SD An Najah 2013 DS-alesandro PDF

Course “Fundamentals of Structural Dynamics”
An-Najah National University Course “Fundamentals of Structural Dynamics” April 19 - April 23, 2013
April 19 - April 23, 2013
Lecturer: Dr. Alessandro Dazio, UME School
4 Schedule of classes
Date Time Topic
Fundamentals of Structural Dynamics 09:00 - 10:30 1. Introduction
2. SDoF systems: Equation of motion and modelling
Day 1
Fri. April 19 11:00 - 12:30 3. Free vibrations
1 Course description 2013 14:30 - 16:00 Assignment 1
Aim of the course is that students develop a “feeling for dynamic problems” and acquire the theoretical 16:30 - 18:00 Assignment 1
background and the tools to understand and to solve important problems relevant to the linear and, in 9:00 - 10:30 4. Harmonic excitation
part, to the nonlinear dynamic behaviour of structures, especially under seismic excitation. Day 2 11:00 - 12:30 5. Transfer functions
The course will start with the analysis of single-degree-of-freedom (SDoF) systems by discussing: (i) Sat. April 20
Modelling, (ii) equations of motion, (iii) free vibrations with and without damping, (iv) harmonic, pe- 2013 14:30 - 16:00 6. Forced vibrations (Part 1)
riodic and short excitations, (v) Fourier series, (vi) impacts, (vii) linear and nonlinear time history anal- 16:30 - 18:00 6. Forced vibrations (Part 2)
ysis, and (viii) elastic and inelastic response spectra.
09:00 - 10:30 7. Seismic excitation (Part 1)
Afterwards, multi-degree-of-freedom (MDoF) systems will be considered and the following topics will
Day 3 11:00 - 12:30 7. Seismic excitation (Part 2)
be discussed: (i) Equation of motion, (ii) free vibrations, (iii) modal analysis, (iv) damping, (v) Rayleigh’s Sun. April 21
quotient, and (vi) seismic behaviour through response spectrum method and time history analysis. 2013 14:30 - 16:00 Assignment 2
To supplement the suggested reading, handouts with class notes and calculation spreadsheets with se- 16:30 - 18:00 Assignment 2
lected analysis cases to self-training purposes will be distributed.
9:00 - 10:30 8. MDoF systems: Equation of motion
Lecturer: Dr. Alessandro Dazio, UME School
Day 4 11:00 - 12:30 9. Free vibrations
Mon. April 22 14:30 - 16:00 10. Damping
2 Suggested reading 2013 11. Forced vibrations
[Cho11] Chopra A., “Dynamics of Structures”, Prentice Hall, Fourth Edition, 2011. 16:30 - 18:00 11. Forced vibrations
[CP03] Clough R., Penzien J., “Dynamics of Structures”, Second Edition (revised), Computer and 09:00 - 10:30 12. Seismic excitation (Part 1)
Structures Inc., 2003. Day 5 11:00 - 12:30 12. Seismic excitation (Part 2)
[Hum12] Humar J.L., “Dynamics of Structures”. Third Edition. CRC Press, 2012. Tue. April 23
2013 14:30 - 16:00 Assignment 3
16:30 - 18:00 Assignment 3
3 Software
In the framework of the course the following software will be used by the lecturer to solve selected ex-
amples:
[Map10] Maplesoft: “Maple 14”. User Manual. 2010
[Mic07] Microsoft: “Excel 2007”. User Manual. 2007
[VN12] 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
A. Dazio, April 19, 2013 Page 1/2 Page 2/2

Course “Fundamentals of Structural Dynamics” An-Najah 2013 Course “Fundamentals of Structural Dynamics” An-Najah 2013
Table of Contents 3.2 Damped free vibrations ....................................................................... 3-6

3.2.1 Formulation 3: Exponential Functions ....................................................... 3-6
3.2.2 Formulation 1: Amplitude and phase angle............................................. 3-10
Table of Contents...................................................................... i 3.3 The logarithmic decrement .............................................................. 3-12
3.4 Friction damping ............................................................................... 3-15
1 Introduction
1.1 Goals of the course .............................................................................. 1-1 4 Response to Harmonic Excitation
1.2 Limitations of the course..................................................................... 1-1 4.1 Undamped harmonic vibrations ......................................................... 4-3
1.3 Topics of the course ............................................................................ 1-2 4.1.1 Interpretation as a beat ................................................................................ 4-5
4.1.2 Resonant excitation (ω = ωn) ....................................................................... 4-8
1.4 References ............................................................................................ 1-3
4.2 Damped harmonic vibration .............................................................. 4-10
4.2.1 Resonant excitation (ω = ωn) ..................................................................... 4-13
2 Single Degree of Freedom Systems
2.1 Formulation of the equation of motion............................................... 2-1 5 Transfer Functions
2.1.1 Direct formulation......................................................................................... 2-1
2.1.2 Principle of virtual work............................................................................... 2-3 5.1 Force excitation .................................................................................... 5-1
2.1.3 Energy Formulation...................................................................................... 2-3 5.1.1 Comments on the amplification factor V.................................................... 5-4
5.1.2 Steady-state displacement quantities ........................................................ 5-8
2.2 Example “Inverted Pendulum”............................................................ 2-4
5.1.3 Derivating properties of SDoF systems from harmonic vibrations....... 5-10
2.3 Modelling............................................................................................. 2-10
5.2 Force transmission (vibration isolation) ......................................... 5-12
2.3.1 Structures with concentrated mass.......................................................... 2-10
2.3.2 Structures with distributed mass ............................................................. 2-11 5.3 Base excitation (vibration isolation)................................................. 5-15
2.3.3 Damping ...................................................................................................... 2-20 5.3.1 Displacement excitation ........................................................................... 5-15
5.3.2 Acceleration excitation ............................................................................. 5-17
5.3.3 Example transmissibility by base excitation .......................................... 5-20
3 Free Vibrations
5.4 Summary Transfer Functions ........................................................... 5-26
3.1 Undamped free vibrations ................................................................... 3-1
3.1.1 Formulation 1: Amplitude and phase angle............................................... 3-1
6 Forced Vibrations
3.1.2 Formulation 2: Trigonometric functions .................................................... 3-3
3.1.3 Formulation 3: Exponential Functions ....................................................... 3-4 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

6.1.2 Half-sine ........................................................................................................ 6-5 7.5 Elastic response spectra ................................................................... 7-42
6.1.3 Example: “Jumping on a reinforced concrete beam”............................... 6-7 7.5.1 Computation of response spectra ............................................................ 7-42
6.2 Short excitation .................................................................................. 6-12 7.5.2 Pseudo response quantities...................................................................... 7-45
6.2.1 Step force .................................................................................................... 6-12 7.5.3 Properties of linear response spectra ..................................................... 7-49
6.2.2 Rectangular pulse force excitation .......................................................... 6-14 7.5.4 Newmark’s elastic design spectra ([Cho11]) ........................................... 7-50
6.2.3 Example “blast action” .............................................................................. 6-21 7.5.5 Elastic design spectra in ADRS-format (e.g. [Faj99])
(Acceleration-Displacement-Response Spectra) .................................... 7-56
7.6 Strength and Ductility ........................................................................ 7-58

7 Seismic Excitation
7.6.1 Illustrative example .................................................................................... 7-58
7.1 Introduction .......................................................................................... 7-1 7.6.2 “Seismic behaviour equation” .................................................................. 7-61
7.2 Time-history analysis of linear SDoF systems ................................. 7-3 7.6.3 Inelastic behaviour of a RC wall during an earthquake ........................ 7-63
7.2.1 Newmark’s method (see [New59]) .............................................................. 7-4 7.6.4 Static-cyclic behaviour of a RC wall ........................................................ 7-64
7.2.2 Implementation of Newmark’s integration scheme within 7.6.5 General definition of ductility ................................................................... 7-66
the Excel-Table “SDOF_TH.xls”.................................................................. 7-8 7.6.6 Types of ductilities .................................................................................... 7-67
7.2.3 Alternative formulation of Newmark’s Method. ....................................... 7-10
7.7 Inelastic response spectra ............................................................... 7-68
7.3 Time-history analysis of nonlinear SDoF systems ......................... 7-12 7.7.1 Inelastic design spectra............................................................................. 7-71
7.3.1 Equation of motion of nonlinear SDoF systems ..................................... 7-13 7.7.2 Determining the response of an inelastic SDOF system
7.3.2 Hysteretic rules........................................................................................... 7-14 by means of inelastic design spectra in ADRS-format ........................... 7-80
7.3.3 Newmark’s method for inelastic systems ................................................ 7-18 7.7.3 Inelastic design spectra: An important note............................................ 7-87
7.3.4 Example 1: One-storey, one-bay frame ................................................... 7-19 7.7.4 Behaviour factor q according to SIA 261 ................................................. 7-88
7.3.5 Example 2: A 3-storey RC wall .................................................................. 7-23 7.8 Linear equivalent SDOF system (SDOFe) ....................................... 7-89
7.4 Solution algorithms for nonlinear analysis problems .................... 7-26 7.8.1 Elastic design spectra for high damping values ..................................... 7-99
7.4.1 General equilibrium condition................................................................. 7-26 7.8.2 Determining the response of inelastic SDOF systems
by means of a linear equivalent SDOF system and
7.4.2 Nonlinear static analysis ........................................................................... 7-26 elastic design spectra with high damping ............................................. 7-103
7.4.3 The Newton-Raphson Algorithm............................................................... 7-28
7.9 References ........................................................................................ 7-108
7.4.4 Nonlinear dynamic analyses ..................................................................... 7-35
7.4.5 Comments on the solution algorithms for
nonlinear analysis problems ..................................................................... 7-38 8 Multi Degree of Freedom Systems
7.4.6 Simplified iteration procedure for SDoF systems with
idealised rule-based force-deformation relationships............................ 7-41 8.1 Formulation of the equation of motion............................................... 8-1
8.1.1 Equilibrium formulation ............................................................................... 8-1
8.1.2 Stiffness formulation ................................................................................... 8-2
Table of Contents Page iii Table of Contents Page iv

8.1.3 Flexibility formulation ................................................................................. 8-3 10.3Classical damping matrices .............................................................. 10-5
8.1.4 Principle of virtual work............................................................................... 8-5 10.3.1 Mass proportional damping (MpD) ......................................................... 10-5
8.1.5 Energie formulation...................................................................................... 8-5 10.3.2 Stiffness proportional damping (SpD) .................................................... 10-5
8.1.6 “Direct Stiffness Method”............................................................................ 8-6 10.3.3 Rayleigh damping..................................................................................... 10-6
8.1.7 Change of degrees of freedom.................................................................. 8-11 10.3.4 Example....................................................................................................... 10-7
8.1.8 Systems incorporating rigid elements with distributed mass ............... 8-14
11 Forced Vibrations
9 Free Vibrations
11.1Forced vibrations without damping ................................................. 11-1
9.1 Natural vibrations ................................................................................. 9-1 11.1.1 Introduction ................................................................................................ 11-1
9.2 Example: 2-DoF system ...................................................................... 9-4 11.1.2 Example 1: 2-DoF system.......................................................................... 11-3
9.2.1 Eigenvalues ................................................................................................ 9-4 11.1.3 Example 2: RC beam with Tuned Mass Damper (TMD) without damping...
11-7
9.2.2 Fundamental mode of vibration .................................................................. 9-5
9.2.3 Higher modes of vibration ........................................................................... 9-7 11.2Forced vibrations with damping ..................................................... 11-13
9.2.4 Free vibrations of the 2-DoF system .......................................................... 9-8 11.2.1 Introduction .............................................................................................. 11-13
9.3 Modal matrix and Spectral matrix ..................................................... 9-12 11.3Modal analysis: A summary ............................................................ 11-15
9.4 Properties of the eigenvectors.......................................................... 9-13

9.4.1 Orthogonality of eigenvectors .................................................................. 9-13 12 Seismic Excitation
9.4.2 Linear independence of the eigenvectors................................................ 9-16 12.1Equation of motion............................................................................. 12-1
9.5 Decoupling of the equation of motion.............................................. 9-17 12.1.1 Introduction................................................................................................. 12-1
12.1.2 Synchronous Ground motion.................................................................... 12-3
9.6 Free vibration response..................................................................... 9-22
12.1.3 Multiple support ground motion ............................................................... 12-8
9.6.1 Systems without damping ......................................................................... 9-22
9.6.2 Classically damped systems..................................................................... 9-24 12.2Time-history of the response of elastic systems .......................... 12-18
12.3Response spectrum method ........................................................... 12-23
10 Damping 12.3.1 Definition and characteristics ................................................................. 12-23
12.3.2 Step-by-step procedure ......................................................................... 12-27
10.1Free vibrations with damping ........................................................... 10-1
12.4Practical application of the response spectrum method
10.2Example .............................................................................................. 10-2
to a 2-DoF system ............................................................................ 12-29
10.2.1 Non-classical damping ............................................................................ 10-3
12.4.1 Dynamic properties ................................................................................. 12-29
10.2.2 Classical damping .................................................................................... 10-4
12.4.2 Free vibrations.......................................................................................... 12-31
Table of Contents Page v Table of Contents Page vi

12.4.3 Equation of motion in modal coordinates.............................................. 12-38 14 Pedestrian Footbridge with TMD
12.4.4 Response spectrum method ................................................................... 12-41
12.4.5 Response spectrum method vs. time-history analysis ........................ 12-50 14.1Test unit and instrumentation ........................................................... 14-1
14.2Parameters .......................................................................................... 14-4
13 Vibration Problems in Structures 14.2.1 Footbridge (Computed, without TMD) ...................................................... 14-4
14.2.2 Tuned Mass Damper (Computed) ............................................................. 14-4
13.1Introduction ........................................................................................ 13-1
13.1.1 Dynamic action ........................................................................................... 13-2
14.3Test programme ................................................................................. 14-5
13.1.2 References .................................................................................................. 13-3 14.4Free decay test with locked TMD ...................................................... 14-6
13.2Vibration limitation ............................................................................. 13-4 14.5Sandbag test ....................................................................................... 14-8
13.2.1 Verification strategies ................................................................................ 13-4 14.5.1 Locked TMD, Excitation at midspan ....................................................... 14-9
13.2.2 Countermeasures ....................................................................................... 13-5 14.5.2 Locked TMD, Excitation at quarter-point of the span .......................... 14-12
13.2.3 Calculation methods .................................................................................. 13-6 14.5.3 Free TMD: Excitation at midspan .......................................................... 14-15
13.3People induced vibrations................................................................. 13-8 14.6One person walking with 3 Hz......................................................... 14-17
13.3.1 Excitation forces......................................................................................... 13-8
14.7One person walking with 2 Hz......................................................... 14-20
13.3.2 Example: Jumping on an RC beam....................................................... 13-15
14.7.1 Locked TMD (Measured) .......................................................................... 14-20
13.3.3 Footbridges............................................................................................... 13-18
14.7.2 Locked TMD (ABAQUS-Simulation) ...................................................... 14-22
13.3.4 Floors in residential and office buildings .............................................. 13-26
14.7.3 Free TMD .................................................................................................. 14-24
13.3.5 Gyms and dance halls.............................................................................. 13-29
14.7.4 Remarks about “One person walking with 2 Hz” .................................. 14-25
13.3.6 Concert halls, stands and diving platforms........................................... 13-30
14.8Group walking with 2 Hz .................................................................. 14-26
13.4Machinery induced vibrations......................................................... 13-30
14.8.1 Locked TMD ............................................................................................ 14-29
13.5Wind induced vibrations.................................................................. 13-31 14.8.2 Free TMD ................................................................................................. 14-30
13.5.1 Possible effects ........................................................................................ 13-31
14.9One person jumping with 2 Hz ........................................................ 14-31
13.6Tuned Mass Dampers (TMD) ........................................................... 13-34 14.9.1 Locked TMD .............................................................................................. 14-31
13.6.1 Introduction............................................................................................... 13-34 14.9.2 Free TMD ................................................................................................. 14-33
13.6.2 2-DoF system ........................................................................................... 13-35 14.9.3 Remarks about “One person jumping with 2 Hz” ................................. 14-34
13.6.3 Optimum TMD parameters....................................................................... 13-39
13.6.4 Important remarks on TMD...................................................................... 13-39
Table of Contents Page vii Table of Contents Page viii

1 Introduction 1.3 Topics of the course
1) Systems with one degree of freedom

1.1 Goals of the course
- Modelling and equation of motion
• Presentation of the theoretical basis and of the relevant tools;
- Free vibrations with and without damping
• General understanding of phenomena related to structural dy-
- Harmonic excitation
namics;
• Focus on earthquake engineering; 2) Forced oscillations
• Development of a “Dynamic Feeling”; - Periodic excitation, Fourier series, short excitation
• Detection of frequent dynamic problems and application of ap- - Linear and nonlinear time history-analysis
propriate solutions. - Elastic and inelastic response spectra
1.2 Limitations of the course 3) Systems with many degree of freedom

- Modelling and equation of motion
• Only an introduction to the broadly developed field of structural
dynamics (due to time constraints); - Modal analysis, consideration of damping
• Only deterministic excitation; - Forced oscillations,
• No soil-dynamics and no dynamic soil-structure interaction will - Seismic response through response spectrum method and
be treated (this is the topic of another course); time-history analysis
• Numerical methods of structural dynamics are treated only 4) Continuous systems

partially (No FE analysis. This is also the topic of another
- Generalised Systems
course);
• Recommendation of further readings to solve more advanced 5) Measures against vibrations
problems.
- Criteria, frequency tuning, vibration limitation
1 Introduction Page 1-1 1 Introduction Page 1-2

1.4 References
Blank page
Theory
[Bat96] Bathe KJ: “Finite Element Procedures”. Prentice Hall, Upper
Saddle River, 1996.
[CF06] Christopoulos C, Filiatrault A: "Principles of Passive Supple-
mental Damping and Seismic Isolation". ISBN 88-7358-037-
8. IUSSPress, 2006.
[Cho11] Chopra AK: “Dynamics of Structures”. Fourth Edition.
Prentice Hall, 2011.
[CP03] Clough R, Penzien J: “Dynamics of Structures”. Second Edi-
tion (Revised). Computer and Structures, 2003.
(http://www.csiberkeley.com)
[Den85] Den Hartog JP: “Mechanical Vibrations”. Reprint of the fourth
edition (1956). Dover Publications, 1985.
[Hum12] Humar JL: “Dynamics of Structures”. Third Edition. CRC
Press, 2012.
[Inm01] Inman D: “Engineering Vibration”. Prentice Hall, 2001.
[Prz85] Przemieniecki JS: “Theory of Matrix Structural Analysis”. Do-
ver Publications, New York 1985.
[WTY90] Weawer W, Timoshenko SP, Young DH: “Vibration problems
in Engineering”. Fifth Edition. John Wiley & Sons, 1990.
Practical cases (Vibration problems)

[Bac+97] Bachmann H et al.: “Vibration Problems in Structures”.
Birkhäuser Verlag 1997.
1 Introduction Page 1-3 1 Introduction Page 1-4

2 Single Degree of Freedom Systems 2) D’Alembert principle

F+T = 0 (2.4)
2.1 Formulation of the equation of motion
The principle is based on the idea of a fictitious inertia force
that is equal to the product of the mass times its acceleration,
2.1.1 Direct formulation
and acts in the opposite direction as the acceleration
1) Newton's second law (Action principle) The mass is at all times in equilibrium under the resultant
force F and the inertia force T = – mu··.
dI
F = = d ( mu· ) = mu·· ( I = Impulse) (2.1) y = x ( t ) + l + us + u ( t ) (2.5)
dt dt
y·· = x·· + u·· (2.6)
The force corresponds to the change of impulse over time.
T = – my·· = – m ( x·· + u··) (2.7)
F = – k ( u s + u ) – cu· + mg (2.8)
= – ku s – ku – cu· + mg
= – ku – cu·
F+T = 0 (2.9)
– cu· – ku – mx·· – mu·· = 0 (2.10)
mu·· + cu· + ku = – mx·· (2.11)
– f k ( t ) – f c ( t ) + F ( t ) = mu··( t ) (2.2)
Introducing the spring force f k ( t ) = ku ( t ) and the damping

force f c ( t ) = cu· ( t ) Equation (2.2) becomes:
• To derive the equation of motion, the dynamic equilibrium for
each force component is formulated. To this purpose, forces,
mu··( t ) + cu· ( t ) + ku ( t ) = F ( t ) (2.3)
and possibly also moments shall be decomposed into their
components according to the coordinate directions.
2 Single Degree of Freedom Systems Page 2-1 2 Single Degree of Freedom Systems Page 2-2
2.1.2 Principle of virtual work 2.2 Example “Inverted Pendulum”

δu (2.12)
Direct Formulation
• Virtual displacement = imaginary infinitesimal displacement
m
• Should best be kinematically permissible, so that unknown reac- Fm
tion forces do not produce work
δA i = δA a (2.13)
k a sin(ϕ1)
• Thereby, both inertia forces and damping forces must be consid- Fp
ered Fk
( f m + f c + f k )δu = F ( t )δu (2.14) l
a cos(ϕ1)
ϕ1
l
2.1.3 Energy Formulation a
• Kinetic energy T (Work, that an external force needs to pro- sin(ϕ1) ~ ϕ1
vide to move a mass) cos(ϕ1) ~ 1
• Deformation energy U (is determined from the work that an ex- O
ternal force has to provide in order to generate a deformation) ϕ1)
l sin(ϕ
• Potential energy of the external forces V (is determined with Spring force: F k = a ⋅ sin ( ϕ 1 ) ⋅ k ≈ a ⋅ ϕ 1 ⋅ k (2.17)
respect to the potential energy at the position of equilibrium)
··
• Conservation of energy theorem (Conservative systems) Inertia force: Fm = ϕ1 ⋅ l ⋅ m (2.18)
E = T + U + V = T o + U o + V o = cons tan t (2.15) External force: Fp = m ⋅ g (2.19)
dE Equilibrium
= 0 (2.16)
dt F k ⋅ a ⋅ cos ( ϕ 1 ) + F m ⋅ l – F p ⋅ l ⋅ sin ( ϕ 1 ) = 0 (2.20)
2 ·· 2
m ⋅ l ⋅ ϕ1 + ( a ⋅ k – m ⋅ g ⋅ l ) ⋅ ϕ1 = 0 (2.21) Spring force: F k ⋅ cos ( ϕ 1 ) ≈ a ⋅ ϕ 1 ⋅ k (2.24)
··
Circular frequency: Inertia force: Fm = ϕ1 ⋅ l ⋅ m (2.25)
K 2
a ⋅k–m⋅g⋅l
2
a ⋅ k- g External force: F p ⋅ sin ( ϕ 1 ) ≈ m ⋅ g ⋅ ϕ 1 (2.26)
ω = -------1 = ------------------------------------- = ------------ – ---
2 l
(2.22)
M1 2
m⋅l m⋅l Virtual displacement:
The system is stable if: δu k = δϕ 1 ⋅ a , δu m = δϕ 1 ⋅ l (2.27)
2
ω > 0: a ⋅k>m⋅g⋅l (2.23) Principle of virtual work:
( F k ⋅ cos ( ϕ 1 ) ) ⋅ δu k + ( F m – ( F p ⋅ sin ( ϕ 1 ) ) ) ⋅ δu m = 0 (2.28)
Principle of virtual work formulation
··
m ( a ⋅ ϕ 1 ⋅ k ) ⋅ δϕ 1 ⋅ a + ( ϕ 1 ⋅ l ⋅ m – m ⋅ g ⋅ ϕ 1 ) ⋅ δϕ 1 ⋅ l = 0 (2.29)
Fm
Fpsin(ϕ1) After cancelling out δϕ 1 the following equation of motion is ob-
tained:
Fkcos(ϕ1) δum 2 ·· 2
k m ⋅ l ⋅ ϕ1 + ( a ⋅ k – m ⋅ g ⋅ l ) ⋅ ϕ1 = 0 (2.30)
The equation of motion given by Equation (2.30) corresponds to

δuk Equation (2.21).
l
ϕ1
a
δϕ1
sin(ϕ1) ~ ϕ1
cos(ϕ1) ~ 1
O
Energy Formulation for small angles ϕ 1 we have:

m 2 2
(1-cos(ϕ1)) l ϕ1 ϕ1
~ Epot,p cos ( ϕ 1 ) = 1 – ------ and ------ = 1 – cos ( ϕ 1 ) (2.35)
2 2
0.5 l ϕ12 vm
a sin(ϕ1) Ekin,m and Equation (2.33) becomes:
k
Edef,k 2
E pot,p = – ( m ⋅ g ⋅ 0.5 ⋅ l ⋅ ϕ 1 ) (2.36)
l Energy conservation:
ϕ1 E tot = E def,k + E kin,m + E pot,p = constant (2.37)
a
1 2 ·2 1 2 2
sin(ϕ1) ~ ϕ1 E = --- ( m ⋅ l ) ⋅ ϕ 1 + --- ( k ⋅ a – m ⋅ g ⋅ l ) ⋅ ϕ 1 = constant (2.38)
2 2
cos(ϕ1) ~ 1
O Derivative of the energy with respect to time:
dE
1 2 1 2
Spring: E def,k = --- ⋅ k ⋅ [ a ⋅ sin ( ϕ 1 ) ] = --- ⋅ k ⋅ ( a ⋅ ϕ 1 ) (2.31) = 0 Derivation rule: ( g • f )' = ( g' • f ) ⋅ f' (2.39)
2 2 dt
2 · ·· 2 ·
Mass:
1 2 1 ·
E kin,m = --- ⋅ m ⋅ v m = --- ⋅ m ⋅ ( ϕ 1 ⋅ l )
2
(2.32) ( m ⋅ l ) ⋅ ϕ1 ⋅ ϕ1 + ( k ⋅ a – m ⋅ g ⋅ l ) ⋅ ϕ1 ⋅ ϕ1 = 0 (2.40)
2 2
·
After cancelling out the velocity ϕ 1 :
E pot,p = – ( m ⋅ g ) ⋅ ( 1 – cos ( ϕ 1 ) ) ⋅ l (2.33)
2 ·· 2
m ⋅ l ⋅ ϕ1 + ( a ⋅ k – m ⋅ g ⋅ l ) ⋅ ϕ1 = 0 (2.41)
by means of a series development, cos ( ϕ 1 ) can be
expressed as: The equation of motion given by Equation (2.41) corresponds to
2 4 Equations (2.21) and (2.30).
ϕ1 ϕ1 k x
2k
cos ( ϕ 1 ) = 1 – ------ + ------ – … + ( – 1 ) ⋅ ------------- + … (2.34)
2! 4! ( 2k )!
Comparison of the energy maxima 2.3 Modelling

1 · 2
KE = --- ⋅ m ⋅ ( ϕ 1,max ⋅ l ) (2.42) 2.3.1 Structures with concentrated mass
2
Water tank
1 2 1 2
PE = --- ⋅ k ⋅ ( a ⋅ ϕ 1 ) – --- ⋅ g ⋅ m ⋅ l ⋅ ϕ 1 (2.43) Tank:
2 2 Mass=1000t F(t) F(t)
By equating KE and PE we obtain:
RC Walls in the
longitudinal direction
· § 2 ⋅ k – m ⋅ g ⋅ l·
ϕ 1,max = ¨ a------------------------------------
-¸ ⋅ ϕ 1 (2.44) Ground
© m⋅l
2 ¹
· al Tra
ϕ 1,max = ω ⋅ ϕ 1 (2.45) o
din
itu n
ng ctio
dir nsver
ect se
ion
L ire
d
3EI w
• ω is independent of the initial angle ϕ 1 k = … k = 2 -----------
3
-
Bridge in transverse direction H
• the greater the deflection, the greater the maximum velocity. F(t)
Frame with rigid beam
F(t)
3EI w
k = -----------
3
-
H
12EI s
k = 2 -------------
3
-
H
mu·· + ku = F ( t ) (2.46)
2.3.2 Structures with distributed mass L

δA i = ³0 ( EIu'' ⋅ δ [ u'' ] ) dx (2.53)
• Transformations:
u'' = ψ''U and ··

u·· = ψU (2.54)
• The virtual displacement is affine to the selected deformation:
δu = ψδU and δ [ u'' ] = ψ''δU (2.55)

• Using Equations (2.54) and (2.55), the work δA a produced by the
external forces is:
L
·· ⋅ ψδU ) dx + L ( f ⋅ ψδU ) dx
δA a = – ³ ( mψU
Deformation: u ( x, t ) = ψ ( x )U ( t ) (2.47)
0
³ 0
(2.56)
External forces: t ( x, t ) = – mu··( x, t ) ·· L mψ 2 dx + L fψ dx
f ( x, t ) (2.48)
= δU – U ³ 0
³ 0
• Principle of virtual work • Using Equations (2.54) and (2.55) the work δA i produced by the
internal forces is:
δA i = δA a (2.49)
L L
L L δA i = ³0 ( EIψ''U ⋅ ψ''δU ) dx = δU U ³ ( EI ( ψ'' ) 2 ) dx (2.57)
δA a = ³0 ( t ⋅ δu ) dx + ³ ( f ⋅ δu ) dx
0
(2.50) 0
L L • Equation (2.49) is valid for all virtual displacements, therefore:

= – ³ ( mu·· ⋅ δu ) dx + ³ ( f ⋅ δu ) dx L
·· L mψ 2 dx + L fψ dx
U ³ ( EI ( ψ'' ) 2 ) dx = – U ³ ³
0 0
(2.58)
L 0 0 0
δA i = ³0 ( M ⋅ δϕ ) dx where: (2.51)
* ··
m U
*
+k U = F
*
(2.59)
M = EIu'' and δϕ = δ [ u'' ] (2.52)
• Circular frequency • Example No. 1: Cantilever with distributed mass

L
2 k
* ³0 ( EI ( ψ'' ) 2 ) dx
ωn = ------*- = ------------------------------------
L
- (2.60)
m 2
³ mψ dx
0
-> 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 πx π 2 πx
fulfills all boundary conditions; ψ = 1 – cos § -------· , ψ'' = § -------· cos § -------· (2.61)
© 2L¹ © 2L¹ © 2L¹
- The boundary conditions are automatically satisfied if the
deformation figure corresponds to the deformed shape due
§ πx· · dx + ψ 2 ( x = L )M
to an external force; L 2
* §
m = ³0 m © 1 – cos © ------
2L¹ ¹
- (2.62)
- A possible external force is the weight of the structure act-
ing in the considered direction. L
§ 3πx – 8 sin § ------ πx-· πx πx
L + 2 cos § -------· sin § -------· L·
• Properties of the Rayleigh-Quotient 1--- ¨ © 2L¹ © 2L¹ © 2L¹ ¸
= m ¨ ----------------------------------------------------------------------------------------------------¸ +M
- The estimated natural frequency is always larger than the 2 ¨ π ¸
© ¹
exact one (Minimization of the quotient!); 0
- Useful results can be obtained even if the assumed defor-

( 3π – 8 )
mation figure is not very realistic. = -------------------- mL + M = 0.23mL + M
2π
• Example No. 2: Cantilever with distributed mass
π 4 L πx 2
k = EI § -------· ³ § cos § -------· · dx
*
(2.63)
© 2L¹ 0 © © 2L¹ ¹
L
§ πx + 2 cos § ------ πx· § πx· ·
- sin ------- L
π 4 ¨ © 2L ¹ © 2L¹ ¸
§ · 1
= EI ------- ⋅ --- ¨ ---------------------------------------------------------------¸
© 2L¹ 2 ¨ π ¸
© ¹
0
4
π EI EI 3EI
= ------ ⋅ -----3- = 3.04 ⋅ -----3- ≈ --------
3
-
32 L L L
πx π 2 πx
ψ = 1 – cos § -------· , ψ'' = § -------· cos § -------· (2.65)
© 2L¹ © 2L¹ © 2L¹
3EI
ω = -----------------------------------------
- (2.64) *
3 • Calculation of the mass m
( 0.23mL + M )L
§ πx· · dx + ψ 2 § x = L
L 2
• Check of the boundary conditions of the deformation figure * § ---· M 1 + ψ 2 ( x = L )M 2
m = ³0 m © 1 – cos © ------
2L¹ ¹
-
© 2¹
(2.66)
πx
ψ ( 0 ) = 0 ? -> ψ ( x ) = 1 – cos §© -------·¹ : ψ ( 0 ) = 0 OK!
2L ( 3π – 8 ) π 2
m = -------------------- mL + § 1 – cos § ---· · ⋅ M 1 + 1 ⋅ M 2
* 2
© ©
(2.67)
π πx 2π 4¹ ¹
ψ' ( 0 ) = 0 ? -> ψ' ( x ) = ------- sin §© -------·¹ : ψ' ( 0 ) = 0 OK!
2L 2L
( 3π – 8 )
m = -------------------- mL + § -------------------· ⋅ M 1 + M 2
* 3–2 2
π 2 πx © 2 ¹
(2.68)
ψ'' ( L ) = 0 ? -> ψ'' ( x ) = §© -------·¹ cos §© -------·¹ : ψ'' ( L ) = 0 OK! 2π
2L 2L
*
m = 0.23mL + 0.086M 1 + M 2 (2.69)
* 6 13 2
• Calculation of the stiffness k EI = 200000 ⋅ 431.9 ×10 = 8.638 ×10 Nmm (2.75)
4 2
π 4 L πx 2 EI = 8.638 ×10 kNm
k = EI § -------· ³ § cos § -------· · dx (2.76)
*
(2.70)
© 2L¹ 0 © © 2L¹ ¹
By means of Equation (2.73) we obtain:
4
* π EI EI 3EI- 4
k = ------ ⋅ -----3- = 3.04 ⋅ -----3- ≈ -------- (2.71) 8.638 ×10 - = 4.9170
EI = 1.673 ------------------------
32 L L L
3 ω = 1.673 ----------- (2.77)
3 3
ML 10 ⋅ 10
• Calculation of the circular frequency ω
1 4.9170
f = ------ ⋅ ω = ---------------- = 0.783Hz (2.78)
2π 2π
3.04EI
ω = ------------------------------------------------------------------------- (2.72)
3 From Equation (2.74):
( 0.23mL + 0.086M 1 + M 2 )L
1.652 EI 1.652 8.638 ×104

Special case: m = 0 and M 1 = M 2 = M f = ------------- ----------- = ------------- ------------------------
- = 0.773Hz (2.79)
2π ML 3 2π 10 ⋅ 10
3
3.04EI EI
ω = ------------------------------ = 1.673 ----------- (2.73)
3 3 The first natural frequency of such a dynamic system can be cal-
( 1.086M )L ML
culated using a finite element program (e.g. SAP 2000), and it is
equal to:
The exact first natural circular frequency of a two-mass oscillator
with constant stiffness and mass is: T = 1.2946s , f = 0.772Hz (2.80)
3.007EI - = 1.652 -----------

EI
ω = ----------------------------- (2.74) Equations (2.78), (2.79) and (2.80) are in very good accordance.
3 3
( 1.102M )L ML The representation of the first mode shape and corresponding
natural frequency obtained by means of a finite element program
As a numerical example, the first natural frequency of a is shown in the next figure.
L = 10m tall steel shape HEB360 (bending about the strong ax-
is) featuring two masses M 1 = M 2 = 10t is calculated.
2.3.3 Damping
M = 10t
• Types of damping
Damping
Internal External
Contact areas
Material
within the structure
Hysteretic Relative External contact

(Viscous, movements (Non-structural
Friction, between parts of elements, Energy
M = 10t Yielding) the structure radiation in the
(Bearings, Joints, ground, etc.)
etc.)
• Typical values of damping in structures

Material Damping ζ
Reinforced concrete (uncraked) 0.007 - 0.010
Reinforced concrete (craked 0.010 - 0.040
HEB 360 Reinforced concrete (PT) 0.004 - 0.007
Reinforced concrete (partially PT) 0.008 - 0.012
Composite components 0.002 - 0.003
Steel 0.001 - 0.002
Table C.1 from [Bac+97]
SAP2000 v8 - File:HEB_360 - Mode 1 Period 1.2946 seconds - KN-m Units
• Bearings • Dissipators
Source: A. Marioni: “Innovative Anti-seismic Devices for Bridges”.

[SIA03] Source: A. Marioni: “Innovative Anti-seismic Devices for Bridges”.
[SIA03]
3 Free Vibrations • Relationships
“A structure undergoes free vibrations when it is brought out of ωn = k ⁄ m [rad/s]: Angular velocity (3.7)
its static equilibrium, and can then oscillate without any external
dynamic excitation” ωn
f n = ------ [1/s], [Hz]: Number of revolutions per time (3.8)
2π
3.1 Undamped free vibrations 2π
T n = ------ [s]: Time required per revolution (3.9)
ωn
mu··( t ) + ku ( t ) = 0 (3.1)
• Transformation of the equation of motion
3.1.1 Formulation 1: Amplitude and phase angle
2
u··( t ) + ω n u ( t ) = 0 (3.10)
• Ansatz:
u ( t ) = A cos ( ω n t – φ ) (3.2) • Determination of the unknowns A and φ :
2 The static equilibrium is disturbed by the initial displacement

u··( t ) = – A ω n cos ( ω n t – φ ) (3.3)
u ( 0 ) = u 0 and the initial velocity u· ( 0 ) = v 0 :
By substituting Equations (3.2) and (3.3) in (3.1):
v0 2 v0
u 0 + § ------· , tan φ = -----------
2 2
A ( – ω n m + k ) cos ( ω n t – φ ) = 0 (3.4) A = - (3.11)
© ω n¹ u0 ωn
2
– ωn m + k = 0 (3.5) • Visualization of the solution by means of the Excel file given on
the web page of the course (SD_FV_viscous.xlsx)
ωn = k ⁄ m “Natural circular frequency” (3.6)
3 Free Vibrations Page 3-1 3 Free Vibrations Page 3-2

3.1.2 Formulation 2: Trigonometric functions 3.1.3 Formulation 3: Exponential Functions
mu··( t ) + ku ( t ) = 0 (3.12) mu··( t ) + ku ( t ) = 0 (3.19)

• Ansatz: • Ansatz:
u ( t ) = A 1 cos ( ω n t ) + A 2 sin ( ω n t ) (3.13) u(t) = e
λt
(3.20)
2 2
u··( t ) = – A 1 ω n cos ( ω n t ) – A 2 ω n sin ( ω n t ) (3.14) 2 λt
u··( t ) = λ e (3.21)
By substituting Equations (3.13) and (3.14) in (3.12): By substituting Equations (3.20) and (3.21) in (3.19):
2 2 2
A 1 ( – ω n m + k ) cos ( ω n t ) + A 2 ( – ω n m + k ) sin ( ω n t ) = 0 (3.15) mλ + k = 0 (3.22)
2 2 k
– ωn m + k = 0 (3.16) λ = – ---- (3.23)
m
ωn = k ⁄ m “Natural circular frequency” (3.17) k- = ± iω
λ = ± i --- n (3.24)
m
• Determination of the unknowns A 1 and A 2 :
The complete solution of the ODE is:
The static equilibrium is disturbed by the initial displacement iω n t –i ωn t
u ( 0 ) = u 0 and the initial velocity u· ( 0 ) = v 0 : u ( t ) = C1 e + C2 e (3.25)
and by means of Euler’s formulas

v0
A 1 = u 0 , A 2 = ------ (3.18) iα –i α iα –i α
ωn e +e e –e
cos α = ----------------------- , sin α = ----------------------- (3.26)
2 2i
iα –i α
e = cos ( α ) + i sin ( α ) , e = cos ( α ) – i sin ( α ) (3.27)

Equation (3.25) can be transformed as follows: 3.2 Damped free vibrations

u ( t ) = ( C 1 + C 2 ) cos ( ω n t ) + i ( C 1 – C 2 ) sin ( ω n t ) (3.28) mu··( t ) + cu· ( t ) + ku ( t ) = 0 (3.30)
u ( t ) = A 1 cos ( ω n t ) + A 2 sin ( ω n t ) (3.29) - In reality vibrations subside
- Damping exists
Equation (3.29) corresponds to (3.13)!
- It is virtually impossible to model damping exactly
- From the mathematical point of view viscous damping is
easy to treat
s
Damping constant: c N ⋅ ---- (3.31)
m
3.2.1 Formulation 3: Exponential Functions
mu··( t ) + cu· ( t ) + ku ( t ) = 0 (3.32)

• Ansatz:
λt λt 2 λt
u(t) = e , u· ( t ) = λe , u··( t ) = λ e (3.33)
By substituting Equations (3.33) in (3.32):
2 λt
( λ m + λc + k )e = 0 (3.34)
2
λ m + λc + k = 0 (3.35)
c 1 2
λ = – -------- ± -------- c – 4km (3.36)
2m 2m

• Critical damping when: c 2 – 4km = 0 • Types of vibrations

1
c cr = 2 km = 2ω n m (3.37) Underdamped vibration
• Damping ratio Critically damped vibration

Overdamped vibration
c c c 0.5
ζ = ------ = --------------- = -------------- (3.38)
c cr 2 km 2ω nm
u(t)/u0 [-]
• Transformation of the equation of motion
0
mu··( t ) + cu· ( t ) + ku ( t ) = 0 (3.39)
c k
u··( t ) + ---- u· ( t ) + ---- u ( t ) = 0 (3.40) -0.5
m m
2
u··( t ) + 2ζω n u· ( t ) + ω n u ( t ) = 0 (3.41)
-1
• Types of vibrations: 0 0.5 1 1.5 2 2.5 3 3.5 4
t/Tn [-]
c
ζ = ------ < 1 : Underdamped free vibrations
c cr
c
ζ = ------ = 1 : Critically damped free vibrations
c cr
c
ζ = ------ > 1 : Overdamped free vibrations
c cr

Underdamped free vibrations ζ < 1 The determination of the unknowns A 1 and A 2 is carried out as
usual by means of the initial conditions for displacement
By substituting: ( u ( 0 ) = u 0 ) and velocity ( u· ( 0 ) = v 0 ) obtaining:
c c c 2 k v 0 + ζω n u 0
ζ = ------ = --------------- = --------------- and ω n = ---- (3.42)
c cr 2 km 2ω n m m A 1 = u 0 , A 2 = --------------------------- (3.51)
ωd
in:
3.2.2 Formulation 1: Amplitude and phase angle
c 2 k
λ = – -------- ± -------- c – 4km = – -------- ± § --------· – ----
c 1 2 c
(3.43)
2m 2m 2m © 2m¹ m Equation (3.50) can be rewritten as “the amplitude and phase
angle”:
it is obtained:
– ζω n t
u ( t ) = Ae cos ( ω d t – φ ) (3.52)
2 2 2 2
λ = – ζω n ± ω n ζ – ω n = – ζω n ± ω n ζ – 1 (3.44)
with
2
λ = – ζω n ± iω n 1 – ζ (3.45) v 0 + ζω n u 0 2 v 0 + ζω n u 0
u 0 + § ---------------------------· , tan φ = ---------------------------
2
A = (3.53)
© ωd ¹ ωd u0
2
ω d = ω n 1 – ζ “damped circular frequency” (3.46)
λ = – ζω n ± iω d (3.47)
The motion is a sinusoidal vibration with
The complete solution of the ODE is: – ζω n t
circular frequency ω d and decreasing amplitude Ae
( – ζω n + iω d )t ( – ζω n – iω d )t
u ( t ) = C1 e + C2 e (3.48)
– ζω n t iω d t –i ωd t
u(t) = e ( C1 e + C2 e ) (3.49)
– ζω n t
u(t) = e ( A 1 cos ( ω d t ) + A 2 sin ( ω d t ) ) (3.50)

• Notes 3.3 The logarithmic decrement

- The period of the damped vibration is longer, i.e. the vibra- 20
tion is slower Td Free vibration
15 u0
1
u1
10
0.9
Displacement
0.8 5
0.7
0
0.6 2
ωd = ωn 1 – ζ
Tn/Td
-5
0.5
0.4 Tn -10
T d = ------------------
0.3
03 2
1–ζ -15
0.2
-20
0.1 0 1 2 3 4 5 6 7 8 9 10
0 Time (s)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Damping ratio ζ • Amplitude of two consecutive cycles
– ζω t
- The envelope of the vibration is represented by the follow- u0 Ae
n
cos ( ω d t – φ )
ing equation: ----- = --------------------------------------------------------------------------------
– ζω ( t + T )
(3.55)
u1 Ae n d
cos ( ω d ( t + T d ) – φ )
– ζω n t v 0 + ζω n u 0 2
u 0 + § ---------------------------·
2 with
u ( t ) = Ae with A = (3.54)
© ωd ¹
– ζω n ( t + T d ) – ζω n t – ζω n T d
e = e e (3.56)
- Visualization of the solution by means of the Excel file giv-
en on the web page of the course (SD_FV_viscous.xlsx) cos ( ω d ( t + T d ) – φ ) = cos ( ω d t + ω d T d – φ ) = cos ( ω d t – φ ) (3.57)

we obtain: • Evaluation over several cycles

u0 1 ζω T u u u uN – 1 ζω T N Nζω n T d
----- = ----------------- = e n d
u1 – ζω T
(3.58) -----0- = ----0- ⋅ ----1- ⋅ … ⋅ ------------
- = (e n d) = e (3.61)
e n d
uN u1 u2 uN
• Logarithmic decrement δ u0
δ = ---- ln § ------·
1
(3.62)
u0 N © u N¹
δ = ln § -----· = ζω n T d = ------------------ ≅ 2πζ (if ζ small)
2πζ
(3.59)
© u 1¹ 2 • Halving of the amplitude
1–ζ
The damping ratio becomes: u
---- ln § -----0-·
1 1
N © u N¹ ---- ln ( 2 )
δ δ N 1 - ---------
1-
ζ = ------------------------- ≅ ------ (if ζ small) (3.60) ζ = ---------------------- = ------------------ = ------ ≅ (3.63)
2 2 2π
2π 2π 9N 10N
4π + δ
10
Useful formula for quick evaluation
Exact equation
9
Approximation • Watch out: damping ratio vs. damping constant
Logarithmic Decrement δ
8
7 Empty Full
6
5
4
3
2
1 m 1, k 1, c 1 m2>m1, k1, c1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
c1 c1
ζ 1 = -------------------- ζ 2 = -------------------- < ζ 1
Damping ratio ζ 2 k1 m1 2 k1 m2

3.4 Friction damping • Free vibrations

It is a nonlinear problem!
20
Free vibration
15
Friction force
10
Displacement
5
a) b) 0
– f k ( t ) – f μ = mu··( t ) – f k ( t ) + f μ = mu··( t ) -5
mu··( t ) + ku ( t ) = – f μ mu··( t ) + ku ( t ) = f μ -10
• Solution of b) -15
-20
fμ
u ( t ) = A 1 cos ( ω n t ) + A 2 sin ( ω n t ) + u μ with u μ = ---- (3.64) 0 1 2 3 4 5 6 7 8 9 10
k Time (s)
Figure: f=0.5 Hz , u0=10 , v0 = 50, uf = 1
u· ( t ) = – ω n A 1 sin ( ω n t ) + ω n A 2 cos ( ω n t ) (3.65)
• Calculation example:
by means of the initial conditions u ( 0 ) = u 0 , u· ( 0 ) = v 0 we ob-
- Step 1:
tain the constants:
Initial conditions u ( 0 ) = u 0 , u· ( 0 ) = 0
A1 = u0 – uμ , A2 = v0 ⁄ ωn A1 = u0 – uμ , A2 = 0 (3.66)
• Solution of a): Similar, with – u μ instead of +u μ π
u ( t ) = [ u 0 – u μ ] cos ( ω n t ) + u μ 0 ≤ t < ------ (3.67)
ωn
π
End displacement: u § ------· = [ u 0 – u μ ] ( – 1 ) + u μ = – u 0 + 2u μ
© ω n¹

- Step 2: • Comparison Viscous damping vs. Friction damping

Initial conditions u ( 0 ) = – u 0 + 2u μ , u· ( 0 ) = 0
Free vibration: f=0.5 Hz , u0=10 , v0 = 50, uf = 1
A 1 = u ( 0 ) + u μ = – u 0 + 2u μ + u μ = – u 0 + 3u μ , A 2 = 0 (3.68)
Logarithmic decrement:
π
u ( t ) = [ – u 0 + 3u μ ] cos ( ω n t ) – u μ 0 ≤ t < ------ (3.69) δ ζ [%]
ωn U0 UN N
π 1 18.35 14.35 1 0.245 3.91
End displacement: u § ------· = [ – u 0 + 3u μ ] ( – 1 ) – u μ = u 0 – 4u μ
© ω n¹ 2 18.35 10.35 2 0.286 4.56
3 18.35 6.35 3 0.354 5.63
- Step 3: 4 18.35 2.35 4 0.514 8.18
Initial conditions .... Average 5.57
• Important note: Comparison:

The change between case a) and case b) occurs at velocity re- 20
versals. In order to avoid the build-up of inaccuracies, the dis-
Friction damping
placement at velocity reversal should be identified with 15
Viscous damping
adequate precision (iterate!)
10
• Visualization of the solution by means of the Excel file given on
Displacement
5
the web page of the course (SD_FV_friction.xlsx)
0
• Characteristics of friction damping
- Linear decrease in amplitude by 4u μ at each cycle -5
- The period of the damped and of the undamped oscillator -10

is the same: -15
2π
T n = ------ -20
ωn 0 1 2 3 4 5 6 7 8 9 10
Time (s)

4 Response to Harmonic Excitation Linear inhomogeneous differential equation
• Particular solution: u p
2
u··p + 2ζω n u· p + ω n u p = f ( t ) (4.4)
• Solution of the homogeneous ODE: u h
2
u··h + 2ζω n u· h + ω n u h = 0 (4.5)
• Complete solution: u = u p + Cu h
An harmonic excitation can be described either by means of a 2
sine function (Equation 4.1) or by means of a cosine function u·· + 2ζω n u· + ω n u = f o cos ( ωt ) (4.6)
(Equation 4.2):
• Initial conditions
mu·· + cu· + ku = F o sin ( ωt ) (4.1)
u ( 0 ) = u0 , u· ( 0 ) = v 0 (4.7)
mu·· + cu· + ku = F o cos ( ωt ) (4.2)
Here we consider Equation (4.2) which after transformation be-

comes:
2
u·· + 2ζω n u· + ω n u = f o cos ( ωt ) (4.3)
where: ω n : Circular frequency of the SDoF system

ω: Circular frequency of the excitation
f o = F o ⁄ m = ( F o ⁄ k ) ⋅ ω n2
4 Response to Harmonic Excitation Page 4-1 4 Response to Harmonic Excitation Page 4-2
4.1 Undamped harmonic vibrations • Complete solution of the ODE:
2 fo
u·· + ω n u = f o cos ( ωt ) (4.8) - cos ( ωt )
u = A 1 cos ( ω n t ) + A 2 sin ( ω n t ) + ------------------ (4.16)
2 2
ωn – ω
• Ansatz for particular solution
By means of the initial conditions given in Equation (4.7), the
u p = A o cos ( ωt ) (4.9) constants A 1 and A 2 can be calculated as follows:
2 fo v0
u··p = – A o ω cos ( ωt ) (4.10)
A 1 = u 0 – ------------------
- , A 2 = ------ (4.17)
2
ωn – ω
2 ωn
By substituting (4.9) and (4.10) in (4.8):
2 2 • Denominations:
– A o ω cos ( ωt ) + A o ω n cos ( ωt ) = f o cos ( ωt ) (4.11)
- Homogeneous part of the solution: “transient”
2 2
Ao ( – ω + ωn ) = fo (4.12) - Particular part of the solution: “steady-state”
fo F 1 • Visualization of the solution by means of the Excel file given on
- = -----o ⋅ ------------------------------
A o = ------------------ - (4.13) the web page of the course (SD_HE_cosine_viscous.xlsx)
2
ωn – ω
2 k 1 – ( ω ⁄ ωn )2
fo
- cos ( ωt )
u p = ------------------ (4.14)
2 2 • Harmonic vibration with sine excitation
ωn – ω
fo
• Ansatz for the solution of the homogeneous ODE 2
- sin ( ωt )
u = A 1 cos ( ω n t ) + A 2 sin ( ω n t ) + ------------------
2
(see section on free vibrations) ωn – ω
u h = B 1 cos ( ω n t ) + B 2 sin ( ω n t ) (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:
v0 fo ( ω ⁄ ωn )
A1 = u0 , A 2 = ------ – -----------------------
-
ωn ω2 – ω2
n
4.1.1 Interpretation as a beat • Case 1: Natural frequency SDoF 0.2 Hz, excitation frequency 0.4 Hz
2 500
u·· + ω n u = f o cos ( ωt ) with u ( 0 ) = u· ( 0 ) = 0 (4.18) Total response Envelope
400
300
The solution is:
200
Displacement
fo 100
2
- ⋅ [ cos ( ωt ) – cos ( ω n t ) ]
u ( t ) = ------------------
2
(4.19)
ωn – ω 0
-100
and using the trigonometric identity -200
-300
α–β α+β
cos ( α ) – cos ( β ) = – 2 sin § ------------- t· sin § ------------- t· (4.20) -400
© 2 ¹ © 2 ¹
-500
0 2 4 6 8 10 12 14 16 18 20
one gets the equation Time [s]
2f o ω–ω ω+ω
- ⋅ sin § ----------------n t· sin § ----------------n- t·
• Case 2: Natural frequency SDoF 2.0 Hz, excitation frequency 2.2 Hz
u ( t ) = ------------------ (4.21)
2 2 © 2 ¹ © 2 ¹
ω – ωn 80
Total response Envelope
60
that describes a beat with:
40
f + fn
Displacement
Fundamental vibration: f G = ------------ (4.22) 20
2
0
f – fn -20
Envelope: f U = ------------ (4.23)
2
-40
A beat is always present, but is only evident when the natural fre- -60
quency of the SDoF system and the excitation frequency are
-80
close (see figures on the next page) 0 2 4 6 8 10 12 14 16 18 20
Time [s]
• Transition to f = fn 4.1.2 Resonant excitation (ω = ωn)

2
u·· + ω n u = f o cos ( ω n t ) (4.24)
f 2.0500 • Ansatz for the particular solution
---- = ----------------
fn 2.0000
u p = A o t sin ( ω n t ) (4.25)
u· p = A o sin ( ω n t ) + A o ω n t cos ( ω n t ) (4.26)
2
u··p = 2A o ω n cos ( ω n t ) – A o ω n t sin ( ω n t ) (4.27)
---f- = 2.0250
----------------
fn 2.0000 By substituting Equations (4.25) and (4.27) in (4.24):
2 2
2A o ω n cos ( ω n t ) – A o ω n t sin ( ω n t ) + A o ω n t sin ( ω n t ) = f o cos ( ω n t )
(4.28)
2A o ω n = f o (4.29)
---f- = 2.0125
---------------- fo Fo ωn
fn 2.0000 A o = ---------- = ----- ⋅ ------ (4.30)
2ω n k 2
fo
u p = ---------- t sin ( ω n t ) (4.31)
2ω n
f 2.0000 • Ansatz for the solution of the homogeneous ODE
---- = ----------------
fn 2.0000 (see section on free vibrations)
Resonance!
u h = B 1 cos ( ω n t ) + B 2 sin ( ω n t ) (4.32)
• Complete solution of the ODE: 4.2 Damped harmonic vibration

fo 2
u = A 1 cos ( ω n t ) + A 2 sin ( ω n t ) + ---------- t sin ( ω n t ) (4.33) u·· + 2ζω n u· + ω n u = f o cos ( ωt ) (4.37)
2ω n
• Ansatz for particular solution
By means of the initial conditions given in Equation (4.7), the
constants A 1 and A 2 can be calculated as follows: u p = A 3 cos ( ωt ) + A 4 sin ( ωt ) (4.38)
v0 u· p = – A 3 ω sin ( ωt ) + A 4 ω cos ( ωt ) (4.39)
A1 = u0 , A 2 = ------ (4.34)
ωn
2 2
u··p = – A 3 ω cos ( ωt ) – A 4 ω sin ( ωt ) (4.40)
• Special case u 0 = v 0 = 0
(The homogeneous part of the solution falls away) By substitution Equations (4.38) to (4.40) in (4.37):
2 2 2 2
fo [ ( ω n – ω )A 3 + 2ζω n ωA 4 ] cos ( ωt ) + [ – 2ζω n ωA 3 + ( ω n – ω )A 4 ] sin ( ωt ) = f o cos ( ωt )
u = ---------- t sin ( ω n t ) (4.35) (4.41)
2ω n
Equation (4.41) shall be true for all times t and for all
Is a sinusoidal vibration with amplitude:
constants A 3 and A 4 , therefore Equations (4.42) and (4.43)
fo can be written as follows:
A = ---------- t (4.36)
2ω n 2 2
( ω n – ω )A 3 + 2ζω n ωA 4 = f o (4.42)
- The amplitude grows linearly with time (see last picture of 2 2
– 2ζω n ωA 3 + ( ω n – ω )A 4 = 0 (4.43)
interpretation “beat”);
- We have A → ∞ when t → ∞ , i.e. after infinite time the am- The solution of the system [(4.42), (4.43)] allows the
plitude of the vibration is infinite as well. calculations of the constants A 3 and A 4 as:
2 2
• Visualization of the solution by means of the Excel file given on ωn – ω 2ζω n ω
- , A 4 = f o ---------------------------------------------------------
A 3 = f o --------------------------------------------------------- -
the web page of the course (SD_HE_cosine_viscous.xlsx) 2 2 2 2 2 2 2 2
( ω n – ω ) + ( 2ζω n ω ) ( ω n – ω ) + ( 2ζω n ω )
(4.44)
• Ansatz for the solution of the homogeneous ODE • Example 1: fn = 1Hz, f = 0.2Hz, ζ = 5%, fo = 1000, u0 = 0, v0 = fo/ωn
(see Section 3.2 on damped free vibrations) 60
Steady-state response
– ζω n t
uh = e ( B 1 cos ( ω d t ) + B 2 sin ( ω d t ) ) (4.45) 40 Total response
with:
Displacement
20
2 0
ω d = ω n 1 – ζ “damped circular frequency” (4.46)
-20
• Complete solution of the ODE:
– ζω n t -40
u = e ( A 1 cos ( ω d t ) + A 2 sin ( ω d t ) ) + A 3 cos ( ωt ) + A 4 sin ( ωt )
-60
(4.47) 0 2 4 6 8 10 12 14 16 18 20
Time [s]
By means of the initial conditions of Equation (4.7), the con-
stants A 1 and A 2 can be calculated. The calculation is labo- • Example 2: Like 1 but with F(t) = Fosin(ωt) instead of Focos(ωt)
rious and should be best carried out with a mathematics pro- 50
gram (e.g. Maple). 40
Total response
30
• Denominations:
20
Displacement
- Homogeneous part of the solution: “transient” 10
0
- Particular part of the solution: “steady-state”
-10
• Visualization of the solution by means of the Excel file given on -20
20
the web page of the course (SD_HE_cosine_viscous.xlsx) -30
-40
-50
0 2 4 6 8 10 12 14 16 18 20
Time [s]
4.2.1 Resonant excitation (ω = ωn) - The amplitude is limited, i.e. the maximum displacement of
the SDoF system is:
By substituting ω = ω n in Equation (4.44) the constants A 3 and
A 4 becomes: fo Fo u st
u max = ------------2- = --------- = ------ (4.53)
2ζω n 2ζk 2ζ
fo
A 3 = 0 , A 4 = ------------2- (4.48)
2ζω n where u st = F o ⁄ k is the static displacement.
i.e. if damping is present, the resonant excitation is not a special
case any more, and the complete solution of the differential 2
equation is: - For small damping ratios ( ζ ≤ 0.2 ) ω d ≈ ω n and 1–ζ ≈1
hence Equations (4.51) becomes:
– ζω n t fo
u = e ( A 1 cos ( ω d t ) + A 2 sin ( ω d t ) ) + ------------2- sin ( ω n t ) (4.49) fo
u = ------------2- ( 1 – e
– ζω n t
) sin ( ω n t ) = u max ( 1 – e
– ζω n t
) sin ( ω n t ) (4.54)
2ζω n 2ζω n
• Special case u 0 = v 0 = 0 It is a sinusoidal vibration with the amplitude:
fo fo – ζω n t
A = u max ( 1 – e ) (4.55)
A1 = 0 , A 2 = – -------------------------------- = – -------------------- (4.50)
2 2 2ζω n ω d
2ζω n 1 – ζ and the magnitude of the amplitude at each maxima j is
fo § sin ( ω d t ) – ζω n t· uj
----------
– ζω t
- = ( 1 – e n j ) sin ( ω n t j ) (4.56)
u = ------------2- ¨ sin ( ω n t ) – --------------------- e ¸ (4.51) u max
2ζω n © 1–ζ
2 ¹
Maxima occur when sin ( ω n t ) = – 1 , d.h. when
- After a certain time, the homogeneous part of the solution
Tn
subsides and what remains is a sinusoidal oscillation of the t j = ( 4j – 1 ) ⋅ ------ , j = 1…∞ (4.57)
4
amplitude:
Tn π
fo uj – ζω n ( 4j – 1 ) ⋅ ----- – ζ ( 4j – 1 ) ⋅ ---
4 2
A = ------------2- (4.52) ----------
- = 1–e = 1–e (4.58)
2ζω n u max
• Dynamic amplification • Magnitude of the amplitude after each cycle: f(ust)

50 50
ζ = 0.01
40 40
abs(uj) / ust
umax / ust
30 30
0.02
20 20
0.05
10 10
0.10
0 0
0 0.05 0.1 0.15 0.2 0 5 10 15 20 25 30 35 40 45 0.20 50
Damping ratio ζ [-] Cycle
• Magnitude of the amplitude after each cycle: f(umax)

1
0.2
0.1 0.05
0.02
0.8
ζ = 0.01
abs(uj) / umax
0.6
0.4
0.2
0
0 5 10 15 20 25 30 35 40 45 50
Cycle
5 Transfer Functions The maximum dynamic amplitude u max given by Equation (5.5)
can be transformed to:
5.1 Force excitation 2
ωn – ω
2 2
2ζω n ω
§ · § ·2
u max = ¨ f o ---------------------------------------------------------
-¸ + ¨ f o ---------------------------------------------------------
-¸
The steady-state displacement of a system due to harmonic ex- © ( ω 2 – ω 2 ) 2 + ( 2ζω ω ) 2¹ © ( ω 2 – ω 2 ) 2 + ( 2ζω ω ) 2¹
n n n n
citation is (see Section 4.2 on harmonic excitation): (5.7)
u p = a 1 cos ( ωt ) + a 2 sin ( ωt ) (5.1)
2 2 2 2
( ω n – ω ) + ( 2ζω n ω )
with u max = f o ----------------------------------------------------------------
- (5.8)
2 2 2 2 2
[ ( ω n – ω ) + ( 2ζω n ω ) ]
2 2
ωn – ω 2ζω n ω
- , a 2 = f o ---------------------------------------------------------
a 1 = f o --------------------------------------------------------- - (5.2) 1
2 2 2 2 2 2 2 2 u max = f o -------------------------------------------------------------- (5.9)
( ω n – ω ) + ( 2ζω n ω ) ( ω n – ω ) + ( 2ζω n ω ) 2 2 2 2
( ω n – ω ) + ( 2ζω n ω )
By means of the trigonometric identity
fo 1
b
u max = -----2- ---------------------------------------------------------------------------------- (5.10)
2 2
a cos ( α ) + b sin ( α ) = a + b ⋅ cos ( α – φ ) where tan φ = --- (5.3) ω n [ 1 – ( ω ⁄ ω ) 2 ] 2 + [ 2ζ ( ω ⁄ ω ) ] 2
a n n
Equation (5.1) can be transformed as follows: Introducing the maximum static amplitude u o = F o ⁄ k = f o ⁄ ω 2n the
dynamic amplification factor V ( ω ) can be defined as:
u p = u max cos ( ωt – φ ) (5.4)
u max 1
It is a cosine vibration with the maximum dynamic amplitude V ( ω ) = ----------- = ----------------------------------------------------------------------------------- (5.11)
uo 2 2 2
u max : [ 1 – ( ω ⁄ ω n ) ] + [ 2ζ ( ω ⁄ ω n ) ]
2 2 The maximum amplification factor V ( ω ) occurs when its deriva-
u max = a1 + a2 (5.5)
tive, given by Equation (5.12), is equal to zero.
and the phase angle φ obtained from:
a2
tan φ = ----- (5.6)
a1
5 Transfer Functions Page 5-1 5 Transfer Functions Page 5-2

2 2
2ωω n [ ω – ω n ( 1 – 2ζ ) ]
2 2 5.1.1 Comments on the amplification factor V
dV
= --------------------------------------------------------------------------------
- (5.12)
dω 4 2 2 2 4 (3 ⁄ 2)
[ ω – 2 ( 1 – 2ζ )ω ω n + ω n ] 1
V ( ω ) = ----------------------------------------------------------------------------------- (5.20)
2 2 2
dV 2
[ 1 – ( ω ⁄ ω n ) ] + [ 2ζ ( ω ⁄ ω n ) ]
= 0 when: ω = 0 , ω = ± ω n 1 – 2ζ (5.13)
dω
The maximum amplification factor V ( ω ) occurs when:

• ω ⁄ ω n « 1 : Slow variation of the excitation (ζ not important)
2 1
ω = ω n 1 – 2ζ for ζ < ------- ≈ 0.71 (5.14) • V ( ω ) ≈ 1 therefore: u max ≈ u o
2
• φ ≈ 0 : Motion and excitation force are in phase
and we have:
1
ω = ωn : V = ------ (5.15) • ω ⁄ ω n » 1 : Quick variation of the excitation (ζ not important)
2ζ
ωn 2
ω = ω n 1 – 2ζ :
2 1
V = ------------------------- (5.16) • V ( ω ) ≈ § ------·
2 © ω¹
2ζ 1 – ζ
ωn 2
• u max ≈ u o ⋅ § ------· = F o ⁄ ( mω ) : Mass controls the behaviour
2
From Equation (5.6), the phase angle φ is: © ω¹
2ζω n ω 2ζ ( ω ⁄ ω n ) • φ ≈ 180 : Motion and excitation force are opposite
tan φ = ------------------
2
- = ------------------------------
2
-
2
(5.17)
ωn – ω 1 – ( ω ⁄ ωn )
The phase angle has the following interesting property: • ( ω ⁄ ω n ) ≈ 1 : (ζ very important)
2 1
dφ 2ζ [ 1 + ( ω ⁄ ω n ) ] • V ( ω ) ≈ ------
= ---------------------------------------------------------------------------------------------------
- (5.18) 2ζ
d ( ω ⁄ ωn ) 2
1 – 2 ( ω ⁄ ω n ) + ( ω ⁄ ω n ) + 4ζ ( ω ⁄ ω n )
4 2 2
• u max ≈ u o ⁄ ( 2ζ ) = F o ⁄ ( cω n ) : Damping controls the behaviour
dφ 1 1 180 • φ ≈ 90 : zero displacement when excitation force is maximum
at ω ⁄ ω n = 1 we have: = --- (= --- ⋅ --------- when φ in deg)
d ( ω ⁄ ωn ) ζ ζ π
(5.19)

• Amplification factor • Example:

10
9
0.00 An excitation produces the static displacement
0.01
Amplification factor V(ω)
8
7
F o cos ( ωt )
ζ = 0.05
u st = -------------------------- (5.21)
6 k
5
4
and its maximum is:
0.10
3
2
Fo
0.20
0.50
u o = ----- (5.22)
1 k
0.70
0
0 0.5 1 1.5 2 The steady-state dynamic response of the system is:
ω/ωn
• Phase angle u p = u max cos ( ωt – φ ) (5.23)

180
0.00 therefore:
0.01 0.10 0.20 0.50
]= 0.05
u st up
0.70 ------ = cos ( ωt ) ----- = V cos ( ωt – φ )
Phase Angle I
, (5.24)
uo uo
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 φ is always positive and because of the mi-

0 nus sign in Equation (5.24) it shows how much the response
0 0.5 1 1.5 2 to the excitation lags behind.
Z/Zn

Frequency of SDoF System fn = 1Hz (ωn = 6.28rad/s), Damping ζ = 0.1 5.1.2 Steady-state displacement quantities
5 ω ⁄ ω n = 0.9
4 V • Displacement: Corresponds to Equation (5.4)
3
Δt ω = 0.9 ⋅ 6.28
2
= 5.65 rad/s up
1 ------------ = V ( ω ) cos ( ωt – φ ) (5.25)
u / uo
0 V = 3.82 Fo ⁄ k
-1
-2 φ = 43.45°
-3 Excitation = 0.76 rad • Velocity: Obtained by derivating Equation (5.25)
-4 Steady-state response
φ 0.76
-5 Δt = ---- = ----------
ω 5.65 u· p
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 ------------ = – V ( ω )ω sin ( ωt – φ ) (5.26)
Time [s] = 0.14 s Fo ⁄ k
V ω ⁄ ω n = 1.0
5 u· p ω
4 ------------------------ = – V ( ω ) ------ sin ( ωt – φ ) (5.27)
3 ω = 1.0 ⋅ 6.28 ( F o ⁄ k )ω n ωn
2 Δt = 6.28 rad/s
1
u / uo
0 V = 5.00 u· p ω
-1 - = – V v ( ω ) sin ( ωt – φ ) with V v ( ω ) = ------ V ( ω ) (5.28)
---------------------
-2 φ = 90.00° F o ⁄ km ω n
-3 Excitation
= 1.57 rad
-4 Steady-state response φ 1.57
-5 Δt = ---- = ---------- • Acceleration: Obtained by derivating Equation (5.26)
ω 6.28
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
Time [s]
= 0.25 s
u··p 2
------------ = – V ( ω )ω cos ( ωt – φ ) (5.29)
5 ω ⁄ ω n = 1.1 Fo ⁄ k
4 V
3 ω = 1.1 ⋅ 6.28
2 Δt = 6.91 rad/s u··p ω
2
1 ------------------------ = – V ( ω ) ------ cos ( ωt – φ ) (5.30)
u / uo
2 2
0 V = 3.28 ( F o ⁄ k )ω n ωn
-1
φ = 133.66°
-2 = 2.33 rad
-3 Excitation u··p ω
2
-4 φ 2.33
Δt = ---- = ----------
-------------- = – V a ( ω ) cos ( ωt – φ ) with V a ( ω ) = ------ V ( ω ) (5.31)
Fo ⁄ m
2
-5 ω 6.91 ωn
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 = 0.34 s
Time [s]

• Amplification factors 5.1.3 Derivating properties of SDoF systems from

10
0.00 Resonant harmonic vibrations
9
0.01
displacement
Amplification Factor V(ω)
8
• Half-power bandwidth
7
ζ = 0.05
6 10
5
V(Resonance)
2 9
4 ω = ω n 1 – 2ζ
plification factor V(ω)

0.10 8
3
V(Resonance)
2 0.20 1 7
1
0.50 V = ---------------------- 2
0
0.70 2ζ 1 – ζ 6
0 0.5 1 1.5 2
ω/ωn
5
10 4
9
0.00 Resonant
0.01
velocity 3
Amp
cation factor Vv(ω)
7 2
6
ζ = 0.05 ζ = 0.05
1 2ζ
5 ω = ωn
0
ωa
Amplificati
4
0 10
0.10 ωb
3 0 0.5 1 1.5 2
1
2 0.20 V = ------ ω/ωn
1
0.50 2ζ
0
0.70 Condition:
0 0.5 1 1.5 2
ω/ωn 2
V ( ω ⁄ ω n = 1 – 2ζ ) 1 1
10
0.00 Resonant V ( ω ) = ----------------------------------------------------- = ------- ⋅ ---------------------- (5.32)
9 2 2 2ζ 1 – ζ
8
0.01
acceleration
Amplification factor Va(ω)
1 1
1- ----------------------
7
ζ = 0.05
---------------------------------------------------------------------------------- = ------ ⋅ (5.33)
6 2 2 2 2 2ζ 1 – ζ
5 ωn [ 1 – ( ω ⁄ ω n ) ] + [ 2ζ ( ω ⁄ ω n ) ]
4 ω = ----------------------
ω ·4
§ ----- ω 2
- – 2 ( 1 – 2ζ 2 ) § ------· + 1 – 8ζ 2 ( 1 – ζ 2 ) = 0
0.10 2
3 1 – 2ζ (5.34)
2 0.20
© ω n¹ © ω n¹
0.50
1 1
V = ---------------------- ω-· 2
0
0.70
2ζ 1 – ζ § ----- 2
= 1 – 2ζ ± 2ζ 1 – ζ
2
(5.35)
0 0.5 1 1.5 2 © ω n¹
ω/ωn

2
For small damping, the terms featuring ζ can be neglected: 5.2 Force transmission (vibration isolation)
ω-
----- ≈ 1 ± 2ζ ≈ 1 ± ζ (5.36)
ωn
This yield the solution for the half-power bandwidth: The mass-spring-damper system,
shown here on the right, is excited
ωb – ωa by the harmonic force
2ζ = ------------------- (5.37)
ωn
F ( t ) = F o cos ( ωt )
• Remarks on the frequency response curve
What is the reaction force F T ( t ) ,
• The natural frequency of the system can be derived from the res-
which is introduced in the founda-
onant response. However, it is sometimes problematic to build the
whole frequency response curve because at resonance the sys- tion?
tem could be damaged. For this reason it is often better to deter-
mine the properties of a system based on vibration decay tests
(see section on free vibration)
• The natural frequency ω n can be estimated by varying the Excita- The reaction force F T ( t ) results from the sum of the spring force
F s and the damper’s force F c
tion until a 90° phase shift in the response occurs.
• Damping can be calculated by means of Equation (5.15) as: F T ( t ) = F s ( t ) + F c ( t ) = ku ( t ) + cu· ( t ) (5.38)
1 uo The steady-state deformation of the system due to harmonic ex-

ζ = --- ⋅ -----------
2 u max citation F ( t ) is according to Equation (5.4):
However, it is sometimes difficult to determine the static deflection Fo
u o , therefore, the definition of half-power bandwidth is used to u p = u max cos ( ωt – φ ) with u max = u o V ( ω ) = ----- V ( ω ) (5.39)
k
estimate the damping.
• Damping can be determined from the slope of the phase angle
By substituting Equation (5.39) and its derivative into Equation
curve using Equation (5.19). (5.38) we obtain:
Fo
F T ( t ) = ----- V ( ω ) [ k cos ( ωt – φ ) – cω sin ( ωt – φ ) ] (5.40)
k

with the trigonometric identity from Equation (5.3): • Representation of the transmissibility TR
Fo 2 2 2 10
F T ( t ) = ----- V ( ω ) [ k + c ω cos ( ωt – φ ) ] (5.41) 0.00
k
9
0.01
and by substituting the identity c = ( 2ζk ) ⁄ ω n : 8
ζ = 0.05
Trrasmissibility TR
7
ω 2
F T ( t ) = F o V ( ω ) 1 + § 2ζ ------· cos ( ωt – φ ) (5.42)
© ω n¹ 6
5
the maximum reaction force becomes: 0.10
4
F T,max
--------------- = TR ( ω ) (5.43) 3
0.20
Fo
2
0.50
where the quantity TR ( ω ) is called Transmissibility and it is 1
0.70 2
equal to: 0
0 0.5 1 1.5 2 2.5 3 3.5 4
ω 2 ω/ωn
TR ( ω ) = V ( ω ) 1 + § 2ζ ------· (5.44)
© ω n¹
- When ω ⁄ ω n > 2 then TR < 1 : Vibration isolation
2
1 + [ 2ζ ( ω ⁄ ω n ) ]
= ------------------------------------------------------------------------------ - When ω ⁄ ω n > 2 damping has a stiffening effect
2 2 2
[ 1 – ( ω ⁄ ω n ) ] + [ 2ζ ( ω ⁄ ω n ) ]
- High tuning (sub-critical excitation)
Special case: - Low tuning (super-critical excitation):
2 Pay attention to the starting phase!
ω
TR § ------ = 1· = ----------------------
1 + 4ζ
© ωn ¹
(5.45)
2ζ

5.3 Base excitation (vibration isolation) The right hand side of the ODE (5.48) can be interpreted as an
external excitation force F ( t ) = ky + cy· :
5.3.1 Displacement excitation
F ( t ) = ky go cos ( ωt ) – cy go ω sin ( ωt ) (5.49)
ω
= ky go cos ( ωt ) – 2ζ ------ sin ( ωt )
ωn
ω 2
= ky go 1 + § 2ζ ------· cos ( ωt + φ )
© ω n¹
The external excitation force F ( t ) is harmonic with amplitude:
ω 2
F o = ky go 1 + § 2ζ ------· (5.50)
© ω n¹
The mass-spring-damper system, shown here above is excited
According to Equations (5.10) and (5.11) the maximum displace-
by the harmonic vertical ground displacement
ment of the system due to such a force is equal to:
y g ( t ) = y go cos ( ωt ) (5.46)
Fo ω 2
u max = ----- V ( ω ) = y go 1 + § 2ζ ------· V ( ω ) (5.51)
What is the absolute vertical displacement u ( t ) of the system? k © ω n¹
The differential equation of the system is: By substituting Equation (5.44) we obtain:
mu·· + c ( u· – y· ) + k ( u – y ) = 0 (5.47) u max

----------- = TR ( ω ) (5.52)
after rearrangement: y go
mu·· + cu· + ku = ky + cy· (5.48) where again TR ( ω ) is the transmissibility given by Equation
(5.44).

5.3.2 Acceleration excitation The steady-state relative deformation u rel of the system due to
Pay attention: the harmonic ground acceleration y··g is given by Equation (5.1):
This base excitation, like u rel = a 1 cos ( ωt ) + a 2 sin ( ωt ) (5.59)
the excitation discussed
in the previous Section with the constants a 1 and a 2 according to Equation (5.2), and
5.3.1, is an harmonic ex- with:
citation and not an arbi-
trary excitation like e.g. Fo – my··go
an earthquake (see Sec- f o = ----- = ---------------- = – y··go (5.60)
m m
tion 7).
By double derivation of Equation (5.59), the relative acceleration
The mass-spring-damper system, shown above here, is excited u··rel can be calculated as:
by the harmonic vertical ground acceleration.
2 2
u··rel = – a 1 ω cos ( ωt ) – a 2 ω sin ( ωt ) (5.61)
y··g ( t ) = y··go cos ( ωt ) (5.53)
The desired absolute acceleration is:
What is the absolute vertical acceleration u··( t ) of the system?
2 2
u·· = u··rel + y··g = – a 1 ω cos ( ωt ) – a 2 ω sin ( ωt ) + y··go cos ( ωt ) (5.62)
The differential equation of the system is:
By substituting the constants a 1 , a 2 and f o given by Equations
mu·· + c ( u· – y· ) + k ( u – y ) = 0 (5.54)
(5.2) and (5.60), and after a long but simple rearrangement, the
after rearrangement: equations for the maximum absolute vertical acceleration of the
system is obtained as:
mu·· + c ( u· – y· ) + k ( u – y ) – my·· = – my·· (5.55)
m ( u·· – y··) + c ( u· – y· ) + k ( u – y ) = – my·· (5.56) u··max
----------- = TR ( ω ) (5.63)
y··go
mu··rel + cu· rel + ku rel = – my··g (5.57)
where again TR ( ω ) is the transmissibility given by Equation
mu··rel + cu· rel + ku rel = – my··go cos ( ωt ) (5.58) (5.44).

• Additional derivation: 5.3.3 Example transmissibility by base excitation

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:
Vertical base excitation:
fo – y··go
u rel, max = -----2- V ( ω ) = -------------
2
-V(ω) (5.64) y··g ( t ) = A 0 cos ( ω 0 t )
ωn ωn
u rel, max
- = V(ω)
---------------------- (5.65)
2
( y··go ⁄ ω n )
• Natural frequency SDoF system: f n = 0.5Hz
• Excitation frequency: f 0 = 2.0Hz
• Excitation amplitude: A 0 = 10m ⁄ s 2
Sought is the maximum absolute acceleration u··max of the SDoF

system for ζ = 2% and for ζ = 20% .
• The steady-state maximum absolute acceleration is:
2
• ζ = 2% , ω 0 ⁄ ω n = 4 : TR = 0.068 and u··max = 0.68m ⁄ s
• ζ = 20% , ω 0 ⁄ ω n = 4 : TR = 0.125 and u··max = 1.25m ⁄ s 2
• Is the steady-state maximum absolute acceleration really the

maximum absolute acceleration or at start even larger abso-
lute accelerations may result?
• Assumptions: starting time t a = 80s , sinusoidal start function for
excitation frequency and excitation amplitude.
• Numerical computation using Newmark’s Method (see Section 7)

• Case 1: Initial situation with ζ = 2% • Case 2: Increase of the damping rate from ζ = 2% to ζ = 20%
2 2
Excitation freq. [Hz]

1 1
f=fn=0.5Hz f=fn=0.5Hz
0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Excitation amplitude [m/s2]

10 10
5 5
0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
10 10
Excitation [m/s2]
Excitation [m/s2]
5 5
0 0
-5 -5
-10 -10
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
20 20
Aabs SDoF [m/s2]
Aabs SDoF [m/s2]

10 10
0 0
-10 -10
-20 -20
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time [s] Time [s]

• Case 3: Reduction of starting time from ta = 80s to ta = 20s (ζ = 2%) • Case 4: Change the start function for the amplitude (ζ = 2%)
2 2

1 1
f=fn=0.5Hz f=fn=0.5Hz
0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100

10 10
5 5
0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
10 10
Excitation [m/s2]
Excitation [m/s2]
5 5
0 0
-5 -5
-10 -10
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
20 20
Aabs SDoF [m/s2]
Aabs SDoF [m/s2]

10 10
0 0
-10 -10
-20 -20
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time [s] Time [s]

• Notes 5.4 Summary Transfer Functions

The excitation function in the starting phase has the form:
1
y··g ( t ) = A ( t ) cos ( Ω ( t ) ⋅ t ) (5.66) V ( ω ) = ----------------------------------------------------------------------------------- (5.72)
2 2 2
[ 1 – ( ω ⁄ ω n ) ] + [ 2ζ ( ω ⁄ ω n ) ]
The excitation angular frequency varies with time, and is:
2
ω(t) = d (Ω(t) ⋅ t ) (5.67) 1 + [ 2ζ ( ω ⁄ ω n ) ]
dt TR ( ω ) = ------------------------------------------------------------------------------ (5.73)
2 2 2
• Linear variation of the excitation circular frequency
[ 1 – ( ω ⁄ ω n ) ] + [ 2ζ ( ω ⁄ ω n ) ]
ω0 t u max
Ω ( t ) = ------- ⋅ t : ω ( t ) = ω 0 ⋅ --- ( 0 ≤ t ≤ t a ) (5.68) • Force excitation: ----------- = V ( ω )
2t a ta uo
• Parabolic variation of the excitation circular frequency
F T,max
ω0 2 t 2 • Force transmission: --------------- = TR ( ω )
Ω ( t ) = ------2- ⋅ t : ω ( t ) = ω 0 ⋅ § ---· ( 0 ≤ t ≤ t a ) (5.69) Fo
© t a¹
3t a
u max
• Sinusoidal variation of the excitation circular frequency • Displacement excitation: ----------- = TR ( ω )
y go
2ω 0 t a π t π t
Ω ( t ) = – -------------- cos § --- ⋅ ---· : ω ( t ) = ω 0 ⋅ sin § --- ⋅ ---· ( 0 ≤ t ≤ t a )
πt © 2 t a¹ © 2 t a¹ u··max
• Acceleration excitation ----------- = TR ( ω )
(5.70) y··go
• Double-sinusoidal variation of the excitation circular frequency u rel, max
- = V(ω)
----------------------
·· 2
sin § π ⋅ ---· t a
t ( y go ⁄ ω n )
ω0 © t a¹
Ω ( t ) = ------ 1 – ----------------------------- : ω ( t ) = ω 0 1 – cos § π ⋅ ---·
t
©
(5.71)
2 πt t a¹ • For further cases check the literature.
• Visualization of the solution by means of the Excel file given on

the web page of the course (SD_HE_Starting_Phase.xlsx)

6 Forced Vibrations with the fundamental frequency

2π
ω 0 = ------ (6.3)
6.1 Periodic excitation T0
4.0 Taking into account the orthogonality relations:
Half-sine excitation
T0
3.5
T0 0 for n ≠ j
³0 sin ( nω 0 t ) sin ( jω 0 t ) dt = ® (6.4)
3.0 ¯ T 0 ⁄ 2 for n = j
Force F(t) [kN]
2.5
T0 0 for n ≠ j
³0 cos ( nω 0 t ) cos ( jω 0 t ) dt = ® (6.5)
2.0 ¯ T 0 ⁄ 2 for n = j
1.5 T0
³0 cos ( nω 0 t ) sin ( jω 0 t ) dt = 0 (6.6)
F
1.0
0.5 the Fourier coefficients a n can be computed by multiplying Equation

(6.2) by cos ( jω 0 t ) first, and then integrating it over the period T 0 .
0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 • j = 0
Time (s)
T0 T0
An excitation is periodic if: ³0 F ( t ) cos ( jω 0 t ) dt = ³0 a 0 cos ( jω 0 t ) dt (6.7)
F ( t + nT 0 ) = F ( t ) for n = – ∞, …, – 1, 0, 1, …, ∞ (6.1) ∞ T0 T0
+ ¦ ³
n=1 0
a n cos ( nω 0 t ) cos ( jω 0 t ) dt + ³ b n sin ( nω 0 t ) cos ( jω 0 t ) dt
0
The function F ( t ) can be represented as a sum of several har-

T0 T0
monic functions in the form of a Fourier series, namely: ³0 F ( t ) dt = ³0 a 0 dt = a 0 T 0 (6.8)
∞
1 T0
F ( t ) = a0 + ¦ [ an cos ( nω0 t ) + bn sin ( nω0 t ) ] (6.2) a 0 = ------ ⋅ ³ F ( t ) dt
T0 0
(6.9)
n=1
6 Forced Vibrations Page 6-1 6 Forced Vibrations Page 6-2

• j = n 6.1.1 Steady state response due to periodic excitation
T0 T0 mu·· + cu· + ku = F ( t ) (6.14)

³0 F ( t ) cos ( jω 0 t ) dt = ³0 a 0 cos ( jω 0 t ) dt (6.10) 2 F(t)
u·· + 2ζω n u· + ω n u = ---------- (6.15)
∞ T0 T0
m
+ ¦ ³0
n=1
a n cos ( nω 0 t ) cos ( jω 0 t ) dt + ³ b n sin ( nω 0 t ) cos ( jω 0 t ) dt
0 ∞
F ( t ) = a0 + ¦ [ a n cos ( nω 0 t ) + b n sin ( nω 0 t ) ] (6.16)
T0
n=1
T
³0 F ( t ) cos ( nω 0 t ) dt = a n ⋅ -----0
2
(6.11)
• Static Part ( a 0 )
2 T0 a0
a n = ------ ⋅ ³ F ( t ) cos ( nω 0 t ) dt (6.12) u 0 ( t ) = -----
k
(6.17)
T0 0
• Harmonic part “cosine” (see harmonic excitation)
Similarly, the Fourier coefficients b n can be computed by first
multiplying Equation (6.2) by sin ( jω 0 t ) and then integrating it Co sin e a n 2ζβ n sin ( nω 0 t ) + ( 1 – β n ) cos ( nω 0 t )
2
nω
un (t) - , β n = ---------0
= ----- ⋅ ----------------------------------------------------------------------------------------
over the period T 0 . k 2 2 2 ωn
( 1 – β n ) + ( 2ζβ n )
2 T0 (6.18)
b n = ------ ⋅ ³ F ( t ) sin ( nω 0 t ) dt (6.13)
T0 0 • Harmonic part “sine” (similar as “cosine”)
2
• Notes Sine b n ( 1 – β n ) sin ( nω 0 t ) – 2ζβ n cos ( nω 0 t ) nω
un ( t ) - , β n = ---------0
= ----- ⋅ ----------------------------------------------------------------------------------------
k 2 2 2 ωn
- a 0 is the mean value of the function F ( t ) ( 1 – β n ) + ( 2ζβ n )
(6.19)
- The integrals can also be calculated over the interval
[– T 0 ⁄ 2,T 0 ⁄ 2] • The steady-state response u ( t ) of a damped SDoF system un-
der the periodic excitation force F ( t ) is equal to the sum of the
- For j = 0 no b-coefficient exists terms of the Fourier series.
∞ Co sin e ∞ Sine
u ( t ) = u0 ( t ) + ¦ un (t) + ¦ un (t) (6.20)
n=1 n=1

6.1.2 Half-sine 2
2A t p πt 4Aτ cos ( nπτ ) -
a n = ------- ⋅ ³ sin § -----· cos ( nω 0 t ) dt = ------------------------------------ (6.23)
T0 0 © tp ¹ 2 2
A series of half-sine functions is a good model for the force that π ( 1 – 4n τ )
is generated by a person jumping.
2A tp πt 4Aτ sin ( nπτ ) cos ( nπτ )-
4.0 b n = ------- ⋅ ³ sin § -----· sin ( nω 0 t ) dt = --------------------------------------------------------
T0 0 © tp ¹ 2 2
3.5
T0
π ( 1 – 4n τ )
tp (6.24)
3.0
The approximation of the half-sine model for T 0 = 0.5s and
Force F(t) [kN]
2.5
t p = 0.16s by means of 6 Fourier terms is as follows:
2.0
4.0
Static term (n=0)
1.5 T0 First harmonic (n=1)
F
3.0 Second harmonic (n=2)

1.0 Third harmonic (n=3)
Total (6 harmonics)
0.5
Force F(t) [kN]

2.0
0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.0
Time (s)
0.0
F
§ πt-· for 0 ≤ t < t
° A sin © ----
tp ¹ p -1.0
F(t) = ® (6.21)
° 0 for t p ≤ t < T 0 -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: • Note
The static term a 0 = 2Aτ ⁄ π = G corresponds to the weight of
A tp πt tp
a 0 = ------ ⋅ ³ sin § -----· dt = ----------
2Aτ
with τ = ------ (6.22) the person jumping.
T0 0 © tp ¹ π T0

6.1.3 Example: “Jumping on a reinforced concrete beam” • Maximum deflections

• Beam Static:
G
u st = ------
Kn
• Young’s Modulus:
E = 23500MPa Dynamic: u max = max ( u ( t ) ) with u ( t ) from Equation (6.20)
• Density:
ρ = 20.6kN ⁄ m 3 u max
Ratio: V = -----------
u st
• Bending stiffness:
EI = 124741kNm 2 • Investigated cases
• Damping rate
Length Frequency fn umax V
ζ = 0.017
[m] [Hz] [m] [-]
• Modal mass
M n = 0.5M tot 26.80 1 0.003 1.37
• Modal stiffness 19.00 2 0.044 55.94
π 4 EI 15.50 3 0.002 3.62
K n = ----- ⋅ -----3-
2 L
13.42 4 0.012 41.61
• Excitation (similar to page 186 of [Bac+97]) 12.01 5 0.001 4.20
4.0
• Jumping frequency: 10.96 6 0.004 25.02
3.5 f 0 = 2Hz
3.0
• Period: T 0 = 0.5s • Notes
Force F(t) [kN]
2.5
2.0
• Contact time: - When the excitation frequency f 0 is twice as large as the
t p = 0.16s
1.5
natural frequency f n of the beam, the magnification factor
F
1.0
• Person’s weight:
0.5
G = 0.70kN V is small.
0.0 • Amplitude: - Taking into account the higher harmonics can be impor-
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
A = 3.44kN
Time (s) tant!

• Case1: f0 = 2Hz, fn = 1Hz • Case 3: f0 = 2Hz, fn = 3Hz

0.0035 0.0020
Static term (n=0)
0.0030 First harmonic (n=1)

Total (6 harmonics)
Displacement [m]
Displacement [m]
0.0020 Static term (n=0)
First harmonic (n=1) 0.0010
Third harmonic (n=3)
0.0010 Total (6 harmonics)

0.0005
0.0005
0.0000
D
D
0 0000
0.0000
-0.0005
-0.0005
-0.0010
-0.0015 -0.0010
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s) Time (s)
• Case 2: f0 = 2Hz, fn = 2Hz • Case 4: f0 = 2Hz, fn = 4Hz

0.0500 0.0150
Static term (n=0)
First harmonic (n=1)
0.0400
Second harmonic (n=2)
0.0300 Total (6 harmonics)
Displacement [m]
Displacement [m]
0.0200
0.0050
0.0100
0.0000 0.0000
-0.0100
-0.0050
D
D
0 0200
-0.0200
Static term (n=0)
First harmonic (n=1)
-0.0300
Second harmonic (n=2) -0.0100
-0.0400 Total (6 harmonics)
-0.0500 -0.0150
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s) Time (s)

• Case 5: f0 = 2Hz, fn = 5Hz 6.2 Short excitation

0.0010
Static term (n=0)
First harmonic (n=1) 6.2.1 Step force
0.0006 Total (6 harmonics) The differential equation of an undamped SDoF System loaded
Displacement [m]
0.0004 with a force F 0 which is applied suddenly at the time t = 0 is:

0.0002
mu·· + ku = F 0 (6.25)
0.0000
-0.0002 There is a homogeneous and a particular solution

D
-0.0004 u h = A 1 cos ( ω n t ) + A 2 sin ( ω n t ) (see free vibrations) (6.26)

-0.0006
up = F0 ⁄ k (6.27)
-0.0008
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s) The overall solution u ( t ) = u h + u p is completely defined by the in-
itial conditions u ( 0 ) = u· ( 0 ) = 0 and it is:
• Case 6: f0 = 2Hz, fn = 6Hz
F0
0.0050
Static term (n=0) u ( t ) = ----- [ 1 – cos ( ω n t ) ] (6.28)
First harmonic (n=1) k
0.0030 Total (6 harmonics) • Notes
Displacement [m]
0.0020 • The damped case can be solved in the exact same way. On the
0.0010
web page of the course there is an Excel file to illustrate this exci-
tation.
0.0000
• The maximum displacement of an undamped SDoF System under
-0.0010
a step force is twice the static deflection u st = F 0 ⁄ k .
D
-0.0020
• The deflection at the time t = ∞ of a damped SDoF System under
-0.0030
a step force is equal to the static deflection u st = F 0 ⁄ k .
-0.0040
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)

• Step force: Tn=2s, Fo/k=2, ζ=0 6.2.2 Rectangular pulse force excitation
4.5
Dynamic response
4 Excitation
3.5
3
Displacement
2.5
1.5
1 The differential equation of an undamped SDoF system under a

0.5 rectangular pulse force excitation is:
0 ·· for t ≤ t 1
0 1 2 3 4 5 6 7 8 9 10 mu + ku = F 0
® ·· (6.29)
Time (s)
¯ mu + ku = 0 for t > t 1
• Step force: Tn=2s, Fo/k=2, ζ=10% Up to time t = t 1 the solution of the ODE corresponds to Equa-
4
Dynamic response
tion (6.28). From time t = t 1 onwards, it is a free vibration with
3.5 Excitation the following initial conditions:
3 F0
u ( t 1 ) = ----- [ 1 – cos ( ω n t 1 ) ] (6.30)
Displacement
2.5 k
2 F0
u· ( t 1 ) = ----- ω n sin ( ω n t 1 ) (6.31)
1.5 k
1 The free vibration is described by the following equation:

0.5 u h = A 1 cos ( ω n ( t – t 1 ) ) + A 2 cos ( ω n ( t – t 1 ) ) (6.32)
0
0 1 2 3 4 5 6 7 8 9 10 and through the initial conditions (6.30) and (6.31), the constants
Time (s) A 1 and A 2 can be determined.

• Short duration of excitation ( t 1 ⁄ T n is small) The equation of an undamped free vibration is:
The series expansion of sine and cosine is: v0 2 v0

u ( t ) = A cos ( ω n t – φ ) with A = u 0 + § ------· and tan φ = ------------
© ω n¹ ωn u0
( ωn t1 )2
cos ( ω n t 1 ) = cos § ------ t 1· = 1 – ------------------ + …
2π (6.38)
© Tn ¹
(6.33)
2
therefore, the maximum amplitude of a short excitation is:
( ωn t1 )3
sin ( ω n t 1 ) = sin § ------ t 1· = ω n t 1 + ------------------ + …
2π
© Tn ¹
(6.34) v0
6
A = ------ (6.39)
ωn
and for small t 1 ⁄ T n the expressions simplifies to:
• Notes
cos ( ω n t 1 ) ≅ 1 , sin ( ω n t 1 ) ≅ ω n t 1 (6.35)
• The damped case can be solved in the exact same way. On the web
By substituting Equation (6.35) in Equations (6.30) and (6.31) it page of the course there is an Excel file to illustrate this excitation.
follows that: • Rectangular pulse force excitation: When t 1 > T n ⁄ 2 , the maximum
response of the SDoF system is equal to two times the static de-
F0 2 F0 t1 flection u st = F 0 ⁄ k .
u ( t1 ) = 0 , u· ( t 1 ) = ----- ω n t 1 = ---------- (6.36)
k m • 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
Equation (6.36) shows, that a short excitation can be interpreted system can even be as large as 4F 0 ⁄ k .
as a free vibration with initial velocity
• Rectangular pulse force excitation: try yourself using the provided
v0 = I ⁄ m (6.37) Excel spreadsheet.
• Short excitation: The shape of the excitation has virtually no effect
where I is the impulse generated by the force F 0 over the time t 1 . on the maximum response of the SDoF system. Important is the
impulse.
• Rectangular pulse force excitation: I = F0 t1
• Short excitation: Equation (6.59) is exact only for t 1 ⁄ T n → 0
• Triangular pulse force excitation: I = 0.5F 0 t 1 and ζ = 0 . For all other cases, it is only an approximation, which
overestimates the actual maximum deflection.
t1
• Arbitrary short excitation: I = ³0 F ( t ) dt

• Rectangular pulse: Tn=2s, t1=0.5s (t1/Tn=0.25), Fo/k=2, ζ=0% • Rectangular pulse: Tn=2s, t1=2s (t1/Tn=1.00), Fo/k=2, ζ=0%
4 4.5
Dynamic response Dynamic response
3 Excitation 4 Excitation
3.5
2
3
Displacement
Displacement
1 2.5
0 2
-1 1.5
1
-2
0.5
-3 0
-4 -0.5
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Time (s) Time (s)
• Rectangular pulse: Tn=2s, t1=1s (t1/Tn=0.50), Fo/k=2, ζ=0% • Rectangular pulse: Tn=2s, t1=3s (t1/Tn=1.50), Fo/k=2, ζ=0%
5 5
4 Excitation 4 Excitation
3 3
2 2
Displacement
Displacement
1 1
0 0
-1 -1
-2 -2
-3 -3
-4 -4
-5 -5
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Time (s) Time (s)

• Rectangular pulse: Tn=2s, t1=3.5s (t1/Tn=1.75), Fo/k=2, ζ=0% • Short rectangular pulse: Tn=2s, t1=0.05s, Fo/k=2, ζ=0%
5 2.5
4 Excitation Excitation
2
3
2 1.5
Displacement
Displacement
1
1
0
-1 0.5
-2
0
-3
-4 -0.5
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Time (s) Time (s)
• Rectangular pulse: Tn=2s, t1=4s (t1/Tn=2.00), Fo/k=2, ζ=0% • Short rectangular pulse: Tn=2s, t1=0.05s, Fo/k=2, ζ=5%
4.5 2.5
4 Excitation Excitation
2
3.5
3
1.5
Displacement
Displacement
2.5
2 1
1.5
0.5
1
0.5
0
0
-0.5 -0.5
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Time (s) Time (s)

6.2.3 Example “blast action” Mass: m = 3.05 ⋅ 0.276 ⋅ 2.45 = 2.06t ⁄ m
• Test Concrete f c' = 41.4MPa , E c = 5000 ⋅ f c' = 32172MPa

3 6 4
Stiffness I o = ( 3050 ⋅ 276 ) ⁄ 12 = 5344 ×10 mm
E c I o = 171.9kNm 2
E c I = 0.30E c I o = 52184kNm 2 (due to cracking!)
- Action
t 1 ≈ 0.3ms is by sure much

shorter than the period
T n = 64ms of the slab (see
Equation (6.51)). There-
• Modelling option 1 fore, the excitation can be
considered as short.
Within a simplified modelling approach, it is assumed that the
slab remains elastic during loading. Sought is the maximum de-
flection of the slab due to the explosion. - Equivalent modal SDoF system (see Section “Modelling”)
- Simplified system
Cross-Section
Ansatz for the deformed shape:
ψ = C1 ⋅ sin ( βx ) + C2 ⋅ cos ( βx ) + C3 ⋅ sin h ( βx ) + C4 ⋅ cos h ( βx )

(6.40)

Boundary conditions: L 2 EI
k* = ³0 ( EI ⋅ ( ψ'' ) ⋅ dx ) = 104.37 ⋅ -----3-
L
(6.45)
ψ ( 0 ) = 0 , ψ ( L ) = 0 , ψ' ( 0 ) = 0 , ψ'' ( L ) = 0 (6.41)
L 2 =3.45m
By means of the mathematics program “Maple” Equation (6.40) P* = ³L =1.55m ( p ⋅ ψ ⋅ dx )
1
= 0.888 ⋅ P tot (6.46)
can be solved for the boundary conditions (6.41) and we get:
For this example, the modal properties characterizing the equiv-
[ sin ( βL ) + sin h ( βL ) ] ⋅ [ cos ( βx ) – cos h ( βx ) ]
1.508 ⋅ ψ = sin ( βx ) – sin h ( βx ) + ------------------------------------------------------------------------------------------------------------------ alent modal SDoF system are:
– cos ( βL ) – cos h ( βL )
(6.42)
m * = 0.439 ⋅ 2.06 ⋅ 5 = 4.52t (6.47)
with
52184-
k * = 104.37 ⋅ -------------- = 43571kN/m (6.48)
βL = 3.927 (6.43) 53
The shape of the function ψ is: P * = 0.888 ⋅ 192000 = 170496kN (6.49)
1 ω = k* ⁄ m* = 43571 ⁄ 4.52 = 98.18rad/s (6.50)

0.8
T n = 2π ⁄ ω = 0.064s (6.51)
0.6
\ [-]
0.4 The maximum elastic deformation of the SDoF system can be

0.2
calculated using the modal pulse as follows:
0 –3
I * = 0.5 ⋅ P * ⋅ t 0 = 0.5 ⋅ 170496 ⋅ 0.3 ×10 = 25.6kNs (6.52)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x/L [-]
The initial velocity of the free vibration is:
And with the equations given in Section “Modelling”, the modal
I* 25.6
properties of the equivalent SDoF system are determined: v 0 = ------*- = ---------- = 5.66m/s (6.53)
m 4.52
L 2
m* = ³0 m ⋅ ψ ⋅ dx = 0.439mL (6.44) The maximum elastic deflection is:
Δ m, e = v 0 ⁄ ω = 5.66 ⁄ 98.18 = 0.058m (6.54)

• Modelling option 2 1
0.8
0.6
Within a simplified modelling approach, it is assumed that the 0.4
slab remains elastic during loading. Sought is the maximum de- 0.2
ψ [-]
0
flection of the slab due to the explosion. -0.2
-0.4
-0.6
- Simplified system -0.8
-1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x/L [-]
And with the equations given in Section “Modelling”, the modal

properties of the equivalent SDoF system are determined:
L 2
m* = ³0 m ⋅ ψ ⋅ dx = 0.5mL (6.57)
- Equivalent modal SDoF system (see Section “Modelling”)
L 2 4 EI EI
k* = ³0 ( EI ⋅ ( ψ'' ) ⋅ dx ) = 8π ⋅ -----3- = 779.27 ⋅ -----3-
L L
(6.58)
L 2 =8.45m
P* = ³L =6.55m ( p ⋅ ψ ⋅ dx )
1
= 0.941 ⋅ P tot (6.59)
For this example, the modal properties characterizing the equiv-

Ansatz for the deformed shape: alent modal SDoF system are:
m * = 0.5 ⋅ 2.06 ⋅ 10 = 10.3t (6.60)

ψ = – sin § ----------·
2πx
© L ¹
(6.55)
52184-
k * = 779.27 ⋅ -------------- = 40666kN/m (6.61)
Boundary conditions: 10 3
ψ ( 0 ) = 0 , ψ ( L ) = 0 , ψ'' ( 0 ) = 0 , ψ'' ( L ) = 0 (6.56) P * = 0.941 ⋅ 192000 = 180672kN (6.62)
The shape of the function ψ is: ω = k* ⁄ m* = 40666 ⁄ 10.3 = 62.83rad/s (6.63)

T n = 2π ⁄ ω = 0.10s (6.64) • Modelling option 3
The maximum elastic deformation of the SDoF system can be As a third option, the slab is modelled using the commercial finite
calculated using the modal pulse as follows: element software SAP 2000.
–3 - Numerical Model
I * = 0.5 ⋅ P * ⋅ t 0 = 0.5 ⋅ 180672 ⋅ 0.3 ×10 = 27.1kNs (6.65)
The initial velocity of the free vibration is:
I* 27.1
v 0 = ------*- = ---------- = 2.63m/s (6.66)
m 10.3
The maximum elastic deflection is:
Δ m, e = v 0 ⁄ ω = 2.63 ⁄ 62.83 = 0.042m (6.67)
The distributed load q is replaced by n = 19 concentrated forc-

es F i :
192000
F i = ------------------ = 10105kN (6.68)
19
The first period of the system is:
T 1 = 0.100s (6.69)
which corresponds to Equation (6.64).

And the time-history of the elastic deflection is:

Blank Page
0.08
0.06
c deformation [m]
0.04
0.02
-0.02
Elastic
-0.04
-0.06
-0.08
0.00 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
m* k* P* T Δ m, e
System
[t] [kN/m] [P] [s] [m]
4.52 43571 0.888 0.064 0.058
10.30 40666 0.941 0.100 0.042
- - - 0.100 0.064

7 Seismic Excitation From the previous figure it can be clearly seen that the time-his-
tory of an earthquake ground acceleration can not be described
by a simple mathematical formula. Time-histories are therefore
7.1 Introduction
usually expressed as sequence of discrete sample values and
The equation of motion for a base point excitation through an ac- hence Equation (7.1) must therefore be solved numerically.
celeration time-history u··g ( t ) can be derived from the equilibrium 2
of forces (see Section 2.1.1) as:
1
xg [m/s2]
mu·· + cu· + f s (u,t) = – mu··g (7.1) 0
Δüg
where u·· , u· and u are motion quantities relative to the base point
ü
-1
of the SDoF system, while f s (u,t) is the spring force of the system -2
that can be linear or nonlinear in function of time and space. The 9.5 9.6
2
time-history of the motion quantities u·· , u· and u for a given SDoF Zeit [s]
system are calculated by solving Equation (7.1). 1
x [m/s2]
Δü
4 0
Ground acceleration
ü
2 -1
üxg [m/s2]
0 -2
t t+Δt
9.5 9.6
-2
Zeit [s]
Time
-4
0 10 20 30 40 The sample values of the ground acceleration u··g ( t ) are known
Zeit [s]
Time [s] from beginning to end of the earthquake at each increment of
4
Response of a T=0.5s SDoF system time Δt (“time step”). The solution strategy assumes that the mo-
2 tion quantities of the SDoF system at the time t are known, and
x [m/s ]
2
0 that those at the time t + Δt can be computed. Calculations start

at the time t = 0 (at which the SDoF system is subjected to
ü
-2
known initial conditions) and are carried out time step after time
-4
step until the entire time-history of the motion quantities is com-
0 10 20 30 40
Zeit [s]
Time [s] puted, like e.g. the acceleration shown in the figure on page 7-1.
7 Seismic Excitation Page 7-1 7 Seismic Excitation Page 7-2

7.2 Time-history analysis of linear SDoF systems 7.2.1 Newmark’s method (see [New59])
• Incremental formulation of the equation of motion
mΔu·· + cΔu· + kΔu = – mΔu··g (7.4)
t + Δt t t + Δt · t t + Δt·· t
u = u + Δu , u = u· + Δu· , u = u·· + Δu·· (7.5)
• Assumption of the acceleration variation over the time step:
In the case of a linear SDoF system Equation (7.1) becomes:

mu·· + cu· + ku = – mu··g (7.2)
and by introducing the definitions of natural circular frequency 1 t t + Δt·· t Δu··

u··( τ ) = --- ( u·· + u ) = u·· + ------- (7.6)
ω n = k ⁄ m and of damping ratio ζ = c ⁄ ( 2mω n ) , Equation (7.1) 2 2
can be rearranged as:
τ
Δu··
u· ( τ ) = u + ³ u··( τ ) dτ = u· + § u·· + -------· ( τ – t )
2 t· t t
u·· + 2ζω n u + ωn u = – u··g (7.3) © 2¹
(7.7)
t
The response to an arbitrarily time-varying force can be comput-
τ τ
ed using:
Δu··
u ( τ ) = u + ³ u· ( τ ) dτ = u + ³ u· + § u·· + -------· ( τ – t ) dτ
t tt t
(7.8)
• Convolution integral ([Cho11] Chapter 4.2) © 2¹
• Numerical integration of the differential equation of motion t t
([Cho11] Chapter 5)

2
Δu·· ( τ – t )
u ( τ ) = u + u· ( τ – t ) + § u·· + -------· -----------------
t t t
(7.9)
© 2¹ 2 Linear Acceleration:
Δt-
---- ≤ 0.551
1 1 T
The increments of acceleration, velocity and displacement dur- β = --- , γ = ---
ing the time step are: 6 2
t + Δt·· t It is important to note that the “average acceleration”-method is

Δu·· = u – u·· = Δu·· (7.10)
unconditionally stable, while the “linear acceleration”-method is
t + Δt · Δu·· only stable if the condition Δt ⁄ T ≤ 0.551 is fulfilled.
u – u· = § u·· + -------· Δt
t t
Δu· = (7.11)
© 2¹
However, the “linear acceleration”-method is typically more ac-
2 curate and should be preferred if there are no stability concerns.
Δu·· Δt
Δu = u· Δt + § u·· + -------· --------
t t
(7.12) For a discussion on stability and accuracy of the Newmark’s
© 2¹ 2 methods see e.g. [Cho11] and [Bat96].
Introducing the parameters γ and β into Equations (7.11) and • Solution of the differential equation: Option 1
(7.12) for Δu· and Δu , respectively, can be generalized as follows:
Substituting Equations (7.13) and (7.14) into Equation (7.4)
t·· gives Equation (7.15), which can be solved for the only remain-
Δu· = ( u + γΔu··)Δt (7.13)
ing variable Δu·· :
2
t· t·· Δt
Δu = u Δt + ( u + 2βΔu··) -------- (7.14) t Δt
2
( m + cγΔt + kβΔt )Δu·· = – mΔu··g – c u··Δt – k § u· Δt + u··--------·
2 t t
2 ©
(7.15)
2 ¹
where different values of the parameters γ and β correspond to
or in compact form:
different assumptions regarding the variation of the acceleration
within the time step: m̃Δu·· = Δp
˜ (7.16)
Substituting Δu·· into Equations (7.11) and (7.12) gives the incre-
Average Acceleration: ments of the velocity Δu· and of the displacement Δu . In conjunc-
Δt
----- ≤ ∞ tion with Equation (7.5), these increments yield the dynamic re-
1 1 T
β = --- , γ = --- sponse of the SDoF system at the end of the time step t + Δt .
4 2

• Solution of the differential equation: Option 2 7.2.2 Implementation of Newmark’s integration scheme
within the Excel-Table “SDOF_TH.xls”
Equation (7.14) can be transformed to:
t· t·· Equation (7.15) is implemented in the Excel-Table as follows:
Δu u u
Δu·· = -----------2 – --------- – ------ (7.17)
βΔt βΔt 2β
2
( m + cγΔt + kβΔt )
2 Δu·· – mΔu··g t··
u Δt § tu· Δt + tu··Δt
--------·
–k © 2 ¹

®
¯
= –c

®
¯

°
°
°
®
°
°
°
¯

®
¯

°
°
®
°
°
¯
and substituting Equation (7.17) into (7.13) we obtain: meq da ΔF ( t ) dv
dd
t·
γΔu γ u γ t
Δu· = --------- – ------- + Δt § 1 – ------· u·· (7.18) • In the columns C to E the so-called “predictors” dd, dv and da
βΔt β © 2β¹ are computed first:
Substituting Equations (7.17) and (7.18) into Equation (7.4) 2
t· t··Δt
gives Equation (7.19), which can be solved for the only remain- dd = u Δt + u -------- (“delta-displacement”)
2
ing variable Δu :
t
γc γc t γ dv = u··Δt (“delta-velocity”)
§ k + -----------
m
+ ---------· Δu = – mΔu··g + § --------- + -----· u· + § ------ – Δt § 1 – ------· c· u··
m m t
© 2 βΔt¹ © βΔt β ¹ © 2β © 2β ¹ ¹
βΔt – mΔu··g – c ⋅ dv – k ⋅ dd
(7.19) da = ------------------------------------------------------ = Δu·· (“delta-acceleration”)
meq
or in compact form:
• Afterwards, in the columns F to H the ground motion quantities
kΔu = Δp (7.20) at the time step t + Δt are computed by means of so-called
Substituting Δu into Equations (7.18) and (7.17) gives the incre- “correctors:
ments of the velocity Δu· and of the acceleration Δx··. In conjunc- t + Δt·· t
u = u·· + da
tion with Equation (7.5), these increments yield the dynamic re-
sponse of the SDoF system at the end of the time step t + Δt . t + Δt · t
u = u· + dv + ( da ⋅ γ ⋅ Δt )

°
°
®
°
°
¯
For linear systems we have: Δu
• m , c and k are constant throughout the whole time-history.
t + Δt t 2
• m̃ in Equation (7.15), as well as k in Equation (7.20), are also u = u + dd + ( da ⋅ β ⋅ Δt )

°
°
°
®
°
°
°
¯
constant and have to be computed only once. Δu

• Finally, in column I the absolute acceleration u··abs at the time 7.2.3 Alternative formulation of Newmark’s Method.
step t + Δt is computed as follows:
The formulation of the Newmark’s Method presented in Section
t + Δt·· t + Δt·· t + Δt·· 7.2.1 corresponds to an incremental formulation. It is possible to
uabs = u+ ug
rearrange the methodology to obtain a total formulation.
Observations about the use of the Excel-Table The equation of motion at the time t + Δt can be written as:
t + Δt·· t + Δt · t + Δt t + Δt··
• Only the yellow cells should be modified: m u+c u+k u = –m ug (7.21)
• The columns A and B contain the time vector and the ground ac-
celeration u··g ( t ) at intervals Δt ; for this ground motion the re- where
sponse of a Single-Degree-of-Freedom (SDoF) system will be t + Δt·· t
computed. To compute the response of the SDoF system for a dif- u = u·· + Δu·· (7.22)
ferent ground motion u··g ( t ) , the time and acceleration vector of the t + Δt · t
new ground motion have to be pasted into columns A and B. u = u· + Δu· (7.23)
• For a given ground motion u·· ( t ) , the response of a linear SDoF
g Using the expressions for Δu·· and Δu· given by Equations (7.17)
system is only dependent on its period T = 2π ⁄ ω n and its damp- and (7.18), the acceleration and the velocity at the time t + Δt
ing ζ . For this reason, the period T and the damping ζ can be cho-
can be written as:
sen freely in the Excel-Table.
t + Δt·· 1 t + Δt
u – u ) – --------- u· – § ------ – 1· u··
• The mass m is only used to define the actual stiffness of the SDoF t 1 t 1 t
2 u = -----------2 ( (7.24)
system k = m ⋅ ω n and to compute from it the correct spring force βΔt © 2β ¹
βΔt
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.
t + Δt · γ t + Δt γ t γ t
u – u ) + § 1 + ---· u· + Δt § 1 – ------· u··
t
u = --------- ( (7.25)
• In the field “Number of periods” (cell V19) one can enter the βΔt © β¹ © 2β¹
number of periods T i for which the response of the SDoF is to be
computed in order to draw the corresponding response spectra. Introducing Equations (7.24) and (7.25) into Equation (7.21) and
t + Δt
• The response spectra are computed if the button “compute re- solving for the only unknown u we obtain:
sponse spectra” is pressed. The macro pastes the different pe- t + Δt t + Δt
riods T i into cell S3, computes the response of the SDoF system, k⋅ u = p (7.26)
reads the maximum response quantities from the cells F6, G6, where:
H6 and I6 and writes these value into the columns L to P.

k = k + a1 (7.27) 7.3 Time-history analysis of nonlinear SDoF systems
t + Δt t + Δt·· t t t f el um
p = –m ug + a 1 u + a 2 u· + a 3 u·· (7.28) f y = ------ , μ Δ = ------
Ry uy
m γc
a 1 = -----------2 + --------- (7.29)
βΔt βΔt
γ
a 2 = --------- + § --- – 1· c
m
(7.30)
βΔt © β ¹
γ
a 3 = § ------ – 1· m + Δt § ------ – 1· c
1
(7.31)
© 2β ¹ © 2β ¹
This formulation corresponds to the implementation of New-

mark’s method presented in [Cho11].
• Strength f y of the nonlinear SDoF system
f el k el ⋅ u el
f y = ------ = ------------------ (7.32)
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 ζ would experience if submitted to
the same ground motion u··g
• Maximum deformation u m of the nonlinear SDoF system
um = μΔ ⋅ uy , hence μΔ = um ⁄ uy (7.33)
• u y = yield displacement
• μ Δ = displacement ductility

7.3.1 Equation of motion of nonlinear SDoF systems 7.3.2 Hysteretic rules

In the equation of motion for a base point excitation through an The next figure shows typical hysteretic rules (or models) for
acceleration time-history u··g ( t ) nonlinear SDoF systems.
mu·· + cu· + f s (u,t) = – mu··g (7.34)
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 + Δu,t + Δt) at the time t + Δt can eas-
ily be computed.
This description of the cyclic force-deformation relationship is
known as hysteretic rule. In the literature many different hyster-
etic rules for nonlinear SDoF systems are available (See e.g.
[Saa91]).
In the following section a few hysteretic rules are presented and
discussed.

In the following the often used Takeda hysteretic model is dis- Unloading occurs along a straight line with stiffness k u . This un-
cussed in some detail. loading stiffness is computed by means of Equation (7.35) as a
function of the elastic stiffness k el and taking into softening ef-
The Takeda model was first described in [TSN70] and later mod-
fects in proportion of the previously reached maximum displace-
ified by various authors. The assumed force-deformation rela-
ment ductility μ Δ . The parameter α controls the unloading stiff-
tionship shown in the following figure was derived from the mo-
ness reduction and varies from structural element to structural
ment-curvature relationship described in [AP88].
element.
Large amplitude cycles Small amplitude cycles
–α –α
ξ+
p k u+ = k el ⋅ ( max { μ Δ+ } ) , k u- = k el ⋅ ( max { μ Δ- } ) (7.35)
+ + +
fs β⋅ξp β⋅ξp fs β⋅ξp
A+ B+ A+ A+ B+
Reloading follows a straight line which is defined by the force re-
k pl =r o ·kel fy k pl =r o ·kel
C+ versal point (u rev,0) and the point A. The location of point A is de-
β>0 β<0 termined according to the figure on the previous page as a func-
+ +
ku
-
ku
tion of the last reversal point B, the plastic strain ξ p and the dam-
k el
u rev
uy u rev
u
u rec u rev
+
u rec
u age influence parameter β . The parameter β allows taking into
account softening effects occurring during the reloading phase.
- -
ku ku
β<0
C- Again, in the case of RC walls meaningful parameters α and β
β>0
X- B- X- lay in the following ranges: α = 0.2…0.6 and β = 0.0… – 0.3 .
x- x-
-
β⋅ξp
-
β⋅ξp
These rules, which are valid for cycles with large amplitude, are
typically based on observations of physical phenomena made
The initial loading follows the bilinear force-deformation relation- during experiments.
ship for monotonic loading mentioned in the previous section.
On the other hand, rules for small amplitude cycles are based on
The exact definition of this so-called skeleton curve depends on
engineering considerations rather than on exact observations.
the structural element at hand. For example, in the case of Re-
They are designed to provide reasonable hysteresis lops during
inforced Concrete (RC) structural walls the elastic stiffness k el
an earthquake time-history, thus avoiding clearly incorrect be-
corresponds to 20 to 30% of the uncracked stiffness, while the
haviours like e.g. negative stiffnesses.
plastic stiffness k pl = r o ⋅ k el is approximated assuming an hard-
ening factor r o = 0.01…0.05 . The rules for small amplitude cycles are shown on the right of the
figure on the previous page.

If a reloading phase starts from a force reversal point (u rev,0) lay- 7.3.3 Newmark’s method for inelastic systems
+
ing between the two extreme force reversal point (u rec,0) and
- The Newmark’s numerical method discussed in Sections 7.2.1
(u rec,0) , then reloading does no longer occurs towards point A,
to 7.2.3 can be easily modified for application to nonlinear sys-
but towards a newly calculated point X, which lies between
tems. The following modifications are required:
points A and B. The position of point X is calculated using the
+ -
auxiliary variables x and x defined in Equation (7.36). • The mass m and the damping c are typically constant through-
+ -
out the whole time-history.
+ § u rev – u rec· + - § u rev – u rec· - • The stiffness k changes during the time-history, hence m̃ , re-
-¸ ⋅ βξ p ,
x = ¨ ------------------------ -¸ ⋅ βξ p
x = ¨ ------------------------ (7.36)
- + + -
© u rec – u rec¹ © u rec – u rec¹ spectively k , are no longer constant.
• If the stiffness changes within the time step iterations are need-
When a load reversal occurs before point X is reached, a new
ed (e.g. Newton-Raphson).
point C is defined as a temporary maximum and minimum. The
reloading in the subsequent cycles, which are smaller than the • For nonlinear systems the second solution strategy presented
temporary maximum and minimum is then always in the direc- in Section 7.2.1 (Option 2) has the advantage that the factors
tion of point C.
m γc γ
a = § --------- + -----· and b = ------ – Δt § 1 – ------· c
m
These rules for cycles with a small amplitude are a simplification © βΔt β ¹ 2β © 2β¹
of those described in [AP88], however they lead to very satisfac-
tory results and can be programmed very easily and efficiently. 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 New-
mark’s time stepping method, it is suggested to use the formula-
tion presented in Section 7.4.4 in conjunction with the Newton-
Raphson algorithm described in Section 7.4.3.

7.3.4 Example 1: One-storey, one-bay frame • Frames with T = 3.75s , ζ = 5% subjected to “El Centro”
Δ=28.1cm 40
Displacement [cm]
20
Shear force [kN]

20
0 0
−20
−20 Vel=39.5kN
−40
0 10 20 30 −20 0 20
40
Displacement [cm]
20
Shear force [kN]

20
0 0
−20
12EI s My Vy −20
Δ=25.9cm Vy=19.8kN
- , T = 2π m
k = 2 ⋅ ------------- ---
- , V = 2 ⋅ ----------
- , Δ = ------ (7.37) −40
H
3 k y H⁄2 y k 0 10 20 30 −20 0 20
40
Displacement [cm]
• Parameters 20
Shear force [kN]

20
Columns k [kN/m] T [s] fy [MPa] Vy [kN] Vy/Vel [-] Δy [cm]
0 0
HEA 100 141 3.75 595 39.5 1.00 28.1
HEA 100 141 3.75 298 19.8 0.50 14.0 −20
−20
HEA 100 141 3.75 149 9.9 0.25 7.0 Δ=21.6cm −40
Vy=9.9kN
HEA 100 141 3.75 99 6.6 0.167 4.7 0 10 20 30 −20 0 20

HEA 220 2181 0.95 543 246.7 1.00 11.3 40
Displacement [cm]
20
HEA 220 2181 0.95 272 123.4 0.50 5.7
Shear force [kN]

20
HEA 220 2181 0.95 136 61.7 0.25 2.8
HEA 220 2181 0.95 91 41.1 0.167 1.9 0 0
IPE 550 27055 0.27 185 411.4 1.00 1.52 −20
IPE 550 27055 0.27 93 205.7 0.50 0.76 −20
Δ=21.2cm −40
Vy=6.6kN
IPE 550 27055 0.27 43 102.9 0.25 0.38
0 10 20 30 −20 0 20
IPE 550 27055 0.27 31 68.6 0.167 0.25 Time [s] Displacement [cm]

• Frames with T = 0.95s , ζ = 5% subjected to “El Centro” • Frames with T = 0.27s , ζ = 5% subjected to “El Centro”
400
200 2
Displacement [cm]
Displacement [cm]
10
Shear force [kN]
Shear force [kN]

1 200
0 0 0 0
−1 −200
−10 Δ=11.3cm −200 Vel=246.7kN −2
Δ=1.5cm Vel=411.4kN
−400
0 10 20 30 −10 0 10 0 10 20 30 −2 −1 0 1 2
400
200 2
Displacement [cm]
Displacement [cm]
10
Δ=7.7cm Shear force [kN]
Shear force [kN]

1 200
0 0 0 0
−1 −200
−10
Δ=1.2cm
−200 Vy=123.4kN −2 Vy=205.7kN
−400
0 10 20 30 −10 0 10 0 10 20 30 −2 −1 0 1 2
400
200 2
Displacement [cm]
Displacement [cm]
10
Δ=1.5cm
Shear force [kN]
Shear force [kN]

1 200
0 0 0 0
−1 −200
−10 Δ=9.4cm −200 Vy=61.7kN Vy=102.9kN
−2
−400
0 10 20 30 −10 0 10 0 10 20 30 −2 −1 0 1 2
400
200 2
Displacement [cm]
Displacement [cm]
10
Shear force [kN]
Shear force [kN]

1 200
0 0 0 0
Δ=2.2cm
−1 −200
−10
Δ=10.4cm −200 Vy=41.1kN −2 Vy=68.6kN
−400
0 10 20 30 −10 0 10 0 10 20 30 −2 −1 0 1 2
Time [s] Displacement [cm] Time [s] Displacement [cm]

7.3.5 Example 2: A 3-storey RC wall Time-history of the top displacement

As a second example, the behaviour of the RC wall WDH4 pre- The time-history of the top displacement (see below) shows that:
sented in Section 7.6.3 is simulated. Wall WDH4 is actually a 3- (i) plastic phenomena affect the behaviour of the wall WDH4 sig-
DoF system and its behaviour is simulated by means of an equiv- nificantly, and that (ii) the Takeda-SDoF System is able can de-
alent SDoF system. For this reason the relative displacement of scribe the global behaviour of the wall WDH4 quite accurately.
the SDoF system shall be multiplied by the participation factor
80
Γ̃ = 1.291 to obtain an estimation of top displacement of WDH4. WDH4: Experiment
Relative displacement [mm]

60
To simulate the behaviour of WDH4 a nonlinear SDoF system 40 Linear SDoF System
with Takeda hysteretic model is used. The parameter used to fn =1.2Hz, ζ=5%
20
characterise the SDoF system are:
0
ζ = 5% , f n = 1.2Hz , R y = 2.5 , r o = 0.03 , -20
α = 0.2 , β = – 0.15 (7.38) -40
where ζ is the damping rate, f n is the natural frequency of the -60
SDoF system for elastic deformations (i.e. with k = k el ), and R y -80

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
is the force reduction factor. The parameters given in (7.38) were Time [s]
subsequently adjusted to obtain the best possible match be- 80
tween the simulation and the experiment. In a first phase, a cal- WDH4: Experiment
Relative displacement [mm]

60
culation using a linear SDoF system is performed. The latter has Takeda SDoF System
40
the same damping rate and natural frequency as the Takeda fn =1.2Hz, ζ=5%, R y =2.5,
20 ro =0.03, α =0.2, β=-0.15

SDoF system and allows an estimation of the maximum elastic
spring force f el and of the maximum elastic deformation u el . The 0
yield force f y and the yield displacement u y of the Takeda SDoF -20
system are then estimated using the force reduction factor R y as -40
follows: -60
f el u el -80
0 3 8 11 14 17
R y = ----- = ------ (7.39) 1 2 4 5 6 7 9 10 12 13 15 16 18
fy uy Time [s]

Hysteretic behaviour of the nonlinear SDoF System 7.4 Solution algorithms for nonlinear analysis
Force-deformation relationship of the Takeda-SDoF System
problems
subjected to the same ground motion as Wall WDH4. In both di- In this section the more general case of system with multiple de-
agrams the same curve is plotted: On the left in absolute units grees of freedom is discussed. SDoF system can be seen as a
and on the right in normalised units. special case thereof.
80
Normalised spring force fs/fy [kN]

60 1.0 7.4.1 General equilibrium condition
Spring force fs [kN]
40
0.5 The general equilibrium condition for elastic and inelastic static
20
and dynamic systems is:
0 0.0
-20 F(t) = R(t) (7.40)
-0.5
-40 In this equation F ( t ) is the time-dependent vector of the internal
-60 -1.0 forces of all DoFs of the structure and R ( t ) the time-dependent
-80 vector of the external forces.
-80 -60 -40 -20 0 20 40 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
Relative displacement x [mm] Displacement dutility mD=u/uy [mm] 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 there-
fore be rewritten as
KU = R (7.42)

In Equation (7.42) R is known and K is also known, therefore The actual solution of the equilibrium conditions of Equation
the unknown vector U can be computed by means of Equation (7.40) is often obtained by the numerical method for the iterative
(7.43): solution of nonlinear equations that was originally developed by
Newton.
–1
U = K R (7.43)
In the next section the so-called Newton-Raphson Algorithm for
The equilibrium condition of Equation (7.40) can only be solved the solution of Equation (7.40) will be discussed.
for linear-elastic systems by means of Equation (7.43).
For inelastic systems, due to successive yielding of the struc- 7.4.3 The Newton-Raphson Algorithm
ture, the stiffness matrix K is not constant over the course of the The Newton-Raphson Algorithm allows the solution of the load-
loading. ing of nonlinear springs with the following equilibrium condition:
For this reason Equation (7.40) must be solved in increments F(U(t) ) = R(t) (7.47)
(=small load steps) and iteratively. The approach is as follows:
t F ( U ( t ) ) = R ( t ) represents the internal spring force, which is a
• The nodal displacements U at the time t are known from the given nonlinear function of U ( t ) . The external force R ( t ) is a
previous load step; function of the time t .
t + Δt
• The nodal displacements U at the end of the load step Δt
For a system with 1 DOF the solution method of the Newton-
are determined by means of n iterations of Equations (7.44)
Raphson Algorithm can be illustrated by the figure on page 7-29.
and (7.45).
The algorithm consists of the following steps:
t + Δt i–1 i t + Δt i–1
KT ΔU = ΔR (7.44)
0) Up to time step t a solution was obtained and at time step t
t t
t + Δt i t + Δt i – 1 i the system is in equilibrium with ( U, R) ;
U = U + ΔU (7.45)
1) The initial conditions at the beginning of the iteration are de-
where: termined. The iteration commences with the nodal displace-
t + Δt i–1 t + Δt t + Δt i – 1 ment, the tangent stiffness and the internal force that have
ΔR = R– F (7.46) resulted at the end of the previous time step t . The external
t + Δt 0
and K T is the tangent stiffness matrix of the structure. initial loading increment ΔR within the time step is de-
termined by Equation (7.51).

R
t+ΔtΔU1 t+ΔtΔU2 t+ΔtΔU3
t+ΔtR
t+ΔtΔR2
t+ΔtK 2
t+ΔtΔR1 T
t+ΔtK 1
t+ΔtΔR0 T
t+ΔtK 0
T
tR
tR=t+ΔtF0 t+ΔtF1 t+ΔtF2 t+ΔtFn ~ t+ΔtF = t+ΔtR
tU=t+ΔtU0 t+ΔtU1 t+ΔtU2 t+ΔtU=t+ΔtUn

U
Newton-Raphson algorithm to solve systems of nonlinear equations (according to [AEM86])
Page 7-29
An-Najah 2013
R
t+ΔtΔU1 t+ΔtΔU2 t+ΔtΔU3
t+ΔtR
t+ΔtΔR2
t+ΔtΔR1
t+ΔtK 0
T
t+ΔtK 0
t+ΔtΔR0 T
t+ΔtK 0
T
tR
tR=t+ΔtF0 t+ΔtF1 t+ΔtF2 t+ΔtFn ~ t+ΔtF = t+ΔtR
tU=t+ΔtU0 t+ΔtU1 t+ΔtU2 t+ΔtU=t+ΔtUn

U
Modified Newton-Raphson algorithm to solve systems of nonlinear equations (according to [AEM86])

Page 7-30
An-Najah 2013
t + Δt 0 t Comments on the Newton-Raphson Algorithm

Displacement: U = U (7.48)
t + Δt 0 t • The Newton-Raphson Algorithm for systems with several or
Tangent stiffness: KT = KT (7.49)
many DoFs follows exactly the same procedure as the algo-
t + Δt 0 t rithm for SDoF systems. Only difference: Scalar values are re-
Internal force: F = F (7.50)
placed by the corresponding vectorial quantities.
t + Δt 0 t + Δt t + Δt 0
External force: ΔR = R– F (7.51) • Apart from the Newton-Raphson Approach (“Full Newton-
2) Computation of the ith
-displacement increment ΔU by i Raphson iteration”) the “Modified Newton-Raphson iteration”
means of Equation (7.52) (i starts from 1). is often applied. This algorithm is illustrated on page 7-30.
t + Δt i – 1
For systems with more DoFs, KT is a matrix (tangent • Unlike in the Newton-Raphson Algorithm, in the Modified Newton-
stiffness matrix) and Equation (7.52) is best solved by means Raphson Algorithm the tangent stiffness matrix K T is updated only
t + Δt i – 1
of a LDL T -decomposition of the matrix KT . at the beginning of the time step and is kept constant over all the
iterations within this time step.
t + Δt i – 1 i t + Δt i–1 t + Δt
KT ΔU = ΔR (7.52) • To reach the target displacement U more iterations are re-
t + Δt i quired for the Modified Newton-Raphson Algorithm than for the
3) Computation of the displacement U at the end of the ith-
Full Newton-Raphson Algorithm. However, these can be comput-
iteration ed more quickly since assembling the tangent stiffness matrix K T
T
t + Δt i t + Δt i – 1 i (Step 6) and in particular its LDL -decomposition (Step 2) are
U = U + ΔU (7.53)
only required at the beginning of a time step and not at each itera-
t + Δt i
4) Computation of the internal force F and the new external tion within the time step. This is particularly advantageous for sys-
t + Δt i tems with many DoFs.
force (residual force) R
t + Δt i t + Δt t + Δt i
• In most FE-analysis programs both Newton-Raphson Algo-
ΔR = R– F (7.54) rithms as well as other algorithms are typically combined in a
t + Δt i general solver in order to obtain a successful convergence of
5) If ΔU i and/or ΔR are so small that they can be neglected:
Continue with Step 7; the iteration process for many structural analysis problems.
t + Δt i • Other algorithms for the solution of the equilibrium conditions
6) Determine the new tangent stiffness KT and repeat from
can be found in Chapters 8 (static analysis) and 9 (dynamic
Step 2;
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.

Convergence criteria • For this reason the displacement criterion is typically used in con-
junction with other convergence criteria.
• A in-depth discussion of the convergence criteria can be found • A force criterion, which checks the residual forces, is given in
in [Bat96] and [AEM86]. This section provides only a short Equation (7.56).
overview.
t + Δt t + Δt i
• In Step 5) criteria are required in order to decide whether con- R– F-
---------------------------------------
t + Δt
≤ εF (7.56)
vergence of the iteration was obtained. Possible convergence R– F
t
criteria can be based on displacements, force or energy con-
siderations. ε F is the force convergence tolerance, which checks the mag-
nitude of the residual force after the ith-iteration against the first
• Since within the time step the unknown target displacement
t + Δt load increment of the time step.
U needs to be determined, it makes sense to prescribe that
the target displacement is reached within a certain tolerance • As for the displacement criterion, this force criterion should not be
interval. For this reason a possible displacement criterion for applied on its own because in some cases the target displacement
t + Δt
U may not have been reached. This may happen for systems
the convergence is:
with small post-yield stiffness.
i
ΔU - • The energy criterion in Equation (7.57) has the advantage
------------------
t + Δt
≤ εD (7.55)
U that it checks the convergence of the displacements and the
forces simultaneously.
where ε D is the displacement convergence tolerance.
t + Δt i T t + Δt t + Δt i – 1
• The vector U is actually unknown and must therefore be ap- ( ΔU ) ( R– F )
t + Δt i ----------------------------------------------------------------- ≤ ε E (7.57)
proximated. Typically U is used in conjunction with a suffi- 1 T t + Δt t
( ΔU ) ( R – F)
ciently small value of ε D .
• It is important to note that in some cases – although the criteria de- ε E is the energy convergence tolerance, which checks the
scribed by Equation (7.55) is satisfied – the wanted target dis- work of the residual forces of the ith-iteration against the work
t + Δt
placement U has not been reached. of the residual forces of the first load increment of the time
• This is the case when the computed displacements vary only step.
slightly during one iteration but these small increments are repeat- Choosing the tolerances ε D , ε F or ε E too large, can yield wrong
ed over many iterations. results, which can lead to the divergence of the solution in the
• Such a situation can result when the modified Newton-Raphson following load steps.
Algorithm is used (see page 7-30).

Choosing the tolerances ε D , ε F or ε E too small, results in an un- The numerical time integration of Equation (7.58) is often per-
necessary increase of the required iteration steps. The in- formed by means of the Newmark’s Algorithms ([New 59]). Ac-
creased accuracy is typically not useful. cording to these algorithms the displacement vector at the time
For numerical reasons it can also happen that too small con- t + Δt is estimated as follows:
vergence tolerances do not allow to reach convergence at all. t + Δt t t ·· t + Δt ·· Δt
2
U = U + U· Δt + [ ( 1 – 2β ) U
t
+ 2β U ] ------- (7.61)
For strongly inelastic systems it is recommended to check 2
the sensitivity of the results to the chosen convergence t + Δt t t ·· t + Δt ··
criteria and the chosen tolerances. U· = U· + [ ( 1 – γ ) U +γ U ] Δt (7.62)
From Equation (7.61):
7.4.4 Nonlinear dynamic analyses
t + Δt ·· 1 – 2 β t ··
U – U ) – -------- U· – § -------------· U
1 t + Δt t 1 t
U = ----------2- ( (7.63)
Similar to Equation (7.44) the equilibrium condition for nonlinear β Δ t © 2β ¹
β Δt
dynamic analyses is:
Substituting Equation (7.63) into (7.62):
t + Δt ··i t + Δt · i t + Δt i – 1 t + Δti–1 i t + Δt
M U +C U +( F + KT ΔU ) = R (7.58) t + Δt γ t + Δt γ t γ t ··
U· = -------- ( U – U ) + § 1 – ---· U· + § 1 – ------· U
t
Δt (7.64)
β Δt © β¹ © 2β¹
t + Δt i t + Δt i – 1 i
U = U + ΔU (7.59)
The expressions for the displacement, the velocity and the acceler-
For base excitation by means of ground accelerations the vector ation at the time t + Δt from Equations (7.59), (7.64) and (7.63) can
of the external forces is computed according to Equation (7.60). be substituted into the differential equation of motion (7.58) which
t + Δt t + Δt can then be solved for the only remaining unknown ΔU i :
R = –M 1 ag (7.60)
§ t + ΔtKi – 1 + ---------- γ
- M + --------- C· ΔU i =
1
where M is the mass matrix of the structure, 1 the norm vector © T 2 βΔt ¹
βΔt
with entries of unity for all DoFs in the direction of the excitation t + Δt t + Δt i – 1
t + Δt R– F –
and ag the ground acceleration at the time t + Δt .
– U ) – --------- U· – § ------ – 1· U
1 t + Δt i – 1 t 1 t 1 t ··
For this type of excitation the differential equation of motion M ----------2- ( U –
βΔt βΔt © 2β ¹
(7.58) has to be integrated numerically and – due to the inelas- γ t + Δt i – 1 t γ t γ t ··
C --------- ( U – U ) + § 1 – ---· U· + § 1 – ------· U Δt
tic behaviour of the system – the equation must be solved itera- βΔt © β¹ © 2β¹
tively and incrementally. (7.65)

or in a more compact format: 7.4.5 Comments on the solution algorithms for nonlinear
analysis problems
t + Δt i–1 i t + Δt i–1
K T ΔU = ΔR (7.66)
˜ ˜ • Contrary to the analysis of elastic systems, the analysis of in-
Equation (7.66) corresponds exactly to Equation (7.44) and is elastic systems is often interrupted before the targeted load or
also solved iteratively by the Newton-Raphson Algorithm. deformation state is reached.
When dynamic analyses are carried out, typical convergence • This situation arises if in one of the time steps convergence cannot
criteria also consider the inertia forces and, if present, damping be reached.
forces. Possible, often used convergence criteria are: • Typical causes for the failing convergence
t + Δt t + Δt i – 1 t + Δt ··i – 1 t + Δt · i – 1 • The convergence tolerances are too small or too large;
R– F –M U –C U
------------------------------------------------------------------------------------------------------------------- ≤ ε F (7.67) • The chosen values for the parameters of the solution algorithm are
RNORM
not appropriate;
i T t + Δt t + Δt i – 1 t + Δt i – 1 ·· t + Δt i – 1
( ΔU ) ( R– F –M U –C U· ) • The solution algorithm is not suitable.
------------------------------------------------------------------------------------------------------------------------------------- ≤ ε E (7.68)
1 T t + Δt t t ·· t· Typical example: If special measures are not taken, the algorithms
( ΔU ) ( R – F – M U – C U)
of the Newton family are not able to solve a system with a global
with negative post-yield stiffness;
• Apart from the algorithms by Newmark, many other algorithms
RNORM = ¦ m ij ⋅ g (7.69)
have been developed for solving the differential equation of
As alternative, depending at which point during the iteration proc- motion (7.58) (e.g. “Houbolt Method”, “Wilson θ Method”, “α-
ess convergence is checked, both criteria can be rewritten as: Method”). These methods are described in detail in [Bat96].
t + Δt t + Δt i t + Δt ··i t + Δt · i
The “α-Method” allows to introduce numerical damping, which
R– F –M U –C U- can be useful. The “α-Method” is described in detail in
---------------------------------------------------------------------------------------------- ≤ εF (7.70)
RNORM [HHT77].
i T t + Δt t + Δt i t + Δt i t + Δt i
( ΔU ) ( R– F –M U·· – C U· ) • Choice of the time step Δt for static analyses:
----------------------------------------------------------------------------------------------------------------- ≤ ε E (7.71)
1 T t + Δt t ·· t
– C U· ) • For static analyses the time t has no physical meaning. For this rea-
t
( ΔU ) ( R– F–MU
son the size of the time step Δt can be chosen almost arbitrarily.
As for static analyses, different convergence criteria exist also • As long as the algorithm converges and the variation of the exter-
for dynamic analyses and a discussion of these can also be nal loads is captured correctly, the size of the time step Δt has only
found in [Bat96]. a minor influence on the results.

• However, when geometric nonlinearities are considered or when • An in-depth discussion on the accuracy of time-integration meth-
the constitutive laws are a function of the strain history, care should ods and on so-called “conditionally stable” integration methods
be taken also for static analyses when choosing the size of the can be found in [Bat96].
time step. • Example:
• The time step size influences the convergence of the algorithm: The time-integration method by Newmark with γ = 1 ⁄ 2 and
For small Δt the algorithm converges more quickly; however, more β = 1 ⁄ 6 (linear variation of the acceleration over the length of the
steps are required. time step, see Section 7.2.1) is only stable if the criterion given in
• Time steps of a variable size can be advantageous. If the system Equation (7.73) is met for all natural periods T n of the system. For
is elastic or almost elastic, large times steps can be chosen. If the systems with many DoFs, higher modes can be especially prob-
system is close to its capacity, small time steps should be chosen. lematic and a very short time step is generally required when this
time-integration method is used.
• Certain analysis programs (see for example [HKS03]) determine
the time step size within chosen limits as a function of the conver- Δt-
----- ≤ 0.551 (7.73)
gence and the number of required iterations. Tn
• Choice of the time step Δt for dynamic analyses
• For this reason the “unconditionally stable” time-integration algo-
• The objective of the dynamic analyses is the solution of the differ- rithm by Newmark with γ = 1 ⁄ 2 and β = 1 ⁄ 4 is often used in
ential equation of motion (7.58) between the time t and the time seismic engineering.
t + Δt . For this reason the choice of the time step Δt plays always
an important role regarding the accuracy of the solution.
• The accuracy of the integration of the differential equation of mo-
tion (7.58) depends on the chosen time-integration algorithm and
on the ratio given by Equation (7.72), where T n are the natural pe-
riods of the system. To capture the motion components due to
higher modes the time step Δt has therefore to be reduced.
Δt
------ (7.72)
Tn
• Certain time-integration algorithms can become instable if the time

step size Δt is too large.

7.4.6 Simplified iteration procedure for SDoF systems with 7.5 Elastic response spectra
idealised rule-based force-deformation relationships
7.5.1 Computation of response spectra
In the case that the hysteretic behaviour is described by a se-
quence of straight lines, a so called “idealized rule-based force-
deformation relationships” like the Takeda model presented in
Section 7.3.2, it is possible to avoid implementing a Newton-
Raphson 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 fig- • Response spectra are used to represent the seismic demand
ure: on structures due to a ground motion record and design spec-
tra 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/

• “El Centro”: Linear response spectra • Limits of response spectra
ζ = 0%
Absolute acceleration [m/s2]
40
ζ = 2%
30 ζ = 5%
ζ = 10%
20 ζ = 20%
10
0
0.01 0.10 1.00 10.00 100.0 u··a + 2ζωu· + ω 2 u = 0 u··a + 2ζωu· + ω 2 u = 0
3
u = 0 u··a ( t ) = u··( t ) + u··g ( t ) = u··g ( t ) u··( t ) = – u··g ( t ) u ( t ) = – u g ( t )
ζ = 0%
ζ
Relative velocity [m/s]
= 2%
2 ζ = 5% 2.0
ζ
ag [m/s2]
= 10%
ζ = 20% 0.0
1 −2.0
ag,max = 3.13 m/s2
0 10 20 30
0.4
0
vg,max = 36.1cm/s
0.01 0.10 1.00 10.00 100.0 0.2
vg [m/s]
0.6 0.0
ζ = 0%
Relative displacement [m]
0.5 ζ = 2% −0.2
0.4 ζ = 5% 0 10 20 30
ζ = 10% dg,max = 21.1cm
0.3 0.2 Baseline
ζ = 20%
correction!
dg [m]
0.2 0.1
0.1 0.0
−0.1
0.01 0.10 1.00 10.00 100.0 0 10 20 30
Period [s] Time [s]

7.5.2 Pseudo response quantities • True vs. Pseudo response quantities

• Pseudo-velocity S pv 10
Acceleration
Absolute acceleration [m/s2]

Pseudo-acceleration
S pv = ωS d (7.74)
ζ = 5%
- S pv has units of a velocity 5
ζ = 50%
- S pv is related to the peak value of the strain energy E s
2 2 2
kS d k ( S pv ⁄ ω ) mS pv 0
E s = --------- = --------------------------- = ------------- (7.75) 0.01 0.10 1.00 10.00 100.00
2 2 2
1.0
Velocity
Pseudo-velocity
• Pseudo-acceleration
Relative velocity [m/s]

ζ = 5%
S pa = ω 2 S d (7.76)
0.5 ζ = 50%
- S pa has units of an acceleration
- S pa is related to the peak value of the base shear V

0.0
0.01 0.10 1.00 10.00 100.00
V = kS d = k ( S pa ⁄ ω 2 ) = mS pa (7.77) Periode [s]
• For ζ = 0 are acceleration and pseudo-acceleration identical.

• For T ∞ the pseudo-velocity tends to zero.
• Pseudo-velocity and pseudo-acceleration match well the true mo-
tion of a SDOF system with ζ < 20% and T < 1s

• Remarks on the Pseudo-acceleration • Combined D-V-A spectra

S pv = ωS d log ( S pv ) = log ( ω ) + log ( S d )
1.0
Pseudo-acceleration ζ=0 log ( S pv ) = log ( f ) + log ( 2π ) + log ( S d )
Pseudo-velocity log ( S pv ) = – log ( T ) + log ( 2π ) + log ( S d )
... / ... max [−]
S pa
S pv = -------- log ( S pv ) = – log ( ω ) + log ( S pa )
0.0 ω
log ( S pv ) = – log ( f ) – log ( 2π ) + log ( S pa )
log ( S pv ) = log ( T ) – log ( 2π ) + log ( S pa )
-1 x Displacement
−1.0
0 5 10 15
1.0
Pseudo-acceleration ζ>0 100
Spv = 71 cm/s
Pseudo-velocity
10
00
0
... / ... max [−]
/s 2
Pseudo−velocity Spv [cm/s]
cm
0.0
10
0
Sd
44
10
00
=
=
11
pa
10
.
S
2
cm
-1 x Displacement
10
10
Ps
−1.0
0
eu
do
0 5 10 15
−a
]
cc
m
Time [s]
[c
.S
d
tS
pa
10
1
[c
en
u··a ( t ) = – ω 2 u ( t ) – 2ζωu· ( t )
m
(7.78)
em
/s
2
]
ac
1
pl
Time-history of the pseudo-acceleration A(t)
is
D
1
1
0.
• For ζ = 0 : u(t) = A(t)
• For ζ > 0 : At u max : u a = A however A < A max T=1s
Shift of the location of the maxima through damping 0.1 1.0 10.0
Period [s]

7.5.3 Properties of linear response spectra 7.5.4 Newmark’s elastic design spectra ([Cho11])
ζ = 0%
ζ = 2%
Spa / ag [-]
ζ = 5%
100 ζ = 10%
10
00
0
vg = 36.1 cm/s
T [s]
10
0
10
dg
00
=
21
10
.1
cm
10
10
Ps
0
eu
/s 2
do
[-]
−a
cm
]
cc
g
/d
[c
Sp
.S
3
31
d
tS
pa
10
d
a
1
S
=
[c
/a
en
m
g
a
em
/s
g
2
[-]
]
ac
1
pl
Spv / vg [-]
is
D
1
1
0.
0.1 1.0 10.0

Period [s]
• Response spectra typically show spectral regions where the re-

sponse 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. T [s]

• Newmark’s elastic design spectra • Elastic design spectra according to SIA 261 (Art. 16.2.3)
6
ζ = 5% Zone 3b, ζ = 5%
5
Pseudo−acceleration Spa [m/s2]

100
αvvg D 4
10
00
αd
g
C
aa
0
dg
α
vg 3
10
0
E
dg
10
00
g
a
10 B 2 Ground Class A
Ground Class B
10
10
Ground Class C
Ps
F 1
eu
Ground Class D
do
−a
]
cc
Ground Class E
m
[c
.S
A 0
d
tS
pa
10
1
[c
0.01 0.10 1.00 10.00

en
m
em
/s
Period [s]
2
]
ac
1
pl
is
• Ground class A: firm or soft rock with a maximum soil cover of 5m

D
1
1
0.
• Ground class B: deposit of extensive cemented gravel and sand

TA=1/33s TB=1/8s TE=10s
with a thickness >30m.
0.1 1.0 10.0

• Ground class C: deposits of normally consolidated and unce-
TF=33s
Period [s] mented gravel and sand with a thickness >30m.
• Ground class D: deposits of unconsolidated fine sand, silt and
Median(50%) One sigma (84%)
clay with a thickness >30m.
Damping ζ αa αv αd αa αv αd
2% 2.74 2.03 1.63 3.66 2.92 2.42 • Ground class E: alluvial surface layer of GC C or D, with a thick-
5% 2.12 1.65 1.39 2.71 2.30 2.01 ness of 5 to 30m above a layer of GC A or B.
10% 1.64 1.37 1.20 1.99 1.84 1.69 • Ground class F: deposits of structurally-sensitive and organic
20% 1.17 1.08 1.01 1.26 1.37 1.38 deposits with a thickness >10m.

• Displacement elastic design spectra according to SIA 261 • Elastic design spectra according to SIA 261 (linear)
25 6
Zone 3b, ζ = 5% Ground Class A

5 Ground Class B
Pseudo−acceleration Spa [m/s2]

20 Ground Class A Ground Class C
Ground Class B 4 Ground Class D
Ground Class E
Displacement Sd [cm]
Ground Class C
15 Ground Class D 3
Ground Class E
2
10
1
Zone 3b, ζ = 5%
5
0
0.00 0.50 1.00 1.50 2.00 2.50 3.00
Period [s]
0
25
0.01 0.10 1.00 10.00
Period [s] Ground Class A
Ground Class B
• The displacement spectra are computed from the acceleration 20
Ground Class C
spectra using equation (7.79)
Displacement Sd [cm]
Ground Class D
S pa 15 Ground Class E
S d = -------2- (7.79)
ω
10
• Displacement spectra are an important design tool (even within
force-based design procedures) because they allow a quick esti-
mate of the expected deformations, hence of the expected dam- 5
age. Zone 3b, ζ = 5%

0
0.00 0.50 1.00 1.50 2.00 2.50 3.00
Period [s]

• Elastic design spectra: Newmark vs. SIA 261 7.5.5 Elastic design spectra in ADRS-format (e.g. [Faj99])
(Acceleration-Displacement-Response Spectra)
3.5
Elastic design spectrum according to
SIA 261, Ground Type A S pa = ω 2 S d
3.0
B C ζ = 5%
2.5 0.5 15
0.4
10
Spa / Ag [−]
Sd [cm]
2.0 0.3
Spa [g]
0.2
5
1.5
0.1
0.0 0
1.0 A 0 1 2 3 4 5 0 1 2 3 4 5
Period [s] Period [s]
Elastic design spectrum according to
SIA 261, Ground Type B D
0.5
Newmark’s elastic design spectrum 0.5
0.0 0.4
0.01 0.10 1.00 10.00 T = const.
Period [s] 0.3
Spa [g]
• The SIA 261 spectra, like the spectra of the majority of the stand- 0.2
ards worldwide, were defined using the same principles as New- 0.1
mark’s spectra.
0.0
• However, different ground motion were used: 0 5 10 15
Sd [cm]
- SIA 261 takes into account different ground classes;
- Different seismic sources were considered; Periods T correspond to lines running through the origin of the
- A larger number of ground motions was considered. axes, because:
• Note: in SIA 261 the corner period T A is not defined.
S pa = ω 2 S d and after reorganizing: T = 2π S d ⁄ S pa

• Elastic design spectra in ADRS-format 7.6 Strength and Ductility

0.5
Elastic design spectrum according to 7.6.1 Illustrative example
SIA 261, Ground Type B
Elastic design spectrum of an Comparison of the time history analyses of an elastic and an in-
0.4 earthquake with equal ag,max elastic single-degree-of-freedom system (SDOF system):
T=0.77s: Overestimation
Telastic = Tinelastic
0.3 T=0.87s: Underestimation
Spa [g]
0.2 ζ = 5%
0.1
Where:
0.0 f el
0 5 10 15 R y = ----- : Force reduction factor (7.80)
Sd [cm] fy
Maximum restoring force that the elastic SDOF system
• Design spectra are defined based on averaged response spectra. f el : (7.81)
reaches over the course of the seismic excitation u··g ( t )
For this reason, the spectral values of single response spectra
fy : Yield force of the inelastic SDOF system (7.82)
may differ significantly from the design spectra.
• This is a crucial property of design spectra and should be um
μ Δ = ------ : Displacement ductility (7.83)
kept in mind during design! uy
Maximum displacement that the inelastic SDOF system
um : (7.84)
reaches over the course of the seismic excitation u··g ( t )
uy : Yield displacement of the inelastic SDOF system (7.85)

• Results Quantity Elastic SDOF Inela. SDOF Ry=2 Inela. SDOF Ry=6
0.15
Elastic SDOF system T [s] 2.0 2.0 2.0
0.10 Inelastic SDOF system (Ry=2) Fmax [kN] 134.70 67.35 22.45
Inelastic SDOF system (Ry=6) Ry [-] – 2.0 6.0
0.05 uy [m]
Displacement [m]
– 0.068 0.023
um [m] 0.136 0.147 0.126
0.00
μΔ [-] – 2.16 5.54
-0.05
• Comments
-0.10
• Both inelastic SDOF systems show a stable seismic response.
-0.15
0 5 10 15 20 25 30 35 40
Time [s]
150
100
50
Force [kN]
-50
Elastic SDOF system

-100 Inelastic SDOF system (Ry=2)
Inelastic SDOF system (Ry=6)
-150
-0.15 -0.10 -0.05 0.00 0.05 0.10 0.15
Displacement [m]

7.6.2 “Seismic behaviour equation” • More realistic representation of the decision possibilities
For seismic collapse prevention, the following approximate rela-
tionship applies
Δ
″quality″ of seismic behaviour ≈ strength × ductility (7.86) F
To survive an earthquake different combination of strength and

ductility are possible:
Δ
F
• If the strength of the structure reduces, the stiffness typically re-

duces 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 sub-
jected to larger deformations, i.e., the deformation demand is larg-
er.

7.6.3 Inelastic behaviour of a RC wall during an earthquake 7.6.4 Static-cyclic behaviour of a RC wall
• Moment-curvature-relation-
Wall WDH4
ship at the base of the plas- 100
tic hinge zone.
50
Moment [kNm]
• Despite reaching and ex-

ceeding its elastic limit the 0
wall did not collapse.

−50
• The plastic deformation ca-
pacity of structures can real- −100
ly be taken into account for

−30 −20 −10 0 10 20 30
seismic design purposes. Curvature [1/km] Wall WSH6 [DWB99]

• Hysteretic behaviour of the RC wall under static-cyclic loading 7.6.5 General definition of ductility
500 …m
Ductility demand μ … = ---------
400
Wall WSH3 Fy …y
0.75 Fy
300
…u
Ductility capacity: μ … = --------
200 …y
Actuator force [kN]
100
0
μΔ=6 μΔ=5 μΔ=4 μΔ=3 μΔ=2 μΔ=1 μΔ=1 μΔ=2 μΔ=3 μΔ=4 μΔ=5 μΔ=6
−100
−200
−300 −0.75 Fy
−400 −Fy
−500
−100 −80 −60 −40 −20 0 20 40 60 80 100
Top displacement [mm]
• 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:
Ductility capacity ≥ Ductility demand

• The structural member fractures when locally the deformation ca-
pacity of the structural materials (i.e., their strain capacities) are
reached. The ductility capacity is therefore exhausted.
Plastic region of test unit WSH6 (left) and close-up of the left boundary region (right). Both pho-
tos were taken at displacement ductility 6.

7.6.6 Types of ductilities 7.7 Inelastic response spectra
f el um
R y = ----- , μ Δ = ------
fy uy
strain εu
μ ε = -----
ductility εy μΔ ( Ry ) = ?
10
ζ = 5% μΔ=1
φu μΔ=2
Sa [m/s2]
curvature
μ φ = ----- 5 μΔ=4
ductility φy μΔ=6
0
0.01 0.10 1.00 10.00 100.0
1.0
μΔ=1 ζ = 5%
μΔ=2
Sv [m/s]
rotation θu 0.5 μΔ=4
μ θ = -----
ductility θy μΔ=6
0.0
0.01 0.10 1.00 10.00 100.0
0.4
μΔ=1 ζ = 5%
0.3
μΔ=2
Sd [m]
displacement Δu 0.2 μΔ=4
μ Δ = ------
ductility Δy 0.1
μΔ=6
0.01 0.10 1.00 10.00 100.0

Period [s]

Force reduction factor Ry Displacement ductility μ Δ

TA TB TC TD TE TF
TA TB TC TD TE TF
ζ = 5% ζ = 5%
Displacement ductility [−]

10
10
μΔ=6 Ry=6
Ry [−]
Ry=4
μΔ=4
μΔ=2
Ry=2
Ry=(2μΔ-1)0.5 Ry=μΔ
1 1
0.1 1.0 10.0
0.1 1.0 10.0
Period [s]
Period [s]
• 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
Ry = μΔ Ry = 2μ Δ – 1 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 princi-
ple” applies – large discrepancies between real and estimated ductility
demand can occur.
• The “equal displacement principle” and the “equal energy principle”
are “historical” Ry-μΔ-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], ...)

7.7.1 Inelastic design spectra • Newmark’s inelastic design spectra [NH82]

g
• Inelastic design spectra in combined D-V-A format
g ζ = 5%
ζ = 5% μΔ=1 Elastic design

spectrum
μΔ=2 100
μΔ=4 V=αvvg D
100 μΔ=6
D
C
=α
aa
α
d
10
dg
A=
vg
00
0
g
a
0
E
dg
10
Vy [cm/s]
B C’ V/μΔ
D’
10
0
10
10
00
Vy [cm/s]
10
.5
1) 0
00
10
/μ Δ
10
F
Δ-
μ
(2
A/
E’
10
10
Ay
D
B’ Inelastic
0
=d
[c
10
m
g
design spectrum
0
A
/s
2
1
]
A’ for μΔ=4
]
1
m
10
a
1
A=
Ay
[c
F’
y
[c
10
D
m
1
/s
]
0.
2
m
]
[c
D
y
/μ Δ
D
1
1
0.
TA=1/33s TB=1/8s TE=10s
1
0.1 1.0 10.0 TF=33s
Period [s]
0.1 1.0 10.0
Period [s] • Maximum displacement of the SDOF system
um = μΔ ⋅ Dy
Note the new axes: D y = u y , V y = ω n u y , A y = ω n2 u y
• Yield strength of the SDOF system:
where: u y = yield displacement
fy = m ⋅ Ay

Construction of the spectra using Ry-μΔ-Tn relationships • Inelastic design spectra according to [NH82] (log. x-axis)
The inelastic design spectra are computed by means of Ry-μΔ- TA TB TC TD TE TF

Tn relationships:
2 μΔ=1
1
A y = S pa, inelastic = ------ ⋅ S pa, elastic (7.87) μΔ=2
Ry
μΔ=4
μΔ μΔ=6
Ay / ag [−]
D = S d, inelastic = ------ ⋅ S d, elastic (7.88)
Ry
1
It should be noted that: Equal displacement principle
S pa, inelastic ≠ ω 2 ⋅ S d, inelastic (7.89)
• Ry-μΔ-Tn relationship according to [NH82]
0
0.01 0.10 1.00 10.00 100.00

° 1 Tn < Ta
° TA TB TC TD TE TF
° ( 2μ – 1 ) β ⁄ 2 T < T < T
° Δ a n b
° 2μ Δ – 1 T b < T n < T c' (EE principle) 2 μΔ=1
Ry = ® (7.90)
° T
μΔ=2
° -----n- μ Δ T c' < T n < T c μΔ=4
° Tc
° μΔ=6
Sd / dg [−]
° μΔ Tn > Tc (ED principle)
¯
1
Where: β = log ( T n ⁄ T a ) ⁄ log ( T b ⁄ T a ) (7.91)
T a = 1 ⁄ 33s , T b = 1 ⁄ 8s (7.92)
T c = Corner period between the constant Spa and the Equal displacement principle
constant Spv regions
T c' = Corner period between the constant Spa and the 0
0.01 0.10 1.00 10.00 100.00
constant Spv regions of the inelastic spectrum Period [s]

• Inelastic design spectra according to [NH82] (linear x-axis) • Ry-μΔ-Tn relationships according to [VFF94]
In [VFF94] Ry-μΔ-Tn relationships are defined as follows:
5.0 TB TC TD ζ = 5%

° cR T n
° c 1 ( μ Δ – 1 ) ⋅ ----- Tn ≤ T0
4.0 -+1
μΔ=1 Ry = ® T0 (7.93)
°
3.0
μΔ=2 c
° c1 ( μΔ – 1 ) R + 1 Tn > T0
Ay [m/s2]
μΔ=4 ¯
μΔ=6 Where:
2.0 T 0 = c 2 ⋅ μ ΔcT ⋅ T c ≤ T c (7.94)
T c = Corner period between the constant Spa and the
1.0
The parameters c 1 , c 2 , c R and c T are defined as follows for 5%

0.0 damping:
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.15 Model
TB TC TD ζ = 5%
Hysteresis Damping c1 cR c2 cT
Q Mass 1.0 1.0 0.65 0.30
0.10 Q Tangent stiffness 0.75 1.0 0.65 0.30

Bilinear Mass 1.35 0.95 0.75 0.20
D [m]
μΔ=1 Bilinear Tangent stiffness 1.10 0.95 075 0.20

μΔ=2
0.05 μΔ=4 and where the Q-hysteretic rule is a stiffness degrading rule sim-
μΔ=6 ilar to the Takeda-hysteretic rule presented in Section 7.3.2.
The table shows the dependency of the Ry-μΔ-Tn relationships
both on damping and hysteretic model.
0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Period [s]

For the Q-hysteretic model and mass-proportional damping, the • Inelastic design spectra according to [VFF94]
Ry-μΔ-Tn relationships by [VFF94] specialise as:
5.0 TB TC TD ζ = 5%
T
° ( μ Δ – 1 ) ⋅ -----n- + 1 Tn ≤ T0
Ry = ® T0 (7.95) 4.0
° μΔ=1
¯ μΔ Tn > T0 (ED principle)
μΔ=2
Where: T 0 = 0.65 ⋅ μ Δ0.3 ⋅ T c ≤ T c
3.0 μΔ=4
Ay [m/s2]
(7.96)
μΔ=6
T c = Corner period between the constant Spa and the
2.0
The spectra depicted on the following pages correspond to this

1.0
case.
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.15
TB TC TD ζ = 5%
0.10
D [m]
μΔ=1
0.05
μΔ=2
μΔ=4
μΔ=6
0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Period [s]

• Inelastic design spectra in ADRS-format 7.7.2 Determining the response of an inelastic SDOF sys-
tem by means of inelastic design spectra in ADRS-
5.0
TB TC format
ζ = 5%
In this section the response of two example inelastic SDOF sys-
4.0
[NH82] tems is determined by means of inelastic design spectra in
μΔ=1 ADRS-format.
μΔ=2
Spa [m/s2]
3.0
• SDOF system 1 with Tn = 0.9 s
μΔ=4
μΔ=6 • SDOF system 2 with Tn = 0.3 s
2.0
• The spectra according to [VFF94] will be used (see Section 7.7.1)
TD • Example 1: SDOF system with Tn= 0.9 s
1.0
m fs
0.0 properties:
0.00 0.02 0.04 0.06 0.08 0.10 0.12
Sd [m] C m = 100 t
5.0
TB TC fy = 80 kN
ζ = 5% k kel = 4874 kN/m
uy = 0.016 m
4.0 kpl = 0 kN/m
[VFF94] fy
μΔ=1 ζ = 5%
kpl
kel
μΔ=2
Spa [m/s2]
3.0
u
μΔ=4 uy um
2.0
μΔ=6
• Response of the elastic SDOF system 1:
TD
T n = 2π m 100- = 0.9s
1.0
---- = 2π -----------
k 4874
S pa = 2.62m/s 2
0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.12
S d = 0.054m
Sd [m]
f el = 261.7kN

• Response of the inelastic SDOF system 1: • Example 2: SDOF system with Tn= 0.3 s
f el 261.7
R y = ----- = ------------- = 3.27 m fs
fy 80 properties:
μ Δ = R y = 3.27 (From Equation (7.95) since T n > T c = 0.5s ) fy,a = 120 kN
C m = 100 t uy,a = 0.0027 m
u m = u y ⋅ μ Δ = 0.016 ⋅ 3.27 = 0.054m = S d
Representation of the inelastic SDOF system 1 in the inelastic k kel = 43‘865 kN/m fy,b = 300 kN
design spectrum in ADRS-format: uy,b = 0.0068 m
kpl = 0 kN/m
fy
kpl
5.0
ζ = 5%
μΔ = 1 kel
u
uy um
4.0
• Response of the elastic SDOF system 2
T n = 2π m 100 - = 0.3s
3.0 ---- = 2π --------------
Spa [m/s2]
μΔ = 3.27 k 43865
Tn = 0.9s
Performance point S pa = 4.71m/s 2
2.0 S d = 0.011m
f el = 471kN
1.0 In this second example two different inelastic SDOF systems will
Capacity curve
um = 0.054m be considered: (a) A SDOF system with a rather low fy and (b) a
0.0
SDOF system with a rather high fy.
0.00 0.02 0.04 0.06 0.08 0.10 0.12
Sd [m] • Response of the second inelastic SDOF system 2a
f el 471
• If the force-deformation relationship of the inelastic SDOF system R y = ----- = --------- = 3.93
fy 120
is divided by its mass m, the “capacity curve” is obtained, which
can be plotted on top of the spectrum in ADRS-format. In this case the resulting displacement ductility μ Δ is so large, that
• The capacity curve and the inelastic spectrum intersect in the “per- Equation (7.96) T 0 = T c = 0.5s results. After rearranging Equa-
tion (7.95), the displacement ductility μ Δ can be computed as:
formance point”.

Tc 0.5 Now consider the SDOF system 2b:

μ Δ = ( R y – 1 ) ⋅ ------ + 1 = ( 3.93 – 1 ) ⋅ ------- + 1 = 5.88
Tn 0.3 • Response of the inelastic SDOF system 2b
Check that T 0 > T c : f el 471
0.3 0.3
R y = ----- = --------- = 1.57
T 0 = 0.65 ⋅ μΔ ⋅ T c = 0.65 ⋅ 5.88 ⋅ 0.5 = 0.553s > T c fy 300
The maximum displacement response is therefore: In this case the displacement ductility μ Δ will be such that Equation
(7.96) yields T 0 < T c = 0.5s . To compute μ Δ insert therefore Equa-
u m = u y ⋅ μ Δ = 0.0027 ⋅ 5.88 = 0.016m > S d tion (7.96) in Equation (7.95). This results in following expression:
Representation of the inelastic SDOF system 2a in the inelastic Tn
( μ Δ – 1 ) ⋅ ---------------------------------
- + 1 = Ry (7.97)
design spectrum in ADRS-format: 0.65 ⋅ μ Δ ⋅ T c
0.3
5.0 Equation (7.97) needs to be solved numerically.

Tn = 0.3s
μ Δ ( T n = 0.3 ,T c = 0.5 ,R y = 1.57 ) = 1.73
4.0 Check that T 0 < T c :

(*) 0.3 0.3
T 0 = 0.65 ⋅ μ Δ ⋅ T c = 0.65 ⋅ 1.73 ⋅ 0.5 = 0.383s < T c
Performance point Tn
3.0 0.3
( μ Δ – 1 ) ⋅ ------ + 1 = ( 1.73 – 1 ) ⋅ ------------- + 1 = 1.57 = R y
Spa [m/s2]
T0 0.383
The maximum displacement response is therefore:
2.0
u m = u y ⋅ μ Δ = 0.0068 ⋅ 1.73 = 0.012m > S d
Capacity curve μΔ = 1
1.0 um = 0.016m
μΔ = 5.88
0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.12
Sd [m]
• Note that the line (*) is no longer vertical as in Example 1, but in-
clined according to the equation μ Δ = ( R y – 1 ) ⋅ ( T c ⁄ T n ) + 1 .

Representation of the inelastic SDOF system 2b in the inelastic T

design spectrum in ADRS-format: ° ( μ Δ – 1 ) ⋅ -----n- + 1 Tn ≤ Tc
Ry = ® Tc (7.98)
°
5.0 Tn = 0.3s ¯ μΔ Tn > Tc (ED principle)
“Performance Point”
This approximation is particularly satisfactory, if the large uncer-
4.0
tainties associated with smoothed spectra are considered.
Capacity curve
• Comments
• A discussion of similar examples can be found in [Faj99].
3.0
Spa [m/s2]
• For computing the response of inelastic SDOF systems by means

um = 0.012m of inelastic design spectra, the Ry-μΔ-Tn relationships in Section
2.0 7.7.1 are sufficient. The spectra in ADRS-format are not absolutely
necessary, but they illustrate the maximum response of inelastic
μΔ = 1 SDOF systems very well.
(*)
1.0 • Ry-μΔ-Tn relationships should only be used in conjunction with
smoothed spectra. They should not be used to derive the inelastic
μΔ = 1.73
response spectra of a single ground motion
0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.12 • Remember:
Sd [m]
- Design spectra are very useful tools to design structures for the expected
seismic demand. Design spectra represent the average effect of an
• Note that the curve (*) is no longer a straight line as in Examples earthquake with design intensity.
1 and 2a.
- If a single earthquake is considered, the spectra may underestimate the
• In Example 2b the curve (*) needs to be computed numerically. seismic demand for a certain period range (... overestimate ...).
• In Example 2a the curve (*) is only an approximation of the curve - This characteristic of design spectra should be considered when design-
(*) in Example 2b. As soon as T 0 = T c both curves are identical. ing structures: The seismic design should aim at structures that are as
robust as possible.
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 differ-
ence is, however, small. For this reason, in most cases Equation
(7.95) can be approximated as:

7.7.3 Inelastic design spectra: An important note 7.7.4 Behaviour factor q according to SIA 261
The “equal displacement” and the “equal energy”

principles represent a strong simplification of the real
inelastic behaviour of SDOF systems. q = Φo 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 struc- Φ0 = Overstrength
tures.
• Design spectra according to SIA 261

• However, if single ground motions are considered, then it can
easily be the case that design spectra significantly underesti- 4
mate the expected peak behaviour of structures. Ground Class B TB TC TD
3
• This characteristic of the design spectra shall be taken into ac- Elastic design
Sd / agd [-]
count during design by aiming at robust structures. q=1.5 spectrum
2
q=2
1
Φ0=1.5 q=3
q=4
0
0.01 0.10 1.00 10.00
Period [s]

7.8 Linear equivalent SDOF system (SDOFe) • Example: Inelastic SDOF system with Takeda-hyst. rule [TNS70]
u The properties of the inelastic SDOF system are:
..
fs -mug(t) fs • Damping: ζ = 5% (constant, proportional to kel)
..
mu Elastic • Mass: m = 100t
. (with elastic stiffness kel)
cu • Stiffness: k el = 4874kN/m
• Yield force: f y = 80kN
• Hysteresis: Takeda-hysteresis with r o = 0.05 , α = 0.5 , β = 0.0
kel Elastic
(with „effective stiffness“
ug(t) or „secant stiffness“ keff) + + –α
fm k u = k el ( max { μ Δ } )
.. fy - - –α
ug(t) kpl Inelastic k u = k el ( max { μ Δ } )
kel keff
u
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: k eff = f m ⁄ u m (7.99)
Damping: ζe (7.100)
The differential equation of the SDOFe is: The maximum response of the SDOF system when subjected to the NS-
component of the 1940 El Centro Earthquake is:
2
u··( t ) + 2ζ e ω e u· ( t ) + ω e2 u ( t ) = – u··g ( t ) with ω e = k eff ⁄ m (7.101) x m = 0.073m , f m = 93.0kN
The question is how the viscous damping ζ e of the SDOFe can The properties of the corresponding SDOFe are:
be determined so that max ( u ( t ) ) = u m . fm 93.0 kN m- = 2π -----------
100- = 1.76s
k eff = ------ = ------------- = 1274 ------- , T e = 2π -------
um 0.073 m k eff 1274
ζ e = 22.89% , the viscous damping ζ e was determined iteratively!

Comparison between inelastic SDOF and SDOFe Comments regarding the example:
• The damping ζ e is in general larger than the damping ζ , since ζ e
100
of the SDOFe needs to compensate for the hysteretic energy ab-
ζ = 5%
sorption of the inelastic SDOF system.
ζe = 22.89%
• However, in rare cases it happens that ζ e < ζ . This shows again
50
the difficulties that are associated with the prediction of the seismic
response of inelastic SDOF systems.
Force [kN]
• In the example, the viscous damping ζ e was determined iteratively

0
until a value for ζ 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 vis-
-50 cous damping ζ e , then the maximum response of the inelastic
Inelastic SDOF SDOF system could indeed be estimated by means of the linear
fm Equivalent SDOF equivalent SDOF system.
-100
-0.08 • The stiffness k eff and the period T e of the SDOFe system are only
um -0.06 -0.04 -0.02 0.00
Deformation [m]
0.02 0.04 0.06 0.08
known once the maximum response of the inelastic SDOF system
0.08 are known. Section 7.8.2 shows how the equivalent viscous damp-
ζ = 5% ing ζ e can be estimated without knowing the stiffness k eff and the
0.06
ζe = 22.89% period T e of the SDOFe system a priori.
0.04 • Estimating the damping ζ e
Deformation [m]
0.02 In particular in the sixties significant research has been dedicated to es-
timating the damping ζ e (see for example [Jac60], [Jen68] and [IG79]).
0.00 At that time the interest in linear equivalent systems was big because the
numerical computation of the response of inelastic systems was ex-
-0.02
tremely expensive. The basic idea behind estimating the damping ζ e
-0.04 was:
Inelastic SDOF The inelastic SDOF system dissipates energy due to ζ and due to the
-0.06
um Equivalent SDOF inelastic deformations, which are a function of its inelastic force-defor-
-0.08 mation relationship. The equivalent SDOF system, however, dissipates
0 5 10 15 20 25 30 35 40 energy solely due to its viscous damping. For this reason the following
Time [s]
relationship applies:

ζ e = ζ + ζ eq (7.102) fm = [ 1 + ro ( μΔ – 1 ) ] ⋅ fy
fo = ( 1 – ro ) ⋅ fy
where ζ eq is the viscous damping equivalent to the hysteretic energy ab-
sorption of the inelastic system. u1
f 1 = ----------------- ⋅ f 2
u1 + u3
The simplest method for estimating the equivalent viscous damping is to
f2 = [ 1 + ro ( μΔ – 1 ) ( 1 – β ) ] ⋅ fy
assume that the inelastic system and the linear equivalent system dissi-
pate the same energy within one displacement cycle. According to this u1 = μΔ uy – u2
assumption [Cho11] defines the equivalent elastic damping as: 1 + ro ( μΔ – 1 )
- ⋅ uy
u 2 = ---------------------------------
–α
μΔ
1 Ah
ζ eq = ------ ⋅ ------ (7.103) u3 = [ μΔ – β ( μΔ – 1 ) ] ⋅ uy
4π A e
Where: The equivalent viscous damping of other important hysteresis rules is:
A h : Energy dissipated by the inelastic SDOF system due to the ine-
lastic deformation of the system. The dissipated energy corre-
sponds to the area of the force-displacement hysteresis of the Elasto-plastic (EP) rule:
considered displacement cycle;
2 μΔ – 1
A e : Potential energy of the equivalent SDOF system at maximum dis- ζ eq, EP = --- ⋅ --------------- (7.105)
π μΔ
placement:
2
k eff ⋅ u m
A e = --------------------
2
Bilinear (BL) rule:
The inelastic force-deformation relationship of many structural RC ele-
ments can be described by the “Takeda”-hysteresis rule. According to 2 ( μΔ – 1 ) ( 1 – ro )
ζ eq, BL = --- ⋅ ------------------------------------------ (7.106)
Equation (7.103) the equivalent viscous damping of this hysteresis rule π μΔ ( 1 + ro μΔ – ro )
is:
1 ( f m + f o )μ Δ u y + f 1 u 1 – f m u 2 – ( f m + f o )u 3
ζ eq,Tak = ------ ⋅ ----------------------------------------------------------------------------------------------------- (7.104)
4π ( fm μΔ uy ) ⁄ 2
Rule according to Clough (Clo) [CP75]:
Where:
2 3 μΔ – 1
ζ eq, Clo = --- ⋅ ------ ⋅ --------------- (7.107)
π 2π μΔ

ζ eq - μ Δ -relationship for these important hysteresis rules: Comments regarding the comparison of the theoretical value with the
computed value of ζ e for the Takeda-SDOF system when excited by the
70
2/π El Centro earthquake:
60 Elasto-plastic • The computed value of ζ e was determined iteratively. Eight differ-
50 ent inelastic SDOF systems with different periods T n were consid-
Clough ered. The strength of each inelastic SDOF system was varied in
ζeq [-]
40 such a way that seven different displacement ductilities resulted

30
Bilinear (ro=5%) ( μ Δ =2 to 8).
Bilinear (ro=20%)
• The results show that ζ e is not only dependent on μ Δ but also on
20
the period T n of the SDOF system. This effect is not considered by
Equations (7.103) and (7.104), respectively.
10 Takeda (α=0.5, β=0.0, ro=5%)
0 • In some cases the difference between the theoretical value and
0 1 2 3 4 5 6 7 8 9 10 the computed vale for ζ e is considerable. For this reason there are
μΔ [-] also considerable differences between μ Δ, t arg et (target ductility)
and μ Δ, actual (actual ductility obtained from the time-history analy-
The next figure compares the theoretical value for ζ e for the Takeda-SDOF
sis of the SDOFe system with the viscous damping ζ e according to
(Equ. (7.104), ro=0.05, α=0.5, β=0) with the computed value (for El Centro):
Equation (7.104)).
50
11 Theory (Eq. (3.46)) • Typically these differences increase as the target ductility increas-
Tn=4.00s es.
10
Tn=2.00s
40 9 • Similar observations were made when the computation of the ine-
Tn=1.33s
8 Tn=1.00s lastic spectra was discussed.
μΔ,actual [-]
Tn=0.67s
30 7 Tn=0.50s • This shows again the difficulties associated with the prediction of
ζe [-]
6 Tn=0.33s the seismic response of inelastic SDOF system.

Tn=0.25s
5 • Improved estimate for ζ e
20
4 Over the last years some researchers suggested improved formulas for ζ e by
3 carrying out statistical analyses of time-history responses of inelastic SDOF
10 systems (see [PCK07]). [GBP05] suggest for example the Equation (7.108).
2
1
§ 1 ·§ 1 ·
0 0 ζ e = ζ + ζ eq where: ζ eq = a ¨ 1 – -----b-¸ ¨ 1 + ---------------------d-¸ (7.108)
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 © μ Δ¹ © ( Te + c ) ¹
μΔ,target [-] μΔ,target [-]

The constants a to d are: • Takeda Thin (TT): Hysteretic rule that characterises RC struc-
tures 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”, β=0.35 (FS): Hysteretic rule that characterises
prestressed structures with unbonded tendons.
ζ eq - μ Δ -relationships for the most important hysteresis rules according to
[GBP05]:
20
1) EPP
2) BL (ro=0.2)
Hysteresis rule a b c d 15 3) Takeda Thin
1) Elasto-Plastic (EPP) 0.224 0.336 -0.002 0.250 4) Takeda Fat
ζeq [-]
2) Bilinear, ro=0.2 (BI) 0.262 0.655 0.813 4.890 5) FS (beta=0.35)
10 6) RO
3) Takeda Thin (TT) 0.215 0.642 0.824 6.444
4) Takeda Fat (TF) 0.305 0.492 0.790 4.463
5) Ramberg-Osgood (RO) 0.289 0.622 0.856 6.460 5
6) “Flag-Shaped”, β=0.35 (FS) 0.251 0.148 3.015 0.511
Te=2.0s
from [GBP05] 0
0 1 2 3 4 5 6
The hysteresis rules 1) to 6) were chosen because they can be used to
μΔ [-]
represent the hysteretic behaviour of typical structural types:
• Elasto-Plastic (EPP): Hysteretic rule that characterises systems Important comments:
for the seismic isolation of structures (sliding systems that are • With these relationships an in a statistical sense improved esti-
based on friction). mate of the damping ζ e is obtained.
• Bilinear, ro=0.2 (BI): Hysteretic rule that also characterises sys- • For single systems subjected to a specific ground motion differenc-
tems for the seismic isolation of structures. The value of the post- es between the maximum response of the inelastic system and the
yield stiffness rokpl may vary significantly between different sys- maximum response of the equivalent SDOF with ζ e according to
tems. these improved ζ eq - μ Δ -relationships can still be significant!

7.8.1 Elastic design spectra for high damping values 1.0

0.9
Correction factor η [-]

To compute the response of the equivalent SDOF systems, elas- 0.8
tic design spectra can be used. 0.7
0.6
The damping values of equivalent SDOFe systems are in gener-
0.5
al larger than the typical 5%. For this reason the design spectra
0.4
needs to be computed for higher damping values.
0.3
The design spectra for higher values of damping are often ob- [BE99]
0.2
tained by multiplying the design spectra for 5% damping with a [TF99]
0.1
correction factor η : 0.0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50
S pa (T n,ζ) = η ⋅ S pa (T n,ζ = 5%) (7.109) Damping ζ [-]
Comments:
The literature provides different estimates for this correction fac-
tor η . Two of these are: • A discussion of the different approaches for computing the design
spectra for high values of damping can be found in [PCK07].
1.5
[TF99]: η = ------------------ where 0.05 ≤ ζ ≤ 0.5 (7.110) • Equations (7.110) and (7.111) were derived for ground motions
1 + 10ζ without near-field effects.
• Equations (7.110) and (7.111) were derived from the statistical
1 - where 0.05 ≤ ζ ≤ 0.3
[BE99]: η = ---------------------- (7.111) analysis of several response spectra for different ground motions.
0.5 + 10ζ For this reason Equ.s (7.110) and (7.111) should only be used in
Equation (7.111) corresponds to Equation (29) in the Swiss conjunction with smoothed response or design spectra.
Code SIA 261 [SIA03]. • As for all statistical analyses the resulting design spectra corre-
spond only in average with the true highly damped spectral ordi-
The correction factors η obtained with Equ.s (7.110) and (7.111) nates. For single periods and ground motions the differences
are plotted for different damping values ζ in the next figure: 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.

• Elastic design spectra according to [BE99] • Elastic design spectra in ADRS-format
5.0 TB TC TD 5.0 TB TC
[BE99]
ζ = 5%
4.0 10% 4.0
20%
Spa [m/s2]
3.0 3.0
Ay [m/s2]
30% ζ = 5%
2.0 2.0 10%
20% TD
1.0 1.0 30%
0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.00 0.02 0.04 0.06 0.08 0.10 0.12
Sd [m]
0.15
TB TC TD
5.0 TB TC
[TF99]
ζ = 5%
4.0
0.10 10%
Spa [m/s2]
3.0
20%
D [m]
ζ = 5%
30%
2.0
10%
0.05
20%
30 TD
40
1.0 50%
0.00 0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.00 0.02 0.04 0.06 0.08 0.10 0.12
Period [s] Sd [m]

7.8.2 Determining the response of inelastic SDOF systems • Response of the elastic SDOF system
by means of a linear equivalent SDOF system and
T n = 2π m 100- = 0.9s
elastic design spectra with high damping ---- = 2π -----------
k 4874
The computation of the seismic response of inelastic systems by S pa = 2.62m/s 2
means of linear equivalent systems was studied by Sozen and S d = 0.054m
his co-workers in the seventies (see for example [GS74], [SS76] • Response of the inelastic SDOF system
and [SS81]). The maximum response of the inelastic SDOF system will be com-
puted by means of the ADRS-spectra (page 7-106).
Today this approach gains new attention since the “Direct Dis-
placement-Based Design (DDBD)” approach, which was devel- Step 1: The capacity curve of the SDOF system is plotted on top
of the ADRS spectra.
oped by Priestley and his co-workers, is based on the idea of the
linear equivalent system ([PCK07]). Step 2: By means of Equ.s (7.102) and (7.104) the nonlinear
scale, which represents the damping ζ e as a function of the maxi-
This sections outlines the procedure for computing the response mum response of the SDOF system, is plotted along the capacity
of an inelastic SDOF system by means of an linear equivalent curve.
SDOF system and elastic design spectra with high damping. Step 3: Several spectra for different values of damping are plotted.
• Example: SDOF system with Tn= 0.9 s Step 4: The “Performance Point” is the point where the spectrum
with damping ζ e intersects the capacity curve at the same value of
m fs ζe .
Properties: Takeda hysteresis For the considered example the maxmimum response of the ine-
(α=0.5, β=0.0, ro=0.05) lastic SDOF system is:
C m = 100 t
S d = 0.065m
fy = 80 kN
k kel = 4874 kN/m Comments regarding the example:
uy = 0.016 m
kpl = 244 kN/m • To determine the “Performance Point” exactly, an iterativ approach
fy is typically required.
kpl
ζ = 5% kel • The linear equivalent SDOF system is fully defined by the period
u T e and the damping ζ e . The period T e results from the slope of the
uy um
line that connects the origin with the “Performance Point”.
For the example the spectra according to [BE99] will be used (Section 7.8.1).

• The damping values ζ eq used in the figures on page 7-106 were Determining the SDOF behaviour by means of elastic ADRS-spectra
determined according to Equation (7.104). In the figures on page 5.0 ζ = 5%
7-107 the damping ζ eq was determined using Equation (7.108).
The difference is, however, rather small.
• It should be noted that in both cases the computed maximum re- 4.0 10% ζ = 18.5%
sponse of the inelastic SDOF system does not comply with the 15%
Performance point
“equal displacement principle”.
Spa [m/s2]
3.0
20%
• The linear equivalent SDOF system leads often to results that do 25% Capacity curve
not agree with the “equal displacement principle”. This applies in
2.0
particular to SDOF systems with long periods or systems with
large ductility demands.
1.0 ζe = 5%
• A second example is presented on page 7-107. It is a SDOF sys-
20%
tem with a shorter period and a smaller ductility demand than in 10% 15%
ζe = 18.5%
Example 1. In this second example the “equal displacement prin- 0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.12
ciple” is approximately confirmed. Sd [m]
Alternative representation:
5.0 ζ = 5%
4.0
ζ = 18.5% Performance point
Spa [m/s2]
3.0
Tn = 0.9s
2.0
Te = 1.67s
1.0
μΔ = 3.98 ζe= 18.5%

0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.12
uy = 0.016m Sd [m] um = 0.065m

Recalculate the example with ζ eq according to [GBP05]: 7.9 References

5.0 ζ = 5%
[AEM 86] Anderheggen E., Elmer H., Maag H.: “Nichtlineare Finite El-
ement Methoden: Eine Einführung für Ingenieure”. IBK Vor-
4.0 lesungsskript. Institut für Baustatik und Konstruktion (IBK),
ETH Zürich. 1986.
ζ = 17.8%
[AP88] Allahabadi R., Powell G.: “Drain-2DX, User Guide”. Report
Spa [m/s2]
3.0 Performance point

Tn = 0.90s UBC/EERC-88/06. Earthquake Engineering Research Cent-
er, University of California, Berkeley 1988.
2.0
[Bat96] Bathe K-J.: Finite Element Procedures. ISBN 0-13-301458-4.
Te = 1.69s
Prentice-Hall, Upper Saddle River, New Jersey 1996.
1.0
[BE99] Bommer J.J., Elnashai A.S.: “Displacement spectra for seis-
μΔ = 4.07 ζe= 17.8% mic design”. Journal of Earthquake Engineering vol. 3, No. 1,
0.0 1999.
0.00 0.02 0.04 0.06 0.08 0.10 0.12
uy = 0.016m Sd [m] um = 0.067m
[Cho11] Chopra A.K.: “Dynamics of Structures”. Fourth Edition. Pren-
Second example with smaller ductility demand: tice Hall, 2011.
5.0 ζ = 5% [DWB99] Dazio A., Wenk T., Bachmann H.: “Versuche an Stahlbeton-
tragwänden unter zyklisch-statischer Einwirkung. (Tests on
Performance point RC Structural Walls under Cyclic-Static Action)”. IBK-Report
4.0
ζ = 12.8% Tn = 0.63s No. 239, IBK-ETH Zurich. Birkhäuser Verlag, Basel 1999.
(http://e-collection.ethbib.ethz.ch/view/eth:23296)
Spa [m/s2]
3.0 [Faj99] Fajfar P.: “Capacity Spectrum Method Based on Inelastic De-
Te = 0.88s
mand Spectra”. Earthquake Engineering and Structural Dy-
namics, 28, 979-993, 1999.
2.0
[HHT77] Hilber H.M., Hughes T.J.R, Taylor R.L.: “Improved Numerical
Dissipation for Time Integration Algorithms in Structural Dy-
1.0
namics”. Journal of Earthquake Engineering and Structural
μΔ = 1.97 ζe= 12.8% Dynamics, Vol. 5, 1977.
0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.12 [HKS03] Hibbit Karlsson & Sorensen: “ABAQUS Version 6.4 - User’s
uy = 0.02m um = 0.039m Sd [m] Manual”. Pawtucket, RI, 2003.

[KN92] Krawinkler H., Nassar A.A.: “Seismic design based on ductil- [PCK07] Priestley M.J.N., Calvi G.M., Kowalsky M.J.: “Displacement-
ity and cumulative damage demands and capacities”. Nonlin- Based Seismic Design of Structures”. IUSS Press, 2007.
ear Seismic Analysis and Design of Reinforced Concrete
[Saa91] Saatcioglu M.: “Modeling Hysteretic Force-Deformation Re-
Buildings, eds P. Fajfar and H. Krawinkler. New York: Elsevier
lationships for Reinforced Concrete Elements”. In Publication
Applied Science. 1992.
SP 127 “Earthquake-Resistant Concrete Structures, Inelastic
[GBP05] Grant D.N., Blandon C.A., Priestley M.J.N.: “Modelling Ine- Response and Design”, ACI, Detroit 1991.
lastic Response in Direct Displacement-Based Design”.
[SIA03] Swiss standard SIA 261: “Actions on structures”. SIA, Zürich
Rose School, Report 2005/03, IUSS Press, Pavia, 2005.
2003.
[GS74] Gulkan P., Sozen M.: “Inelastic Responses of Reinforced
[SS81] Saiidi A., Sozen M.: “Simple Nonlinear Seismic Analysis of
Concrete Structures to Earthquake Motions”. ACI Journal, Ti-
R/C Structures”. ASCE, Journal of the Structural Division,
tle No. 71-41. December 1974.
Vol. 107, No. 5, May 1981.
[IG79] Iwan W.D., Gates N.C.: “Estimating Earthquake Response of
[SS76] Shibata A., Sozen M.: “Substitute Structure Method for Seis-
Simple Hysteretic Structures”. ASCE, Journal of the Engi-
mic Design in R/C”. ASCE, Journal of the Structural Division,
neering Mechanics Division, Vol. 105, EM3, June 1979.
Vol. 102, No. 1, January 1976.
[Jac60] Jacobsen L.S.: “Damping in composite structures”. Proceed-
[TF99] Tolis S.V., Faccioli E.: “Displacement design spectra”. Jour-
ings of the 2nd World Conference in Earthquake Engineering.
nal of Earthquake Engineering vol. 3, No. 1, 1999.
Vol. 2. Tokio and Kyoto, Japan, 1960.
[TSN70] Takeda T., Sozen M.A., Nielsen N.N.: “Reinforced Concrete
[Jen68] Jennings P.C.: “Equivalent Viscous Damping for Yielding
Response to Simulated Earthquakes”. ASCE, Journal of
Structures”. ASCE, Journal of the Engineering Mechanics Di-
Structural Engineering, Vol. 96, No. 12, December 1970.
vision, Vol. 94, No. EM1, February 1968.
[VFF94] Vidic T., Fajfar P., Fischinger M.: “Consistent inelastic design
[Mir01] Miranda E.: “Estimation of inelastic defromation demands of
spectra: strength and displacement”. Earthquake Engineer-
SDOF Systems”. ASCE, Journal of structural engineering,
ing and Structural Dynamics 23(5) 507-521. 1994.
Vol. 127, No. 9, 2001.
[New59] Newmark N.M.: “A Method of Computation for Structural Dy-
namics”. ASCE, Journal of the Engineering Mechanics Divi-
sion, Vol. 85, No. 3, July 1959.
[NH82] Newmark N.M., Hall W.J.: “Earthquake Spectra and Design”.
Earthquake Engineering Research Institute (www.eeri.org),
1982.

8 Multi Degree of Freedom Systems 8.1.2 Stiffness formulation
8.1 Formulation of the equation of motion
8.1.1 Equilibrium formulation 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 11 k 12 ( k1 + k2 ) –k2
K = = (8.5)
k 21 k 22 –k2 k2
Unit displacement u 1 = 1 Unit displacement u 2 = 1

·· · · ·
m1 u1 + c1 u1 + k1 u1 = f1 ( t ) + c2 ( u2 – u1 ) + k2 ( u2 – u1 )
® (8.1)
¯ m 2 u··2 + c 2 ( u· 2 – u· 1 ) + k 2 ( u 2 – u 1 ) = f 2 ( t )
·· · ·
m 1 u 1 + ( c 1 + c 2 )u 1 – c 2 u 2 + ( k 1 + k 2 )u 1 – k 2 u 2 = f 1 ( t )
® (8.2)
¯ m 2 u··2 – c 2 u· 1 + c 2 u· 2 – k 2 u 1 + k 2 u 2 = f 2 ( t )
m1 0 u··1 ( c1 + c2 ) –c2 u· 1 ( k1 + k2 ) –k2 u1 f (t)

+ + = 1 (8.3)
0 m2 ··
u2 –c2 c2 ·u – k k u f ( t )
2 2 2 2 2
·· + Cu· + Ku = f ( t )
Mu (8.4)
8 Multi Degree of Freedom Systems Page 8-1 8 Multi Degree of Freedom Systems Page 8-2
• Mass matrix M By means of the principle of virtual forces the vertical displace-
ment Δ at location d due to a unit force F = 1 acting at location
m1 0 a can be readily determined.
M = (8.6)
0 m2 FL 3 a d
Δ (α,δ) = – α δ ( α 2 + δ 2 – 1 ) ⋅ --------- with α = --- and δ = --- (8.9)
6EI L L
• Equation of motion
The flexibilty matrix consists of the following elemnts:
m1 0 u··1 ( k1 + k2 ) –k2 u1
+ = 0 (8.7) u = DF (8.10)
0 m2 ··
u2 –k2 k2 u2 0
u1 d 11 d 12 F1
·· + Ku = 0
Mu (8.8) = ⋅ (8.11)
u2 d 21 d 22 F2
8.1.3 Flexibility formulation The d ij factors can be computed by means of Equation (8.9) as
follows:
4 L3
d 11 = Δ (1 ⁄ 3,2 ⁄ 3) = --------- ⋅ ------ (8.12)
243 EI
7 L3
d 12 = Δ (2 ⁄ 3,2 ⁄ 3) = --------- ⋅ ------ (8.13)
486 EI
7 L3
d 21 = Δ (1 ⁄ 3,1 ⁄ 3) = --------- ⋅ ------ (8.14)
• Flexibility matrix D 486 EI
4 L3
d 22 = Δ (2 ⁄ 3,1 ⁄ 3) = --------- ⋅ ------ (8.15)
243 EI
and the flexibilty matrix D becomes:
L3
D = --------------- ⋅ 8 7 (8.16)
486EI 7 8
• Stiffness matrix K 8.1.6 “Direct Stiffness Method”

• Stiffness matrix of a beam element
162 EI
K = D – 1 = --------- ⋅ -----3- ⋅ 8 – 7 (8.17)
5 L –7 8
• Mass matrix M
m1 0
M = (8.18)
0 m2
The stiffness matrix K of a beam element with constant flexural
• Equation of motion
and axial stiffness is well known:
m1 0 u··1 162 EI u F = Ku (8.21)
+ --------- ⋅ -----3- ⋅ 8 – 7 1 = 0 (8.19)
0 m2 u··2 5 L –7 8 u2 0
EA EA
·· + Ku = 0 -------- 0 0 – -------- 0 0
Mu (8.20) L L
12EI-
----------- 6EI
--------- 12EI- 6EI
F1 0 0 – ----------- --------- u1
8.1.4 Principle of virtual work L3 L2 L3 L2
F2 u2
6EI 4EI 6EI- 2EI
• See e.g. [Hum12] F3
0 --------- --------- 0 – -------- ---------
u3
= L2 L L2 L
⋅ (8.22)
F4 EA EA u4
8.1.5 Energie formulation – -------- 0 0 -------- 0 0
F5 L L u5
• See e.g. [Hum12] F6 0
12EI 6EI
– -----------
- – --------- 0
12EI 6EI
------------ – --------- u6
L3 L2 L3 L2
6EI 2EI 6EI 4EI
0 --------- --------- 0 – -------- - ---------
L2 L L2 L
If the axial elongation of the beam is not considered, the matrix

can be further simplified as follows:
Equation of motion:
mL 0 0 0 u··1 24 0 – 12 6L u1 0
0 0 0 0 u·· u2
⋅ 2 EI
+ -----3- ⋅ 08L 2 2L 2
– 6L ⋅ = 0 (8.25)
mL u··3 L u3 0
0 0 -------- 0 – 12 – 6L 12 – 6L
2
u··
4
2
6L 2L – 6L 4L 2 u4 0
12 6L – 12 6L 0 0 0 0
EI 2 2
K = -----3- ⋅ 6L 4L – 6L 2L (8.23)
L – 12 – 6L 12 – 6L mL 0 0 0 u··1 24 – 12 0 6L 0 u1
6L 2L 2 – 6L 4L 2 mL u··EI – 12 12 – 6L – 6L u
0 -------- 0
2
0 ⋅ 3 + -----3- ⋅ ⋅ 3 = 0 (8.26)
• Example: Cantilever ··
u2 L 0 – 6L 8L 2 2L 2 u2 0
0 0 0 0
u··4 6L – 6L 2L 2 4L 2 u4 0
0 0 0 0
Static condensation:
mL 0 0 0 u··1 24 – 12 0 6L 0 u1
mL u··EI u
0 -------- 0
2
0 ⋅ 3 + -----3- ⋅ – 12 12 – 6L – 6L ⋅ 3 = 0 (8.27)
u··2 L 0 – 6L 8L 2 2L 2 u2 0
Assemblage of the stiffness matrix 0 0 0 0
u··4 6L – 6L 2L 2 4L 2 u4 0
0 0 0 0
F1 12 + 12 – 6 L + 6L – 12 6L u1
F2 EI 2 2 2L 2 u2 ··
= -----3- ⋅ – 6 L + 6L 4L + 4L – 6L ⋅ (8.24) m tt 0
⋅
u t + k tt k t0 ⋅ u t = 0 (8.28)
F3 L – 12 – 6L 12 – 6L u3 ··
0 0 u0 k 0t k 00 u0 0
F4 6L 2L 2 – 6L 4L 2 u4
with L = L ⁄ 2
··
m tt u t + k tt u t + k t0 u 0 = 0 EI 6
® (8.29) k̂ tt = -----3- ⋅ --- ⋅ 16 – 5 (8.36)
¯ k 0t u t + k 00 u 0 = 0 L 7 –5 2
From the second row of Equation (8.29) the following expression after substituting L = L ⁄ 2 :
can be derived:
EI 48
–1 k̂ tt = -----3- ⋅ ------ ⋅ 16 – 5 (8.37)
u0 = – k 00 k 0t u t (8.30) L 7 –5 2
Substituting Equation (8.30) in the first line of Equation (8.29) we The final equation of motion of the cantilever is therefore:
obtain:
mL
·· + k u – k k k u = 0
m tt u
–1
(8.31) -------- 0 u·· u
2 EI 48
⋅ 1 + -----3- ⋅ ------ ⋅ 16 – 5 ⋅ 1 = 0
t tt t t0 00 0t t
(8.38)
·· + ( k – –1 mL u··3 L 7 – 5 2 u3 0
m tt u t tt k t0 k 00 k 0t )u t = 0 (8.32) 0 --------
4
T
and with k t0 = k 0t :
• Notes
T
·· + ( k – k k k )u = 0 –1
m tt u t tt 0t 00 0t t (8.33) • The “Direct Stiffness Method” is often used in the Finite Element
Method.
·· + k̂ u = 0 with k̂ = k – k k k T –1
m tt u t tt t tt tt 0t 00 0t (8.34) • The derivation of the stiffness matrix K for a beam element and in-
structions for assembling the stiffness matrix of entire structures
can be found e.g. in the following references:
Where k̂ tt is the condensed stiffness matrix, and in our case it is
equal to: [Prz85] Przemieniecki J.S.: “Theory of Matrix Structural Analy-
sis”. Dover Publications, New York 1985.
§ 1 1 ·
¨ ---------
2
– -----------2- ¸
EI ¨ 24 – 12 14L ⋅ 0 – 6 L ¸
k̂ tt = -----3- ⋅ – 0 6L ⋅ 7L [Bat96] Bathe K-J.: “Finite Element Procedures”. Prentice Hall,
L ¨ ¸ Upper Saddle River, 1996.
¨ – 12 12 –6 L –6 L 1
– -----------2-
2-
-------- 6L – 6 L ¸
© 14L 7L 2 ¹
(8.35)
8.1.7 Change of degrees of freedom or:

The equation of motion for free vibration of the 2-DoF system de-
u··1 u
picted in the following can be immediately set up if the DoFs u 1 M⋅ +K⋅ 1 = 0 (8.40)
·· θ1
and θ 1 are considered. θ1 0
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 (8.41)
° u 2 = --- ⋅ θ 1
¯ 2
which in matricial form yields the following system of equations:
u1 1 0 u1
= ⋅ or u = Au (8.42)
θ1 0 2⁄L u2

°
®
°
¯
A
Using Equation (8.23), the equation of motion for free vibrations The matrix A is called coordinate transformation matrix and can
of the system becomes be used to transform the mass matrix, the stiffness matrix and
the load vector from one set of DoF to the other, i.e.
3m
-------- 0 T
2 u·· EI 12 – 6L u K = A KA (8.43)
⋅ 1 + -----3- ⋅ ⋅ 1 = 0 (8.39)
2 ·· 2 θ1
mL θ1 L – 6L 4L 0 M = A MA
T
(8.44)
0 -----------
8
T
F = A F (8.45)
For the example at hand, the stiffness matrix K expressed in the 8.1.8 Systems incorporating rigid elements with distribut-
set of DoFs u 1 and u 2 becomes: ed mass
The 2-DoF system depicted in the following incorporates a rigid
EI 12 – 6L
K = A KA = 1 0 ⋅ -----3- ⋅ ⋅ 1 0
T
(8.46) element with distributed mass μ .
0 2⁄L L – 6L 4L
2 0 2⁄L
EI
K = -----3- ⋅ 12 – 12 (8.47)
L – 12 16
while the mass matrix M becomes:
3m
-------- 0
2
M = A MA = 1 0 ⋅ ⋅ 1 0
T
(8.48)
0 2⁄L mL
2 0 2⁄L
0 -----------
8
3m
-------- 0
M = 2 (8.49)
m
0 ----
2 The elements of the 2x2 mass matrix can be determined by
imparting a unit acceleration üa=1 to one degree of freedom
which yields the equation of motion of the 2-DoF systems ex-
while keeping the acceleration of the other degree of free-
pressed in terms of the DoFs u 1 and u 2
dom equal to zero (üb=0).
3m The resulting inertia forces are then applied as static forces
-------- 0 u··1 EI u
2 acting onto the system, and the elements of the mass matrix
⋅ + -----3- ⋅ 12 – 12 ⋅ 1 = 0 (8.50)
m u··2 L – 12 16 u2 0 are computed as the reactions to these static forces.
0 ----
2
In the example at hand, if the DoFs u 1 and u 2 are considered,

μL u·· EI- 28 – 10 u1
the elements of the mass matrix can be easily computed as fol- ------- ⋅ 2 1 ⋅ 1 + ----- ⋅ ⋅ = 0 (8.57)
6 ·· 3
lows: 1 2 u2 L – 10 4 u2 0
2
m 11 = --- μL (8.51)
3
1
m 21 = --- μL (8.52)
3
1
m 12 = --- μL (8.53)
3
2
m 22 = --- μL (8.54)
3
Hence the mass matrix becomes:
μL
M = ------- ⋅ 2 1 (8.55)
6 1 2
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
K = -----3- ⋅ 28 – 10 (8.56)
L – 10 4
and the equation of motion of the system for free vibration be-
comes:
9 Free Vibrations Also in the case of Equation (9.7), there is always the trivial so-
lution φ n = 0 , which corresponds to an absence of movement.
9.1 Natural vibrations If the matrix A has an inverse A –1 , then Equation (9.7) can be
rearranged as follows:
·· + Ku = 0
Mu (9.1)
–1 –1
Ansatz: A Aφ n = A 0 (9.8)
u ( t ) = q n ( t )φ
φ n where q n ( t ) = A n cos ( ω n t ) + B n sin ( ω n t ) (9.2) and therefore
The double derivation of Equation (9.2) yields: φn = 0 (9.9)

2 2
q··n ( t ) = – ω n [ A n cos ( ω n t ) + B n sin ( ω n t ) ] = – ω n q n ( t ) (9.3) 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).
··( t ) = – ω 2 q ( t ) φ
u (9.4)
n n n
The inverse of Matrix A has the form:
and by substituting Equations (9.2) and (9.4) in (9.1) we obtain:
–1 1 ˆ
2
A = ------- A (9.10)
[– ω n Mφ n + Kφ n ]q n ( t ) = 0 (9.5) A
Equation (9.5) is satisfied if q n ( t ) = 0 , which is a trivial solution If the determinant A is equal to zero, then the matrix is singular
meaning that there is no movement, because u ( t ) = q n ( t )φφ n = 0 . and has no inverse.
To obtain a nontrivial solution the term in brackets in Equation Therefore, Equation (9.6) has a nontrivial solution only if:
(9.5) must be equal to zero, i.e.:
2
2 – ωn M + K = 0 (9.11)
[– ωn M + K ]φ
φn = 0 (9.6)
2
The determinant yields a polynomial of order N in ω n which is
or:
called characteristic equation. The N roots of the characteris-
Aφ n = 0 with A = – ω n M + K
2
(9.7) tic equation are called eigenvalues and allow the calculation of
the N natural circular frequencies ω n of the system.

As soon as the natural circular frequencies ω n are computed, 9.2 Example: 2-DoF system
also the vectors φ n can be computed within a multiplicative con-
stant by means of Equation (9.6). There are N independent Vec-
tors which are called eigenvectors or natural modes of vibra-
tion of the system. We consider a regular 2-DoF
oscillator with
Summary m1 = m2 = m
• A MDoF system with N degrees of freedom has N circular fre- and

quencies ω n ( n = 1, 2, 3, …, N ) and N eigenvectors. Each ei- k1 = k2 = k
genvector has N elements. The circular frequencies are
arranged in ascending order, i.e.: ω 1 < ω 2 < … < ω n .
• Natural circular frequencies and eigenvectors are properties of The equation of motion of the system corresponds to equation
the MDoF system and depends only from its mass and stiff- (8.7):
ness properties.
u·· u
• The index n refers to the numbering of the eigenvectors and m 1 0 1 + k 2 –1 1 = 0 (9.12)
0 1 u··2 –1 1 u2 0
the first mode of vibration ( n = 1 ) is commonly referred to as
the fundamental mode of vibration.
9.2.1 Eigenvalues
The eigenvalues are calculated from the determinant:
2
2 2k – ω n m –k
K – ωn M = = 0 (9.13)
2
–k k –ωn m
which gives a quadratic equation in ω 2n

2 2 2 4 2 2
( 2k – ω n m ) ⋅ ( k – ω n m ) – ( – k ) ⋅ ( – k ) = m ω n – 3kmω n + k = 0 (9.14)

and both solutions yield the following eigenvalues: ( – 1 + 5 )k ( 1 + 5 )

– kφ 11 + ----------------------------- § --------------------- φ 11· = 0 (9.19)
2 © 2 ¹
2 2 2 2
2 3km ± 9k m – 4k m 3± 5 k
ω n = -----------------------------------------------------------
2
- = ---------------- ⋅ ---- (9.15) – kφ 11 + kφ 11 = 0
2m 2 m
φ 11 = φ 11
For each eigenvalue ω 2n we can now compute an eigenvector
and a natural circular frequency.
As expected, the eigenvector is determined within a multiplica-
9.2.2 Fundamental mode of vibration tive constant, and can therefore be arbitrarily normalized as fol-
lows:
3– 5 k
With the smallest eigenvalue ω 21 = ---------------- ⋅ ---- we obtain the
2 m • so that the largest element of the eigenvector is equal to 1
3– 5 k k • so that one particular element of the eigenvector is equal to 1
1. circular frequency ω 1 = ---------------- ⋅ ---- = 0.618 ---
- (9.16)
2 m m • so that the norm of the eigenvector is equal to 1
2 • ...
By substituting this eigenvalue ω 1 into the system of equations
Fundamental mode:
2k – § ---------------- ⋅ ----· m
3– 5 k
–k φ 11
© 2 m¹
= 0
2
[ K – ω 1 M ]φ
φ1 = ⋅
φ 21 0 3---------------
– 5- --- k k
ω1 = ⋅ - = 0.618 ----
k – § ---------------- ⋅ ----· m
3– 5 k
–k 2 m m
© 2 m¹
(9.17)
“Mode” φ 11 2 -
---------------
we obtain two independent equations that can be used to deter- φ1 = = 1 + 5 = 0.618
φ 21 1
mine the elements of the first eigenvector φ 1 . The first row of the “Degree of freedom” 1
system yield the equation:
(------------------------
1 + 5 )k (1 + 5)
φ 11 – kφ 21 = 0 and φ 21 = --------------------- φ 11 (9.18)
2 2
and by substituting this into the second row we obtain:

9.2.3 Higher modes of vibration 9.2.4 Free vibrations of the 2-DoF system
Additionally to the fundamental mode of vibration, the consid- According to Equation (9.2), the free vibration of the 2-DoF sys-
ered 2-DoF system has a second mode of vibration. tem is:
The properties of this second mode of vibration can be comput- u = [ C 1 cos ( ω 1 t ) + C 2 sin ( ω 1 t ) ]φ
φ 1 + [ C 3 cos ( ω 2 t ) + C 4 sin ( ω 2 t ) ]φ
φ2
ed in analogy to the fundamental mode and the following results (9.20)
are obtained:
u1 φ 11
= [ C 1 cos ( ω 1 t ) + C 2 sin ( ω 1 t ) ] (9.21)
Second mode u2 φ 21
φ 12
3---------------
+ 5- --- k k- + [ C 3 cos ( ω 2 t ) + C 4 sin ( ω 2 t ) ]
ω2 = ⋅ - = 1.618 --- φ 22
2 m m
The still unknown constants C 1 to C 4 can be computed using the
“Mode” φ 12 1
initial conditions given by Equation (9.24) and become:
φ2 = = 1– 5 = 1
φ 22 ---------------- – 0.618
“Degree of freedom” 2 φ 22 u1 – φ 12 u2 φ 22 v1 – φ 12 v2
C 1 = -------------------------------------- , C 2 = ------------------------------------------------- (9.22)
φ 11 φ 22 – φ 21 φ 12 ( φ 11 φ 22 – φ 21 φ 12 )ω 1
φ 11 u2 – φ 21 u1 φ 11 v2 – φ 21 v1
C 3 = -------------------------------------- , C 4 = ------------------------------------------------- (9.23)
φ 11 φ 22 – φ 21 φ 12 ( φ 11 φ 22 – φ 21 φ 12 )ω 2
u1 ( 0 ) = u1
°
° u2 ( 0 ) = u2
Initial conditions: ® · (9.24)
° u1 ( 0 ) = v1
°·
¯u (0) 2 = v2
For an alternative methodology to compute the constants C 1 to

C 4 see Section 9.6.

• Case 1: u1 = 0.618 , u2 = 1.000 , v1 = v2 = 0 • Case 2: u1 = 1.000 , u2 = – 0.618 , v1 = v2 = 0

1.2 1.2
First mode First mode
1.0 1.0 Second mode
Second mode
0.8 Total displacement 0.8 Total displacement
splacement u2 [m]
splacement u2 [m]
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
-0.2 -0.2
-0.4 -0.4
Dis
Dis
-0.6 -0.6
-0.8 -0.8
-1.0 -1.0
-1.2 -1.2
0 5 10 15 20 0 5 10 15 20
Time [s] Time [s]
1.2 1.2
1.0 Second mode 1.0 Second mode
0.8 Total displacement 0.8 Total displacement
splacement u1 [m]
splacement u1 [m]
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
-0.2 -0.2
-0.4 -0.4
Dis
Dis
-0.6 -0.6
-0.8 -0.8
-1.0 -1.0
-1.2 -1.2
0 5 10 15 20 0 5 10 15 20
Time [s] Time [s]

• Case 3: u1 = 0.618 , u2 = 0.000 , v1 = v2 = 0 9.3 Modal matrix and Spectral matrix

0.8
First mode All N eigenvalues and all N eigenvectors can be compactly rep-
0.6 Second mode resented in matricial form:
Total displacement
splacement u2 [m]
0.4 • Modal matrix

0.2
φ 11 φ 12 … φ 1N
0.0
φ 21 φ 22 … φ 2N
Φ = [ φ jn ] = (9.25)
-0.2
… … … …
Dis
-0.4 φ N1 φ N2 … φ NN
-0.6
• Spectral matrix
-0.8
0 5 10 15 20
ω 12 0 … 0
Time [s]
2 0 ω 22 … 0
Ω = (9.26)
0.8
First mode … … … …
0.6 Second mode
0 0 … ω N2
Total displacement
splacement u1 [m]
0.4
Equation (9.6) can therefore be rearranged as follows:
0.2
Kφ n = Mφ n ω n2 (9.27)
0.0
-0.2
and it is immediately apparent that the equation for all eigenval-
ues and all eigenvectors can be expressed in terms of modal
Dis
-0.4 and the spectral matrices, as follows:

-0.6 2
KΦ = MΦΩ (9.28)
-0.8
0 5 10 15 20
Time [s]

9.4 Properties of the eigenvectors vectors are linearly independent and can be chosen so that they
are orthogonal (proof complicated).
9.4.1 Orthogonality of eigenvectors So far we have shown that φ Tn Mφ r = 0 for n ≠ r . By means of
The orthogonality conditions of the eigenvectors are: Equation (9.32) we can prove also the φ Tn Kφ r = 0 for n ≠ r . We
have already seen that for n ≠ r the right hand side of Equation
T T
φ n Kφ r = 0 and φ n Mφ r = 0 for n ≠ r (9.29) (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 veri-
and can be proven by means of Equation (9.27). Equation (9.27) fication.
is first to be set up for the eigenvector vector n , and then pre-
T
multiplied with φ r on both sides: Example: 2-DoF system
T T
φ r Kφ n = ω n2 φ r Mφ n (9.30) In the following the orthogonality of the eigenvectors of the 2-
Afterwards, Equation (9.30) shall be transposed making use of DoF system presented in Section 9.2 is checked:
T T
the symmetry properties of the matrices K = K and M = M : • Relative to the mass matrix
T T
φ n Kφ r = ω n2 φ n Mφ r (9.31) 2
T 2 - m 0 ⋅ ---------------- = 2m (5 + 5)
φ 1 Mφ 1 = --------------- 1 ⋅ 1+ 5 ----------------------------- ≅ 1.382m
Now, Equation (9.27) shall be set up for the eigenvector vector 1+ 5 0 m (1 + 5)
2
T
r , and then pre-multiplied with φ n on both sides: 1
(9.34)
T T
φ n Kφ r = ω r2 φ n Mφ r (9.32)
1
Equation (9.32) can now be subtracted from Equation (9.31) T 2 -
φ 1 Mφ 2 = --------------- 1 ⋅
m 0 ⋅ (9.35)
yielding the following equation: 0 m 1– 5 = 0
1+ 5 ----------------
2
T
( ω n2 – ω r2 )φ
φ n Mφ r = 0 (9.33)
2
In the case that the eigenvalues are different, then for n ≠ r we T – 5- ⋅ m 0 ⋅ ---------------- = 0
φ 2 Mφ 1 = 1 1--------------- (9.36)
T 1+ 5
have ( ω n2 – ω r2 ) ≠ 0 and the expression φ n Mφ r must be zero. In 2 0 m
1
the case that an eigenvalue occurs more than once, the eigen-

9.4.2 Linear independence of the eigenvectors

1
T – 5- ⋅ m 0 ⋅ m
φ 2 Mφ 2 = 1 1--------------- 1 – 5 = ---- ( 5 – 5 ) ≈ 1.382m The eigenvectors are linearly independent. To prove this, it
2 0 m ---------------- 2
2 needs to be shown that if
(9.37) α1 φ1 + α2 φ2 + … + αn φn = 0 (9.42)
• Relative to the stiffness matrix
then all scalars α i must be equal to zero.
T
2 To this purpose, we left-multiply Equation (9.42) by φ i M and we
T 2 2k – k ---------------- 2k ( 5 – 5 )
φ 1 Kφ 1 = ---------------- 1 ⋅ ⋅ 1 + 5 = --------------------------
2
- ≅ 0.528k obtain:
1+ 5 –k k (1 + 5)
1 T T
φ i M ( α 1 φ 1 + α 2 φ 2 + … + α n φ n ) = φ i Mφ i α i = 0 (9.43)
(9.38)
T
In Section 9.4.1 we have shown that φ i Mφ i ≠ 0 , therefore
T 2 2k – k
1 α i = 0 meaning that the eigenvectors are linearly independent.
φ 1 Kφ 2 = ---------------- 1 ⋅ ⋅ 1– 5 = 0 (9.39)
1+ 5 –k k ---------------- The property that the eigenvectors are linearly independent,
2
is very important because it allows to represent any dis-
placement vector as a linear combination of the eigenvec-
2
T 1 – 5 ⋅ 2k – k ⋅ ---------------- = 0
φ 2 Kφ 1 = 1 --------------- (9.40) tors.
- 1+ 5
2 –k k
1
1
T 1 – 5- ⋅ 2k – k ⋅ k
φ 2 Kφ 2 = 1 --------------- 1 – 5 = --- ( 5 + 5 ) ≈ 3.618k
2 –k k ---------------- 2
2
(9.41)

9.5 Decoupling of the equation of motion Modal mass:

T
m n* = φ n Mφ n (9.49)
The equation of motion for free vibrations is T
Modal stiffness: k n* = φ n Kφ n (9.50)
·· + Ku = 0
Mu (9.44)
and Equation (9.48) can be rewritten as follows:
and as a possible solution the displacement vector
m n* q··n ( t ) + k n* q n ( t ) = 0 (9.51)
u(t) = ¦ qi ( t )φφ i (9.45)
i
For each n we can set up such an equation, which yields to N
decoupled Single Degree of Freedom systems. The total dis-
can be assumed, where: placement of the system can then be computes as the sum of
φi : linearly independent eigenvectors of the system the contribution of all decoupled SDoF systems, i.e.:
qi : modal coordinates N
u(t) = ¦ q i ( t )φ
φi (9.52)
The displacement vector u ( t ) and its double derivative i=1
··( t ) =
u ¦ q··i ( t )φφ i (9.46) in matricial form:
i
q1 ( t )
can be substituted into Equation (9.44), and the latter can be left-
T u ( t ) = Φ q ( t ) with q = … (9.53)
multiplied by φ n yielding the following equation:
qN ( t )
φ n M § ¦ q··i ( t )φ
φ i· + φ n K § ¦ q i ( t )φ
φ i· = 0
T T
© ¹ © ¹
(9.47) ·· + K * q = 0
i i
M*q (9.54)
Because of the orthogonality properties of the eigenvectors only with

one term of the summations remains, i.e.:
m 1* … 0 k 1* … 0
T T T T
φ n Mφ n q··n ( t ) + φ n Kφ n q n ( t ) = 0 (9.48) M * = Φ MΦ = … … … and K * = Φ KΦ = … … …
0 … m n* 0 … k n*
where:
(9.55)

In Equations (9.52) and (9.53) are rewritten Example: 2-DoF system

N
The modal masses and modal stiffness of the 2-DoF system of
u(t) = ¦ q i ( t )φ
φ i = Φq ( t ) (9.56) Section 9.2 were already checked during the verification of the
i=1 orthogonality of the eigenvectors (See Equations (9.34), (9.37),
the LHS and the RHS of the resulting equation can be pre-mul- (9.38) and (9.41)). They are:
T
tiplied by φ n M and we obtain:
T 2m ( 5 + 5 )
m 1* = φ 1 Mφ 1 = -----------------------------
2
≈ 1.382m (9.61)
N
(1 – 5)
T T
φ n Mu ( t ) = ¦ φn Mφi qi ( t ) (9.57)
T m
i=1
m 2* = φ 2 Mφ 2 = ---- ( 5 – 5 ) ≈ 1.382m (9.62)
2
Because of the orthogonality of the eigenvectors (see Section
T 2k ( 5 – 5 -)
9.4.1), Equation can be further simplified to: k 1* = φ 1 Kφ 1 = --------------------------
2
≅ 0.528k (9.63)
(1 – 5)
T T
φ n Mu ( t ) = φ n Mφ n q n ( t ) (9.58) T k
k 2* = φ 2 Kφ 2 = --- ( 5 + 5 ) ≈ 3.618k (9.64)
2
which yields the following relationship between q n ( t ) and u ( t )
• First modal SDoF system:
T
φ n Mu ( t )
q n ( t ) = ----------------------
T
- (9.59) m 1* q··1 ( t ) + k 1* q 1 ( t ) = 0 (9.65)
φ n Mφ n
1.382mq··1 ( t ) + 0.528kq 1 ( t ) = 0 (9.66)
or introducing the definition of the modal mass given by Equation
(9.49) we obtain the equivalent expression k* 0.528k- = 0.618 ---
k-
ω1 = ------1- = ----------------- (9.67)
T m 1* 1.382m m
φ n Mu ( t )
qn ( t ) = ----------------------- (9.60) The natural frequency corresponds to equation (9.16) of this
m n*
chapter
Equations (9.59) and (9.60) will be later used to compute the re-
sponse of MDoF systems.

• Second modal SDoF system: 9.6 Free vibration response

m 2* q··2 ( t ) + k 2* q 2 ( t ) = 0 (9.68)
9.6.1 Systems without damping
1.382mq··2 ( t ) + 3.618kq 2 ( t ) = 0 (9.69)
The equation of motion for free vibration of a system without
damping is
k* 3.618k- = 1.618 ---
k-
ω2 = ------2- = ----------------- (9.70) ·· + Ku = 0
m 2* 1.382m m Mu (9.71)
The natural frequency corresponds to the result shown on page and making use of the possibility of decoupling of the equation
9-7. of motion, the total deformation u ( t ) under free vibration can be
computed as the sum of the contribution of all modes. The equa-
tion of motion of the nth decoupled SDoF system is:
m n* q··n ( t ) + k n* q n ( t ) = 0 (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 ( ω n t ) + B n sin ( ω n t ) (9.73)
The the total deformation u ( t ) under free vibration is hence

N N
u(t) = ¦ φi qi ( t ) = ¦ φ i [ A i cos ( ω i t ) + B i sin ( ω i t ) ] (9.74)
i=1 i=1
The 2 ⋅ N constants A i and B i can be computed by means of the

initial conditions u ( 0 ) = u 0 and u· ( 0 ) = v 0 .
To this purpose, the vector of the velocity is needed and can
easily be computed by deriving Equation (9.74), i.e.:

N N 9.6.2 Classically damped systems

u· ( t ) = ¦ φ i q· i ( t ) = ¦ φ i ω i [ – A i sin ( ω i t ) + B i cos ( ω i t ) ] (9.75)
i=1 i=1
The equation of motion for free vibration of a system with damp-
ing is
Considering Equations (9.74) and (9.75) at the time t = 0 , we
have. ·· + Cu· + Ku = 0
Mu (9.80)
N N As it will be shown in Chapter 10, if the MDoF system is classi-
u(0) = ¦ φ i q i ( 0 ) and u· ( 0 ) = ¦ φ i q· i ( 0 ) (9.76) cally damped, the equation of motion can be decoupled analo-
i=1 i=1 gously to system without damping, and the total deformation u ( t )
under free vibration can be computed again as the sum of the
Making use of Equation (9.59) we can now write the equations
contribution of all modes. The equation of motion of the nth de-
to compute the initial conditions of the nth decoupled SDoF sys-
coupled SDoF system is:
tem as:
T m n* q··n ( t ) + c n* q· n ( t ) + k n* q n ( t ) = 0 (9.81)
φ n Mu ( 0 )
qn ( 0 ) = ------------------------
T
(9.77) or
φ n Mφ n
2
T q··n ( t ) + 2ω n ζ n q n ( t ) + ω n q n ( t ) = 0 (9.82)
φ n Mu· ( 0 )
q· n ( 0 ) = ------------------------
T
(9.78) where
φ n Mφ n
c n*
In Section 3.1.2 (see Equation 3.18) it as been shown that the c n* =
T
φ n Cφ n and ζ n = ----------------- , respectively. (9.83)
constants A n and B n are equal to q n ( 0 ) and q· n ( 0 ) ⁄ ω n , respec- 2m n* ω n
tively, hence Equation (9.74) can be rewritten as:
The solution of Equation (9.82) can be computed as discussed
N
q· i ( 0 ) in Chapter 3 for SDoF systems. According to Equation (3.50) we
u(t) = ¦ φ i q i ( 0 ) cos ( ω i t ) + ------------ sin ( ω i t ) (9.79) have:
ωi
i=1
– ζω n t
qn ( t ) = e [ A n cos ( ω nd t ) + B n sin ( ω nd t ) ] (9.84)
where:

2
ω nd = ω n 1 – ζ “damped circular frequency of the nth mode” Blank Page
(9.85)
The the total deformation u ( t ) under free vibration is hence
N N
– ζω n t
u(t) = ¦ φi qi ( t ) = ¦ φi e [ A i cos ( ω id t ) + B i sin ( ω id t ) ] (9.86)
i=1 i=1
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
u ( 0 ) = u 0 and u· ( 0 ) = v 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 in-
itial conditions of the modal coordinate q as follows:
An = qn ( 0 ) (9.87)
q· n ( 0 ) + ζω n q n ( 0 )
B n = -------------------------------------------- (9.88)
ω nd
where the initial displacement q n ( 0 ) and the initial velocity q· n ( 0 )

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
– ζω n t q· i ( 0 ) + ζω i q i ( 0 )
u(t) = ¦ φi e q i ( 0 ) cos ( ω id t ) + ----------------------------------------- sin ( ω id t ) (9.89)
ω id
i=1
For nonclassically damped system see e.g. [Cho11], Chapter 14.

10 Damping 10.2 Example
10.1 Free vibrations with damping

The differential equation to compute the free vibrations of a
MDoF system is:
·· + Cu· + Ku = 0
Mu (10.1)
with the initial conditions:
The properties of the 2-DoF system are:
u ( 0 ) = u0 and u· ( 0 ) = v 0 (10.2)
m 1 = 2m , m2 = m (10.6)
The displacement vector u ( t ) may be expressed as linear com-
bination of the eigenvectors, i.e. u ( t ) = Φ q ( t ) , and Equation k 1 = 2k , k2 = k (10.7)
(10.1) becomes:
while the damping characteristics will be defined later.
·· + CΦq· + KΦq = 0
MΦq (10.3)
• Natural frequencies and eigenvectors
Equation (10.3) may be further multiplied by Φ T yielding the fol-
lowing equations: The natural frequencies and the eigenvectors of the 2-DoF sys-
tem can be easily computed as:
T ·· + Φ T CΦq· + Φ T KΦq = 0
Φ MΦq (10.4)
k- , 2k
·· + C * q· + K * q = 0
M*q (10.5) Natural frequencies: ω1 = ------- ω2 = ------ (10.8)
2m m
Definition:
Eigenvectors: φ1 = 1 ⁄ 2 , φ2 = –1 (10.9)
1 1
• A system is classically damped if the matrix C * is diagonal
• A system is non-classically damped if the matrix C * is not di-
agonal
10 Damping Page 10-1 10 Damping Page 10-2

10.2.1 Non-classical damping 10.2.2 Classical damping

The damping characteristics of the 2-DoF system are chosen as: The damping characteristics of the 2-DoF system are chosen as:
c1 = c , c 2 = 4c (10.10) c 1 = 4c , c 2 = 2c (10.15)
The equation of motion of the system can be easily assembled The equation of motion of the system can be easily assembled
by means of the equilibrium formulation: by means of the equilibrium formulation:
u·· u· u
m 2 0 1 + c 5 –4 1 + 3 –1 1 = 0 (10.11) u·· u· 1 u
0 1 u··2 – 4 4 u· 2 –1 1 u2 0 m 2 0 1 + c 6 –2 + 3 –1 1 = 0 (10.16)
0 1 u··2 –2 2 ·u
2 –1 1 u2 0
It is now attempted to decouple the equations by computing the
It is now attempted to decouple the equations by computing the
modal properties of the 2-DoF system:
modal properties of the 2-DoF system:
1--- 1--- 3---
1 –1 m 0 3--- 3k
M*
T
= Φ MΦ = 2 ⋅ 2m 0 ⋅ 2 = 2 (10.12) m 0 ------ 0
–1 1
0 m
1 1 0 3m M* = 2 , K* = 4 (10.17)
0 3m 0 6k
1--- 1--- 3---

1 –1 k 0
K * = Φ KΦ = 2
T
⋅ 3k – k ⋅ 2 = 4 (10.13) 1--- 1--- 3---
1 –1 c 0
C * = Φ CΦ = 2 ⋅ 6c – 2c ⋅ 2 2
–k k T
–1 1 1 1 0 6k = (10.18)
– 2c 2c
–1 1 1 1 0 12c
1--- 1--- 5--- 7---

1 –1 c c The Matrix C * is diagonal, hence it is possible to decouple the
T
C * = Φ CΦ = 2 ⋅ 5c – 4c ⋅ 2 = 4 2 (10.14)
– 4c 4c 7--- equations!
–1 1 1 1 c 17c
2
The Matrix C * is not diagonal, hence it is not possible to decou-

ple the equations!

10.3 Classical damping matrices 10.3.3 Rayleigh damping

C = a0 M + a1 K (10.25)
10.3.1 Mass proportional damping (MpD)
The damping constant of each mode of vibration is therefore:
C = a0 M (10.19)
2
The damping constant of each mode of vibration is therefore: c n* = a 0 m n* + a 1 k n* = ( a 0 + a 1 ω n )m n* (10.26)
c n* = a 0 m n* (10.20) and using the results for MpD and SpD damping ratio ζ n be-
comes:
and the corresponding damping ratio ζ n becomes (see Section 3.2):
a0 a1
ζ n = --------- + ----- ω n (10.27)
c n* a 0 m n* a0 2ω n 2
ζ n = ----------------*- = ----------------*- = --------- (10.21)
2ω n m n 2ω n m n 2ω n
The coefficients a 0 and a 1 may be computed for vibration modes
i and j by means of equation (10.28):
10.3.2 Stiffness proportional damping (SpD)
C = a1 K (10.22) a0 1 a1
° ----- ⋅ ----- + ----- ⋅ ω i = ζ i
° 2 ωi 2
The damping constant of each mode of vibration is therefore: ® (10.28)
° ----
a0 1 a1
2 ° 2- ⋅ ω ----- + ----- ⋅ ω j = ζ j
2
c n* = a 1 k n* = a 1 ω n m n* (10.23) ¯ j
and the corresponding damping ratio ζ n becomes: In the case that ζ i = ζ j = ζ , coefficients a 0 and a 1 can be com-
2 puted as follows:
c n* a 1 ω n m n* a1
ζ n = ----------------*- = ------------------- = ----- ω n (10.24)
2ω n m n 2ω n m n * 2 2ω i ω j 2
a 0 = ζ ⋅ ----------------- a 1 = ζ ⋅ ----------------- (10.29)
ωi + ωj ωi + ωj
Remark
Both MpD and SpD, taken alone, are not a good approximation
of the behaviour of real structures. Studies have shown that dif-
ferent modes of vibration exhibit similar damping ratios.

10.3.4 Example If we choose values for m , k and ζ so that

A damping matrix shall be assembled so that in the case of the ω 1 = 2rad ⁄ s , ω 2 = 5rad ⁄ s , ζ = 5% (10.34)
2-DoF system shown on page 10-2 both modes of vibration are
the coefficients a 0 and a 1 become:
characterised by the same damping ratio ζ .
a 0 = 14.287 , a 1 = 1.429 (10.35)
The natural frequencies are:
and the representation of the damping ratio in function of the nat-
k 2k
ω1 = -------- , ω2 = ------ (10.30) ural circular frequency is:
2m m
10
hence the coefficients a 0 and a 1 become: Mass proportional damping
9
Stiffness proportional damping
4ζ m k 4ζ m 8
a 0 = ------ ⋅ ------ ⋅ ---- , a 1 = ------ ⋅ ------ (10.31) Rayleigh Damping
mping ratio ζn [%]

3 2k m 3 2k 7
6
yielding the damping matrix C = a 0 M + a 1 K equal to:
5
k- 4
--- ⋅ 2m + 3k 0–k
Dam
4ζ m 4ζ mk
C = ------ ⋅ ------ ⋅ m = ------ ⋅ -------- ⋅ 5 – 1 3
3 2k k- 3 2 –1 2
0–k --- ⋅m+k 2
m
1
(10.32)
0
0 1 2 3 4 5 6 7 8 9 10
Check:
Circular frequency ω
1
--- 1 4ζ
1
--- – 1
Remarks
mk 5 – 1 mk
= ζ ⋅ -------- ⋅ 3 0
T
C * = Φ CΦ = 2 ⋅ -----
- ⋅ -------
- ⋅ ⋅ 2
3 2 –1 2 2 0 12 - If there are more than 2 modes of vibrations, then not all of
–1 1 1 1
them will have the same damping ratio.
(10.33)
- If more than 2 modes of vibrations should have the same
The damping matrix is indeed diagonal. damping, then a different damping modal shall be used. To
this purpose see e.g. “Caughey-Damping” in [Cho11].

CCourse “Fundamentals of Structural Dynamics” An-Najah 2013 CCourse “Fundamentals of Structural Dynamics” An-Najah 2013
11 Forced Vibrations T ·· + Φ T KΦq = Φ T F ( t )

Φ MΦq (11.4)
·· + K * q = F * ( t )
M*q (11.5)
11.1 Forced vibrations without damping
Where
11.1.1 Introduction - 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 rear-
ranged as:
* ·· * *
m1 q1 + k1 q1 = F1
® * ·· (11.6)
Sought is the response of the 2-DoF system as a result of the ex- ¯ m 2 q 2 + k 2* q 2 = F 2*
ternal excitation force F ( t ) given by Equation (11.1)
or as alternative:
F1 ( t )
F(t) = (11.1) ·· 2 F 1*
F2 ( t ) ° q 1 + ω 1 q 1 = ------*-
° m1
® (11.7)
The equation of motion of the system is: ° ·· 2 F 2*
° 2
q + ω q
2 2 = ------ -
··( t ) + Ku ( t ) = F ( t ) ¯ m 2*
Mu (11.2)
The displacement vector u ( t ) can be represented as a linear The two equations of the system (11.7) are decoupled and can
combination of the eigenvectors of the 2-DoF system, be solved independently. The constants resulting from the solu-
u ( t ) = Φ q ( t ) , and Equation (11.2) becomes: tion of the system can be determined by means of the initial con-
ditions u ( 0 ) = u 0 and u· ( 0 ) = v 0 .
·· + KΦq = F ( t )
MΦq (11.3)
T
We can now multiply Equation (11.3) by Φ obtaining:

11.1.2 Example 1: 2-DoF system F 0 sin ( ωt ) F 0 sin ( ωt )

° q··1 + ω 21 q 1 = -------------------------- - = f 1 sin ( ωt )
- = ------------------------
° 2 ⋅ ( 3 ⁄ 2 )m 3m
® (11.13)
° ·· 2 – F 0 sin ( ωt ) – F 0 sin ( ωt )
° q 2 + ω 2 q 2 = ---------------------------
3m
- = f 2 sin ( ωt )
- = ---------------------------
3m
¯
with
F0 F0
f 1 = -------- and f 2 = – -------- (11.14)
3m 3m
The properties of the 2-DoF system are:
Each equation of the system (11.13) corresponds to the equation
m 1 = 2m , m2 = m (11.8) of motion of an undamped SDoF system under an harmonic sine
k 1 = 2k , k2 = k (11.9) excitation. The complete solution of these differential equations
has been discussed in Chapter 4 and it is:
c1 = 0 , c2 = 0 (11.10)
fn
2
- sin ( ωt )
q n = A 1 cos ( ω n t ) + A 2 sin ( ω n t ) + ------------------
2
(11.15)
The external excitation is: ωn – ω
F 0 sin ( ωt ) The two equations have the following solutions:

F(t) = (11.11)
0 f1
q = A cos ( ω t ) + A sin ( ω t ) + ------------------ - sin ( ωt )
° 1 1 1 2 1 2 2
and the modal excitation force is calculated using the modal ma- ° ω 1 – ω
trix: ® (11.16)
° f2
2
- sin ( ωt )
° q 2 = A 3 cos ( ω 2 t ) + A 4 sin ( ω 2 t ) + ------------------
2
¯ ω2 – ω
1 F 0 sin ( ωt )
--- 1
T 2 F 0 sin ( ωt ) ------------------------
-
F*( t ) = Φ F( t ) = = 2 (11.12) The 4 constants A 1 to A 4 can be easily computed for the initial
0 – F 0 sin ( ωt )
–1 1 conditions u ( 0 ) = u· ( 0 ) = 0 by means of the mathematical soft-
ware “Maple”. They are:
The system of equations becomes:

A1 = 0 (11.17) u(t) = Φq(t) = ¦ φn qn ( t ) = φ1 q1 ( t ) + φ2 q2 ( t ) (11.23)

n
ω ⁄ ω1 F0 ω ⁄ ω1
- ⋅ -------- = – ------------------
A 2 = – ------------------ - ⋅ f1 (11.18)
2 2 2 2
ω 1 – ω 3m ω1 – ω 1 § sin ( ωt ) – ( ω ⁄ ω 1 ) sin ( ω 1 t ) ·
---
u = 2 ¨ f 1 -----------------------------------------------------------------
2 2
- ¸+ (11.24)
A3 = 0 (11.19) © ω1 – ω ¹
1
ω ⁄ ω2 F0 ω ⁄ ω2 sin ( ωt ) – ( ω ⁄ ω 2 ) sin ( ω 1 t ) ·
– 1 § f -----------------------------------------------------------------
- ⋅ -------- = – ------------------
A 4 = ------------------ - ⋅ f2 (11.20) ¨ 2 - ¸
2 2 3m 2 2 2 2
ω2 – ω ω2 – ω 1 © ω2 – ω ¹
The displacements q n becomes:

where:
§ ω ⁄ ω1 · f1 F0 F0
° q = ¨ – ------------------ - ⋅ f 1¸ sin ( ω 1 t ) + ------------------
- sin ( ωt ) f 1 = -------- , f 2 = – -------- , ω 1 =
k
-------- , ω 2 =
2k
------ (11.25)
° 1
© ω1 – ω2 2
¹ ω
2
– ω
2 3m 3m 2m m
° 1
® (11.21)
° § ω ⁄ ω2 · f2
- ⋅ f 2¸ sin ( ω 2 t ) + ------------------
° q 2 = ¨ – ------------------ - sin ( ωt )
2 2 2 2
° © ω2 – ω ¹ ω2 – ω
¯
or
sin ( ωt ) – ( ω ⁄ ω 1 ) sin ( ω 1 t )
°q = f -----------------------------------------------------------------
-
° 1 1 2
ω1 – ω
2
°
® (11.22)
° sin ( ωt ) – ( ω ⁄ ω 2 ) sin ( ω 2 t )
° q2 = f2 -----------------------------------------------------------------
2 2
-
° ω2 – ω
¯
Therefore, the total displacement u ( t ) becomes:

11.1.3 Example 2: RC beam with Tuned Mass Damper (TMD) • RC beam with TMD
without damping
• RC Beam
• Damping ratio
ζ n = 0.0
• Modal mass
M n = 5.626t
• Modal stiffness
K n = 886kN ⁄ m
• Natural frequency
f n = 2Hz
• Excitation
• TMD (In this case damping is neglected)
As excitation a vertical harmonic sine force acting only on the
• Damping ratio
ζ T = 0.0 beam is assumed.
• Mass F ( t ) = F o sin ( ωt ) (11.26)
M T = 0.310t
with: ω : excitation frequency
• Stiffness
F o : static excitation force: F o = 0.8kN
K T = 44kN ⁄ m
• Natural frequency
• Solution
f T = 1.90Hz Both the transient and the steady-state part of the solution are
considered.

• Case 1: K T = 10000kN ⁄ m , excitation frequency f = 2Hz 140

Displacement of the beam (u1)
Displacement of the TMD (u2)
0.04 120
plification factor A [-]

Displacement of the beam
0.03 100
placement u1(t) [m]
0.02 80
0.01
60
0
Amp
40
-0.01
Disp
20
-0.02
0
-0.03
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
Excitation frequency / Natural frequency of the beam [-]
-0.04
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Time [s] • Remarks
0.04 - The amplification factor A is defined as:

Displacement of the TMD
0.03 A TMD = u 2 ⁄ u st , A Beam = u 1 ⁄ u st
splacement u2 [m]
0.02
where u st = F o ⁄ K n
0.01
- The solution was computed by means of the Excel file giv-
0 en on the web page of the course (SD_MDOF_TMD.xlsx)
-0.01
- The Tuned Mass Damper (TMD) is blocked
Dis
-0.02
- The natural frequency of the beam with TMD is: f n = 1.94Hz
-0.03
- At f = f n resonance occurs. In the diagram above the am-
-0.04 plification factor is limited, because the response of the
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Time [s] system was only calculated during 60 seconds.

• Case 2: K T = 44kN ⁄ m , excitation frequency f = 2Hz 350

Displacement of the beam (u1)
Displacement of the TMD (u2)
0.008 300
plification factor A [-]

Displacement of the beam
0.006 250
placement u1(t) [m]
0.004 200
0.002
150
0
Amp
100
-0.002
Disp
50
-0.004
0
-0.006
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
Excitation frequency / Natural frequency of the beam [-]
-0.008
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Time [s] • Remarks
0.04 - The amplification factor A is defined as:

Displacement of the TMD
0.03 A TMD = u 2 ⁄ u st , A Beam = u 1 ⁄ u st
splacement u2 [m]
0.02
where u st = F o ⁄ K n
0.01
- The solution was computed by means of the Excel file giv-
0 en on the web page of the course (SD_MDOF_TMD.xlsx)
-0.01
- The Tuned Mass Damper (TMD) is free to move
Dis
-0.02
- No resonance at f = f n occurs. Resonance occurs in corre-
-0.03 spondence of the first and of the second natural frequen-
-0.04
cies of the 2-DoF system.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 In the diagram above the factor A is limited, because the
Time [s]
response of the system was only calculated during 60s.

11.2 Forced vibrations with damping Where:

- M * : Diagonal matrix of the modal masses m n*
11.2.1 Introduction
- 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 ex- * ·· *· *
m1 q1 + c1 q1 + k1 q1 = F1
*
ternal excitation force F ( t ) given by Equation (11.27) ® * ·· (11.32)
¯ m 2 q 2 + c 2* q· 2 + k 2* q 2 = F 2*
F1 ( t ) or as alternative:
F(t) = (11.27)
F2 ( t )
·· · 2 F 1*
° 1
q + 2ζ ω q + ω q = ------
-
The equation of motion of the system is: °
1 1 1 1 1
m 1*
® (11.33)
··( t ) + Cu· ( t ) + Ku ( t ) = F ( t )
Mu (11.28) ° ·· F 2*
· 2
° 2
q + 2ζ ω q
2 2 2 + ω q
2 2 = ------
-
The displacement vector u ( t ) can be represented as a linear ¯ m 2*
combination of the eigenvectors of the 2-DoF system,
The two equations of the system (11.33) are decoupled and can
u ( t ) = Φ q ( t ) , and Equation (11.28) becomes:
be solved independently. The constants resulting from the solu-
·· + CΦq· + KΦq = F ( t )
MΦq (11.29) tion of the system can be determined by means of the initial con-
T ditions u ( 0 ) = u 0 and u· ( 0 ) = v 0 .
We can now multiply Equation (11.29) by Φ obtaining:
T ·· + Φ T CΦq· + Φ T KΦq = Φ T F ( t )
Φ MΦq (11.30)
·· + C * q· + K * q = F * ( t )
M*q (11.31)

11.3 Modal analysis: A summary

Blank Page
The dynamic response of a Multi-Degree of Freedom (MDoF)
system due to an external force F ( t ) can be computed by means
of modal analysis. The required steps are:
1) Compute the properties of the MDoF system
- Compute the mass matrix M and the stiffness matrix K .
- Estimate the modal damping ratios ζ n*
2) Compute the natural circular frequencies ω n and the eigen-
vectors φ 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
F n*
q··n + 2ζ n* ω n q· n + ω n2 q n = ------*- and solve it for q n
mn
- Compute the modal displacements u n ( t ) = φ n q n

- Compute the sectional forces by means of the static equiv-
alent forces F n ( t ) = Ku n ( t ) = Kφ n q n = ω n2 Mφ n q n ( t )
4) Sum up (respectively combine) the contribution from all
modes of vibration to obtain the total response of the system.

12 Seismic Excitation The system of equations governing the motion of the system is
F1 + T1 = 0
12.1 Equation of motion ® (12.1)
¯ F2 + T2 = 0
12.1.1 Introduction ·· ·· · ·
– m 1 ( x + u 1 ) – ( c 1 + c 2 )u 1 + c 2 u 2 – ( k 1 + k 2 )u 1 + k 2 u 2 = 0
In analogy to Section 2.1.1, the equation of motion of the system ® (12.2)
¯ – m 2 ( x·· + u··2 ) + c 2 u· 1 – c 2 u· 2 + k 2 u 1 – k 2 u 2 = 0
depicted here can be formulated by means of the d’Alembert
principle F + T = 0 applied to each one of the masses. and in matricial form:
y 1 = x ( t ) + l 1 + u s1 + u 1 ( t )
m1 0 x·· + u··1 c 1 + c 2 – c 2 u· 1 k 1 + k 2 – k 2 u 1
– – – = 0 (12.3)
y··1 = x··( t ) + u··1 ( t ) 0 m2 ··
x + u2 ·· –c2 c2 u2 · –k2 k2 u2 0
T 1 = – m 1 y··1 = – m 1 ( x·· + u··1 ) or:
F 1 = – k 1 ( u s1 + u 1 ) – c 1 u· 1 + m 1 g m1 0 x·· + u··1 c + c 2 – c 2 u· 1 k + k2 –k2 u1

+ k 2 ( u s2 + u 2 – u 1 ) + c 2 ( u· 2 – u· 1 ) + 1 + 1 = 0 (12.4)
0 m2 ··
x + u2 ·· –c2 c2 u2 · –k2 k2 u2 0
F 1 = – ( k 1 + k 2 )u 1 + k 2 u 2 – ( c 1 + c 2 )u· 1 + c 2 u· 2
or:
y 2 = x ( t ) + l 1 + u s1 + l 2 + u s2 + u 2 ( t )
m1 0 u··1 c + c 2 – c 2 u· 1 k + k2 –k2 u1 m 0 x··
y··2 = x··( t ) + u··2 ( t ) + 1 + 1 = – 1
0 m2 u··2 – c 2 c 2 u· 2 –k2 k2 u2 0 m2 x··
T 2 = – m 2 y··2 = – m 2 ( x·· + u··2 ) (12.5)
F 2 = – k 2 ( u s2 + u 2 – u 1 ) – c 2 ( u· 2 – u· 1 ) + m 1 g which is similar to Equation (8.3) meaning that the base point ex-
citation x ( t ) can be considered equivalent to two external forces
F 2 = k 2 u 1 – k 2 u 2 + c 2 u· 1 – c 2 u· 2 f 1 ( t ) = m 1 x··( t ) and f 2 ( t ) = m 2 x··( t ) acting on the masses m 1 and
m 2 . This is the same interpretation given in Section 2.1.1 for
SDoF systems.

12.1.2 Synchronous Ground motion Influence vector for some typical cases
As shown in the previous section, the equation of motion of a • Planar system with translational ground motion (Case 1)
system subjected to a base excitation is:
In this case all DoFs of the system undergo
·· + Cu· + Ku = 0
Mu (12.6) static displacements u s ( t ) which are equal
a
to the ground displacement u g ( t ) , hence:
where u·· is vector of the absolute accelerations of the DoFs of
a
the system while u· and u are the vectors of the relative veloci- 1
ties and of the relative displacements of the DoFs of the system,
respectively. ι = 1 = 1 (12.11)
…
The absolute displacement u a of the system can be expressed 1
as:
where 1 is a vector of order N , i.e. the
ua = us + u (12.7) number of DoFs, with all elements equal to 1.
where u s is displacement of the DoFs due to the static applica-
tion (i.e. very slow so that no inertia and damping forces are gen- • Planar system with translational ground motion (Case 2)
erated) of the ground motion, and u is again the vector of the rel- The axial flexibility of the elements
ative displacements of the DoFs of the system. of the depicted system can be ne-
glected, hence 3 DoFs are defined.
The “static displacements” u s ( t ) can now be expressed in func-
In this case DoFs 1 and 2 undergo
tion of the ground displacement u g ( t ) as follows:
static displacements which are
us ( t ) = ι ug ( t ) (12.8) equal to the ground displacement,
while the static displacement of
where ι is the so-called influence vector. Equation (12.6) can DoF 3 is equal to 0, i.e.:
now be rewritten as:
··) + Cu· + Ku = 0
M ( ιu··g + u (12.9) 1
ι = 1 (12.12)
·· + Cu· + Ku = – Mιu·· ( t )
Mu (12.10)
g 0

• Planar system with rotational ground motion. • Spatial system with multiple translational ground motion
Consider the spatial frame depicted here:
U6
U5 U4
U3
U2 U1
The depicted system is subjected to a rotational ground motion θ g
which generates the following static displacements of the DoFs:
Excitation in Excitation in
h1 h1 the y-direction
the x-direction
u s ( t ) = h θ g ( t ) hence ι = h (12.13)
2 2 ugx(t)
L L ugy(t)
Remark
Picture from: Chaudat T., Pilakoutas K., Papastergiou P., Ciupala M. A. (2006) “Shaking Table
If the planar system with rotational ground motion has more than Tests on Reinforced Concrete Retrofitted Frame With Carbon Fibre Reinforced Polymers
one support and every support is subjected only to the base ro- (CFRP),” Proceedings of the First European Conference on Earthquake Engineering and
Seismology, Geneva, Switzerland, 3-8 September 2006
tation θ g , then the static application of the base rotations typical-
ly create stresses within the system. Such a case must be con- The equation of motion of the frame structure for the ground mo-
sidered like a multiple support excitation (see Section 12.1.3). tions u gx ( t ) and u gy ( t ) neglecting damping is:

12.1.3 Multiple support ground motion

u··1 u1
u··
2 u2 Structures with a significative spatial extension may be subject-
u··3 u ed to ground motion time-histories that are different from support
M + K 3 = – Mι x u··gx ( t ) – Mι y u··gy ( t ) (12.14) to support. A typical example for such structures is the bridge
··
u4 u4
shown in the following figure.
u··
5 u 5
u··6 u6
u··1 u1
§ 0 1 ·
u·· u2 ¨ ¸
2
¨ 1 0 ¸ Example of structure where often multiple support excitation is applied: Plan view of the
u·· u ¨ 0 u·· ( t ) + 0 u·· ( t )¸ dynamic model for the seismic analysis of a bridge in the transverse direction.
M 3 +K 3 = –M ¨ gx gy ¸ (12.15) The springs represent the piers.
u··4 u4 ¨ 0 1 ¸
¨ ¸ In this case it is distinguished between the DoFs of the structure
u·· u
5 5 ¨ 1 0 ¸
© 0 0 ¹ u a , which are free to move and whose displacements are ex-
u··6 u6
pressed in absolute coordinates, and those of the ground u g ,
and with which undergo the displacements imposed to the support. The
vector containing the displacements of all DoFs is hence:
m1 u··1 u1 – m 1 u··gy
m2 0 u·· 2 u2 – m u·· ua
2 gx u = (12.17)
I3 u·· u 0 ug
M = we obtain M 3 + K 3 =
m4 ··
u4 u4 – m 4 u··gy
The equation of motion of the system can hence be expressed
0 m5 u··5 u5 – m 5 u··gx
as (see [Cho11]):
I6 u·· u 06 6
(12.16) m mg ··
u c c g u· k kg u 0
a a a
+ + = (12.18)
Remarks T
m g m gg ··
ug T ·
u
c g c gg g
T u
k g k gg g p g ( t )
• For other cases see [Cho11] Sections 9.4 to 9.6.

Where p g ( t ) are the forces resulting at the supports when the

m mg ·· + u
u ·· c c g u· s + u· k kg us + u
supports undergo the displacements u g ( t ) . s
+ + =
0
g(t)
T
m g m gg ··
ug T
c g c gg ·
ug T
k g k gg ug p
In Equation (12.18) the different matrices … g , … gg are not com-
puted separately, they just result form the partition of the overall (12.21)
system of equations when the DoFs of the structure and of the The first line of the system can be rearranged to:
ground are collected as it is shown in the example of page 12-12.
·· + cu· + ku = – ( mu
·· + m u
·· · ·
mu s g g ) – ( cu s + c g u g ) – ( ku s + k g u g ) (12.22)
The vector of Equation (12.17) can be rewritten as:
According to the first line of Equation (12.20) ku s + k g u g = 0
ua us and hence Equation (12.22) becomes:
u = = + u (12.19)
·· + cu· + ku = – ( mu
·· + m u
·· · ·
g g ) – ( cu s + c g u g )
ug ug 0 mu s (12.23)
where u s is the vector of the displacements of the DoFs of the If we now express the vector u s in function of the vector u g as
structure when the ground displacements u g ( t ) are applied stat-
u s = ιu g (12.24)
ically, and u is the vector of the relative displacements of the
DoFs of the structure. the so-called influence matrix ι can be computed, again mak-
The relationship between u s and u g ( t ) is given by the following ing use of the first line of Equation (12.20), i.e.:
system of equation: – k g u g = ku s = kιu g (12.25)
k kg us 0 and after rearranging we obtain:

= (12.20)
T p g, s –1
k g k gg u g ι = –k kg (12.26)
where p g, s is the vector of the support forces needed to impose The influence matrix ι is a N × N g matrix where N is the number
the displacements u g statically. If the system is statically deter- of DoFs of the structure and N g is the number of DoFs of the
minated, p g, s is equal to zero (See example of page 12-12). supports.
By introducing Equation (12.19) into Equation (12.18) we obtain By introducing Equation (12.26) into Equation (12.23) the final
the new system of equations: equation of motion of the system is obtained:

·· + cu· + ku = – ( mι + m )u
·· ·
g g – ( cι + c g )u g
mu (12.27) Example: 2-DoF system
Analogously, with the second line of Equation (12.21), an equa- The following 2-DoF system is subjected to multiple support
tion for the computation of the forces at the supports p g ( t ) can ground motion. Two different ground motions are applied to the
be setup and solved. degrees of freedom u 1 and u 5 . Sought is the equation of motion
of the system:
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 = – mιu
mu ·· – ( cι + c )u· (12.28)
g g g
The stiffness matrix of the system is assembled by means of the
And considering that in most cases the damping forces on the Direct Stiffness Method and the following degrees of freedom
LHS of the equation are small (and they are zero if no damping are considered:
is present) compared to the inertia forces (see [Cho11]), Equa-
tion (12.28) can be further simplified to: u3
u = : displacements of the structure (12.32)
·· + cu· + ku = – mιu
mu ·· (12.29) u4
g
In the case that the movement of the supports is the same at all u1
supports, u g becomes: ug = : displacements of the supports (excited, massless)
u5
u g = 1u g (12.30) (12.33)
and with ι = ι1 Equation (12.29) becomes
u2
·· + cu· + ku = – mιu·· u0 = : displacements of the supports (not excited, massless)
mu g (12.31) u6
which is the same as Equation (12.10). (12.34)

The stiffness matrix K of the system is:

By swapping DoF 2 and 4 we obtain:

12 6L – 12 6L 0 0 u1
2 2
6L 4L – 6L 2L 0 0 u2 24 0 – 12 – 6L – 12 6L u3
EI – 12 – 6L 12 + 12 – 6L + 6L – 12 6L u 2 2 2
u4
K = -----3- ⋅ ,U = 3 0 8L 6L 2L – 6L 2L
2 2 2 2 u4
L 6L 2L – 6L + 6L 4L + 4L – 6L 2L EI – 12 6L 12 6L 0 0 u1
K = -----3- ⋅ ,U = (12.38)
0 0 – 12 – 6L 12 – 6L u5 2 2 u2
L – 6L 2L 6L 4L 0 0
2 2 u6
0 0 6L 2L – 6L 4L – 12 – 6L 0 0 12 – 6L u5
(12.35) 6L 2L
2
0 0 – 6L 4L
2 u6
12 6L – 12 6L 0 0 u1 By swapping DoF 2 and 5 we obtain:

2 2
6L 4L – 6L 2L 0 0 u2
24 0 – 12 – 12 – 6L 6L u3
EI – 12 – 6L 24 0 – 12 6L u
K = -----3- ⋅ ,U = 3 (12.36) 0 8L
2
6L – 6L 2L 2L
2 2
u4
2 2 2 u4
L 6L 2L 0 8L – 6L 2L
EI – 12 6L 12 0 6L 0 u1
0 0 – 12 – 6L 12 – 6L u5 K = -----3- ⋅ ,U = (12.39)
2 2 L – 12 – 6L 0 12 0 – 6L u5
0 0 6L 2L – 6L 4L u6
2 2
– 6L 2L 6L 0 4L 0 u2
with L = L ⁄ 2 6L 2L
2
0 – 6L 0 4L
2 u6
By swapping DoF 1 and 3 we obtain:

By means of static condensation we can now eliminate DoF 2
and 6. We racall that:
24 – 6L – 12 0 – 12 6L u3
2 2
– 6L 4L 6L 2L 0 0 u2 EI k k ˆ EI T –1
K = -----3- ⋅ tt t0 , k tt = -----3- ⋅ ( k tt – k 0t k 00 k 0t ) (12.40)
EI – 12 6L 12 6L 0 0 u L k 0t k 00 L
K = -----3- ⋅ ,U = 1 (12.37)
2 2 2 u4
L 0 2L 6L 8L – 6L 2L
– 12 0 0 – 6L 12 – 6L u5
2 2 u6
6L 0 0 2L – 6L 4L

By performig the needed calculations we obtain: The equation of motion of the system becomes:
1
--------- 0 M 0 0 0 u··3 6 0 –3 –3 u3 0
2 2 ·· 2
1 ˆ 1 0 ⋅ u 4 + ----- 0 6L 3L – 3L ⋅ u 4 0
= ----------- ⋅ k 00 = -----------4- ⋅ 4L 0 = 4L
–1 0 I 0 EI-
k 00 (12.41) ⋅ = (12.46)
k 00 16L 0 4L
2 1 0 0 0 0 u1 L3
·· – 3 3L 3 0 u1 p1 ( t )
0 --------2-
4L 0 0 0 0 u··5 – 3 – 3L 0 3 u5 p5 ( t )
– 6L 6L 1 We racall that:
--------- 0
2 2 2 2
T –1
k 0t k 00 k 0t = 2L 2L ⋅ 4L ⋅ – 6L 2L 6L 0 (12.42) ˆ EI k k g
6L 0 1
0 --------2-
2
6L 2L 0 – 6L k tt = -----3- ⋅ , ι = –k–1 kg (12.47)
T
L k g k gg
0 – 6L 4L
By performig the needed calculations we obtain:
18 0 –9 –9
T –1 2 1
k 0t k 00 k 0t = 0 2L 3L 3L (12.43) 3 --- 0
–1 1 ˆ L 6
– 9 3L 9 0 k = ------ ⋅ k = ------ ⋅ (12.48)
k EI 1
– 9 3L 0 9 0 --------2-
6L
§ 24 0 – 12 – 12 – 9 – 9 ·¸
18 0
¨
1---
ˆ EI ¨ 0 8L 2 6L – 6L – 0 2L 2 3L 3L ¸¸ 0 1
---
1
---
k tt = -----3- ⋅ ¨ (12.44) –1 L 6
3
EI – 3 – 3 2 2
L ¨ – 12 6L 12 0 – 9 3L 9 0 ¸ ι = – k k g = – ------ ⋅ ⋅ -----3- ⋅ = (12.49)
¨ ¸ EI 1 3L 3L 1 1
0 --------2- L – ------- -------
© – 12 – 6L 0 12 – 9 3L 0 9 ¹ 2L 2L
6L
6 0 –3 –3
2
ˆ EI
k tt = -----3- ⋅ 0 6L 3L – 3L (12.45)
L – 3 3L 3 0
– 3 – 3L 0 3

The vector of the effective forces becomes: 12.2 Time-history of the response of elastic systems
M
----- M
-----
As discussed in the previous sections, the equation of motion of
·· 2 2 u·· a MDoF system under base excitation is:
p eff ( t ) = – ( mι + m g )u g ( t ) = – ⋅ g1 (12.50)
I
– -------
I-
------ u··g5 ·· + Cu· + Ku = – Mιu·· ( t )
Mu g (12.52)
2L 2L
And the equation of motion of the system finally becomes: Where:

M: Mass matrix (symmetric and positive-definite, or diag-
M M onal if only lumped masses are present)
·· ----- -----
M 0 ⋅ u 3 + -----
EI- 6 0 u 2 2 u··
3
⋅ ⋅ 3 = – ⋅ g1 (12.51) K: Stiffness matrix (symmetric and positive-definite)
0 I ··
u4 L 0 6L
2
u4 I I u··g5
– ------- ------- C: Damping matrix (Classical damping: C is typically a
2L 2L
linear combination of M and K )
The following drawings show the interpretation of the elements ··
u g ( t ) : Ground acceleration
of the influence matrix ι :
ι: Influence vector of order N . In the simplest case of a
if u g1 = 1 then: planar system under translational ground motion ι = 1 .
1 1 If the damping of the MDoF system is classical, Equation (12.52)
u 3 = --- and u 4 = – -------
2 2L can be written in the form of N decoupled modal equations,
if u g5 = 1 then: where N is the number of modes of the system. The modal
equations are of the following form:
1 1
u 3 = --- and u 4 = ------- T
2 2L m n* q··n + c n* q· n + k n* q n = – φ n Mιu··g (12.53)
Remarks: or:
• See Section 9.7 of [Cho11] for an example with a statically in- T
determinated system. 2 φ n Mι
q··n + 2ζ n* ω n q· n + ω n q n = – -----------------
T
- u··g (12.54)
• In Finite Element analysis, when applying multiple support ex- φ n Mφ n
citation, support displacements instead of support acceleration
The dynamic response of the MDoF system can be written as:
are often used. For more details see [Bat96].

N Unlike the modal mass m n* and the modal participation factor Γ n ,

u(t) = ¦ φn qn ( t ) (12.55) the effective modal mass m n*, eff is independent of the normaliza-
n=1 tion of the eigenvectors. The following equation holds:
φn : nth eigenvector of the MDoF system N N
q n ( t ) : nth modal coordinate of the MDoF system ¦ m n*, eff = ¦ mn = m tot (12.60)
n=1 n=1
Further variables in Equation (12.53) are the modal mass m n*
and the modal stiffness k n* of the nth mode. These parameters where m tot is the total mass of the dynamic system.
are defined as follows: *
The effective modal height h n of the nth mode is:
T
m n* = φ n ⋅ M ⋅ φ n (12.56) θ N
* Ln θ T
k n* =
T
φn ⋅ K ⋅ φn =
2
ωn ⋅ m n* (12.57)
hn = ------ with L n =
Ln ¦ hj ⋅ mj ⋅ φjn and Ln = φ n ⋅ Mι (12.61)
j=1
ωn : nth modal circular frequency of the MDOF system • Significance of the effective modal mass m n*, eff
The modal participation factor Γ n is a measure for the contribu- The effective modal mass m n*, eff is the lumped mass of a single-
tion of the n-th mode to the total response of the system. It is de- storey substitute system which is subjected to a base shear
fined as follows: force V bn equal to the nth modal base shear force of a multi-sto-
T
rey system.
φ n Mι
Γ n = -----------------
T
- (12.58) If in addition the height of the single storey substitute system with
*
φ n Mφ n the lumped mass m n*, eff equals the modal height h n , the single-
storey system is subjected to a base moment M bn which is equal
In addition the so-called effective modal mass of the nth mode
to the nth modal base moment of the multi-storey system.
is defined as:
2
The following holds:
m n*, eff = Γ n ⋅ m n* (12.59)
N
V bn = m n*, eff ⋅ S pa, n = ¦ fjn (12.62)
j=1

N • MDoF system with eigenmodes and equivalent SDoF systems

*
M bn = m n*, eff ⋅ S pa, n ⋅ h n = ¦ fjn ⋅ hj (12.63) MDOF Eigenmodes and
j=1
System equivalent static forces
where S pa, n is the pseudo-acceleration of the nth mode. m3
f31 f32 f33
hs
• Distribution of the internal forces
m2
f21 f22 f23
If the internal forces of the entire system are to be determined,
h hs h3
the modal equivalent static forces f jn should be computed first:
m1 h2 f11 f12 f13
f n = s n ⋅ S pa, n (12.64) hs h1
EI
where
Vb Vb1 Vb2 Vb3
Mb Mb1 Mb2 Mb3
f n = f 1n f 2n … f nn (12.65)
The excitation vector s n is defined according to equation (12.66)

and specifies the distribution of the inertia forces due to excita- MDOF Set of equivalent SDOF systems
tion of the nth mode: System and equivalent static forces
s n = Γ n Mφ n (12.66) m3
m1*=2.180m
m1*Sa1
s n is independent on the normalization of the eigenvector φ n and hs
m2
we have that: m2*=0.646m
m3*=0.174m
h1*=2.50hs
h hs
N
m1 m2*Sa2
¦ sn (12.67)
h2*=0.72hs
h3*=0.48hs
= Mι m3*Sa3
n=1 hs
EI EI1* EI2* EI3*
Vb Vb1 Vb2 Vb3
Mb Mb1 Mb2 Mb3

12.3 Response spectrum method • “Absolute Sum (ABSSUM)” Combination Rule

N
12.3.1 Definition and characteristics u i, max ≤ ¦ φ in ⋅ q n, max (12.70)
If the maximum response only and not the response to the entire n=1
time history according to Equation (12.55) is of interest, the re-
The assumption that all maxima occur at the same instant and in
sponse spectrum method can be applied.
the same direction yields an upper bound value for the response
The response spectrum can be computed for the considered quantity. This assumption is commonly too conservative.
seismic excitation and the maximum value of the modal coordi-
• “Square-Root-of Sum-of-Squares (SRSS)” Combination Rule
nate q n, max can be determined as follows:
N
2
q n, max = Γ n ⋅ S d (ω n,ζ n*)
1
= Γ n ⋅ -----2- ⋅ S pa (ω n,ζ n*) (12.68) u i, max ≈ ¦ ( φin ⋅ qn, max ) (12.71)
ωn n=1
where: This rule is often used as the standard combination method and
yields very good estimates of the total maximum response if the
Γn : modal participation factor of the n-th mode
modes of the system are well separated. If the system has sev-
S d (ω n,ζ n*) : Spectral displacement for the circular eigenfre- eral modes with similar frequencies the SRSS rule might yield
quency ω n and the modal damping rate ζ n* . estimates which are significantly lower than the actual total max-
S pa (ω n,ζ n*) : Spectral pseudo-acceleration for the circular ei- imum response.
genfrequency ω n and the modal damping rate ζ n* .
• “Complete Quadratic Combination (CQC)” Combination Rule
The contribution of the nth mode to the total displacement is:
N N
(j) (k)
u n, max = φ n ⋅ q n, max (12.69) u i, max ≈ ¦ ¦ ui, max ⋅ ρjk ⋅ ui, max (12.72)
j = 1k = 1
The maxima of different modes do not occur at the same instant.
An exact computation of the total maximum response on the ba- where
sis of the maximum modal responses is hence impossible. Dif- (j) (k)
u i, max and u i, max are the max. modal responses of modes j and k
ferent methods have been developed to estimate the total max-
imum response from the maximum modal responses. ρ jk is the correlation coefficient between nodes j and k :

8 ζ i ζ k ( ζ i + rζ k )r 3 ⁄ 2 ω • Number of modes to be considered.

ρ jk - with r = -----k-
= ---------------------------------------------------------------------------------------------------
( 1 – r 2 ) 2 + 4ζ i ζ k r ( 1 + r 2 ) + 4 ( ζ i2 + ζ k2 )r 2 ωj All modes which contribute to the dynamic response of the sys-
(12.73) tem should be considered. In practical applications, however,
only those modes are considered which contribution to the total
This method based on random vibration theory gives exact re- response is above a certain threshold. It should be noted that in
sults if the excitation is represented by a white noise. If the fre- order to achieve the same accuracy for different response meas-
quencies of the modes are well spaced apart, the result converge ures (e.g. displacements, shear forces, bending moments, etc.)
to those of the SRSS rule. More detailed information on this and different numbers of modes might need to be considered in the
other combination rules can be found in [Cho11] Chapter 13.7. computation.
• Internal forces For a regular building the top displacement can be estimated
The aforementioned combination rules cannot only be applied fairly well on the basis of the fundamental mode only. To esti-
on displacements but also on internal forces. mate the internal forces, however, higher modes need to be con-
sidered too.
The maximum modal internal forces can be determined from
equivalent static forces According to Eurocode 8 “Design of Structures for Earthquake
Resistance” [CEN04] all modes should be considered (starting
F n, max = K ⋅ u n, max , (12.74) from the lowest) until the sum of the effective modal masses
m n, eff of all considered modes corresponds to at least 90% of
which, as a first option, are computed from the equivalent static the total mass m tot . As an alternative, Eurocode 8 allows the de-
displacements. Alternatively, the equivalent static forces can be signer to show that all modes with m n*, eff > 0.05m tot were consid-
determined from the inertia forces: ered in the computation.
F n, max = s n ⋅ S pa (ω n,ζ n*) = Γ n Mφ n ⋅ S pa (ω n,ζ n*) (12.75)
with s n being the excitation vector which represents the distribu-

tion 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 .

12.3.2 Step-by-step procedure response spectrum for pseudo-accelerations. (The spectral dis-
placement S d (ω n,ζ n*) should be determined in the same manner)
The maximum response of a N-storey building can be estimated • Computation of the maximum displacements
according to the following procedure:
u n, max = φ n ⋅ Γ n ⋅ S d (ω n,ζ n*)
1) Determine the properties of the MDOF system
• Choose DOFs • Computation of the maximum equivalent static forces
• Determine mass matrix M and stiffness matrix K . F n, max = s n ⋅ S pa ( ω n, ζ n ) = Γ n Mφ n ⋅ S pa ( ω n, ζ n )

• Estimate modal damping ratios ζ n*
• Computation of the maximum internal forces on the basis of the
2) Carry out modal analysis of the MDOF system forces F n, max
• Determine circular eigenfrequencies ω n and eigenvectors φ n 4) Estimate the total response in terms of displacements and in-
ternal forces by means of suitable combination rules. Differ-
2
( K – ωn M ) ⋅ φn = 0 ent combination rules might be applied (ABSSUM, SRSS,
CQC).
• Compute the modal properties of the MDOF system ( M * , K * )
T T Comment
m n* = φ n Mφ n , k n* = φ n Kφ n
In order to consider the non-linear behaviour of the structure the
• Compute the modal participation factor Γ n
equivalent lateral static forces F n, max can be determined from
φ n Mι
T the spectral ordinate S pa ( ω n, ζ n, q ) of the design spectrum for
Γ n = -----------------
T
- pseudo-accelerations:
φ n Mφ n
F n, max = s n ⋅ S pa ( ω n, ζ n, q ) = Γ n Mφ n ⋅ S pa ( ω n, ζ n, q ) (12.76)
3) The maximum response of the n-th mode should be deter-
mined 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 ζ n* ,
the spectral response S a ( ω n, ζ n ) should be determined from the

12.4 Practical application of the response spectrum • Stiffness

method to a 2-DoF system The horizontal stiffness of each story is k = 1 , hence the stiff-
ness matrix K is:
12.4.1 Dynamic properties
k 11 k 12 1 –1
K = = (12.78)
k 21 k 22 –1 2
1. unit displacement u 1 = 1
2. unit displacement u 2 = 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
• Damping
the mass matrix M is:
Damping is small and is neglected, hence the damping matrix C is:
m1 0
M = = 1 0 (12.77) C = 0 0 (12.79)
0 m2 0 1 0 0

12.4.2 Free vibrations 2) Natural modes of vibration and natural frequencies

2
1) Matrix eigenvalue problem For each eigenvalue ω a natural mode of vibration and a natu-
ral frequency can be computed.
2 φ1 • Fundamental mode (first natural mode of vibration)
(K – ω M) ⋅ =0 (12.80)
φ2
2 3– 5
The smallest eigenvalue ω 1 = ---------------- leads to the
φ1 2
The nontrivial solution for the eigenvector φ = ≠ 0 exists if
the determinant is equal to zero: φ 2
1. circular natural frequency ω 1 = 3 – 5- = 0.62 (12.86)
---------------
det ( K – ω M ) = 0
2
(12.81) 2
2
When the eigenvalue ω 1 is known, the system
2
det ( K – ω M ) = det 1 – ω –1
2
= 0 (12.82) φ1
2 2
–1 2 – ω ( K – ω1 M ) ⋅ =0 (12.87)
φ2
2
This leads to the quadratic equation in ω : φ
can be solved for the corresponding vector 11 (fundamental
2 2 2 4
( 1 – ω ) ⋅ ( 2 – ω ) – ( – 1 ) ⋅ ( – 1 ) = 2 – 3ω + ω – 1 = 0 (12.83) mode) to within a multiplicative constant: φ 21
or
3– 5
4 2 1 – ---------------- –1 φ 11
ω – 3ω + 1 = 0 (12.84) 2 ⋅ = 0 (12.88)
3– 5 φ 21 0
The solution of the quadratic equation yield the eigenvalues: –1 2 – ----------------
2
2 3− + 9 – 4- = 3 −
+ 5-
ω = ------------------------ --------------- (12.85) The first row yields following equation:
2 2
2 –( 3 – 5 )
--------------------------- φ 11 – 1φ 21 = 0 (12.89)
2

Normalizing the largest coordinate of the eigenvector to unity • Higher mode of vibration
( φ 11 = 1 ), φ 21 becomes: 2 3+ 5
The largest eigenvalue ω 2 = ---------------- leads to the
2
2 –( 3 – 5 )
--------------------------- – φ 21 = 0 (12.90) 3+ 5
2 2. circular natural frequency ω 2 = ---------------- = 1.62 (12.93)
2
or
In analogy to the fundamental mode, the second mode of vibra-
2
5–1 tion can be computed introducing the second eigenvalue ω 2 into
φ 21 = ---------------- = 0.62 (12.91)
2 the system of equations:
Hence the first natural mode of vibration is:
3+ 5
1 – ---------------- –1 φ 12
2 ⋅ = 0 (12.94)
3+ 5 φ 22 0
–1 2 – ----------------
2
The first row yields following equation:

φ 11 1
= 1 (12.92) 2 – ( 3 + 5 )-
φ 21 5 – 1 = 0.62 --------------------------- φ 12 – 1φ 22 = 0 (12.95)
---------------- 2
2
Normalizing the largest coordinate of the eigenvector to unity
( φ 22 = 1 ), φ 12 becomes:
2 –( 3 + 5 )
---------------------------- φ 12 – 1 = 0 (12.96)
2
or:
–2 1– 5
φ 12 = ---------------- = ---------------- = – 0.62 (12.97)
1+ 5 2

Hence the second natural mode of vibration is: • Orthogonality with respect to the mass matrix
The modal mass matrix M* is:
5 – 1- 5–1
1 --------------- 1 – ----------------
T 2 1 0 2
M* = Φ MΦ = ⋅ ⋅
5–1 0 1 5 – 1-
– ---------------- 1 --------------- 1
φ 12 1– 5
- = – 0.62
2 2
= ------------
2 (12.98)
φ 22 1 5 – 1- 5–1
1 1 --------------- 1 – ----------------
= 2 ⋅ 2
5–1 5–1
– ---------------- 1 ---------------- 1
2 2
2
1 + § ----------------·
5–1
0
© 2 ¹
3) Orthogonality of modes = = 1.38 0
2 0 1.38
1 + § ----------------·
5–1
In the following the orthogonality of the modes of vibration 0
© 2 ¹
should be checked.
(12.100)
Hence, following matrix of the eigenvectors is needed:
i.e. the matrix M* is diagonal.
5–1
φ 11 φ 12 1 – ----------------
Φ = = 2 (12.99)
φ 21 φ 22 5 – 1-
--------------- 1
2

• Orthogonality with respect to the stiffness matrix 5–1 2

k* 22 = 2 + ( 5 – 1 ) + §© ----------------·¹ = 1 + 5 + ---------------------------- = 3.618
5–2 5+1
The modal stiffness matrix K* is: 2 4
(12.105)
5 – 1- 5–1
1 --------------- 1 – ----------------
K* = Φ KΦ =
T 2 ⋅ 1 – 1 ⋅ 2 K* = 0.528 0 (12.106)
–1 2 0 3.618
5–1 5 – 1-
– ---------------- 1 --------------- 1
2 2
i.e. the matrix K* is diagonal.
5–1 5–1
1 – ---------------- – 1 + ( 5 – 1 ) 1 – ----------------
= 2 ⋅ 2 12.4.3 Equation of motion in modal coordinates
5–1 5–1 5–1
– ---------------- – 1 ---------------- + 2 ---------------- 1 The equation of motion in modal coordinates of a system without
2 2 2
damping ( C * = 0 ) is:
2 5 – 1-· 2
1 – ( 5 – 1 ) + 2 § ----------------·
5–1 5 – 1- § ---------------
--------------- + –1 ·· + K* ⋅ q = – L ⋅ u·· ( t )
© 2 ¹ 2 © 2 ¹ M* ⋅ q g (12.107)
=
5 – 1-· 2
5 – 1- § --------------- 2
2 + ( 5 – 1 ) + § ----------------·
--------------- 5–1 where:
+ –1
2 © 2 ¹ © 2 ¹
(12.101) L1 q1
L = , and q = is the vector of the modal coordinates
Computation of the single elements of K* : L2 q2
5–1 2
k* 11 = 1 – ( 5 – 1 ) + 2 §© ----------------·¹ = – 5 + ---------------------------- = 0.528
5–2 5+1
Computation of the elements of the vector L :
2 2
(12.102)
5–1
----------------
1
5–1 2 2 ⋅ 1 0 ⋅ 1 = 1.62
T
k* 12 = ---------------- + §© ----------------·¹ – 1 = --------------------------------------------------------------- = 0
5–1 2 5–2+5–2 5+1–4 L = Φ ⋅M⋅ι =
2 2 4 0 1 1 0.382
5–1
(12.103) – ---------------- 1
2
k* 21 = k* 12 = 0 (12.104) (12.108)

The influence vector ι represents the dis- • Additional important modal quantities
placement of the masses resulting from the
The modal participation factor Γ n is defined as:
static application of a unit ground displace-
ment u g = 1 : Ln
Γ n = ---------- (12.114)
mn *
ι = 1 (12.109)
1 and substituting L n and m n * into this definition gives following
values for Γ 1 and Γ 2 :
The substitution of L into the equation of
motion in modal coordinates L1 1.62
Γ 1 = ---------- = ---------- = 1.17 (12.115)
m1 * 1.38
·· + K* ⋅ q = – L ⋅ u·· ( t )
M* ⋅ q (12.110)
g
L2 0.382
leads to: Γ 2 = ---------- = ------------- = 0.28 (12.116)
m2 * 1.38
··
1.38 0 ⋅ q 1 + 0.528 0 ⋅ q 1 = – 1.62 ⋅ u·· ( t ) The effective modal mass is defined as:
g
0 1.38 q··2 0 3.618 q2 0.382 2
m n*, eff = Γ n ⋅ m n* (12.117)
(12.111)
and substituting Γ n and m n * into this definition gives following
Checking the circular natural frequency computed using modal
values for m 1*, eff and m 2*, eff :
coordinates against the results of Section 12.4.2:
2
k 11 * m 1*, eff = Γ 1 ⋅ m 1* = 1.17 2 ⋅ 1.38 = 1.894 (12.118)
ω1 = 0.528
------------ = ------------- = 0.62 OK! (12.112)
m 11 * 1.38 2
m 2*, eff = Γ 2 ⋅ m 2* = 0.28 2 ⋅ 1.38 = 0.106 (12.119)
k 22 * 3.618
ω2 = ------------ = ------------- = 1.62 OK! (12.113) m 1, eff + m 2*, eff = 1.894 + 0.106 = 2.000 OK! (12.120)
m 22 * 1.38

12.4.4 Response spectrum method 2) Natural modes of vibration and natural frequencies
The 2-DoF system analysed in the previous Sections shall be The results of the previous Sections computed using unit mass-
used to illustrate the response spectrum method. For this reason es and unit stiffnesses shall be multiplied by the factor:
real masses and stiffnesses shall be assumed. The seismic ac-
2
tion on the 2-DoF system is represented by the elastic response k 100kg ⁄ s –1
(12.123)
---- = ------------------------ = 100s
spectrum of the “El Centro” earthquake. m 1kg
1) Model • Fundamental mode
Similar to Section 12.4.1, –1
Natural frequency: ω 1 = 0.62 ⋅ 100s = 6.2 Hz
however with a new definition
of masses and stiffnesses: 2π 2π
Natural period: T 1 = ------ = ---------------- = 1.02 s
m 1 = m 2 = 1kg ω1 6.2 Hz
Hence, the mass matrix is: • Higher vibration mode

–1
Natural frequency: ω 2 = 1.62 ⋅ 100s = 16.2 Hz
m1 0
M = = 1 0 kg
0 m2 0 1 2π 2π
Natural period: T 2 = ------ = ------------------- = 0.39 s
ω2 16.2 Hz
The stiffness chosen for each story is k 1 = k 2 = k = 100 N/m The eigenvectors are dimensionless quantities and remain un-
and an appropriate units transformation leads to: changed:
k = 100 N/m = 100 kgm/s 2 m –1 = 100 kg/s 2 (12.121)

5–1
Hence, the stiffness matrix is: φ 11 φ 12 1 – ----------------
Φ = = 2 (12.124)
φ 21 φ 22 5 – 1-
k 11 k 12 100 – 100 kg/s 2 --------------- 1
K = = (12.122) 2
k 21 k 22 – 100 200

3) Modal analysis 4) Peak modal response

Equation of motion in u 1 - u 2 -coordinates (without damping): The peak modal response of both vibration modes can be com-
puted like in the case of SDoF systems using the spectral value
M ⋅ u·· + K ⋅ u = – M ⋅ 1 ⋅ u··g t (12.125)
given by the relevant response spectrum.
·· The peak value of the modal coordinate q 1 is:

1 0 kg ⋅ u 1 + 100 – 100 kg ⁄ s 2 ⋅ u 1 = – 1 0 kg 1 ⋅ u·· ( t )
g
01 u··2 – 100 200 u2 0 1 1 L1
q 1, max = ---------- ⋅ S d ( ω 1, ζ 1 ) = Γ 1 ⋅ S d ( ω 1, ζ 1 ) (12.130)
(12.126) m1 *
Variables transformation in modal coordinates q 1 and q 2 : where:
u = Φ⋅q (12.127) S d ( ω 1, ζ 1 ) : spectral displacement for a natural frequency

ω 1 and a damping ζ 1 (here ζ 1 = 5% )
where:
If an acceleration instead of a displacement response spectrum
Φ : Modal Matrix, i.e. the matrix of the eigenvectors is used, then the peak value of the modal coordinate q 1 is:
The equation of motion in modal coordinates q 1 and q 2 (without L1 1 Γ1
damping) is: q 1, max = ---------- ⋅ -----2- ⋅ S pa ( ω 1, ζ 1 ) = -----2- ⋅ S pa ( ω 1, ζ 1 ) (12.131)
m1 * ω ω1
·· + K* ⋅ q = – L ⋅ u·· ( t ) 1
M* ⋅ q g (12.128)
where:
··
1.38 0 kg ⋅ q 1 + 52.8 0 kg ⁄ s 2 ⋅ q 1 = – 1.62 kg ⋅ u·· ( t ) S pa ( ω 1, ζ 1 ) : spectral pseudo-acceleration for a natural fre-
g
0 1.38 q··2 0 362 q2 0.382 quency ω 1 and a damping ζ 1 (here ζ 1 = 5% )
(12.129)
yielding the equation of motion in modal coordinates of two inde-
pendent SDoF systems.

The spectral values given by the elastic acceleration response 5) Inverse transformation
spectrum of the “El Centro” earthquake for the periods T 1 and
The peak deformations and internal forces belonging to each mode of vibra-
T 2 are: tion in the original reference system are obtained by multiplying the relevant
eigenvector with the corresponding peak value of the modal coordinate.
S pa1 = 4.25 m/s 2 and (12.132)
• Fundamental mode
S pa2 = 7.34 m/s 2 (12.133)
(1) 1 = 130 mm
u max = q 1, max ⋅ φ 1 = 0.130m ⋅ (12.136)
10 0.62 81
ζ = 5%
Pseudo acceleration [m/s2]
(1) (1)
Spa2 = 7.34 m/s2 f max = K ⋅ u max = 100 – 100 ⋅ 0.130 = 13.0 – 8.1 = 4.9 N
100 200 0.081 – 13.0 + 16.2 3.2
(12.137)
Alternatively (allow an approximate consideration of nonlinearities):
5 S 2
pa1 = 4.25 m/s
s 1 = Γ 1 Mφ 1 = 1.17 ⋅ 1 0 ⋅ 1 = 1.17 (12.138)
0 1 0.62 0.725
(1)
f max = s 1 ⋅ S pa1 = 4.25 1.17 = 4.9 N (12.139)
0.725 3.2
T2 = 0.39 s T1 = 1.02 s
0
0.01 0.10 1.00 10.0
Period [s]
1.62kg 1
q 1, max = ----------------- ⋅ ---------------------2- ⋅ 4.25 m/s 2 = 0.130m (12.134)
1.38kg ( 6.2Hz )
0.38kg 1
q 2, max = ----------------- ⋅ ------------------------2- ⋅ 7.34 m/s 2 = 0.008m (12.135)
1.38kg ( 16.2Hz )

• Higher vibration mode 6) Combination
(2) The total peak response is obtained from the peak response of
u max = q 2, max ⋅ φ 2 = 0.008m ⋅ – 0.62 = – 5.0 mm (12.140) the single vibration modes using e.g. the SRSS combination rule
1 8.0
(SRSS = Square Root of the Sum of Squares).
(2) (2) • Peak displacements
f max = K ⋅ u max = 100 – 100 ⋅ – 0.005 = – 0.5 – 0.8 = – 1.3 N
100 200 0.008 0.5 + 1.6 2.1
2
(12.141) (n) 2 2 2
u 1, max = ¦ ( u1 ) = ( 130mm ) + ( – 5 mm ) = 130mm
Alternatively (allow an approximate consideration of nonlinearities): k=1
(12.144)
s 2 = Γ 2 Mφ 2 = 0.28 ⋅ 1 0 ⋅ – 0.62 = – 0.173 (12.142)
01 1 0.28 2
(n) 2 2 2
u 2, max = ¦ ( u2 ) = ( 81mm ) + ( 8mm ) = 81mm
(1)
f max = s 2 ⋅ S pa2 = 7.34 – 0.173 = – 1.3 N (12.143) k=1
0.28 2.1 (12.145)
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.

• Peak sectional forces (Shear force V) 12.4.5 Response spectrum method vs. time-history analysis
Upper shear force: 1) Model
2 2
V 1, max = ( 4.9N ) + ( – 1.3N ) = 5.1N
(12.146)
Lower shear force:
2 2
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.
Case study 1 Case study 2
Masses: m 1 = 1.0kg Masses: m 1 = 0.1kg
Pay attention to following pitfall!
It is wrong to compute the total peak sectional forces using the m 2 = 1.0kg m 2 = 1.0kg
total peak displacements:
Stiffnesses: k 1 = 100N/m Stiffnesses: k 1 = 10N/m
V 1, max ≠ 100N ⁄ m ⋅ ( 0.130m – 0.081m ) = 4.9N (12.148)
k 2 = 100N/m k 2 = 100N/m
V 2, max ≠ 100N ⁄ m ⋅ 0.081m = 8.1N (12.149)
Case study 1 corresponds to the model analysed in Section
The sectional forces would be underestimated.
12.4.4. Case study 2 represents a dynamic system where the
second vibration mode is important.

2) Results • Demand
• Dynamic properties Case study 1 Case study 2

Displacements: Displacements:
Case study 1 Case study 2
Periods: T 1 = 1.02s Periods: T 1 = 0.74s 1: Δ = 0.129m 1: Δ = 0.130m
T 2 = 0.39s T 2 = 0.54s 2: Δ = 0.005m 2: Δ = 0.072m

Eigenvectors: Eigenvectors:
Sum: Δ = 0.134m Sum: Δ = 0.202m
1: φ 11 = 1 , φ 21 = 0.62 1: φ 11 = 1 , φ 21 = 0.27
SRSS: Δ = 0.130m SRSS: Δ = 0.148m
2: φ 12 = 1 , φ 22 = – 1.62 2: φ 12 = 1 , φ 22 = – 0.37
Time-history: Δ = 0.130m Time-history: Δ = 0.165m
Part. factors: Γ 1 = 1.17 Part. factors: Γ 1 = 2.14
Upper shear force: Upper shear force:
Γ 2 = – 0.17 Γ 2 = – 1.14
SRSS: V = 5.10N SRSS: V = 1.36N
• Note that in this case the eigenvectors are normalized to yield Time-history: V = 5.69N Time-history: V = 1.51N
unit displacement at the top of the second story. Therefore, the
eigenvectors and the participation factors of case study 1 differ Lower shear force: Lower shear force:
from the values obtained in previous sections.
SRSS: V = 8.05N SRSS: V = 4.40N
Time-history: V = 8.44N Time-history: V = 4.92N

• Time-histories: Case study 1 • Time-histories: Case study 2

20 20
Displacement [cm]
Displacement [cm]
10 10
0 0
−10 −10
−20 −20
0 5 10 15 20 0 5 10 15 20
20 20
Second mode Second mode
Displacement [cm]
Displacement [cm]
10 10
0 0
−10 −10
−20 −20
0 5 10 15 20 0 5 10 15 20
20 20
Sum Sum
Displacement [cm]
Displacement [cm]
10 10
0 0
−10 −10
Peak values don’t occur at the same time!

−20 −20
0 5 10 15 20 0 5 10 15 20
Time [s] Time [s]

• Time-histories: Summary
Blank page
20
Case study 1
10
Displacement [cm]
−10 First mode

Second mode
Sum
−20
0 5 10 15 20
20
Case study 2
10
Displacement [cm]
−10 First mode

Second mode
Sum
−20
0 5 10 15 20
Time [s]

13 Vibration Problems in Structures 13.1.1 Dynamic action
a) People-induced vibrations
13.1 Introduction
- Pedestrian bridges
There are more and more vibration problems in structures - Floors with walking people
because: - Floors for sport or dance activities
• Higher quality materials with higher exploitation - Floor with fixed seating and spectator galleries
- slender constructions - High-diving platforms
- smaller stiffnesses and masses
b) Machinery-induced vibrations
• More intensive dynamic excitations
- Machine foundations and supports
• Increased sensitivity of people - Bell towers
- Structure-borne sound
- Ground-transmitted vibrations
Nevertheless vibration sensitive structures are
often designed for static loads only c) Wind-induced vibrations
- Buildings
Goal of this chapter - Towers, chimneys and masts
- Bridges
• Give an overview of possible causes of vibration problems in
buildings and of potential countermeasures - Cantilevered roofs
• Description of practical cases with vibration rehabilitation d) Vibrations induced by traffic and construction activity
- Roads and bridges
- Railways
- Construction works
13 Vibration Problems Page 13-1 13 Vibration Problems Page 13-2

13.1.2 References 13.2 Vibration limitation

[Bac+97] Bachmann et al.:”Vibration Problems in Structures”.
ISBN 3-7643-5148-9. Birkhäuser Verlag, Basel 1997. 13.2.1 Verification strategies
[BB88] Baumann K., Bachmann H.: “Durch Menschen verursachte • Frequency tuning
dynamische Lasten und deren Auswirkungen auf Balkentrag-
werke”. IBK Bericht Nr. 7501-3, 1988. 10
0.00
[BW95] Bachmann H., Weber B.: “Tuned Vibration Absorbers for 9
0.01
Damping of Lively Structures”. Structural Engineering Inter- 8
national, No. 1, 1995. ζ = 0.05
Trrasmissibility TR
7
[SIA06] SIA: “Wind – Kommentar zum Kapitel 6 der Normen SIA 261 6
und 261/1 (2003) Einwirkungen auf Tragwerke”. SIA Doku-
mentation D0188. Zürich 2006. 5
0.10
4
[SIA03] SIA: “Aktuelle Probleme der Brückendynamik”. SIA Doku-
mentation D0198. Zürich 2003. 3
0.20
[SIA96] SIA: “Dynamische Probleme bei Brücken- und Hochbauten”. 2
0.50
SIA Dokumentation D0138. Zürich 1996. 1
0.70 2
[VB87] Vogt R., Bachmann H.: “Dynamische Kräfte beim Klatschen, 0
Fussstampfen und Wippen”. IBK Bericht Nr. 7501-4, 1987. 0 0.5 1 1.5 2 2.5 3 3.5 4
ω/ωn
- High tuning (subcritical excitation)

- Low tuning (supercritical excitation)
• Amplitude limitation

13.2.2 Countermeasures 13.2.3 Calculation methods

• Change of the natural frequency • Computation of the natural frequencies
Strategy Effects The natural frequencies of structures have to be determined by
means of realistic models. Approximate formulas that are often
Stiffness: I ∼ bh ½
3 found in design codes and literature shall be checked carefully.
I
¾ f ∼ ---- ∼ h
Mass: A ∼ bh ¿ A
• Computation of the Amplitude
If the frequency of a harmonic of the excitation coincides with a
2
π natural frequency of the structure (resonance), the maximum de-
f a, 1 = ------------2- ⋅ EI
------ ( μ = distributed mass)
2πL μ flection of the structure can be estimated as follows (See Chap-
ter 5):
2 Fo
3.93 EI
f b, 1 = ------------2- ⋅ ------ = 1.56f a, 1 ( 2.45 ⋅ EI ) u p = ----- ⋅ V ( ω ) ⋅ cos ( ωt – φ ) (13.1)
μ k
2πL
for ω = ω n we have V ( ω ) = 1 ⁄ ( 2ζ ) and:
2 Fo 1
4.73 EI u max = ----- ⋅ ------
f c, 1 = ------------2- ⋅ ------ = 2.27f a, 1 ( 5.14 ⋅ EI ) (13.2)
2πL μ k 2ζ
The maximum velocity and the maximum acceleration can be

2 determined from Equation (13.2) as follows:
π
f d, 1 = ------------------------2- ⋅ EI
------ = 4f a, 1 ( 16 ⋅ EI )
2π ( L ⁄ 2 ) μ Fo 1
u· max = ω ⋅ u max = ω ⋅ ----- ⋅ ------ (13.3)
k 2ζ
• Increase of the damping 2 2 Fo 1

u··max = ω ⋅ u max = ω ⋅ ----- ⋅ ------ (13.4)
- Installation of dampers or absorbers k 2ζ
- Plastic energy dissipation

• Tuned Mass Dampers

The amplitude of the nth harmonic component of a force gener- 13.3 People induced vibrations
ated by people excitation is proportional to the mass of the per-
son ( F o = G ⋅ α n = g ⋅ M ⋅ α n , see Equation (13.8)). 13.3.1 Excitation forces
2 g ⋅ M ⋅ αn 1 k g ⋅ M ⋅ αn 1 M g ⋅ αn In Chapter 6 “Forced Vibrations” it has been already mentioned
u··max = ω ⋅ ----------------------- ⋅ ------ = ---- ⋅ ----------------------- ⋅ ------ = ----- ⋅ ------------- (13.5)
k 2ζ m k 2ζ m 2ζ that excitation due to people, like e.g. walking, running, jumping,
and so on, can be represented as Fourier-series:
M g ⋅ αn
u··max = ----- ⋅ ------------- (13.6)
m 2ζ ∞
F ( t ) = a0 + ¦ [ a n cos ( nω 0 t ) + b n sin ( nω 0 t ) ] (13.7)
• Remarks
n=1
- A soft structure is more prone to vibration than a rigid one.
Equation (13.7) can also be represented in a form according to
See Equations (13.2) to (13.4).
Equation (13.8):
- The acceleration amplitude is directly proportional on the
N
ratio of the mass of the people to the building mass.
F(t) = G + ¦ G ⋅ α n ⋅ sin ( n ⋅ 2πf 0 ⋅ t – φ n ) (13.8)
n=1
Where:
- G = Weight of the person
- α n = Fourier coefficient for the nth harmonic
- G ⋅ α n = Amplitude of the nth harmonic of the excitation force
- f 0 = Step frequency of the excitation force
- φ n = Phase shift of the nth harmonic ( φ 1 = 0 )
- n = Number of the nth harmonic
- N = Number of considered harmonics

The steady-state response of a SDoF system under periodic ex- • Jumping (left) and walking (right), see [BB88]
citation can be computed in analogy to Chapter 6 as:
N
u ( t ) = u0 ( t ) + ¦ un ( t ) (13.9)
n=1
Where
G
u 0 ( t ) = ---- (Static displacement) (13.10)
k
2
G ⋅ α n ( 1 – β n ) sin ( nω 0 t – φ n ) – 2ζβ n cos ( nω 0 t – φ n )
u n ( t ) = --------------- ⋅ ---------------------------------------------------------------------------------------------------------------
- (13.11)
k 2 2 2
( 1 – β n ) + ( 2ζβ n ) • Clapping, foot stomping and rocking, see [VB87]
nω 0
ω 0 = 2πf 0 , β n = --------- (13.12)
ωn
• Measurement of forces (Example)

• Walking (see [Bac+97] Figure G.1) • Jumping (see [Bac+97] Figure G.2)
• Clapping (see [Bac+97] Figure G.3) • Jumping: Fourier amplitude spectrum (see [Bac+97] Figure G.2)

Fourier-coefficients for human activities according to [Bac+97] Table G.2

Characterisation of human activities according to [Bac+97] Table G.1

• Remarks regarding Table G.2 • RC Beam

- Coefficients and phase angles represent averages. The RC beam has a length of 19 meters. The natural frequency
- Phase angles have strong scattering and therefore, in is thus:
many cases, it is difficult to provide reasonable values. In f n = 2Hz . (13.13)
such cases (e.g,. running and dancing) in Table G.2 no val-
ues are specified. • Excitation
- Decisive are cases in which resonance occurs. In such Here “jumping” is described by means of the Fourier-series giv-
cases the phase angle no longer plays a role. en in Table G.2. In Section 6.1 “periodic excitation”, “jumping”
- Coefficients and phase were checked and discussed inter- was described by means of a half-sine function.
nationally. Jumping frequency: f 0 = 2Hz (13.14)
13.3.2 Example: Jumping on an RC beam Contact time: t p = 0.16s (phase angle computation)
(13.15)
Here the same example as in Section 6.1.3 is considered again:
Weight of the person: G = 0.70kN (13.16)
• Results
Excel Table: u max = 0.043m (13.17)
F 1 1.8 ⋅ 0.70 1
Equation (13.2): u max = --- ⋅ ------ = ---------------------- ⋅ --------------------- = 0.042m
k 2ζ 886 2 ⋅ 0.017
(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

• Excitation 13.3.3 Footbridges

3.5
Static component (n=0)
• Frequency tuning
3.0 First harmonics (n=1)
Second harmonics (n=2) - Vertical: Avoid natural frequencies between 1.6 and
2.5 Third harmonics (n=3)
Total (3 harmonics)
2.4Hz. In the case of structures with low damping (Steel),
2.0
avoid also natural frequencies from 3.5 to 4.5Hz.
Force F(t) [kN]
1.5
- Horizontal transverse: Avoid natural frequencies be-
1.0
tween 0.7 and 1.3Hz (absolutely safe: fht,1 > 3.4Hz).
0.5
0.0
00
- Horizontal longitudinal: Avoid natural frequencies be-
-0.5
tween 1.6 and 2.4Hz.
-1.0 • Amplitude limitation
-1.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
- Calculation of the acceleration maximum amplitude.
Time (s)
2
a max < ca. 0.5m ⁄ s = 5% g (13.19)
• Response
• Special features of the amplitude limitation
0.0500
Static component (n=0)
0.0400 First harmonics (n=1) - When walking or running, the effectiveness of people is lim-
Second harmonics (n=2)
0.0300 Third harmonics (n=3)
ited, because the forces are not always applied at midspan;
Total (3 harmonics)
[m]
0.0200
System
Displacement
0.0100
0.0000
-0.0100
-0.0200
0 0200
D
-0.0300 Equivalent
-0.0400 SDoF system
-0.0500
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)

- 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
0.1 0.05
0.02
0.8
ζ = 0.01
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 vibra- The acceleration a max given in Equation (13.20) is the accelera-
tions → Synchronisation effect) tion generated by one person crossing the footbridge. If n people
To take into account the specificities of the amplitude limitation, are on the bridge at the same time, the maximum acceleration is
sophisticated methods are available. From [Bac+97] the follow- typically less than n ⋅ a max because not all people walk in step
ing one is adopted: across the bridge.
2 2 2 .
a max = 4π ⋅ f ⋅ y ⋅ α ⋅ Φ [ m ⁄ s ] (13.20)
Where: The square root of the number of people is often
- y : Static deflection at half the span chosen as the multiplication factor, i.e. n ⋅ a max
- α : Fourier coefficient
- Φ : dynamic amplification factor

Example: “Walking on an RC beam” • Rough estimate of maximum displacement and acceleration
• Situation The maximum displacement, and the maximum acceleration can

be estimated by means of Equations (13.2) and (13.4):
System
G 1
u max,st = ------ = --------- = 0.001128m = 0.11cm (13.21)
Kn 886
G ⋅ α1 1 1 ⋅ 0.4 1
u max,1 = --------------- ⋅ ------ = --------------- ⋅ --------------------- = 0.0133m = 1.33cm
K n 2ζ 886 2 ⋅ 0.017
(13.22)
u max = 0.11 + 1.33 = 1.44cm (13.23)
Discretisation for FE Analysis
2 2 2
a max = ω u max,1 = ( 2π ⋅ 2 ) ⋅ 0.0133 = 2.10m ⁄ s (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 Walking-velocity: v = S ⋅ f 0 = 0.7 ⋅ 2 = 1.4m ⁄ s (13.25)

- Natural frequency: f n = 2Hz Crossing time: Δt = L ⁄ v = 19 ⁄ 1.4 = 13.57s (13.26)
- Damping: ζ = 0.017
Number of cycles: N = Δt ⋅ f n = 13.57 ⋅ 2 = 27 (13.27)
• Excitation
Amplification factor: Φ = 23 (13.28)
- Walking with f 0 = 2Hz according to Table G.2.
From Equation (13.20):
- Step length: S = 0.70m
- Weight of the Person: G = 1kN 2 2 1.00 2
a max = 4π ⋅ 2 ⋅ ---------- ⋅ 0.40 ⋅ 23 = 1.64m ⁄ s (13.29)
886

• Computation of displacements and accelerations by means of - Time history of the displacement

the FE Programme ABAQUS
1.5
Displacement at mid span
Displacements and accelerations are computed by means of
Vertical displacement um [cm]

1
time-history analysis:
- Excitation 0.5
-0.5
-1
-1.5
0 5 10 15 20 25 30 35 40
Time [s]
- Time history of the acceleration

1.6
2
Acceleration at mid span
1.4 F1
acceleration am [m/s2]
1.5
ce / Static weight [-]
F2
1.2
1
F3
1
F15 0.5
0.8 F28 0
0.6
-0.5
Forc
Vertical a
0.4
-1
0.2
-1.5
0
-2
0 2 4 6 8 10 12 14 16
0 5 10 15 20 25 30 35 40
Time [s]
Time [s]

• Remarks 13.3.4 Floors in residential and office buildings

- The refined method and the time history calculations show • Frequency tuning
lower values compared to the rough method;
- If the excitation is generated by walking ( f max ≅ 2.4Hz ), the
- The refined method and the time history calculations are in following natural frequencies shall be exceeded:
good agreement;
Natural frequency
- The time history of the displacement is not symmetric Damping Remark
[Hz]
compared to the time axis, because of the static compo-
Avoid resonance due to the
nent of the displacement caused by the weigh of the cross- > 5% >5
second harmonic
ing person;
Avoid resonance due to the
- The time history of the acceleration is symmetric com- < 5% > 7.5
third harmonic
pared to the time axis, because there is no static compo-
nent of the acceleration.
• Amplitude limitation
- Calculation of the acceleration maximum amplitude
• Swinging footbridge on the Internet
2
http://www.londonmillenniumbridge.com/ a max < ca. 0.05m ⁄ s = 0.5% g (13.30)
- Because of the many non-structural components (wallpa-
http://www.youtube.com per, furniture, suspended ceilings, technical floors, parti-
tions, ....) 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 parame-

ters:

- Position (standing, sitting, lying) • G2 limit for vibrations parallel to the spinal column
- Direction of the action compared to the spinal column
- Activity (resting, walking, running, ...)
- Type of vibration
- ....
Frequency 1 to 10 Hz Frequency 10 to 100 Hz
Description
amax [m/s2] vmax [m/s]
Barely noticeable 0.034 0.0005
Clearly noticeable 0.1 0.0013
Disturbing 0.55 0.0068
Not tolerable 1.8 0.0138
Vertical harmonic vibration action on a standing person. Accepted averages; scat-
ters up to a factor of 2 is possible (from [Bac+97])
• G2 limit for vibrations transverse to the spinal column

• ISO 2631 standard
T
--1- ³ a ( t ) dt
2
a eff = (13.31)
T 0
Where T is the period of time over which the effective accelera-

tion was measured.
3 limits are defined:
- G1: Reduced comfort boundary
- G2: Fatigue-decreased proficiency boundary ~ 3 x G1
- G3: Exposure limit ~ 6 x G1

13.3.5 Gyms and dance halls • Amplitude limitation

Due to gymnastics or dancing, very large dynamic forces are - Calculation of the acceleration maximum amplitude
generated. This is readily understandable when the Fourier co- 2
efficients in Table G2 are considered: a max < ca. 0.5m ⁄ s = 5% g (13.32)
- Walking: α 1 = 0.4 , α 2 = 0.1 , α 3 = 0.1 - Limits depend on the activity, if e.g. people are sitting in the
dance hall, as well, this limit shall be reduced.
- Running: α 1 = 1.6 , α 2 = 0.7 , α 3 = 0.2
- Because of the large forces that can be generated through
- Jumping: α 1 = 1.9 , α 2 = 1.6 , α 3 = 1.1
these activities, the dynamic characteristics of the structure
- Dancing: α 1 = 0.5 , α 2 = 0.15 , α 3 = 0.1 shall be estimated as precisely as possible.
(however: a - many people moving rhythmically. b - certain
dances are very similar to jumping) 13.3.6 Concert halls, stands and diving platforms
• Frequency tuning
See [Bac+97].
- If the excitation is generated through jumping ( f max ≅ 3.4Hz )
or dancing ( f max ≅ 3.0Hz ), then the following natural frequen- 13.4 Machinery induced vibrations
cies shall be exceeded:
Gyms Dance halls It is not possible to carry out here a detailed treatment of machin-
Construction Natural frequency Natural frequency ery induced vibrations. Therefore, reference to [Bac+97] is
[Hz] [Hz] made.
Reinforced concrete > 7.5 > 6.5
Prestressed concrete > 8.0 > 7.0
Composite structures > 8.5 > 7.5
Steel > 9.0 > 8.0

13.5 Wind induced vibrations • Vortex shedding: Periodic action transversely to the wind di-
rection
Wind-induced vibrations cover a challenging and wide area. It is
not possible to carry out here their detailed treatment. Therefore, - Vortex are not shedded left and right at the same time. If the
reference is made to the relevant literature: time-interval of the vortex shedding is equal to the oscilla-
tion period of the structure, resonance excitation occurs.
• [Bac+97]
• Simiu E., Scanlan R.H.: “Wind Effects on Structures”. Third
Edition. John Wiley & Sons, 1996.
13.5.1 Possible effects

• Gusts: Stochastic effects in wind direction fe ⋅ d
u crit = ----------- (13.33)
S
- Turbulent wind with spatially and temporally variable wind
speed. Where
u crit : Critical wind velocity
fe : Natural frequency of the structure transverse to the
wind direction
d: Diameter of the structure
S: Strouhal number (about 0.2 for circular cross sections)
• Buffeting: Periodic action in wind direction
- Vortex detached from an obstacle hit the structure

• “Gallopping” and “Flutter”: 13.6 Tuned Mass Dampers (TMD)

Unstable interaction between wind flow and structural motion
- Gallopping: Motion of the structure transversely to the flow 13.6.1 Introduction
direction. When discussing MDoF systems, in Section 11.1.3 a Tuned
- Flutter: Combined flexural-torsional motion of the structure. Mass Damper (TMD) has already been discussed. However, in
that case zero damping was assumed for both the structure and
Work done by wind forces during flutter the TMD.
There it was possible to solve the equation of

motion simply by means of modal analysis.
Here the theory of the TMD with damping is treated. As we shall

see, the damping of the two degrees of freedom is a design pa-
rameter, and it shall be possible to chose it freely, therefore:
In the case of TMD with damping

Stability curves for bridge cross-sections modal analysis can not be used
• References
[BW95] Bachmann H., Weber B.: “Tuned Vibration Absorbers for
Damping of Lively Structures”. Structural Engineering Inter-
national, No. 1, 1995.
[Den85] Den Hartog J.P.: “Mechanical Vibrations”. ISBN 0-486-
64785-4. Dover Publications,1985. (Reprint of the original
fourth edition of 1956)

13.6.2 2-DoF system 2

° [ – ω m H + iω ( c H + c T ) + ( k H + k T ) ]U H + [ – iωc T – k T ]U T = F H
®
° 2
[ – iωc T – k T ]U H + [ – ω m T + iωc T + k T ]U T = 0
¯
(13.36)
To facilitate the solution of the system, some dimensionless pa-
rameters are now introduced:
γ = mT ⁄ mH : Mass ratio (TMD Mass/Mass of the structure)
ωT = kT ⁄ mT : Natural frequency of the TMD
ωH = kH ⁄ mH : Natural frequency of the structure without TMD
β = ωT ⁄ ωH : Ratio of the natural frequencies
Ω = ω ⁄ ωH : Ratio of the excitation frequency to the natural
frequency of the structure
The equations of motion of the 2-DoF system shown above are:
ζT : Damping ratio of the TMD
·· · · ·
mH uH + cH uH + cT ( uH – uT ) + kH uH + kT ( uH – uT ) = F ( t ) ζH : Damping ratio of the structure
® (13.34)
¯ m T u··T + c T ( u· T – u· H ) + k T ( u T – u H ) = 0 U H0 = F H ⁄ k H : Static deformation of the structure
For an harmonic excitation of the type F ( t ) = F H cos ( ωt ) , a possi- Substituting these dimensionless parameters into Equation
ble ansatz for the steady-state part of the solution is: (13.36) we obtain:
iωt iωt iωt
uH = UH e , uT = UT e , F ( t ) = FH e , (13.35) 2 2 2
° [ – Ω + 2iΩ ( ζ H + βγζ T ) + ( 1 + β γ ) ]U H + [ – 2iΩβγζ T – β γ ]U T = U H0
®
Using the complex numbers formulation allows a particularly el- ° 2 2 2
[ – 2iΩβγζ T – β γ ]U H + [ – Ω γ + 2iΩβγζ T + β γ ]U T = 0
egant solution to the problem. The equations of motion become: ¯
(13.37)

The system of equation can be easily solved using “Maple”, and Thus, the norm of the dynamic amplification function U H ⁄ U H0
we obtain the following expression for the amplification function can be easily calculated:
U H ⁄ U H0 : 2 2 2 2
UH = ( β – Ω ) + ( 2Ωβζ T )
2 2 --------
- --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
-2
( β – Ω ) + 2iΩβζ T 2 2 2 2 2 2 2 2 2 2 2
UH U H0 [ ( β – Ω ) – Ω β ( 1 – γ ) + Ω ( Ω – 4βζ H ζ T ) ] + 4 [ ( β – Ω )ζ H + ( 1 – Ω – Ω γ )βζ T ]
--------
U H0
- = ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
2 2 2 2 2 2 2 2 2 2
- (13.38)
[ ( β – Ω ) – Ω β ( 1 – γ ) + Ω ( Ω – 4βζ H ζ T ) ] + 2i [ ( β – Ω )ζ H + ( 1 – Ω – Ω γ )βζ T ]
(13.44)
The complex expression given in Equation (13.38) shall now be A similar procedure can be followed to compute the dynamic am-
converted into the form: plification function U T ⁄ U H0 .
z = x + iy or U H = U H0 ( x + iy ) (13.39) Next figure show a representation of Equation (13.44) in function
of Ω for an undamped structure ζ H = 0 . Curves for different val-
The displacement U H has therefore two components: 1) One
ues of the parameters β , γ and ζ T are provided.
that is in phase with the displacement U H0 and 2) one with a
phase shift equal to π ⁄ 4 . From the vectorial sum of these two 16
components the norm of U H can be computed as: ]T=0 ]T=inf.
U H = U H0 x + y
2 2
(13.40) J = 1/20
12
E=1
Equation (13.38) has however the form
UH/UH0 [-]
P
( A + iB ) 8
UH = U H0 --------------------- (13.41)
( C + iD ) ]T=0.32
and must be first rearranged as follows:
4
( A + iB ) ⋅ ( C – iD ) ( AC + BD ) + i ( BC – AD ) ]T=0.10 Q
UH = U H0 ----------------------------------------------- = U H0 ----------------------------------------------------------------
-
( C + iD ) ⋅ ( C – iD ) C +D
2 2
(13.42) 0
2 2 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3
A +B : [-]
U H = U H0 -------------------
2 2
(13.43)
C +D

13.6.3 Optimum TMD parameters • Amplification function with TMD: Variation of TMD frequency
Based on observations and consideration at the previous image 50
No TMD (ζT=inf.)
Den Hartog found optimum TMD parameters for an undamped
γ = 0.01
structure: 40
ζΗ = 0.01
fH fH 1
f T, opt = ----------------------------- = ----------- or β opt = ----------- (13.45)
UH/UH0 [-]
30
1 + mT ⁄ mH 1+γ 1+γ
20
3m T ⁄ m H 3γ -
ζ T, opt = --------------------------------------- = --------------------- (13.46) fT=1.02f
1.02fT,opt
T opt
fT=0.98fT,opt
3 3
8 ( 1 + mT ⁄ mH ) 8(1 + γ) 10
fT=fT,opt
These optimum TMD parameters can be applied also to lowly 0

damped structures providing good response results. 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3
Ω [-]
13.6.4 Important remarks on TMD
• Amplification function with TMD: Variation of TMD damping
• The frequency tuning of the TMD shall be quite precise 50
No TMD (ζT=inf.)
• The compliance with the optimum damping is less important
40
γ = 0.01
• Design charts for TMDs shall be computed numerically ζΗ = 0.01
• TMDs are most effective when the damping of the structure is low
UH/UH0 [-]
30
• It is not worth increasing the mass ratio too much

20
• For large mass ratios, the amplitude of the TMD oscillations reduce ζT=1
1.5ζ
5ζT,opt
T t
ζT=0.5ζ
0 5ζT,opt
T opt
10 ζT=ζT,opt
• Meaningful mass ratios γ are 3-5%
• The exact tuning of the TMD occurs experimentally, therefore 0
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3
great care should be paid to construction details.
Ω [-]

• Design charts: Displacement of the structure (from [BW95])

Blank page
• Design charts: Relative TMD displacement (from [BW95])

14 Pedestrian Footbridge with TMD
14.1 Test unit and instrumentation

The test unit is a post-tensioned RC beam. The beam is made
of lightweight concrete and the post-tensioning is without bond.
The dimension of the beam were chosen to make it particularly
prone to vibrations induced by pedestrians. A Tuned Mass
Damper (TMD) is mounted at midspan.
Figure 14.1: View of the test setup.

On the test unit the following quantities are measured:
• Displacement at midspan
• Acceleration at midspan
• Acceleration at quarter point of the span
14 Pedestrian Footbridge with TMD Page 14-1 14 Pedestrian Footbridge with TMD Page 14-2
• Characteristics of the TMD 14.2 Parameters

A close-up of the TMD is shown in Figure 14.2. We can see:
14.2.1 Footbridge (Computed, without TMD)
• The 4 springs that define the stiffness K T of the TMD
• The 4 viscous dampers that define the damping constant c opt of Modal mass: M H = 5300kg
the TMD
Modal stiffness: K H = 861kN ⁄ m
• 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
Natural frequency: f H = 2.03Hz
the fine-tuning of the TMD.
(Computed with TMD mass: f = 1.97Hz )
The properties of the TMD are given in Section 14.2.
14.2.2 Tuned Mass Damper (Computed)
Mass: M T = 310kg
MT
Mass ratio: μ = -------- = 0.0585 = 5.85%
MH
fH
Natural frequency: f opt = ------------ = 1.92Hz
1+μ
(Measured: f T = 1.91Hz )
2
Stiffness: K T = M T ⋅ ( 2πf opt ) = 50.9kN/m
3μ
Damping rate ζ opt =
----------------------- = 0.14 = 14%
8( 1 + μ )3
(Measured: ζ T = 13% )
Damping constant: c opt = 2ζ opt K T M T = 1.18kNs/m

Figure 14.2: Close-up of the TMD.
14.3 Test programme 14.4 Free decay test with locked TMD
Following tests are carried out: Time history of the displacement at midspan
No. Test Action location TMD
40
Displacement at midspan [mm]

1 Free decay Midspan Locked
30
2 Sandbag Midspan Locked 20
3 Sandbag Quarter-point Locked 10

0
4 Sandbag Midspan Free
-10
5 Walking 1 Person 3Hz Along the beam Locked -20
6 Walking 1 Person 2Hz Along the beam Locked -30
-40
7 Walking 1 Person 2Hz Along the beam Free
0 5 10 15 20
8 Walking in group 2Hz Along the beam Locked Time [s]
Figure 14.3: Free decay test with locked TMD: Displacement at
9 Walking in group 2Hz Along the beam Free
midspan.
10 Jumping 1 Person 2Hz Midspan Locked
Evaluation:
11 Jumping 1 Person 2Hz Midspan Free
Logarithmic
Damping ratio
decrement
Typical results of the experiments are presented and briefly
Region 1 1 41.36 0.081
commented in the following sections. δ = --- ln ------------- = 0.081 ζ H = ------------- = 1.29%
Average amplitude: ~30mm 8 21.66 2π
Region 2 1 19.91 0.090
δ = --- ln ------------- = 0.090 ζ H = ------------- = 1.43%
Region 3 1 8.13 0.092
δ = --- ln ---------- = 0.092 ζ H = ------------- = 1.46%
Fourier-spectrum of the displacement at midspan 14.5 Sandbag test

2.0 The sandbag test consists in hanging a 20 kg sandbag 1 meter
above the footbridge, letting it fall down and measuring the re-
sponse of the system.
"Spectral displacement"
1.5
In order to excite the different modes of vibration of the foot-
bridge, the test is repeated several times changing the position
1.0 of the impact of the sandbag on the bridge. The considered lo-
cations are:
0.5 - at midspan (Section 14.5.1)
- at quarter-point of the span (Section 14.5.2).
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 These tests are carried out with locked TMD. In order to investi-
Frequency [Hz] gate the effect of the TMD on the vibrations of the system, the
Figure 14.4: Free decay test with locked TMD: Fourier-spectrum of the test of Section 14.5.1 is repeated with free TMD (see Section
displacement at midspan. 14.5.3).
The measured natural frequency of the footbridge with locked Remark

TMD is equal to:
• The results presented in Section 14.5.1 and those presented
f = 1.89Hz (14.1)
in Section 14.5.2 and 14.5.3 belongs to two different series of
This value is less than the value given in Section 14.2.1. This tests carried out at different point in time. Between these test
can be explained with the large amplitude of vibration at the start series the test setup was completely disassembled and reas-
of the test, which causing the opening of cracks in the web of the sembled. Slight differences in the assemblage of the test setup
beam, hence reducing its stiffness. (support!) may have led to slightly different natural frequencies
of the system.
The second peak in the spectrum corresponds to f = 1.98Hz ,
wich is in good agreement with Section 14.2.1.
14.5.1 Locked TMD, Excitation at midspan

Acceleration an midspan [m/s2]
1.0 1.0
Acc. at quarter-point [m/s2]

0.5 0.5
0.0 0.0
-0.5 -0.5
-1.0 -1.0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Time [s] Time [s]
Figure 14.5: Sandbag test with locked TMD: Acceleration at midspan. Figure 14.7: Sandbag test with locked TMD: Acceleration at quarter-
point of the span.
0.040 0.040
f1=2.00Hz
"Spectral acceleration"
0.030 0.030
f1=2.00Hz
0.020 0.020
0.010 f3=18.06Hz 0.010 f3=18.06Hz
0.000 0.000
0 5 10 15 20 0 5 10 15 20
Frequency [Hz] Frequency [Hz]
Figure 14.6: Sandbag test with locked TMD: Fourier-spectrum of the Figure 14.8: Sandbag test with locked TMD: Fourier-spectrum of the
acceleration at midspan. acceleration at quarter-point of the span.
Remarks 14.5.2 Locked TMD, Excitation at quarter-point of the span
• With the sandbag test in principle all frequencies can be excit- 1.0
Acceleration at midspan [m/s2]

ed. Figures 14.5 and 14.7 show a high-frequency vibration,
which is superimposed on a fundamental vibration; 0.5
• The Fourier amplitude spectrum shows prominent peaks at the

first and third natural frequencies of the system (Footbridge 0.0
with locked TMD);
-0.5
• The second mode of vibration of the system is not excited, be-
cause the sandbag lands in a node of the second eigenvector.
-1.0
• At midspan, the amplitude of the vibration due to the first mode 0 1 2 3 4 5 6 7 8 9 10
Time [s]
of vibration is greater than at quarter-point. The amplitude of
Figure 14.9: Sandbag test with locked TMD: Acceleration at midspan.
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. 0.040
• The vibration amplitude is relatively small, therefore, the meas-
0.030
ured first natural frequency f 1 = 2.0Hz in good agreement with f1=2.00Hz
the computation provided in Section 14.2.1.
0.020
0.010 f3=18.31Hz
0.000
0 5 10 15 20
Frequency [Hz]
Figure 14.10: Sandbag test with locked TMD: Fourier-spectrum of the
acceleration at midspan.
Remarks
1.0 • When the sandbag lands at quarter-point of the bridge, the

Acc. at quarter-point [m/s2]
second mode of vibration of the system is strongly excited. Its

0.5 contribution to the overall vibration at quarter-point of the foot-
bridge is clearly shown in Figures 14.11 and 14.12.
0.0 • The acceleration sensor located at midspan of the footbridge
lays in a node of the second mode of vibration, and as expect-
-0.5 ed in figures 14.9 and 14.10 the contribution of the second
mode is vanishingly small.
-1.0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Figure 14.11: Sandbag test with locked TMD: Acceleration at quarter-
point of the span.
0.040
0.030 f2=8.84Hz
0.020 f1=2.00Hz
0.010 f3=18.31Hz
0.000
0 5 10 15 20
Frequency [Hz]
Figure 14.12: Sandbag test with locked TMD: Fourier-spectrum of the
acceleration at quarter-point of the span.
14.5.3 Free TMD: Excitation at midspan Remarks
1.0 • With active (free) TMD the “first” and the “third “natural fre-
quencies of the bridge are excited. As expected, these fre-

0.5 quencies are slightly larger than the natural frequencies of the
system (bridge with locked TMD), which are given in Figure
0.0
14.6. This is because the mass of the TMD is no longer locked
and can vibrate freely.
-0.5 • The effect of the TMD is clearly shown in Figure 14.14. The
amplitude of the peak in the “first natural frequency” is much
-1.0 smaller than in Figure 14.6. The amplitude of the peak at the
0 1 2 3 4 5 6 7 8 9 10 “third natural frequency” is practically the same. The “third
Time [s]
mode of vibration” is only marginally damped by the TMD.
Figure 14.13: Sandbag test with free TMD: Acceleration at midspan.
• In the two comments above, the natural frequencies are men-
tioned in quotes, because by releasing the TMD number and
0.040 properties of the natural vibrations of the system change. A di-
rect comparison with the natural vibrations of system with
0.030 locked TMD is only qualitatively possible.
0.020
“f1=2.05Hz”
“f3=19.04Hz”
0.010
0.000
0 5 10 15 20
Frequency [Hz]
Figure 14.14: Sandbag test with free TMD: Fourier-spectrum of the
acceleration at midspan.
14.6 One person walking with 3 Hz Test results
One 65 kg-heavy person (G = 0.64 kN) crosses the footbridge.

2.0
He walks with a frequency of about 3 Hz, which is significantly
larger than the first natural frequency of the bridge. 1.0
Remarks 0.0
• The static deflection of the bridge when the person stands at

-1.0
midspan is:
G 0.69 -2.0
d st = ------- = ---------- = 0.00080m = 0.80mm
KH 861 0 5 10 15 20 25 30 35 40
Time [s]
• The maximum measured displacement at midspan of the Figure 14.15: One person walking with 3 Hz: Displacement at midspan
bridge is about 2 mm (see Figure 14.15), which corresponds to with locked TMD.
about 2.5 times d st . As expected, the impact of dynamic effects
1.0

is rather small.
• In the Fourier spectrum of the acceleration at midspan of the 0.5
bridge (see Figure 14.17), the frequencies that are represent-
ed the most correspond to the first, the second and the third 0.0
harmonics of the excitation. However, frequencies correspond-
ing to the natural modes of vibrations of the system are also -0.5
visible.
-1.0
0 5 10 15 20 25 30 35 40
Time [s]
Figure 14.16: One person walking with 3 Hz: Acceleration at midspan
with locked TMD.
0.015 14.7 One person walking with 2 Hz

One 95 kg-heavy person (G = 0.93 kN) crosses the footbridge. He

0.010 walks with a frequency of 1.95 Hz, which is approximately equal to the
first natural frequency of the bridge. The length of the step is 0.70 m.
Sought is the response of the bridge under this excitation. A
0.005 similar problem was solved theoretically in Section 13.3.3.
14.7.1 Locked TMD (Measured)

0.000
0 5 10 15 20
Frequency [Hz]
First the maximum amplitudes are calculated by hand:
Figure 14.17: One person walking with 3 Hz: Fourier-Spectrum of the G 0.93
Static displacement: d st = ------- = ---------- = 0.00108m = 1.08mm
acceleration at midspan with locked TMD. KH 861
(Measured: d st = 1.22mm )
Walking velocity: v = S ⋅ f 0 = 0.7 ⋅ 1.95 = 1.365m ⁄ s
Crossing time: Δt = L ⁄ v = 17.40 ⁄ 1.365 = 12.74s

Number of cycles: N = Δt ⋅ f n = 12.74 ⋅ 1.95 = 25
Amplification factor: Φ = 22 (From page 13-20 with ζ H = 1.6% )

2 2
Max. acceleration: a max = 4π ⋅ 1.95 ⋅ 0.00108 ⋅ 0.4 ⋅ 22
= 1.43m ⁄ s 2
(Measured: a max = 1.63m ⁄ s 2 )
Max. dyn. displ.: d dyn,max = 1.08 ⋅ 0.4 ⋅ 22 = 9.50mm
Max. displacement: d max = 9.50 + 1.08 = 10.58mm

(Measured: d max = 12.04mm )
Test results 14.7.2 Locked TMD (ABAQUS-Simulation)


10 10
5 5
0 0
-5 -5
-10 -10
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Time [s] Time [s]
Figure 14.18: One person walking with 2 Hz: Displacement at midspan Figure 14.20: One person walking with 2 Hz: Displacement at midspan
with locked TMD. with locked TMD. (ABAQUS-Simulation).
2.0 2.0

1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.0
-0.5 -0.5
-1.0 -1.0
-1.5 -1.5
-2.0 -2.0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Time [s] Time [s]
Figure 14.19: One person walking with 2 Hz: Acceleration at midspan Figure 14.21: One person walking with 2 Hz: Acceleration at midspan
with locked TMD. with locked TMD. (ABAQUS-Simulation).
The curves in Figures 14.20 and 14.21 were computed using the 14.7.3 Free TMD
FE program ABAQUS. A similar calculation is described in detail

in Section 13.3.3. The input data used in that section were only
10
slightly adjusted here in order to better describe the properties of
the test. 5
Maximum vibration amplitude 0
Static displacement: d st = 1.08mm -5

-10
Maximum displacement: d max = 11.30mm 0 5 10 15 20 25 30 35 40

(Measured: d max = 12.04mm ) Time [s]
Figure 14.22: One person walking with 2 Hz: Displacement at midspan
d max 11.30 with free TMD.
Amplification factor: V = ----------- = ------------- = 10.5
d st 1.08
2.0

Maximum acceleration: a max = 1.68m ⁄ s 2 1.5
1.0
0.5
The maximum amplitudes of the numerical simulation and of the 0.0
experiment agree quite well and also the time-histories shown in -0.5
Figures 14.18 and 14.21 look quite similar. -1.0
Please note that during the first 2 seconds of the experiment, -1.5
displacements and accelerations are zero, because the person -2.0
0 5 10 15 20 25 30 35 40
started to walk with a slight delay. time [s]
Figure 14.23: One person walking with 2 Hz: Acceleration at midspan
with free TMD.
Estimate of the maximum vibration amplitude 14.8 Group walking with 2 Hz

Amplification factor: about 5.5 (from page 13-41) All student participating to the test (24 people) cross the foot-
bridge in a continuous flow. A metronome is turned on to ensure
Maximum dyn. displ.: d dyn,max = 1.08 ⋅ 0.4 ⋅ 5.5 = 2.38mm
that all students walk in the same step and with a frequency of
about 2 Hz.
Maximum displacement: d max = 2.38 + 1.08 = 3.46mm
The test is carried out both with locked (Section 14.8.1) and free
(Measured: d max = 3.27mm )
(Section 14.8.2) TMD.
2 2
Maximum acceleration.: a max = 4π ⋅ 1.95 ⋅ 0.00108 ⋅ 0.4 ⋅ 5.5 In Figures 14.24 to 14.27 the first 40 seconds of the response of
= 0.36m ⁄ s 2 the bridge are shown.
(Measured: a max = 0.34m ⁄ s 2 ) Remarks
14.7.4 Remarks about “One person walking with 2 Hz” The results of the experiments with several people walking on
the bridge are commented by using the results of tests with one
• The effect of the TMD can be easily seen in Figures 14.22 and person walking (see Section 14.7) as comparison. For this rea-
14.23. The maximum acceleration at midspan reduces from son the maximum vibration amplitudes shown in Figures 14.18,
1.63m ⁄ s 2 to 0.34m ⁄ s 2 , which corresponds to a permissible 14.19, 14.22, 14.23 and 14.24 to 14.27 are summarised in Ta-
value. bles 14.1 and 14.2.
Case Group 1 person ratio

Maximum acceleration at
2.05 m/s2 1.63 m/s2 1.26
midspan. Locked TMD
Maximum acceleration at
0.96 m/s2 0.34 m/s2 2.82
midspan. Free TMD
Ratio 2.14 4.79
Table 14.1: Comparison of the accelerations at midspan.
Case Group 1 person ratio of the displacement is larger that the amplification factor of the
accelerations, because the static deflection caused by the
Maximum displacement at
20.52 mm 12.04 mm 1.70 group is significantly larger than that caused by the single per-
midspan. Locked TMD
son.
Maximum displacement at
12.28 mm 3.27 mm 3.76 • The activation of the TMD results in a reduction of the maxi-
midspan. Free TMD
mum acceleration caused by the single person by a factor of
Ratio 1.67 3.68
4.79. In the case of the group the reduction factor is only 2.14.
Table 14.2: Comparison of the displacements at midspan. It should be noted here that when the TMD is active (free), the
vibrations are significantly smaller, and therefore it is much
It is further assumed that only about 16 of the 24 persons are on
easier for the group to walk in step. It is therefore to be as-
the footbridge at the same time.
sumed that in the case of the free TMD, the action was strong-
The following remarks can thereby be made: er than in the case of the locked TMD. This could explain the
seemingly minor effectiveness of the TMD in the case of the
• The maximum acceleration measured at midspan of the bridge
group.
with locked TMD is only about 1.26-times greater than the ac-
celeration which has been generated by the single person. Ac-
cording 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 displace-
ment generated by the single person. The amplification factor
14.8.1 Locked TMD 14.8.2 Free TMD


10 10
0 0
-10 -10
-20 -20
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Time [s] Time [s]
Figure 14.24: Group walking with 2 Hz: Displacement at midspan with Figure 14.26: Group walking with 2 Hz: Displacement at midspan with
locked TMD. free TMD.

2.0 2.0
1.0 1.0
0.0 0.0
-1.0 -1.0
-2.0 -2.0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
time [s] Time [s]
Figure 14.25: Group walking with 2 Hz: Acceleration at midspan with Figure 14.27: Group walking with 2 Hz: Acceleration at midspan with
locked TMD. free TMD.
14.9 One person jumping with 2 Hz Test results
One 72 kg-heavy person (G = 0.71 kN) keeps jumping at mid-

60
span of the footbridge. He is jumping with a frequency of 40
1.95 Hz, which is approximately equal to the first natural fre-
quency of the bridge. 20
Sought is the response of the bridge under this excitation. A 0

similar problem was solved theoretically in Section 6.1.3. -20
14.9.1 Locked TMD -40

-60
First the maximum amplitudes are calculated by hand: 0 5 10 15 20 25 30 35 40
Time [s]
G 0.71
Static displacement: d st = ------- = ---------- = 0.0008m = 0.82mm Figure 14.28: One person jumping with 2 Hz: Displacement at midspan
KH 861
with locked TMD.
8.0

1 1 6.0
Amplification factor: V = ------ = ------------------------- = 31.25
2ζ ( 2 ⋅ 0.016 ) 4.0
2 2 2.0
Maximum acceleration: a max = 4π ⋅ 1.95 ⋅ 0.0008 ⋅ 1.8 ⋅ 31.25
0.0
= 6.92m ⁄ s 2
-2.0
-4.0
Max. dyn. displacement: d dyn,max = 0.82 ⋅ 1.8 ⋅ 31.25 = 46.13mm -6.0

-8.0
0 5 10 15 20 25 30 35 40
Maximum displacement: d max = 46.13 + 0.82 = 46.95mm Time [s]
(Measured: d max = 51.08mm ) Figure 14.29: One person jumping with 2 Hz: Acceleration at midspan
with locked TMD.
14.9.2 Free TMD Estimate of the maximum vibration amplitude
Amplification factor: about 5.5 (from page 13-41)

60
40 Maximum dyn. displ.: d dyn,max = 0.82 ⋅ 1.8 ⋅ 5.5 = 8.12mm
20
Maximum displacement: d max = 8.12 + 0.82 = 8.94mm
0 (Measured: d max = 8.12mm )
-20
2 2
Maximum acceleration: a max = 4π ⋅ 1.95 ⋅ 0.0008 ⋅ 1.8 ⋅ 5.5
-40
= 1.22m ⁄ s 2
-60
0 5 10 15 20 25 30 35 40 (Measured: a max = 1.04m ⁄ s 2 )
Time [s]
Figure 14.30: One person jumping with 2 Hz: Displacement at midspan 14.9.3 Remarks about “One person jumping with 2 Hz”
with free TMD.
• When jumping, the footbridge can be much strongly excited
8.0 than when walking.
6.0
4.0 • The achieved acceleration a max = 7.18m ⁄ s 2 = 73% g is very
2.0 large and two jumping people could easily produce the lift-off
0.0 of the footbridge.
-2.0 • The effect of the TMD can be easily seen in Figures 14.30 and
-4.0 14.31. The maximum acceleration at midspan reduces from
-6.0 7.18m ⁄ s 2 to 1.04m ⁄ s 2 , what, however, is still perceived as un-
-8.0 pleasant.
0 5 10 15 20 25 30 35 40
Time [s]
Figure 14.31: One person jumping with 2 Hz: Acceleration at midspan
with free TMD.

SD An Najah 2013 DS-alesandro PDF

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

SD An Najah 2013 DS-alesandro PDF

Transféré par

Droits d'auteur :

Formats disponibles

Course “Fundamentals of Structural Dynamics”

A. Dazio, April 19, 2013 Page 1/2 Page 2/2

Table of Contents 3.2 Damped free vibrations ....................................................................... 3-6

Table of Contents Page i Table of Contents Page ii

7.6 Strength and Ductility ........................................................................ 7-58

Table of Contents Page iii Table of Contents Page iv

9.4 Properties of the eigenvectors.......................................................... 9-13

Table of Contents Page v Table of Contents Page vi

Table of Contents Page vii Table of Contents Page viii

1 Introduction 1.3 Topics of the course

1) Systems with one degree of freedom

1.2 Limitations of the course 3) Systems with many degree of freedom

• Numerical methods of structural dynamics are treated only 4) Continuous systems

1 Introduction Page 1-1 1 Introduction Page 1-2

Practical cases (Vibration problems)

1 Introduction Page 1-3 1 Introduction Page 1-4

2 Single Degree of Freedom Systems 2) D’Alembert principle

Introducing the spring force f k ( t ) = ku ( t ) and the damping

2.1.2 Principle of virtual work 2.2 Example “Inverted Pendulum”

E = T + U + V = T o + U o + V o = cons tan t (2.15) External force: Fp = m ⋅ g (2.19)

The equation of motion given by Equation (2.30) corresponds to

Energy Formulation for small angles ϕ 1 we have:

Comparison of the energy maxima 2.3 Modelling

2.3.2 Structures with distributed mass L

u'' = ψ''U and ··

δu = ψδU and δ [ u'' ] = ψ''δU (2.55)

L L • Equation (2.49) is valid for all virtual displacements, therefore:

• Circular frequency • Example No. 1: Cantilever with distributed mass

• Choosing the deformation figure

- Useful results can be obtained even if the assumed defor-

• Example No. 2: Cantilever with distributed mass

1.652 EI 1.652 8.638 ×104

3.007EI - = 1.652 -----------

Hysteretic Relative External contact

• Typical values of damping in structures

SAP2000 v8 - File:HEB_360 - Mode 1 Period 1.2946 seconds - KN-m Units

Source: A. Marioni: “Innovative Anti-seismic Devices for Bridges”.

3 Free Vibrations • Relationships

2 The static equilibrium is disturbed by the initial displacement

3 Free Vibrations Page 3-1 3 Free Vibrations Page 3-2

3.1.2 Formulation 2: Trigonometric functions 3.1.3 Formulation 3: Exponential Functions

mu··( t ) + ku ( t ) = 0 (3.12) mu··( t ) + ku ( t ) = 0 (3.19)

and by means of Euler’s formulas

3 Free Vibrations Page 3-3 3 Free Vibrations Page 3-4

Equation (3.25) can be transformed as follows: 3.2 Damped free vibrations

3.2.1 Formulation 3: Exponential Functions

mu··( t ) + cu· ( t ) + ku ( t ) = 0 (3.32)

3 Free Vibrations Page 3-5 3 Free Vibrations Page 3-6

• Critical damping when: c 2 – 4km = 0 • Types of vibrations

• Damping ratio Critically damped vibration

3 Free Vibrations Page 3-7 3 Free Vibrations Page 3-8

3 Free Vibrations Page 3-9 3 Free Vibrations Page 3-10

• Notes 3.3 The logarithmic decrement

3 Free Vibrations Page 3-11 3 Free Vibrations Page 3-12

we obtain: • Evaluation over several cycles

3 Free Vibrations Page 3-13 3 Free Vibrations Page 3-14

3.4 Friction damping • Free vibrations

mu··( t ) + ku ( t ) = – f μ mu··( t ) + ku ( t ) = f μ -10

3 Free Vibrations Page 3-15 3 Free Vibrations Page 3-16

- Step 2: • Comparison Viscous damping vs. Friction damping

• Important note: Comparison:

- The period of the damped and of the undamped oscillator -10

3 Free Vibrations Page 3-17 3 Free Vibrations Page 3-18