or
U H = U H0 ( x + iy )
(13.39)
The displacement U H has therefore two components: 1) One that is in phase with the displacement U H0 and 2) one with a phase shift equal to 4 . From the vectorial sum of these two components the norm of U H can be computed as:
U H = U H0 x + y
2 2
]T=0
]T=inf.
(13.40)
UH/UH0 [-]
12
J = 1/20 E=1 P
(13.41)
]T=0.32
4
(13.42)
U H = U H0 A +B -----------------2 2 C +D
2 2
(13.43)
: [-]
13 Vibration Problems
Page 13-37
13 Vibration Problems
Page 13-38
An-Najah 2013
An-Najah 2013
13.6.3 Optimum TMD parameters Based on observations and consideration at the previous image Den Hartog found optimum TMD parameters for an undamped structure:
No TMD (T=inf.)
40
= 0.01 = 0.01
UH/UH0 [-]
or
3 --------------------3 8(1 + )
1 opt = ----------1+
(13.45)
30
20
(13.46)
10
fT=0.98fT,opt fT=fT,opt
These optimum TMD parameters can be applied also to lowly damped structures providing good response results. 13.6.4 Important remarks on TMD The frequency tuning of the TMD shall be quite precise The compliance with the optimum damping is less important Design charts for TMDs shall be computed numerically
UH/UH0 [-]
[-]
No TMD (T=inf.)
40
= 0.01 = 0.01
TMDs are most effective when the damping of the structure is low It is not worth increasing the mass ratio too much For large mass ratios, the amplitude of the TMD oscillations reduce Meaningful mass ratios are 3-5% The exact tuning of the TMD occurs experimentally, therefore great care should be paid to construction details.
30
20
[-]
13 Vibration Problems
Page 13-39
13 Vibration Problems
Page 13-40
An-Najah 2013
An-Najah 2013
13 Vibration Problems
Page 13-41
13 Vibration Problems
Page 13-42
An-Najah 2013
An-Najah 2013
An-Najah 2013
An-Najah 2013
Characteristics of the TMD A close-up of the TMD is shown in Figure 14.2. We can see:
The 4 springs that define the stiffness K T of the TMD The 4 viscous dampers that define the damping constant c opt of the TMD The mass M T , which is made up by a concrete block and two side container filled with lead spheres. The lead spheres are used for the fine-tuning of the TMD.
14.2 Parameters
14.2.1 Footbridge (Computed, without TMD) Modal mass: Modal stiffness: Natural frequency:
M H = 5300kg K H = 861kN m f H = 2.03Hz
(Computed with TMD mass: f = 1.97Hz ) 14.2.2 Tuned Mass Damper (Computed) Mass: Mass ratio:
M T = 310kg MT = -------- = 0.0585 = 5.85% MH fH f opt = ------------ = 1.92Hz 1+
Natural frequency:
Damping constant:
Figure 14.2: Close-up of the TMD.
Page 14-3
Page 14-4
An-Najah 2013
An-Najah 2013
Figure 14.3: Free decay test with locked TMD: Displacement at midspan.
Evaluation:
Logarithmic decrement Region 1 Average amplitude: ~30mm Region 2 Average amplitude: ~14mm Region 3 Average amplitude: ~6mm
1 - ln 41.36 = ------------- = 0.081 8 21.66 1 19.91 - ln ------------ = 0.090 = -8 9.68 1 8.13 - ln --------- = 0.092 = -8 3.90
Damping ratio
0.081 - = 1.29% H = -----------2 0.090 - = 1.43% H = -----------2 0.092 - = 1.46% H = -----------2
Typical results of the experiments are presented and briefly commented in the following sections.
Page 14-5
Page 14-6
An-Najah 2013
An-Najah 2013
1.5
1.0
0.5
0.0 0.0
0.5
1.0
3.0
3.5
4.0
Figure 14.4: Free decay test with locked TMD: Fourier-spectrum of the displacement at midspan.
These tests are carried out with locked TMD. In order to investigate the effect of the TMD on the vibrations of the system, the test of Section 14.5.1 is repeated with free TMD (see Section 14.5.3). Remark The results presented in Section 14.5.1 and those presented in Section 14.5.2 and 14.5.3 belongs to two different series of tests carried out at different point in time. Between these test series the test setup was completely disassembled and reassembled. Slight differences in the assemblage of the test setup (support!) may have led to slightly different natural frequencies of the system.
The measured natural frequency of the footbridge with locked TMD is equal to: f = 1.89Hz (14.1)
This value is less than the value given in Section 14.2.1. This can be explained with the large amplitude of vibration at the start of the test, which causing the opening of cracks in the web of the beam, hence reducing its stiffness. The second peak in the spectrum corresponds to f = 1.98Hz , wich is in good agreement with Section 14.2.1.
Page 14-7
Page 14-8
An-Najah 2013
An-Najah 2013
5 6 Time [s]
10
Figure 14.7: Sandbag test with locked TMD: Acceleration at quarterpoint of the span.
0.040 "Spectral acceleration"
f1=2.00Hz
f3=18.06Hz
f3=18.06Hz
10 Frequency [Hz]
15
20
10 Frequency [Hz]
15
20
Figure 14.6: Sandbag test with locked TMD: Fourier-spectrum of the acceleration at midspan.
Figure 14.8: Sandbag test with locked TMD: Fourier-spectrum of the acceleration at quarter-point of the span.
Page 14-9
Page 14-10
An-Najah 2013
An-Najah 2013
Remarks With the sandbag test in principle all frequencies can be excited. Figures 14.5 and 14.7 show a high-frequency vibration, which is superimposed on a fundamental vibration; The Fourier amplitude spectrum shows prominent peaks at the first and third natural frequencies of the system (Footbridge with locked TMD); The second mode of vibration of the system is not excited, because the sandbag lands in a node of the second eigenvector. At midspan, the amplitude of the vibration due to the first mode of vibration is greater than at quarter-point. The amplitude of the vibration due to the third mode of vibration, however, is about the same in both places. This is to be expected, if the shape of the first and third eigenvectors is considered. The vibration amplitude is relatively small, therefore, the measured first natural frequency f 1 = 2.0Hz in good agreement with the computation provided in Section 14.2.1.
5 6 Time [s]
10
f1=2.00Hz
f3=18.31Hz
10 Frequency [Hz]
15
20
Figure 14.10: Sandbag test with locked TMD: Fourier-spectrum of the acceleration at midspan.
Page 14-11
Page 14-12
An-Najah 2013
An-Najah 2013
Remarks
1.0 Acc. at quarter-point [m/s2] 0.5 0.0 -0.5 -1.0 0
When the sandbag lands at quarter-point of the bridge, the second mode of vibration of the system is strongly excited. Its contribution to the overall vibration at quarter-point of the footbridge is clearly shown in Figures 14.11 and 14.12. The acceleration sensor located at midspan of the footbridge lays in a node of the second mode of vibration, and as expected in figures 14.9 and 14.10 the contribution of the second mode is vanishingly small.
1 2 3 4 5 6 Time [s] 7 8 9 10
Figure 14.11: Sandbag test with locked TMD: Acceleration at quarterpoint of the span.
0.040 "Spectral acceleration" 0.030 0.020 0.010 0.000 0
f2=8.84Hz f1=2.00Hz
f3=18.31Hz
10 Frequency [Hz]
15
20
Figure 14.12: Sandbag test with locked TMD: Fourier-spectrum of the acceleration at quarter-point of the span.
Page 14-13
Page 14-14
An-Najah 2013
An-Najah 2013
Remarks With active (free) TMD the first and the third natural frequencies of the bridge are excited. As expected, these frequencies are slightly larger than the natural frequencies of the system (bridge with locked TMD), which are given in Figure 14.6. This is because the mass of the TMD is no longer locked and can vibrate freely. The effect of the TMD is clearly shown in Figure 14.14. The amplitude of the peak in the first natural frequency is much smaller than in Figure 14.6. The amplitude of the peak at the third natural frequency is practically the same. The third mode of vibration is only marginally damped by the TMD. In the two comments above, the natural frequencies are mentioned in quotes, because by releasing the TMD number and properties of the natural vibrations of the system change. A direct comparison with the natural vibrations of system with locked TMD is only qualitatively possible.
5 6 Time [s]
10
0.010 0.000 0
f3=19.04Hz
10 Frequency [Hz]
15
20
Figure 14.14: Sandbag test with free TMD: Fourier-spectrum of the acceleration at midspan.
Page 14-15
Page 14-16
An-Najah 2013
An-Najah 2013
Test results
Displacement at midspan [mm] 2.0 1.0 0.0 -1.0 -2.0 0
10
15
20 Time [s]
25
30
35
40
Figure 14.15: One person walking with 3 Hz: Displacement at midspan with locked TMD.
Acceleration at midspan [m/s2] 1.0 0.5 0.0 -0.5 -1.0 0
10
15
20 Time [s]
25
30
35
40
Figure 14.16: One person walking with 3 Hz: Acceleration at midspan with locked TMD.
Page 14-17
Page 14-18
An-Najah 2013
An-Najah 2013
0.010
0.005
0.000 0
Figure 14.17: One person walking with 3 Hz: Fourier-Spectrum of the acceleration at midspan with locked TMD.
(Measured: d st = 1.22mm ) Walking velocity: Crossing time: Number of cycles: Amplification factor: Max. acceleration:
v = S f 0 = 0.7 1.95 = 1.365m s t = L v = 17.40 1.365 = 12.74s N = t f n = 12.74 1.95 = 25 = 22
H = 1.6%
An-Najah 2013
An-Najah 2013
Test results
Displacement at midspan [mm] 10 5 0 -5 -10 0 5 10 15 20 Time [s] 25 30 35 40
Figure 14.18: One person walking with 2 Hz: Displacement at midspan with locked TMD.
Acceleration at midspan [m/s2] 2.0 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -2.0 0
Figure 14.20: One person walking with 2 Hz: Displacement at midspan with locked TMD. (ABAQUS-Simulation).
Acceleration at midspan [m/s2] 2.0 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -2.0 0
10
15
20 Time [s]
25
30
35
40
10
15
20 Time [s]
25
30
35
40
Figure 14.19: One person walking with 2 Hz: Acceleration at midspan with locked TMD.
Figure 14.21: One person walking with 2 Hz: Acceleration at midspan with locked TMD. (ABAQUS-Simulation).
Page 14-21
Page 14-22
An-Najah 2013
An-Najah 2013
The curves in Figures 14.20 and 14.21 were computed using the FE program ABAQUS. A similar calculation is described in detail in Section 13.3.3. The input data used in that section were only slightly adjusted here in order to better describe the properties of the test. Maximum vibration amplitude Static displacement:
d st = 1.08mm
Figure 14.22: One person walking with 2 Hz: Displacement at midspan with free TMD.
Acceleration at midspan [m/s2] 2.0 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -2.0 0
(Measured: a max = 1.63m s 2 ) The maximum amplitudes of the numerical simulation and of the experiment agree quite well and also the time-histories shown in Figures 14.18 and 14.21 look quite similar. Please note that during the first 2 seconds of the experiment, displacements and accelerations are zero, because the person started to walk with a slight delay.
10
15
20 time [s]
25
30
35
40
Figure 14.23: One person walking with 2 Hz: Acceleration at midspan with free TMD.
Page 14-23
Page 14-24
An-Najah 2013
An-Najah 2013
Estimate of the maximum vibration amplitude Amplification factor: Maximum dyn. displ.: Maximum displacement: about 5.5 (from page 13-41)
d dyn,max = 1.08 0.4 5.5 = 2.38mm d max = 2.38 + 1.08 = 3.46mm (Measured: d max = 3.27mm )
Maximum acceleration.:
(Measured: a max = 0.34m s 2 ) 14.7.4 Remarks about One person walking with 2 Hz The effect of the TMD can be easily seen in Figures 14.22 and 14.23. The maximum acceleration at midspan reduces from 1.63m s 2 to 0.34m s 2 , which corresponds to a permissible value.
2.05 m/s2 1.63 m/s2 0.96 m/s2 0.34 m/s2 2.14 4.79
Page 14-25
Page 14-26
An-Najah 2013
An-Najah 2013
Case Maximum displacement at midspan. Locked TMD Maximum displacement at midspan. Free TMD Ratio
Group
1 person
of the displacement is larger that the amplification factor of the accelerations, because the static deflection caused by the group is significantly larger than that caused by the single person. The activation of the TMD results in a reduction of the maximum acceleration caused by the single person by a factor of 4.79. In the case of the group the reduction factor is only 2.14. It should be noted here that when the TMD is active (free), the vibrations are significantly smaller, and therefore it is much easier for the group to walk in step. It is therefore to be assumed that in the case of the free TMD, the action was stronger than in the case of the locked TMD. This could explain the seemingly minor effectiveness of the TMD in the case of the group.
It is further assumed that only about 16 of the 24 persons are on the footbridge at the same time. The following remarks can thereby be made: The maximum acceleration measured at midspan of the bridge with locked TMD is only about 1.26-times greater than the acceleration which has been generated by the single person. According to section 13.3.3 we could have expected a larger acceleration from the group ( 16 = 4 ). One reason why the maximum acceleration is still relatively small, is the difficulty to walk in the step when the ground is unsteady. With a little more practice, the group could probably have achieved much larger accelerations. It is further to note that the person who walked of the bridge for the test presented in Section 14.7 was with his 95 kg probably much heavier than the average of the group. The maximum displacement measured at midspan of the bridge with locked TMD is 1.70 times larger than the displacement generated by the single person. The amplification factor
Page 14-27
Page 14-28
An-Najah 2013
An-Najah 2013
-10
-10
Figure 14.24: Group walking with 2 Hz: Displacement at midspan with locked TMD.
Acceleration at midspan [m/s2] 2.0 1.0 0.0 -1.0 -2.0 0 5 10 15 20 time [s] 25 30 35 40
Figure 14.26: Group walking with 2 Hz: Displacement at midspan with free TMD.
Acceleration at midspan [m/s2] 2.0 1.0 0.0 -1.0 -2.0 0 5 10 15 20 Time [s] 25 30 35 40
Figure 14.25: Group walking with 2 Hz: Acceleration at midspan with locked TMD.
Figure 14.27: Group walking with 2 Hz: Acceleration at midspan with free TMD.
Page 14-29
Page 14-30
An-Najah 2013
An-Najah 2013
Test results
Displacement at midspan [mm] 60 40 20 0 -20 -40 -60 0 5 10 15 20 Time [s] 25 30 35 40
Figure 14.28: One person jumping with 2 Hz: Displacement at midspan with locked TMD.
Acceleration at midspan [m/s2] 8.0 6.0 4.0 2.0 0.0 -2.0 -4.0 -6.0 -8.0 0
(Measured: a max = 7.18m s 2 ) Max. dyn. displacement: d dyn,max = 0.82 1.8 31.25 = 46.13mm Maximum displacement: d max = 46.13 + 0.82 = 46.95mm (Measured: d max = 51.08mm )
10
15
20 Time [s]
25
30
35
40
Figure 14.29: One person jumping with 2 Hz: Acceleration at midspan with locked TMD.
Page 14-31
Page 14-32
An-Najah 2013
An-Najah 2013
Estimate of the maximum vibration amplitude Amplification factor: Maximum dyn. displ.: Maximum displacement: about 5.5 (from page 13-41)
(Measured: a max = 1.04m s 2 ) 14.9.3 Remarks about One person jumping with 2 Hz When jumping, the footbridge can be much strongly excited than when walking. The achieved acceleration a max = 7.18m s 2 = 73% g is very large and two jumping people could easily produce the lift-off of the footbridge. The effect of the TMD can be easily seen in Figures 14.30 and 14.31. The maximum acceleration at midspan reduces from 7.18m s 2 to 1.04m s 2 , what, however, is still perceived as unpleasant.
Figure 14.30: One person jumping with 2 Hz: Displacement at midspan with free TMD.
Acceleration at midspan [m/s2] 8.0 6.0 4.0 2.0 0.0 -2.0 -4.0 -6.0 -8.0 0
10
15
20 Time [s]
25
30
35
40
Figure 14.31: One person jumping with 2 Hz: Acceleration at midspan with free TMD.
Page 14-33
Page 14-34