You are on page 1of 114

DYNAMICS OF STRUCTURES

Applications 2nd Edition

Structural Engineering Master Program

Mihail Iancovici
Technical University of Civil Engineering Bucharest
UTCB
[page intentionally left blank]
Mihail Iancovici

DYNAMICS OF STRUCTURES

Applications
Structural Engineering Master Program

[Digital Edition]

Department of Structural Mechanics


Technical University of Civil Engineering of Bucharest (UTCB)

2017
DISCLAIMER:
The author assumes no responsibility for the inappropriate use of tools developed and
reproduced in the textbook.
The book does not contain any privileged information or conflicted with copyright
regulations.
Professional opinions expressed by the author, do not state or reflect the Technical
University of Civil Engineering of Bucharest (UTCB) views or opinions.

ii
Contents Page

Preface to the 2nd [Digital Edition] v

Chapter 1. Signal processing for Structural Engineering applications 1

Chapter 2. Computational Structural Dynamics 15

Chapter 3. Damping in structures 39

Chapter 4. Dynamic response analysis. Step-by-step integration methods 49

Chapter 5. Dynamics of unconventional structures. Vibration control 85

ANNEX: Computational algebra using Matlab 101

References 103

iii
[page intentionally left blank]

iv
Preface to the 2nd [Digital Edition]

The rapid growth of practical applications to be solved using the tools provided by
Dynamics of Structures field requires continuous updates of the accompanying support
materials of the course. Therefore several updates were performed over the 1st Edition in
order to bring new class and individual work examples, and to make the original
examples more practical, readable and understandable.
The progress addressing various current Civil Engineering applications in the vast field of
Dynamics of Structures is nowadays remarkable. This is primarily thanks to (i) the
developments in the digital instrumentation of various load types, either from natural
hazard (wind, earthquake, waves etc.) or man-made (industrial loads, blast, impact etc.)
sources and (ii) the structural response instrumentation. The rapid growth of the
instrumentation techniques allowed thus a more accurate load and structural modeling.
New analytical models and procedures are able to reproduce with a higher degree of
accuracy and realism, the structural behavior.
The numerical examples, accompanying the Lecture Notes are primarily dedicated to the
Master students in the field of Structural Engineering but not necessarily restricted to
them. These could be also used by undergraduate and Ph.D. students as well, to solve
practical civil engineering applications.
The reader is supposed to be familiar with fundamentals, both theory and applications, of
Structural Analysis and Structural Dynamics, from the undergraduate level. Handling
basic computer applications, using Microsoft Excel- and MATLAB-based routines, is also
requested. In the annex of this textbook, a brief introduction to MATLAB basic
computational algebra is given.
The numerical examples presents both, basic and actual topics from the Structural
Dynamics field, serving for a better understanding of vibration phenomena and associated
dynamic analysis tools, embedded in the computer programs, largely used in practice.
However, this textbook is not about using specific commercial software to solve
engineering applications. It is a theoretical and practical platform for higher-level
approaches that will be further developed during Master and Ph.D. programs.
The numerical examples include several routines developed by author. However, based
on these examples, the students are requested to bring further developments for specific
applications, either in a form of immediate tasks or homeworks- that require additional

v
working time. Due to time and space limitations, the applications are restricted to the most
relevant topics.
The author hopes that the reader will find the textbook as a useful tool for a
comprehensive understanding of the nature and characterization of dynamic loads, of the
phenomena associated to the dynamic behavior of civil structures, and gets the ability to
formulate and solve a broad range of applications in the current research and structural
design practice.

Mihail Iancovici

2017

vi
1. Signal processing for Structural Engineering applications

In the Structural Engineering field a wide range of signals, either recorded, scaled or
artificially generated are used for analysis and design purposes. The loads features are
revealed by performing time- and frequency-domain analyses. Thus, parameters as peak
values, amplitudes, frequency content and phase, and duration, as well, will provide
essential information for design purposes. In this chapter, the features of time- and
frequency-domain analysis of typical signals used in practice, are developed, following the
theoretical background developed in the Chapter 1 of the accompanying Lecture Notes.

Numerical example 1.1| Various loads representations can be obtained by using a time-
history signal generator. An example of a signal generator routine by Matlab platform is
given in the Figure 1.1a for the simplest harmonic sine function. Similar signal generator
can be easily constructed using the Microsoft Excel platform. Further on, this routine will
be augmented, in order to generate a wider range of periodic or random signals.

% Signal generator
% coded by Mihail Iancovici,UTCB

clear all
% generate a sine function x(t)
x0=2;% amplitude (force, acceleration, pressure units etc.)
omega=4*pi; %circular frequency, rad/s
tf=10; % final time sequence, s
nint=1000;% number of intervals
dt=tf/nint;%sampling time-step
t=0:dt:tf; % time interval
Npoints=unum(t);% number of discrete data points
durat=(Npoints-1)*dt; % time duration,s
x=x0*sin(omega*t);% sine function generator
% plot function
plot(t,x,'linewidth',2)
xlabel('time,s','fontsize',26);ylabel('x(t)','fontsize',26);

Figure 1.1a – Time-history sine signal generator

By running the above routine, the following sine function yields (Figure 1.1b)

1
2

1.5

0.5

x(t)
0

-0.5

-1

-1.5

-2
0 1 2 3 4 5 6 7 8 9 10
time,s

Figure 1.1b – Time-history sine function

Let us generate a triangular pulse, a simplified model for blast load (see eq. 1 in the Lecture
Notes). The routine in the Figure 1.1a is modified to include the tripuls function from
Matlab library (Figure 1.2a).
% Signal generator
% coded by Mihail Iancovici,UTCB

clear all
% generate a triangular pulse/blast load- x(t)
x0=100;% amplitude (force, acceleration, pressure etc.)
tf=10; % final time sequence, s
nint=1000;% number of time-intervals
dt=tf/nint;%sampling time-step
t=0:dt:tf; % time interval
Npoints=unum(t);% number of discrete data points
durat=(Npoints-1)*dt; % time duration,s

td=2;% blast duration


x= x0*tripuls(t,2*td); % triangular pulse/blast load generator

% plot function
plot(t,x,'linewidth',2)
xlabel('time,s','fontsize',26);ylabel('x(t)','fontsize',26);

Figure 1.2a – Time-history triangular pulse signal generator

By running the above routine, the following triangular pulse function yields (Figure 1.2b)

100

80

60
x(t)

40

20

0
0 1 2 3 4 5 6 7 8 9 10
time,s

Figure 1.2b – Time-history triangular pulse signal


2
Numerical example 1.2| Following the basic Fourier decomposition algorithm given in the
Lecture Notes, an efficient automated algorithm is developed for the frequency
decomposition of periodic or random functions.

This is based on the Fast Fourier Transform (FFT) algorithm, developed by Cooley and
Tukey (1965), widely used in all engineering applications. It requires a number of data
points power of 2, so that 𝑛 = 2 . In most cases the number of recorded data points does
not fulfill this requirement so that either interpolation or zero padding of initial data is
necessary (Smith, 1997).
The circular sampling frequency is therefore

𝜔 = ∆
, rad/s (1.1)

or in terms of sampling frequency

𝑓 = ∆ , Hz (1.2)

where ∆𝑡 is the sampling time-step.


2 data points in the time-domain representation 𝑥(𝑡) will correspond to 2 data points in
the symmetric frequency- domain |𝑋(𝑖𝜔)| representation of Fourier amplitudes (Figure
1.3).

Time domain Frequency domain

|X(iω)|
x[t]

t 0
-ωs/2 +ωs/2

2n discrete points 2n discrete points

Figure 1.3- Time- and frequency-domain representations of a typical signal 𝑥(𝑡)

Right hand side figure is called two-sided Fourier spectrum, where

𝜔 = = ∆ , rad/s (1.3)

is called the Nyquist frequency (Smith, 1997). This defines the frequency limit in the
Fourier spectrum.
In terms of frequencies,
3
𝑓 = = ∆
, Hz (1.4)

In the practical applications, the one-sided spectrum is only retained since negative
frequencies have no engineering meaning.
The corresponding frequency step of the Fourier transform is therefore

∆𝑓 = =∆ ∙
, Hz (1.5)

Obviously, the number of data points 𝑛 controls the signal’s resolution. The influence of
number of data points on the spectral representation is shown in the example included in
the Lecture Notes.
The frequency decomposition of unit amplitude ascending “sawtooth” function having 1 s
period is represented in the Figure 1.5.
The accompanying analysis routine is provided in the Figure 1.4.

% Frequency domain analysis by FFT


% coded by Mihail Iancovici, UTCB

clear all
% signal parameters
omega=2*pi; %circular frequency, rad/s
tf=10; % final time sequence, s
nint=1000;% number of time-intervals
dt=tf/nint;% sampling time-step
t=0:dt:tf; % time interval
Npoints=unum(t);% number of discrete data points
durat=(Npoints-1)*dt; % time duration,s

% signal generator
x0=1; % amplitude
width=1;
f = sawtooth(omega*t,width); % triangular (sawtooth) function

figure (1);
plot(t,f,'linewidth',2)
xlabel ('time,s','fontsize',12);ylabel('Function','fontsize',12);
box off;

% FOURIER ANALYSIS (FFT)


fs = 1/dt; % sampling frequency, Hz
NFFT=4096; % number of points for FFT (resolution)
df=fs/NFFT; % frequency step,Hz
freq = df:df:(NFFT/2)*df; % frequency range for plot
FT = fft(f,NFFT);% FFT function
FA = dt*abs(FT); % FOURIER amplitude
p = unwrap(angle(FT)); % phase angle

[af,bf]=max(FA);
fp=bf*df;%predominant frequency, Hz
FAmax=af;%largest Fourier amplitude
Tp=1/fp;%predominant period,s

4
%PLOTS
limitf = 20; % [V] limit for plot
freqp = df:df:limitf; % frequency range for plot

figure (2);
subplot(2,1,1);plot(freqp,FA(1:unum(freqp)),'linewidth',2);
xlabel ('Frequency,Hz','fontsize',12);
ylabel('FA','fontsize',10);
box off;
subplot(2,1,2);plot(freqp,p(1:unum(freqp)),'linewidth',2);
xlabel ('Frequency,Hz','fontsize',12);
ylabel ('Phase','fontsize',12);
box off;

Figure 1.4 – FFT analysis routine

By running the above routine, the following results are obtained (Figure 1.5).

Time- domain Frequency- domain


1 4

0.8 3
FA

0.6 2

0.4 1

0.2 0
0 2 4 6 8 10 12 14 16 18 20
x(t)

0 Frequency,Hz
-0.2 200

-0.4 0
Phase

-0.6 -200

-0.8 -400

-1 -600
0 1 2 3 4 5 6 7 8 9 10 0 2 4 6 8 10 12 14 16 18 20
time,s Frequency,Hz

Figure 1.5 – Time- and frequency- domain representations of triangular periodic pulse
(sawtooth function)

The Fourier spectrum indicates a predominant frequency of 𝑓 = 1𝐻𝑧 component and


gradually negligible higher frequency components. The numerical results are validated
through available “exact” solution (Ifrim, 1984; Smith, 1997).
 Task|Using the basic Fourier decomposition (see Chapter 1 in the accompanying
Lecture Notes), check the validity of FFT-based decomposition by hand calculation.
Using the routine provided in the Figure 1.4, a large number of typical signals are
analyzed and shown in the accompanying Lecture Notes (Chapter 1, Figure 1.16).

Numerical example 1.3| Given the record of Vrancea Earthquake of 4 March 1977,
consisting of two orthogonal records obtained at Bucharest INCERC station, the time- and
frequency- domain analyses of both components are performed.
A typical ground acceleration input file format is given in the Figure 1.6.

5
VN 1977 NS comp, INCERC Bucharest Station 4/03/1977
ACCELERATION TIME HISTORY: m/s/s.
DT= .02000 SEC

time,s Acc.,m/s/s
0.02 -0.05946
0.04 0.03985
0.06 0.03978
0.08 0.04721
0.1 0.04714
0.12 0.05206 ...

Figure 1.6–Strong motion acceleration record data file (NS component)

The ground velocity and displacement are obtained by successive numerical integration
using the trapezoidal rule scheme (see the Lecture Notes; Chapra et al., 1988). In order to
validate the integration scheme (provided in the Figure 1.7), the testing harmonic function,
given by the amplitude 𝑥 and circular frequency 𝜔, as
𝑥(𝑡) = 𝑥 sin(𝜔𝑡) (1.6)
is integrated first by hand calculation.
Therefore, the closed-form “exact” solution is

𝐼 = ∫ 𝑥(𝜏)𝑑𝜏 = ∫ 𝑥 sin(𝜔𝜏) 𝑑𝜏 =  −𝑥 cos(𝜔𝑡) = (1 − 𝑐𝑜𝑠𝜔𝑡) (1.7)

The numerical result is compared with that given by the “exact” formulation (1.7).
Additionally this is validated through the Matlab embedded function called cumptrapz
(Figure 1.7) which can be used alternatively.

% Numerical integration by trapezoidal rule


% coded by Mihail Iancovici, UTCB

clear all
% generate a sine function a(t)
a0=2;% amplitude, m/s/s
omega=4*pi; %circular frequency, rad/s
tf=10; % final time sequence, s
nint=500;% number of intervals
dt=tf/nint;%time step
t=0:dt:tf; % time interval
Npoints=unum(t);% number of discrete data points
durat=(Npoints-1)*dt; % time duration,s
a=a0*sin(omega*t);% sine function

6
% 1. Trapezoidal rule integration scheme
for i=1
v(i)=a(i)*dt/2;% velocity
end
for i=2:Npoints
v(i)=v(i-1)+(a(i-1)+a(i))*dt/2;
end

% 2. Trapezoidal rule by Matlab embedded function


mtls=cumtrapz(t,a);

% 3. Closed-form solution
vcf=a0*(1-cos(omega*t))/omega;

subplot(3,1,1)
plot(t,a,'linewidth',2)
xlabel('time,s','fontsize',26);ylabel('acc.,m/s/s','fontsize',26);
grid off;box off;
subplot(3,1,2)
plot(t,v,t,vcf,'-.r','linewidth',2)
xlabel('time,s','fontsize',26);ylabel('vel.,m/s','fontsize',26);
legend('code','exact');
grid off;box off;
subplot(3,1,3)
plot(t,v,t,mtls,'-.r','linewidth',2)
xlabel('time,s','fontsize',26);ylabel('vel.,m/s','fontsize',26);
legend('code','mtlb function');
grid off;box off;

Figure 1. 7–Numerical integration by trapezoidal rule

The effect of time- sampling on the accuracy of integration is showed using two different
resolutions (Figure 1.8).

n=50 intervals, 𝛥𝑡=0.2s, Npoints=51 n=500 intervals, 𝛥𝑡=0.02s, Npoints=501


2 2
acc.,m/s/s
acc.,m/s/s

0 0

-2 -2
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
0.4 0.4
vel.,m/s
vel.,m/s

0.2 0.2

0 code 0
code
exact exact
-0.2 -0.2
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
0.2 0.4
vel.,m/s

vel.,m/s

0.1 0.2

0 code 0
code
mtlb function mtlb function
-0.1 -0.2
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 1.8- Sampling effect on accuracy: numerical vs. “exact” solution

 Task| Modify the integration scheme above and perform the integration of acceleration
and velocity time-series, respectively. Additionally, compute the acceleration cumulative
normalized energies and compare the results with those provided by the author, in the
Figures 1.9&1.10.

By numerical integration of ground acceleration and velocity components respectively, the


time-histories of ground velocity and ground displacement are successively obtained
7
(Figure 1.9). In the Figure 1.10, the time-instant ground acceleration cumulative
normalized energy is represented.

NS component EW component
Time-domain

200 200
2

2
Acc,cm/s
Acc,cm/s

0 0

-200 -200
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
t,s t,s
50 50

Vel,cm/s
Vel,cm/s

0
0
-50

-100 -50
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
t,s t,s
100 100

Disp,cm
Disp,cm

50 50

0 0

-50 -50
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
t,s t,s

Figure 1.9- Ground acceleration, velocity and displacement time-histories

1 1

0.8 0.8

0.6 0.6
normEx

normEx

0.4 0.4

0.2 0.2

0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
t,s t,s

Figure 1.10- Ground acceleration normalized cumulative energy time-histories

The peaks of ground acceleration, velocity and displacement are given in Table 1.

Table 1- Peak ground motion parameters

Parameter NS component EW component


PGA, cm/s2 194.96 162.21
PGV, cm/s 70.07 26.63
PGD, cm 90.77 69.54

Note that the ground velocity and displacement in the Figure 1.9, are not corrected.

 Homework 1.1|Develop the algorithm for the computation of Power Spectral Density
(PSD) function and compute the frequency bandwidth indicators given in the Chapter 1
(Lecture Notes). Study the “windowing” effect on the predominant frequency, using five
different frequency window lengths.
Hint: use the pwelch Matlab embedded function.
8
By performing the FFT analysis, using the routine given in the Figure 1.4, the Fourier
spectra are obtained and represented in the Figure 1.11. The normalized PSD spectra for
both components are obtained using the developments from the HW1 (Figure 1.12).

NS component EW component
Frequency- domain
300 200

250
150
200

FA,cm/s
FA,cm/s

150 100

100
50

50

0
0 0 1 2 3 4 5 6 7 8 9 10
0 1 2 3 4 5 6 7 8 9 10 Frequency,Hz
Frequency,Hz

Figure 1.11- Fourier Amplitude Spectra (FAS)

1.8 0.9

1.6 0.8

1.4 0.7

1.2 0.6
normSx
normSx

1 0.5

0.8 0.4

0.6 0.3

0.4 0.2

0.2 0.1

0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Frequency,Hz Frequency,Hz

Figure 1.12- Normalized Power Spectral Density (PSD) functions

The frequency content indicators are given in Table 2.

Table 2- Frequency content indicators

Parameter NS component EW component


fp, Hz 0.66 0.46
Tp,s 1.52 2.16
Ɛ 0.97 0.94
Ω 1.61 2.52
δ 0.70 0.68

The records are characterized by low predominant frequencies, in conjunction with


narrow bandwidths (Lungu et al., 2001). The frequency contents are concentrated around
very different frequencies while the power dispersion is almost same. These results are
essential for the analysis of both, conventional and unconventional structures.

9
In practice however, the initial records require the use of appropriate correction and
filtering techniques in order to remove inherent recording noise. Moreover, the numerical
integration introduces errors that are corrected by zero baseline correction technique
(Trifunac, 1971; Kashima, 2001).

Numerical example 1.4| Given the 50 minutes wind speed record obtained on a guyed
mast port at H=60 m above the ground, in suburban terrain conditions (Figure 1.13), the
time- and frequency- domain analyses are performed for the fluctuating speed component.
The fluctuating wind speed data is provided in the “wdsf.txt” file (Figure 1.15). The
sampling time-step of the record is ∆𝑡 = 0.1𝑠.

Fluctuating wind speed, 𝑣 (𝑡)

Mean wind speed, 𝑣̅ 𝐻


= 60

Figure 1.13- Wind speed record at height H and typical mean speed profile

The wind speed record waveform is given in the Figure 1.14.


40

35

30
Wind speed, m/s

25

20

15

10

0
0 5 10 15 20 25 30 35 40 45 50
t,min

Figure 1.14- Typical wind speed record

The actual record at a given height z above the ground, can be splitted into a mean
component-𝑣̅ and a fluctuating (turbulent) component- 𝑣 (Simiu and Scanlan, 1996;
Holmes, 2001) i.e.
10
𝑣(𝑧, 𝑡) = 𝑣̅ (𝑧) + 𝑣 (𝑧, 𝑡) (1.8)
The mean wind speed over 50 minutes record is 𝑣̅ = 19.9203 m/s. The maximum wind
speed is 35.72 m/s.

Fluctuating component of wind speed record


Suburban B Station, PORT 1
Units: m/s.
DT= 0.1 SEC
Mean speed: 19.9203 m/s

time,sec Speed,m/s
0.0000000e+000 4.3565070e-001
1.0000000e-001 2.2069562e-001
2.0000000e-001 -2.5299182e-001
3.0000000e-001 -1.6544210e+000
4.0000000e-001 -1.4667169e+000
5.0000000e-001 -8.6681320e-001
6.0000000e-001 -1.2876784e+000
7.0000000e-001 -3.4865958e+000
8.0000000e-001 -1.9717971e+000
9.0000000e-001 -3.5876185e-001
1.0000000e+000 -2.9635532e+000…

Figure 1.15- Fluctuating wind speed data file

Using the routine developed to solve the numerical example 1.3, the time-history of the
fluctuating wind speed component and the corresponding normalized PSD function are
given in the Figure 1.16.

Time- domain Frequency- domain


20 45

15 40

35
10
Wind speed, m/s

30
5
normSx

25
0
20
-5
15
-10
10

-15 5

-20 0
0 5 10 15 20 25 30 35 40 45 50 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
t,min Frequency,Hz

Figure 1.16- Fluctuating wind speed component record (left) and normalized PSD function (right)

The Fourier and phase spectra of the fluctuating wind speed are given in the Figure 1.17.

11
1800 40

1600 20

1400 0

1200 -20

-40

Phase
FA

1000

800 -60

600 -80

400 -100

200 -120

0 -140
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Frequency,Hz
Frequency,Hz

Figure 1.17- Fourier Amplitude spectrum (left figure) and phase spectrum (right figure)

The spectral shape parameters are given in Table 3.

Table 3- Spectral shape parameters

fp, Hz 0.005
Ɛ 0.98
Ω 0.64
δ 0. 92

The record is dominated by a very low predominant frequency range which is most likely
to excite flexible structures which are wind-sensitive (tall buildings, long-span cable
suspended bridges etc.) since the aerodynamic pressure is proportional to the squared
wind speed.

 Homework 1.2|Compute the cumulative energy of fluctuating wind speed component.


Find the peak and the corresponding occurrence time-instant.

 Homework 1.3|Perform the FFT analysis of triangular pulse given in the Figure 1.2b.

 Homework 1.4|Given the records in the form of data files (the first column is the time
scale, the second column is the corresponding load value; Figure 1.18)

(i) Identify the load signal type;


(ii) Perform the time- and frequency-domain analyses, extract de peak values and the
frequency content of signals, and identify predominant frequencies. Represent
then graphically the energy content of each signal in both time- and frequency
domains. Compute the statistics of data for the 4th signal (mean, standard
deviation and coefficient of variation);

12
1st signal 2nd signal
150 150

100 100

50 50

Load, kN
Load, kN

0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
-50 -50

-100 -100

-150 -150
time,sec time,sec

3rd signal 4th signal


150 400

300
100
200
50
100
Load, kN

Load, kN
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
-100
-50
-200
-100
-300

-150 -400
time,sec time,sec

Figure 1.18- Typical signals of dynamic loads

(iii) Analyze the change of frequency content parameters when to the initial 2nd
signal (Figure 1.18), two successive identical signals but shifted 10 seconds each
are added (Figure 1.19&Figure 1.20).

2nd signal
150

100

50
Load, kN

0
0 10 20 30 40 50 60
-50

-100

-150
time,sec

150

100

50
Load, kN

0
0 10 20 30 40 50 60
-50

-100

-150
time,sec

13
150

100

50

Load, kN
0
0 10 20 30 40 50 60
-50

-100

-150
time,sec

Figure 1.19- Successive individual load signals (10 seconds time-shift)

The resultant signals are given in the Figure 1.20.

1st resultant signal


300

200

100
Load, kN

0
0 10 20 30 40 50 60
-100

-200

-300
time,sec

2nd resultant signal


400

300

200

100
Load, kN

0
0 10 20 30 40 50 60
-100

-200

-300

-400
time,sec

Figure 1.20- Successive load signals

(iv) Perform the same set of frequency decomposition analysis using a resultant signal of 10
simple harmonic sine functions, their peak amplitudes 𝑥 , circular frequencies 𝜔 and
time-shift/phase angle 𝜑 following a normal statistical distribution.

***

14
2. Computational Structural Dynamics

This chapter introduces the Finite Element Method (FEM) analysis approach to solve
current applications in the Dynamics of Structures field. The FEM procedure features, to
obtain discrete models and loads is discussed in this section. The modal analysis is
performed using frame structures and the effect of discretization and model size
reduction over the modal parameters (eigenvalues and vibration mode shapes) is
studied using benchmark analyses and available „exact” solution. This section allows
further extensions to include the effect of members with variable cross-sections, flexible
connections and elastic supports, 3D modeling for a larger number of structural types.

Numerical example 2. 1| Given the Reinforced Concrete (RC) frame with specified
geometry and physical properties in the Figure 2.1, the stiffness and inertia matrices are
constructed and further the corresponding vibration modes are computed. The loads
considered in the analysis are: (i) the Dead Load (DL)- implicitly computed and (ii) Live
Load (LL) taken as 4 𝑘𝑁/𝑚 . The tributary length in the perpendicular direction of
frame is 6.00 m. The © Matlab routine is developed during classes.
p(LL) = 24 kN/m
Young’s modulus E = 2.5x107 kN/m2
Specific weight 𝛾 =2500 kg/m3
Sections:
Columns|50x50 (cm)
4.00

Beam|30x80 (cm)
7.50

Figure 2.1- RC plane frame and physical properties

The proposed discrete model consists of 3 elements (fixed at both ends), their stiffness and
mass matrices being known (see Chapter 2 in the Lecture Notes). The full model (given in
the Figure 2.2) has therefore 12 DOFs, 6 DOFs effective (n-1-6) and 6 DOF-restricted (r- 7-
12).
2 5
Y 1 4
3 6

11 8
Z 12 10 9 7

0 X
Figure 2.2- Discrete (lumped) model and associated degrees of freedom
in the global reference system (XYZ)

15
The individual members and their local refference axes are presented in the Figure 2.3.

0 e1: m1, k1 x
1 2
z
z y z y
0 0
e3: m3, k3 e2: m2, k2

x x

4 3

Figure 2.3- Discrete elements, joints and local reference axes (xyz)

The stiffness and inertia matrices of members are constructed based on the eqs. (2.23) and
(2.24) given in the Lecture Notes.
The structure of member’s stiffness matrix is
(𝒆) (𝒆)
𝒌𝒊𝒊 𝒌𝒊𝒋
𝒌(𝒆)
= (𝒆) (𝒆)
(2.1)
𝒌𝒋𝒊 𝒌𝒋𝒋
(𝒆) (𝒆) (𝒆) (𝒆)
where, 𝒌𝒊𝒊 and 𝒌𝒋𝒋 are the main stiffness sub-matrices, while 𝒌𝒊𝒋 and 𝒌𝒋𝒊 are the end-
stiffness sub-matrices of the element e (i-origin joint, j-termination joint)-due to elastic
coupling.
The structure of member’s inertia matrix is
(𝒆) (𝒆)
𝒎𝒊𝒊 𝒎𝒊𝒋
𝒎(𝒆) = (𝒆) (𝒆)
(2.2a)
𝒎𝒋𝒊 𝒎𝒋𝒋
(𝒆) (𝒆) (𝒆) (𝒆)
where, 𝒎𝒊𝒊 and 𝒎𝒋𝒋 are the main inertia sub-matrices, while 𝒎𝒊𝒋 and 𝒎𝒋𝒊 are the end-
inertia sub-matrices of the element e (i-origin joint, j-termination joint)-due to inertia
coupling.
Using the transformations (2.27) and (2.31) given in the Lecture Notes i.e.
(𝒆) (𝒆)
𝒌𝑮 = 𝑻(𝒆) 𝒌𝑳 𝑻(𝒆) (2.2b)
(𝒆)
where 𝒌𝑳 is the stiffness matrix of the element (2.1) in the local reference system and 𝑻(𝒆)
is the transformation (rotation) matrix of the element,
(𝒆) (𝒆)
𝒎𝑮 = 𝑻(𝒆) 𝒎𝑳 𝑻(𝒆) (2.2c)
(𝒆)
where 𝒎𝑳 is the inertia matrix of the element (2.2a) in the local reference system

16
Thus, the element’s stiffness matrix and the inertia matrix in the global reference system
are obtained successively. The structure’s stiffness and inertia matrix are obtained by
merging the members’ matrices, according to their coordinates as
(𝒆 )
𝟏 𝟑(𝒆 ) (𝒆 ) (𝒆 )
⎡𝒌𝟏𝟏 + 𝒌𝟏𝟏 𝒌𝟏𝟐𝟏 𝟎 𝒌𝟏𝟒𝟑 ⎤
𝒋𝒐𝒊𝒏𝒕 𝟏
⎢ (𝒆 ) (𝒆 ) (𝒆 ) (𝒆 )
𝒌𝟐𝟏𝟏 𝒌𝟐𝟐𝟏 + 𝒌𝟐𝟐𝟐 𝒌𝟐𝟑𝟐 𝟎 ⎥ 𝒋𝒐𝒊𝒏𝒕 𝟐
𝑲 , =⎢ ⎥ (2.3)
𝟎 ⎥ 𝒋𝒐𝒊𝒏𝒕 𝟑
(𝒆 ) (𝒆 )
⎢ 𝟎 𝒌𝟑𝟐𝟐 𝒌𝟑𝟑𝟐
⎢ (𝒆𝟑 )

(𝒆 ) 𝒋𝒐𝒊𝒏𝒕 𝟒
⎣ 𝒌𝟒𝟏 𝟎 𝟎 𝒌𝟒𝟒𝟑 ⎦
𝒋𝒐𝒊𝒏𝒕 𝟏 𝒋𝒐𝒊𝒏𝒕 𝟐 𝒋𝒐𝒊𝒏𝒕 𝟑 𝒋𝒐𝒊𝒏𝒕 𝟒
The inertia matrix has the same structure
(𝒆 ) 𝟑 (𝒆 ) (𝒆 ) (𝒆 )
𝟏
⎡𝒎𝟏𝟏 + 𝒎𝟏𝟏 𝒎𝟏𝟐𝟏 𝟎 𝒎𝟏𝟒𝟑 ⎤
𝒋𝒐𝒊𝒏𝒕 𝟏
⎢ (𝒆 ) (𝒆 ) (𝒆 ) (𝒆 )
𝒎𝟐𝟏𝟏 𝒎𝟐𝟐𝟏 + 𝒎𝟐𝟐𝟐 𝒎𝟐𝟑𝟐 𝟎 ⎥ 𝒋𝒐𝒊𝒏𝒕 𝟐
𝑴 , = ⎢ ⎥ (2.4)
𝟎 ⎥ 𝒋𝒐𝒊𝒏𝒕 𝟑
(𝒆 ) (𝒆 )
⎢ 𝟎 𝒎𝟑𝟐𝟐 𝒎𝟑𝟑𝟐
⎢ (𝒆 ) (𝒆 ) ⎥ 𝒋𝒐𝒊𝒏𝒕 𝟒
⎣ 𝒎𝟒𝟏𝟑 𝟎 𝟎 𝒎𝟒𝟒𝟑 ⎦

𝒋𝒐𝒊𝒏𝒕 𝟏 𝒋𝒐𝒊𝒏𝒕 𝟐 𝒋𝒐𝒊𝒏𝒕 𝟑 𝒋𝒐𝒊𝒏𝒕 𝟒


In the mass matrix terms the dead load and the live load are properly accounted for.
The stiffness and inertia matrices are partitioned, according to n- the index of effective
DOFs and r- the index of restricted DOFs, by declaring the support conditions in order to
remove the rigid body effect, as follows
𝑲𝒏𝒏 𝑲𝒏𝒓
𝑲 , = (2.5)
𝑲𝒓𝒏 𝑲𝒓𝒓
and
𝑴𝒏𝒏 𝑴𝒏𝒓
𝑴 , = (2.6)
𝑴𝒓𝒏 𝑴𝒓𝒓
The stiffness matrix, expressed according to the effective DOFs (1 to 6) yields
0.8244 0 −0.0488 −0.8000 0 0
⎡ 0 1.5716 −0.0341 0 −0.0091 −0.0341⎤
⎢ ⎥
−0.0488 −0.0341 0.3009 0 0.0341 0.0853 ⎥
𝑲𝒏𝒏 = 10 ⎢ (2.7)
⎢ −0.8000 0 0 0.8244 0 −0.0488⎥
⎢ 0 −0.0091 0.0341 0 1.5716 0.0341 ⎥
⎣ 0 −0.0341 0.0853 −0.0488 0.0341 0.3009 ⎦
and the corresponding inertia matrix is
51.1868 0 −2.1358 23.7003 0 0
⎡ 0 396.9832 −418.9875 0 137.1232 247.5836 ⎤
⎢ ⎥
−2.1358 −418.9875 572.9000 0 −247.5836 −428.5100⎥
𝑴𝒏𝒏 =⎢ (2.8)
⎢ 23.7003 0 0 51.1868 0 −2.1358 ⎥
⎢ 0 137.1232 −247.5836 0 396.9832 418.9875 ⎥
⎣ 0 247.5836 −428.5100 −2.1358 418.9875 572.9000 ⎦

17
Note that the inertia matrix is completely populated. This form is called the consistent mass
matrix and is a consequence of inertia coupling (axial and flexural).

The vibration modes analysis is performed by Matlab routine given in the Figure 2.4. This
is used hereinafter to perform the eigenmodes analysis either for 2D models or 3D models.

% Vibration modes analysis

% define K (stiffness matrix)


% define M (inertia matrix)
[fai,omega2] = eig(K,M);
omega = sqrt(omega2);% natural circular frequency matrix
T = 2*pi*inv(omega); % natural period matrix
f = inv(T); % natural frequency matrix
% fai is the modal matrix

Figure 2.4- Routine for the vibration modes analysis

The following spectral and modal matrices are thus obtained as

14.1923 0 0 0 0 0
⎡ 0 15.4043 0 0 0 0 ⎤
⎢ ⎥
0 0 46.4414 0 0 0 ⎥ , rad/s
𝜴=⎢ (2.9a)
⎢ 0 0 0 135.4614 0 0 ⎥
⎢ 0 0 0 0 188.1209 0 ⎥
⎣ 0 0 0 0 0 243.1639⎦
0.4427 0 0 0 0 0
⎡ 0 0.4079 0 0 0 0 ⎤
⎢ ⎥
0 0 0.1353 0 0 0 ⎥
𝑻=⎢ ,s (2.9b)
⎢ 0 0 0 0.0464 0 0 ⎥
⎢ 0 0 0 0 0.0334 0 ⎥
⎣ 0 0 0 0 0 0.0258⎦

0.0006 0.0806 −0.0137 −0.0004 0.0012 −0.1349


⎡ −0.0019 0.0002 −0.0125 −0.0743 0.0934 −0.0019 ⎤
⎢ ⎥
0.0210 0.0111 0.0424 −0.0501 0.1186 −0.0014⎥
𝜱=⎢ (2.9c)
⎢ −0.0006 0.0806 −0.0137 0.0004 0.0012 0.1349 ⎥
⎢ −0.0003 −0.0002 0.0125 −0.0743 −0.0934 −0.0019 ⎥
⎣ −0.0210 0.0111 0.0424 0.0501 0.1186 0.0014 ⎦

The modal matrix 𝜱 above is normalized with respect to modal mass matrix i.e.
𝜱𝑻 𝑴𝜱 = 𝟏 (2.10)
where 𝟏 is the identity matrix.
Alternatively, 𝜱 can be normalized with respect to the first ordinate in each column. The
normalized mode shapes are given in Table 2 and the schematic representation of the
associated mode shapes is given in the Figure 2.5.

18
Table 2- Modal information

Mode 1 Mode 2 Mode 3 Mode 4 Mode 5 Mode 6


Coordinate T=0.4427 s T=0.4079 s T=0.1353 s T=0.0464 s T=0.0334 s T=0.0258 s
1 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
2 -3.0867 0.0025 0.9160 172.5709 77.2848 0.0140
3 33.4493 0.1375 -3.1004 172.5709 98.1338 0.0107
4 -1.0000 1.0000 1.0000 -1.0000 1.0000 -1.0000
5 -3.0867 -0.0025 -0.9160 172.5709 -77.2848 0.0140
6 -33.4493 0.1375 -3.1004 -172.5709 98.1338 -0.0107

Mode 1 Mode 2

Mode 3 Mode 4

Mode 5 Mode 6

Figure 2.5- Schematic representation of mode shapes

(i) Model size reduction


By removing the rotation-type DOFs, corresponding to both lumped masses, using the
condensation approach, the matrices associated to sway-type DOFs only yield according
to eq. (2.47&2.49) given in the Lecture Notes. Thus, the stiffness matrix is
𝑲𝚫 = 𝑲𝚫𝚫 − 𝑲𝚫𝜽 𝑲𝜽𝜽𝟏 𝑲𝜽𝚫 (2.11)
and the corresponding inertia matrix

19
𝑴𝚫 = 𝑴𝚫𝚫 − 𝑴𝚫𝜽 𝑴𝜽𝜽𝟏 𝑴𝜽𝚫 (2.12)
where the matrix components are the sway-sway "𝛥𝛥", rotation-rotation "𝜃𝜃" and the
coupling sub-matrices associated to sway- and rotation-type DOFs "𝛥𝜃" or "𝜃𝛥",
respectively.
The reduced model thus obtained is given in the Figure 2.6.

2 4
1 3

Figure 2.6- Reduced discrete model

The stiffness matrix associated to sway-type DOFs only is


0.8158 −0.0043 −0.7976 0.0043
−0.0043 1.5656 −0.0043 −0.0031
𝑲𝚫 = 10 (2.13a)
−0.7976 −0.0043 0.8158 0.0043
0.0043 −0.0031 0.0043 1.5656
and the corresponding inertia matrix is
51.1688 −1.9785 23.6868 0.5569
−1.9785 73.4014 −0.5569 17.0135
𝑴𝚫 = (2.13b)
23.6868 −0.5569 51.1688 1.9785
0.5569 17.0135 1.9785 73.4014
By performing the vibration modes analysis, the following spectral and modal matrices
are obtained for the reduced model:

15.5925 0 0 0
0 131.4366 0 0
𝜴= , rad/s (2.14a)
0 0 166.8882 0
0 0 0 242.4326
0.4030 0 0 0
0 0.0478 0 0
𝑻= ,s (2.14b)
0 0 0.0376 0
0 0 0 0.0259

And

1.0000 −0.0216 −0.0300 1.0000


0.0052 1.0000 −1.0000 0.0223
𝜱= (2.14c)
1.0000 0.0216 −0.0300 −1.0000
−0.0052 1.0000 1.0000 −0.0223

By normalizing 𝜱 with respect to the maximum ordinate in each column, the normalized
mode shapes are given in Table 3.
20
Table 3- Modal information (4 DOFs, M consistent)

Mode 1 Mode 2 Mode 3 Mode 4


Coordinate T= 0.4030 s T= 0.0478 s T= 0.0376 s T= 0.0259 s
1 1.0000 -0.0216 -0.0300 1.0000
2 0.0052 1.0000 -1.0000 0.0223
3 1.0000 0.0216 -0.0300 -1.0000
4 -0.0052 1.0000 1.0000 -0.0223

The axial ordinate component is therefore negligible for the lower modes. While the lower
modes are flexural-type modes, the higher ones are dominated by axial effects.

(ii) Inertia matrix diagonalization


By neglecting the inertia cross-terms (with contributions from 0.7% to 33% from the main
coefficients), the diagonal mass matrix yields

51.1688 0 0 0
0 73.4014 0 0
𝑴 \ = (2.15)
0 0 51.1688 0
0 0 0 73.4014
This diagonal structure of inertia matrix is beneficial from the numerical analysis point-of-
view (Clough&Penzien, 1993).
Using the diagonal inertia matrix, the following spectral and modal matrices are obtained:

18.8557 0 0 0
0 145.9009 0 0
𝜴= ,rad/s (2.16a)
0 0 146.1904 0
0 0 0 177.5670

0.3332 0 0 0
0 0.0431 0 0
𝑻= ,s (2.16b)
0 0 0.0430 0
0 0 0 0.0354

1.0000 0.0000 −0.0080 −1.0000


−0.0056 1.0000 1.0000 0.0000
𝜱= (2.16c)
−1.0000 0.0000 −0.0080 1.0000
0.0056 1.0000 −1.0000 0.0000

(iii) Direct merging of inertia matrix


The diagonal mass matrix 𝑴 = 𝑴 \ can be straightforward obtained by hand-calculation,
by direct merging approach i.e. summing the mass contributions at each member’s end in the
adjacent joints, with respect to the corresponding coordinates (Chopra, 2001; Figure 2.7).

21
0.5 (𝑚𝐿) 0.5 (𝑚𝐿)
2 element 1: (𝑚𝐿)
1 4 3 2 4
1 3

0.5(𝑚𝐿)
0.5(𝑚𝐿)

element 2: (𝑚𝐿)
element 3: (𝑚𝐿)

0.5(𝑚𝐿)
0.5(𝑚𝐿)

Figure 2.7- Direct merging approach of inertia matrix

By neglecting the axial inertia and rotation inertia, the mass matrix yields by simple
addition of mass-components at each joint as

72.3751 0 0 0
0 72.3751 0 0
𝑴 \ = (2.17)
0 0 72.3751 0
0 0 0 72.3751
This is slightly different from that obtained by matrix condensation and cross-terms
elimination given by eq. (2.15).

The following spectral and modal matrices are thus obtained:

15.8546 0 0 0
0 146.9316 0 0
𝜴= ,rad/s (2.18a)
0 0 147.2222 0
0 0 0 149.3035
0.3963 0 0 0
0 0.0428 0 0
𝑻= ,s (2.18b)
0 0 0.0427 0
0 0 0 0.0421
−1.0000 0.0000 −0.0056 −1.0000
−0.0056 1.0000 1.0000 0.0000
𝜱= (2.18c)
−1.0000 0.0000 −0.0056 1.0000
0.0056 1.0000 −1.0000 0.0000

For this reduced model, the modal information is given in Table 4.

Table 4- Modal information (4 DOFs, M diagonal by direct merging)

Mode 1 Mode 2 Mode 3 Mode 4


Coordinate T= 0.3963 s T= 0.0428 s T= 0.0427 s T= 0.0421 s
1 1.0000 0.0000 0.0056 1.0000
2 0.0056 1.0000 -1.0000 0.0000
3 1.0000 0.0000 0.0056 -1.0000
4 -0.0056 1.0000 1.0000 0.0000

22
The condensation technique applied to the model may influence the vibration modes in a
certain amount especially for the fundamental mode. By neglecting the rotation inertia and
further, the axial inertia, the modal information will skip therefore the complete
information. For regular structures the influence of condensation and inertia matrix
diagonalization is however low.
In practice, if the size of the model can be handled by the hardware support, it is desirable
to operate without any size reductions that might affect the completeness of analysis
results. For instance the columns axial deformations are essential information in the case
of tall and super-tall buildings (Iancovici et al., 2010).
By additionally neglecting the axial deformation of members, the model can be even more
simplified to a SDOF lateral sway model (Figure 2.8).

“driving” joint
“driven” joint 1

Figure 2.8- Simplified SDOF lateral sway model

For the frame given in the Figure 2.8, the mass and stiffness associated to the lateral SDOF
are 𝑚 = 144.75 kNs /m and the stiffness 𝑘 = 35984 kN/m. Then the lateral sway natural
vibration period yields 𝑇 = 2𝜋 𝑚/𝑘 = 0.3985 𝑠.
By additionally neglecting the axial deformations of members and thus reducing the
model to a SDOF model, the model obtained is stiffer (in this case 10% stiffer when
comparing to the full model). For larger dimension problems, this effect might be more
significant. More refined the model, larger the computational effort, as well as
accumulated numerical errors (Bathe et al., 1976).
The summary of the eigenvalues analysis is given in Table 5.1.

Table 5.1- Eigenvalues analysis summary

Full model Condensation M diagonalization M direct merging SDOF


6DOF 4DOF 4DOF 4DOF
Natural period, 2
1
5
4
2
1
4
3
1
3 6

T1 0.4427 0.4030 0.3332 0.3963 0.3985


T2 0.4079 0.0478 0.0431 0.0428
T3 0.1353 0.0376 0.0430 0.0427
T4 0.0464 0.0259 0.0354 0.0421
T5 0.0334
T6 0.0258

23
If the only the self-weight of the structure is taken into account, the summary of the
eigenvalues analysis is given in Table 5.2.

Table 5.2- Eigenvalues analysis summary (Dead Load only)

Full model Condensation M diagonalization M direct merging SDOF


6DOF 4DOF 4DOF 4DOF
Natural period, 2
1
5
4
2
1
4
3
1
3 6

T1 0.1107 0.1075 0.1029 0.0880 0.0885


T2 0.0810 0.0120 0.0109 0.0095
T3 0.0265 0.0095 0.0104 0.0095
T4 0.0117 0.0087 0.0104 0.0093
T5 0.0087
T6 0.0086

Remarks:

(i) This simplifying modeling approach based on the above assumptions works for
typical structures (i.e. rigid or semi-rigid structures) and is a useful finding for the
direct matrix assemblage and model reduction technique;
(ii) The main effect of condensation and mass matrix diagonalization procedure, in fact
of elimination and neglecting certain components, is that the obtained model will
result stiffer. This must be taken into account when interpreting the analysis results.

Numerical example 2.2| Given the RC frame in the Figure 2.9 the global elements load
vector 𝑸(𝑡) is constructed. The frame is subjected to a horizontal uniformly distributed
pressure load 𝑝(𝑥, 𝑡) = 𝑝 sin(𝜃𝑡), where 𝑝 is the peak pressure and 𝜃 is the circular
frequency of applied pressure, acting on the left column.

p(x,t) L

3L

Figure 2.9- RC plane frame subjected to lateral uniformly distributed time-history pressure

The discrete model consists of 3 elements (fixed at both ends). Recall the full model in the
Figure 2.2, having 12 DOF, 6 DOFs effective (n-1-6) and 6 DOF-restricted (r- 7-12). This is
shown in the Figure 2.10.

24
2 5
Y 1 4
3 6

11 8
Z 12 10 9 7

0 X

Figure 2.10- Discrete (lumped) mass model and associated degrees of freedom

The 𝒒(𝒆) (𝑡) vector’s structure is developed in the eq. (2.39) in the Lecture Notes. The beam
( )
and right column are free of element loads so that 𝑸(𝑡) will be assembled from the 𝒒𝑮 (𝑡)
vector as
( ) ( )
𝒒𝑮 (𝑡) = 𝑻( )
𝒒𝑳 (𝑡) = 〈 −𝑝 0 +𝑝 −𝑝 0 −𝑝 〉 sin(𝜃𝑡) (2.19)

At the end,
( )
𝑸(𝑡) = ∑( ) 𝒒𝑮 (𝑡)=〈 −𝑝 0 +𝑝 𝟎𝟑,𝟏 𝟎𝟑,𝟏 −𝑝 0 −𝑝 〉 sin (𝜃𝑡) (2.20)

Therefore the external loads vector expressed in the effective DOFs is

𝑭(𝑡) = 〈+𝑝 −𝑝 〉 sin (𝜃𝑡) (2.21)

i.e. force- and moment-type loads, acting in the coordinate axes 1 and 2.
The components corresponding to the restricted coordinate axes 10 and 12 are

𝑸𝒓 (𝑡) = 〈−𝑝 −𝑝 〉 sin (𝜃𝑡) (2.22)

that will contribute to the reaction forces computation.


If the rotation inertia of masses is neglected then the horizontal load at left effective joint,
reduces to a single component i.e.

𝑭(𝑡) = 𝑝 sin (𝜃𝑡) (2.23)

that could be obtained also directly by simple hand calculation.


Any other load distribution on the elements can be explicitly considered in an automated
analysis scheme using the FEM approach.

Numerical example 2.3|For the steel cantilever beam (Figure 2.10) having the Young’s
modulus E = 2x108 kN/m2 , specific weight 𝛾 =7850 kg/m3, the effect of continuous
model refinement degree on the vibration modes is shown. The in-plane transverse
vibration component of motion is considered only. The results are then compared with
those obtained by a distributed elasticity FEM model.

25
20 mm

15 mm
800mm

6.00
300 mm

Figure 2.10 – Steel cantilever beam

The damped forced flexural vibration of beam is described by the partial differential
equation of motion (Ifrim, 1984; Clough&Penzien, 1993)
( , ) ( , ) ( , )
𝑚(𝑥) + 𝑐(𝑥) + 𝐸𝐼(𝑥) = 𝑝(𝑥, 𝑡) (2.58)

where 𝑚(𝑥) is the mass distribution function, 𝛾 is the mass density, 𝑐(𝑥) is the damping
distribution function, 𝐼(𝑥) is the moment of inertia and 𝑝(𝑥, 𝑡) is the forcing function (both
space and time variable).
For the example’s purpose, the forcing function and damping are neglected, thus
( , ) ( , )
𝑚(𝑥) + 𝐸𝐼(𝑥) =0 (2.59)

Expressing the transverse deflection function as


𝑦(𝑥, 𝑡) = 𝜙(𝑥)sin(𝜔𝑡) (2.60)
where 𝜙(𝑥) is the shape function and 𝜔 is the natural circular frequency; eq.(2.59)
becomes, for uniform mass distribution
( )
− 𝛽 𝜙(𝑥) = 0 (2.61a)

where
𝛽 = (2.61b)

𝑚 is here the constant mass per unit length.


The general solution of eq. (2.61a) has the form
𝜙(𝑥) = 𝐶 𝑠𝑖𝑛(𝛽𝑥) + 𝐶 𝑐𝑜𝑠(𝛽𝑥) + 𝐶 𝑠𝑖𝑛ℎ(𝛽𝑥) + 𝐶 𝑐𝑜𝑠ℎ(𝛽𝑥) (2.61c)
where the integration constants 𝐶 , 𝐶 , 𝐶 , 𝐶 are boundary condition-dependent.
For the first three vibration modes, the analysis for the uniform mass cantilever beam
yields (Clough&Penzien, 1993)
𝛽 𝐿 = 1.875, 𝛽 𝐿 = 4.694 and 𝛽 𝐿 = 7.855 (2.62)
Therefore the first three natural circular frequencies of continuous model are

𝜔 = 1.875 , 𝜔 = 4.694 and 𝜔 = 7.855 (2.63)

26
where 𝑚 is the mass per unit length.
The mode shape function is given by
𝜙 (𝑥) = [𝑐𝑜𝑠ℎ(𝛽 𝑥) − 𝑐𝑜𝑠(𝛽 𝑥)] − 𝐷 [𝑠𝑖𝑛ℎ(𝛽 𝑥) − 𝑠𝑖𝑛(𝛽 𝑥)] (2.64)
Where 𝛽 𝐿 = 𝑖𝜋, 𝑖 = 1,2, … ∞ and
( ) ( )
𝐷 = ( ) ( )
(2.65a)

For the first three vibration modes eq. (2.65a) yields


𝐷 = 0.7341, 𝐷 = 1.0185 and 𝐷 = 0.9992 (2.65b)
The mass normalized eigenvectors can be obtained so that

∫ 𝛾 𝜙 (𝑥)𝑑𝑥 = 1 , 𝑖 = 1,2, … ∞ (2.66)

The mass mode participation factors are

Γ = 𝛾 ∫ 𝜙 (𝑥) 𝑑𝑥 , 𝑖 = 1,2, … ∞ (2.67a)

The mass participation factors for the first three vibration modes are

Γ = 0.783 𝛾𝐿, Γ = 0.434 𝛾𝐿, Γ = 0.254 𝛾𝐿 (2.67b)


meaning less contribution for the higher modes.
The effective modal masses are

∫ ( )
𝑚 , = , 𝑖 = 1,2, … ∞ (2.68a)
∫ ( )

so that
𝑚 , = 613𝛾𝐿, 𝑚 , = 0.188𝛾𝐿 and 𝑚 , = 0.065𝛾𝐿 (2.68b)
Three discrete models proposed for this illustrative example is shown in the Figure 2.11.
The lumped masses are the consequence of using the FEM direct merging technique for
the transverse components.

SDOF 2DOF 3DOF

0.5(𝑚𝐿) 0.5(𝑚𝐿) 0.25(𝑚𝐿) (𝑚𝐿)/3 (𝑚𝐿)/3 (𝑚𝐿)/6

1 1 2 1 2 3

L/2=3.00 L/2=3.00 L/3=3.00 L/3=3.00 L/3=3.00

Figure 2.11–Steel cantilever beam lumped models

27
The analysis is done based on the routine given in the Figure 2.4.

(i) SDOF model, with a single beam element (Figure 2.13)

The cross-section area of the beam is 𝐴 = 2340 ∙ 10 m and inertia moment under the
strong axis is 𝐼 = 2374 ∙ 10 m .

The total mass of beam is 𝑚 = 𝑚𝐿 = = 1.1235 𝑘𝑁𝑠 /𝑚.

The vibration mass is therefore 𝑚/2 = 0.5617 𝑘𝑁𝑠 /𝑚. The vertical sway stiffness of beam
is

𝑘= = 6924.17 𝑘𝑁/𝑚, so that the natural circular frequency yields

𝜔= .
= 111.02 𝑟𝑎𝑑/𝑠 (2.24)

Hence, the natural period is 𝑇 = 0.057 𝑠 and the natural frequency is 𝑓 = 17.7 𝐻𝑧.

(ii) 2DOF model, with two equal lengths beam elements (Figure 2.14)

Simple hand computation yields the inertia matrix and the stiffness matrix as
0.5617 0
𝑴= , 𝑘𝑁𝑠 /𝑚 (2.25a)
0 0.2809
2.4142 −0.7545
𝑲 = 10 , 𝑘𝑁/𝑚 (2.25b)
−0.7545 0.3018
The vibration modes analysis is performed by Matlab routine given in the Figure 2.4. This
is used hereinafter to perform the eigenmodes analysis either for 2D models or 3D models.

% Vibration modes analysis

% define K (stiffness matrix)


% define M (inertia matrix)
[fai,omega2] = eig(K,M);
omega = sqrt(omega2);% natural circular frequency matrix
T = 2*pi*inv(omega); % natural period matrix
f = inv(T); % natural frequency matrix
% fai is the modal matrix

Figure 2.4- Routine for the vibration modes analysis

The spectral matrices are


𝜔 0 139.6267 0
𝜴= = , rad/𝑠 (2.25c)
0 𝜔 0 719.5224
𝑇 0 0.045 0
𝑻= = ,𝑠 (2.25d)
0 𝑇 0 0.009
28
𝑓 0 22.2223 0
𝒇= = , 𝐻𝑧 (2.25e)
0 𝑓 0 114.5155
The normalized modal matrix, with respect to the first ordinate is
1 1
𝜱 = [𝝓𝟏 𝝓𝟐 ] = (2.25f)
3.05 −0.65
where 𝝓𝟏 and 𝝓𝟐 are the 1st and the 2nd mode vibration eigenvectors, respectively.

(iii) 3DOF model, with three equal lengths beam elements (Figure 2.15)
The inertia matrix and the stiffness matrix are

0.3745 0 0
𝑴= 0 0.3745 0 , 𝑘𝑁𝑠 /𝑚 (2.26a)
0 0 0.1872
1.0965 −0.6309 0.1647
𝑲 = 10 −0.6309 0.6035 −0.2195 , 𝑘𝑁/𝑚 (2.26b)
0.1647 −0.2195 0.0960
The spectral matrices are

148 0 0
𝜴= 0 835.5 0 , 𝑟𝑎𝑑/𝑠 (2.26c)
0 0 2081.4
0.043 0 0
𝑻= 0 0.008 0 ,𝑠 (2.26d)
0 0 0.003
23.555 0 0
𝒇= 0 132.973 0 , 𝑟𝑎𝑑/𝑠 (2.26e)
0 0 331.268
and the corresponding modal matrix is

1 1 1
𝜱 = 3.34 0.97 −0.72 (2.26f)
6.18 −1.37 0.45

(iv) The continuous model

The “exact” first three natural circular frequencies of continuous model are computed
based on the solutions given by eqs. (2.63) in the Lecture Notes (Clough&Penzien, 1993) as

𝜔 = 1.875 , 𝜔 = 4.694 and 𝜔 = 7.855 (2.27)

where 𝑚 is the mass per unit length and 𝐿 is the beam’s length.
Further the “exact” mode shapes are determined using the solutions given by eqs. (2.64) in
the Lecture Notes as
𝜙 (𝑥) = [𝑐𝑜𝑠ℎ(𝛽 𝑥) − 𝑐𝑜𝑠(𝛽 𝑥)] − 𝐷 [𝑠𝑖𝑛ℎ(𝛽 𝑥) − 𝑠𝑖𝑛(𝛽 𝑥)] (2.28a)

29
where 𝛽 𝐿 = 𝑖𝜋, 𝑖 = 1,2,3 and
( ) ( )
𝐷 = ( ) ( )
(2.28b)

The eigenvalues and the corresponding exact-to-discrete ratios are given in Table 6.

Table 6- Natural circular frequencies: continuous vs. discrete models

𝝎 Continuous Discrete
Exact SDOF Exact/discr. 2DOF Exact/discr. 3DOF Exact/discr.
𝝎𝟏 167.26 111.02 1.44 139.63 1.14 148 1.08
𝝎𝟐 1048.3 719.52 1.38 835.5 1.20
𝝎𝟑 2935.57 2081.4 1.34

It is to be noticed that once the refinement degree of discrete model increases, this will be
able to reproduce more accurate, the exact solution. This comparison is valid however for
very few cases only, possible to be handled for the exact solution. The discrete model is
more flexible that the continuous one. The computed mode exact shapes are represented
in the Figure 2.16. The normalized mode shapes, with respect to the first coordinate of
discrete systems, are represented in the Figures 2.17 and 2.18.

8
Mode shape 1

6 Mode shape 2
Mode shape 3
4
Mode shape ordinate

0
0 1 2 3 4 5 6
-2

-4

-6

-8

Length,m

Figure 2.16– First three normalized mode shapes of the continuous model

Mode 1 Mode 2

0 8
0 1 2 3 4 5 6
7 Mode shape 2
-0.5
6
Mode shape ordinate

Mode shape ordinate

5 2DOF mode2
-1
4
-1.5
3
2
-2
1
-2.5 0
Mode shape 1 0 1 2 3 4 5 6
-1
-3
2DOF Mode1
-2
-3.5 -3

Length,m Length,m

Figure 2.17– Continuous vs. discrete mode shapes- 2DOF model


30
Mode 1 Mode 2

0 5
0 1 2 3 4 5 6
-1 4 Mode shape 2

3DOF Mode2
Mode shape ordinate

Mode shape ordinate


-2 3

-3 2

-4 1

-5 0
Mode shape 1 0 1 2 3 4 5 6
-6 -1
3DOF mode1

-7 -2

Length, m Length, m

Mode 3

1
Mode shape ordinate

0
0 1 2 3 4 5 6
-1

-2

-3

-4 Mode shape 3

-5 3DOF Mode3

-6

Length, m

Figure 2.18– Continuous vs. discrete model mode shapes- 3DOF model

Notice from Table 6 that o once the refinement degree increases, the model is able to
reproduce more accurately the exact solution for the fundamental mode. Higher modes
are better reproduced by using more refined models and the discrete model is more
flexible than the continuous one.
Using now a full FEM package provided by ANSYS Student software, the cantilever beam
is modeled using a distributed elasticity approach and two refinement levels. The results
are presented in the Figure 2.19 for the first 10 vibration tridimensional modes.

Model 1: 409 nodes/ 48 elements Model 2: 4950 nodes/ 720 elements

31
Mode 1: 4.873Hz Mode 1: 4.8607 Hz

Mode 2: 7.9714Hz Mode 2: 7.9301 Hz

Mode 3: 24.073 Hz Mode 3: 24.062 Hz

Mode 4: 29.097Hz Mode 4: 28.957 Hz

Mode 5: 38.942 Hz Mode 5: 38.734 Hz

32
Mode 6: 63.566 Hz Mode 6: 59.975 Hz

Mode 7: 65.559 Hz Mode 7: 64.97 Hz

Mode 8: 79.356 Hz Mode 8: 77.295 Hz

Mode 9: 94.485 Hz Mode 9: 79.324 Hz

Mode 10: 102.39 Hz Mode 10: 96.195 Hz

Figure 2.19– Tridimensional mode shapes of cantilever beam


33
Despite not having the capability to reproduce the distributed inertial and stiffness effects,
the discrete model reproduces fairly accurate the lower vibration modes associated with
its in-plane and out-of plane behavior as well as the torsional behavior.
However, this comparison can be done for very few cases only, possible to conveniently
handle for the “exact” solution.

Numerical example 2.4|A RC frame structure with specified geometry and pshysical
properties is given in the Figure 2.20 . An additional Live Load (LL) of 20 kN/m acting on
the beams is considered. The vibration modes are computed using first a simplified 3DOF
lateral sway model and then the influence of full FEM discretization is emphasized.

p(LL)=20kN/m

p(LL)=20kN/m Element w,cm h,cm


3.00
B C1 50 50
C2 70 70
20kN/m 20kN/m
3.00 B 40 90
ATRIUM
𝑘𝑁
3.00 𝐸 = 3 ∙ 10
𝑚

γ= 2500 𝑘𝑔/𝑚
6.00 6.00 8.00 m

C1 C1 C2 C2

Figure 2.20- Building’s view, section and material properties

By computing the elements masses from the DL and LL, and applying the direct merging
procedure, the diagonal mass matrix of structure yields

43.0428 0 0
𝑴= 0 41.1315 0 (2.29a)
0 0 31.1417
Using the shear-type frame model, the stiffness matrix expressed in the lateral sway
degrees of freedom yields

8.1133 −4.0567 0
𝑲 = 10 −4.0567 6.7244 −2.6678 (2.29b)
0 −2.6678 2.7666
By running the routine given in the Figure 2.4, the spectral and modal matrices are
obtained as follows

47.9295 0 0
𝜴= 0 114.6190 0 , rad/𝑠 (2.29c)
0 0 169.2549
34
0.1311 0 0
𝑻= 0 0.0548 0 ,𝑠 (2.29d)
0 0 0.0371
The modal mass normalized modal matrix is
0.0548 −0.0978 0.1032
𝜱 = [𝝓𝟏 𝝓𝟐 𝝓𝟑 ] = 0.0963 −0.0593 −0.1073 (2.29e)
0.1253 0.1194 0.0465
where 𝝓𝟏 , 𝝓𝟐 and 𝝓𝟑 are the 1st , the 2nd and the 3rd mode vibration eigenvectors,
respectively.
The first ordinate-based normalized modal matrix is
1 1 1
𝜱 = [𝝓𝟏 𝝓𝟐 𝝓𝟑 ] = 1.7563 0.6061 −1.0396 (2.29f)
2.2842 −1.2206 0.4506
The schematic representation of the mode shapes (2.29f) is given in the Figure 2.21.
Mode1 Mode2 Mode3
3 3 3

2 2 2
Story

Story

Story

1 1 1

0 0 0
0 1 2 3 -2 -1 0 1 2 -2 -1 0 1 2

Figure 2.21- Vibration mode shapes schematic representation of frame

 Task| Using any available commercial software, construct the full FEM model of the
building in Figure 2.20. Compute the first 20 vibration modes and compare them with
those obtained above. Add your comments on the effect of modeling over the vibration
modes.

 Homework 2.1| Compute the first three vibration modes for the RC cantilever in the
Figure 2.11 when the cross-section changes linearly from the fixed support to the tip- to 50
(height)x20 (width) cm. Compare the results with those obtained in the numerical example
2.3.

 Homework 2.2| Compute the vibration modes for the RC frame in the Figure 2.1, using
a more refined model (e.g. 2 and 3 finite elements for each member, columns and beam).
Compare the results with those obtained in the numerical example 2.1.
35
 Homework 2.3| Based on the theoretical background developed in Chapter 4 of the
Lecture Notes, derive the multi-support excitation analysis framework for the model
obtained in the HW 2.2. The left support is subjected to the El Centro 180 input ground
acceleration record and the right support is subjected to the same input motion but 10
seconds delayed from the first (Figure 2.22). Compare the results with those obtained by
simulatenously applying the same input motion to the fixed supports.

10s
Figure 2.22- Multi-support earthquake ground motion

 Homework 2.4| For the RC-steel mixed structure given in the Figure 2.23, compute the
first 20 vibration modes and compare them with those obtained for the initial full RC
frame structure (Numerical example 2.4).

p(LL)=20kN/m
frame (RC)
truss (steel)
(LL) 𝑘𝑁
p =20kN/m 1.50 𝐸 = 2.1 ∙ 10
B 𝑚
1.50
1.50 𝛾 = 7850 𝑘𝑔/𝑚
20kN/m 20kN/m
1.50
ATRIUM 𝑑 = 60 mm
1.50
𝑡 = 15mm
1.50m

6.00m 6.00 8.00 1.00

C1 C1 C2 C2

Figure 2.23- Building’s view and material properties of steel truss

 Homework 2.5| Compute the in-plane vibration modes of the steel cantilever (see
Numerical Example 2.3) using the discrete elasticity FEM formulation in the presence of
two heavy equipments (Figure 2.24). Evaluate the influence of equipments on the
vibration modes.

36
W1=125 kg W2=90 kg

1.00 3.00 2.00

Figure 2.24– Steel cantilever beam with two mounted equipments

 Homework 2.6| Compute the in-plane vibration modes for the steel cantilever (see
Numerical Example 2.3) supported by soft soil conditions (Figure 2.25). The subgrade
reaction of soil is of 50000 kN/m3 and the influence of its damping over the vibration
modes is neglected.
Hint: model de influence of soil by vertical elastic springs and assign their axial stiffness by
using the Winkler model.

W1=125 kg W2=90 kg

1.00 3.00 2.00

Figure 2.25– Steel cantilever beam with mounted equipments on soft soil

***

37
[page intentionally left blank]

38
3. Damping in structures

This chapter presents numerical examples on MDOF damping modeling using available
proportional damping models. Additionally, damping evaluation by vibration
instrumentation is performed, using structural models subjected to free vibration and the
forced vibration techniques using the shaking table test.

Numerical example 3.1| Given the five story RC frame structure founded on stiff soil
conditions, with specified geometry and physical properties (Figure 3.1), the damping
ratios distribution over the vibration modes is studied, using available proportional
damping models. The slab height is 13 cm.

p(LL)=55.5kN/m

Columns Beams
p(LL)=49.5kN/m
3.00

p(LL)=49.5kN/m 60 cm
3.00 90 cm

60 cm
p(LL)=49.5kN/m 3.00
40 cm

p(LL)= 49.5kN/m
3.00
Young’s modulus 𝐸 = 2x10 𝑘𝑁/𝑚

3.00 Specific weight 𝛾 = 2500 𝑘𝑔/𝑚

7.00 4.00

Figure 3.1- Five story RC plane frame structure

For the sake of simplicity, based on the conclusions depicted from the example 2.3, the
lateral sway-type components only are considered. Additionally, the axial deformation of
beams are neglected, therefore the model will have 5DOF of lateral sway (Figure 3.2).
The loads considered in the analysis are: (i) the Dead Load (DL) and (ii) the Live Load
(LL)- taken as 5 𝑘𝑁/𝑚 for the current floor and 6 𝑘𝑁/𝑚 for the top floor. The tributary
length in the perpendicular direction is 6.00 m.
The sway mass components are computed by direct merging from elements masses as
follows 𝑚 = 73.85 𝑘𝑁𝑠 /𝑚, k=1,2,..,4 and 𝑚 = 76.45 𝑘𝑁𝑠 /𝑚.
39
STRUCTURE MODEL

Figure 3.2- Five story RC frame structure (left figure) and


discrete analysis model (right figure)

Thus, the inertia (mass) matrix has the form

73.8532 0 0 0 0
⎡ 0 73.8532 0 0 0 ⎤
⎢ ⎥
𝑴 , =⎢ 0 0 73.8532 0 0 ⎥, 𝑘𝑁𝑠 /𝑚 (3.1a)
⎢ 0 0 0 73.8532 0 ⎥
⎣ 0 0 0 0 73.8532⎦

The stiffness matrix, easily obtained by typical structural analysis software (e.g. FTOOL
3.1 software; Luiz, 2012), is

5.2288 −2.8773 0.5109 −0.0737 0.0282


⎡−2.8773 4.7318 −2.8134 0.4981 −0.0391⎤
⎢ ⎥
𝑲 , = 10 ⎢ 0.5109 −2.8134 4.7173 −2.7874 0.4375 ⎥,𝑘𝑁/𝑚 (3.1b)
⎢−0.0737 0.4981 −2.7874 4.5393 −2.1865⎥
⎣ 0.0282 −0.0391 0.4375 −2.1865 1.7538 ⎦

Using the ©Matlab routine, given in the Figure 2.4, the spectral matrix, in terms of circular
frequency, natural period and frequency yields

12.0854 0 0 0 0
⎡ 0 37.6019 0 0 0 ⎤
⎢ ⎥
𝜴 , =⎢ 0 0 66.2034 0 0 ⎥ , 𝑟𝑎𝑑/𝑠 (3.2a)
⎢ 0 0 0 94.0736 0 ⎥
⎣ 0 0 0 0 116.2838⎦

40
0.5199 0 0 0 0
⎡ 0 0.1671 0 0 0 ⎤
⎢ ⎥
𝑻 , =⎢ 0 0 0.0949 0 0 ⎥, 𝑠 (3.2b)
⎢ 0 0 0 0.0668 0 ⎥
⎣ 0 0 0 0 0.0540⎦

1.9235 0 0 0 0
⎡ 0 5.9845 0 0 0 ⎤
⎢ ⎥
𝒇 , =⎢ 0 0 10.5366 0 0 ⎥, 𝐻𝑧 (3.2c)
⎢ 0 0 0 14.9723 0 ⎥
⎣ 0 0 0 0 18.5071⎦

The normalized modal matrix, with respect to each first component, is

1.0000 1.0000 1.0000 1.0000 1.0000


⎡2.3464 1.6429 0.5654 −0.5687 −1.3806 ⎤
⎢ ⎥
𝜱 , = ⎢3.5597 1.1076 −0.8613 −0.4755 1.3782 ⎥ (3.2d)
⎢ 4.4990 −0.2450 −0.7254 0.9945 −0.9770⎥
⎣ 5.0810 −1.4630 0.7610 −0.4653 0.3287 ⎦

The vibration mode shapes are represented schematically in the Figure 3.3.

Mode 1: T1=0.52 s Mode 2: T2=0.17 s

5 5

4 4

3 3
Story

Story

2 2

1 1

0 0
0 2 4 6 -2 -1 0 1 2

Mode 3: T3=0.09 s Mode 4: T4=0.07 s

5 5

4 4

3 3
Story

Story

2 2

1 1

0 0
-1.5 -1 -0.5 0 0.5 1 1.5 -1 -0.5 0 0.5 1 1.5

41
Mode 5: T5=0.05 s

Story
2

0
-2 -1 0 1 2

Figure 3.3 – Schematic representation of mode shapes

Three damping proportional models are used in this example (Chopra, 2001):

(i) Mass matrix M proportional damping


Given by
𝑪=𝑎 𝑴 (3.3a)
where 𝑎 is the proportionality constant. By using the fundamental circular frequency 𝜔
and associated damping ratio 𝜁 , this yields
𝑎 = 2𝜁 𝜔 (3.3b)
If the damping ratio for the fundamental mode is set to 5%, then 𝑎 = 1.2085 and the mass
proportional damping matrix has the form of
89.2546 0 0 0 0
⎡ 0 89.2546 0 0 0 ⎤
⎢ ⎥
𝑪 , =⎢ 0 0 89.2546 0 0 ⎥ , 𝑘𝑁𝑚/𝑠 (3.4)
⎢ 0 0 0 89.2546 0 ⎥
⎣ 0 0 0 0 92.3961⎦
The higher modes damping ratio are computed based on the eq. (3.18d) in the Lecture
Notes i.e.

𝜁 = , 𝑖 = 2,3,4,5 (3.5)

(ii) Stiffness matrix K proportional damping

Given by
𝑪=𝑎 𝑲 (3.6a)
where 𝑎 is the proportionality constant. By using the fundamental circular frequency 𝜔
and associated damping ratio 𝜁 , this yields

42
𝑎 = (3.6b)

If the damping ratio for the fundamental mode is set to 5%, then 𝑎 = 0.0083 and the ith
damping ratio has the expression

𝜁 = 𝜔 , 𝑖 = 2,3,4,5 (3.7)

The stiffness proportional damping matrix is


4.3265 −2.3808 0.4227 −0.0610 0.0234
⎡ −2.3808 3.9153 −2.3279 0.4122 −0.0323⎤
⎢ ⎥
𝑪 , = 10 ⎢ 0.4227 −2.3279 3.9033 −2.3064 0.3620 ⎥ , 𝑘𝑁𝑚/𝑠 (3.8)
⎢ −0.0610 0.4122 −2.3064 3.7560 −1.8092⎥
⎣ 0.0234 −0.0323 0.3620 −1.8092 1.4511 ⎦

(iii) Mass matrix M and stiffness matrix K proportional damping (Rayleigh model)

Given by
𝑪 =𝑎 𝑴+𝑎 𝑲 (3.9a)
where 𝑎 and 𝑎 are proportionality constants.
If the damping ratio for the first two modes is set to 5%, then 𝑎 = 0.9146 and 𝑎 =
0.0020, from the eq. (3.27) in the Lecture Notes i.e.
𝑎 𝜔 𝜔
𝑎 = (3.10)
1
Thus the ith damping ratio is given by

𝜁 = + 𝜔 (3.11)

The Rayleigh model- based damping matrix has the form of

1.1199 −0.5791 0.1028 −0.0148 0.0057


⎡ −0.5791 1.0199 −0.5662 0.1003 −0.0079 ⎤
⎢ ⎥
𝑪 , = 10 ⎢ 0.1028 −0.5662 1.0169 −0.5610 0.0880 ⎥ , 𝑘𝑁𝑚/𝑠 (3.12)
⎢ −0.0148 0.1003 −0.5610 0.9811 −0.4401⎥
⎣ 0.0057 −0.0079 0.0880 −0.4401 0.4229 ⎦
The circular frequency corresponding to the minimum damping ratio would be

 𝜔| = = 21.32 𝑟𝑎𝑑/𝑠 (3.13a)

The corresponding minimum damping ratio would be (white dot in the Figure 3.4)

𝜁 = 𝑎 𝑎 = 4.30 % (3.13b)

The results are presented synthetically in Table 1 and represented in the Figure 3.4.

43
Table 1 – Modal damping ratios

Damping ratio, 𝜁 (%)


Vibration mode index M prop. K prop. Rayleigh
1 5.00 5.00 5.00
2 1.61 15.56 5.00
3 0.91 27.39 7.35
4 0.64 38.92 9.95
5 0.52 48.11 12.09

From the Figure 3.4, notice that for the mass proportional damping, the higher modes are
under-damped, while for the stiffness proportional damping, the higher modes are over-
damped. Rayleigh model goes in between the two assumptions.

50
M prop 48.11
45
K prop
40 38.92
Rayleigh
35
30
27.39
25
ζ,%

20
15 15.56 12.09
9.95
10 7.35
5
5 5
0.64
1.61 0.91 0.52
5
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130
ω, rad/s

Figure 3.4- Modal damping ratios

The effect of damping model on the dynamic response by direct integration is presented in
the numerical example 4.5-for a rigid model and in the Figure 3.7 (Lecture Notes)- for a
flexible model.

Numerical example 3.2|For a single story rigid model mounted on a shaking table, from
the test using 24 input harmonic components, with excitation frequencies varying from
1Hz to 5Hz, the following data were recorded* (Table 2). The natural vibration frequency
and damping ratio are determined by forced vibration test free vibration test respectively.
The results are then compared.

44
Recording point Model

INPUT
Shaking table

Actuator

Figure 3.5 – Schematic representation of shaking table test and steady-state recorded response

* test performed at the International Institute of Seismology and Earthquake Engineering


(IISEE), Japan (Kashima, 2001; with permission).

Table 2- Input and recorded data from shaking table test

Harmonic no. Frequency,Hz Base acc. (g) Top acc. (g) Ratio top/base
1 1 0.043 0.048 1.12
2 1.5 0.038 0.040 1.07
3 2 0.033 0.050 1.54
4 2.5 0.025 0.058 2.30
5 2.8 0.026 0.071 2.76
6 3 0.030 0.113 3.75
7 3.1 0.030 0.133 4.42
8 3.2 0.028 0.178 6.45
9 3.3 0.030 0.251 8.37
10 3.35 0.028 0.345 12.55
11 3.4 0.026 0.375 14.71
12 3.45 0.020 0.480 24.00
13 3.5 0.021 0.380 18.10
14 3.55 0.025 0.370 14.80
15 3.6 0.025 0.320 12.80
16 3.65 0.029 0.270 9.31
17 3.7 0.030 0.210 7.00
18 3.8 0.030 0.150 5.00
19 3.9 0.028 0.120 4.29
20 4 0.033 0.090 2.77
21 4.2 0.033 0.070 2.15
22 4.4 0.033 0.050 1.54
23 4.7 0.033 0.040 1.23
24 5 0.033 0.029 0.89

45
The ratio of top-to-input base recorded acceleration vs. excitation frequency is plotted in
the Figure 3.6 i.e. the frequency response curve (resonance curve).

25 24.00

20

Top acc./Base acc.


15

10 𝑥̈
( ) /√2
𝑎

0
1 2 3
𝑓𝑓 4 5

Frequency, Hz

Figure 3.6- Frequency response function from experimental test

The peak amplitude ratio yields the natural frequency of 𝑓 = 3.45 𝐻𝑧. Therefore the
natural vibration period yields 𝑇 = = 0.3 𝑠.
The damping ratio can be obtained (see Figure 3.3 in the Lecture Notes) by three
procedures:

(i) Direct identification, as

𝜁= ̈ = ∙
= 0.021 = 2.1% (3.14)

(ii) Half Power Method,


̈
where 𝑓 = 3.41 𝐻𝑧 and 𝑓 = 3.52 𝐻𝑧 , are the frequencies corresponding to ( ) /√2 ,
determined by linear data interpolation (Figure 3.6).
The damping ratio yields thus

𝜁= = 0.016 = 1.6 % (3.15)

(iii) Logarithmic damping decay approach, using the shaking table technique. Thus, two
available consecutive relative acceleration positive peaks are identified as 𝑥̈ = 0.16𝑔 and
𝑥̈ = 0.14𝑔. Therefore the damping ratio yields by logarithmic damping decay technique
as
.

𝜁= .
= 0,021 = 2.1% (3.16)

46
Small difference between the damping ratios obtained by various techniques, are
attributed to measurement errors and inherent approximations.

 Homework 3.1|Identify the modal parameters (natural frequencies and damping


ratios) for the structure analyzed in the numerical example 3.1, using the frequency-
domain analysis tools, given in the Chapter 5 of the accompanying Lecture Notes.
Hint: Generate a Gaussian white noise acceleration support excitation, perform then the
time-history analysis by direct integration technique (given in the Chapter 5 of the
accompanying Lecture Notes and in the Chapter 4 of present textbook) and get the
acceleration transfer function which yields finally the vibration periods and corresponding
damping ratios.

***

47
[page intentionally left blank]

48
4. Dynamic response analysis.
Step-by-step integration methods

The time-domain analysis approach is the most powerful tool for the structural response
analysis. This chapter is dedicated to the dynamic response analysis of linear and
nonlinear models using step-by-step integration approach. Thus, the time-history response
analysis framework of SDOF and MDOF models is developed, by using (i) the Direct
Integration approach and (ii) the Mode Superposition Method. The effect of physical nonlinear
behavior on the structural response is emphasized, as well as the effect of various
proportional damping models used in the analyses. Other important topics e.g. the use of
smooth hysteretic models (SHM), distributed plasticity, nonlinear damping effect,
aerodynamic damping effect, soil-structure interaction effect, multi-support excitation
effect, energy balance-based analysis etc., are to be further developed using given analysis
routines and the accompanying support Lecture Notes.

Numerical example 4.1|For a SDOF model, given by the mass 𝑚 = 40 𝑘𝑁𝑠 /𝑚, stiffness
𝑘 = 4000 𝑘𝑁/𝑚 and damping ratio 𝜁 = 3%, subjected to a forcing function 𝐹(𝑡) specified
in Table 1 and Figure 4.2, the dynamic response analysis is performed using the Linear
Acceleration Method integration scheme (see Chapter 4 in the Lecture Notes).
The natural period yields 𝑇 = 0.63𝑠, damping coefficient 𝑐 = 𝑚𝜁 = 24𝑘𝑁/𝑚 and the
dynamic stiffness 𝑘 = 28720 𝑘𝑁/𝑚. First, the integration scheme (Chopra, 2001;
Buchholdt et al., 2012) requires validation through a benchmark test, by comparing the
numerical solution with the “exact” solution (see Chapter 3, paragraph 3.2 in the Lecture
Notes), for a testing harmonic input load given 𝐹(𝑡) = 40sin (25𝑡) on 11 s duration.
The time-histories of input load and response parameters are given in the Figure 4.1, on 7
seconds duration.

50
40
30
20
10
F(t),kN

0
-10 0 1 2 3 4 5 6
-20
-30
-40
-50
time,s

49
0.008
exact
0.006
LAM

Displacement, m
0.004

0.002

-0.002

-0.004

-0.006
0 1 2 3 4 5 6
time,s

0.1
exact
0.08
LAM
0.06
0.04
Velocity, m/s

0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
0 1 2 3 4 5 6
time,s

2
exact
1.5 LAM
1
Acceleration, m/s2

0.5
0
-0.5
-1
-1.5
-2
0 1 2 3 4 5 6
time,s

Figure 4.1a- Benchmark analysis: time-history of harmonic input load and SDOF response
functions, exact solution vs. numerical (LAM) solution

The LAM integration scheme provides results in a good agreement with the exact solution.
The integration scheme using the Linear Acceleration Method (LAM) is given in Table 1
and the results obtained by a simple Microsoft Excel platform, are plotted in the Figure
4.1b.
Table 1- LAM integration scheme

𝒕𝒊𝒎𝒆, 𝑠 𝑭, 𝑘𝑁 ∆𝑭, 𝑘𝑁 ∆𝑭𝒅 , 𝑘𝑁 ∆𝒙, 𝑚 𝒙,m/s 𝒙̇ ,m/s 𝒙̈ , 𝑚/𝑠


0.0 0 0 0 0 0 0 0
0.1 4 4 4.00000 0.00014 0.00014 0.00418 0.08357
0.2 6 2 22.45682 0.00078 0.00092 0.01092 0.05133
0.3 8 2 35.22209 0.00123 0.00215 0.01238 -0.02219
0.4 12 4 31.91381 0.00111 0.00326 0.00969 -0.03169
0.5 15 3 23.10249 0.00080 0.00406 0.00634 -0.03513
0.6 22 7 18.42726 0.00064 0.00470 0.00832 0.07453
0.7 16 -6 23.58762 0.00082 0.00553 0.00428 -0.15518

50
0.8 10 -6 -14.22186 -0.00050 0.00503 -0.01566 -0.24370
0.9 4 -6 -74.25103 -0.00259 0.00245 -0.03405 -0.12413
1.0 2 -2 -101.22217 -0.00352 -0.00108 -0.03142 0.17674
1.1 1 -1 -57.25422 -0.00199 -0.00307 -0.00580 0.33572
1.2 0 -1 25.35396 0.00088 -0.00219 0.02130 0.20618
1.3 0 0 77.63307 0.00270 0.00051 0.02819 -0.06827
1.4 0 0 61.41652 0.00214 0.00265 0.01118 -0.27191
1.5 0 0 -5.30995 -0.00018 0.00247 -0.01432 -0.23812
1.6 0 0 -64.25351 -0.00224 0.00023 -0.02658 -0.00704
1.7 0 0 -66.54795 -0.00232 -0.00209 -0.01601 0.21834
1.8 0 0 -13.11624 -0.00046 -0.00254 0.00740 0.24996
1.9 0 0 48.59708 0.00169 -0.00085 0.02346 0.07111
2.0 0 0 66.60577 0.00232 0.00147 0.01910 -0.15819

Note that when the forcing function stops, the system starts the free vibration phase, with
initial conditions given in Table 1.

25
22
20

16
15 15
F(t),kN

12
10 10
8
6
5
4 4
2
1
0 0
0.0 0.5 1.0 1.5 2.0
time,s

0.006
0.005
0.004
0.003
Displacement, m

0.002
0.001
0
-0.001 0.0 0.5 1.0 1.5 2.0

-0.002
-0.003
-0.004
time,s

0.04

0.03

0.02
Velocity, m/s

0.01

0
0.0 0.5 1.0 1.5 2.0
-0.01

-0.02

-0.03

-0.04
time,s

51
0.4

0.3

0.2

Acceleration, m/s2
0.1

0
0.0 0.5 1.0 1.5 2.0
-0.1

-0.2

-0.3

-0.4
time,s

Figure 4.1b- Input load and time-history response functions

The resistant force components are plotted in the Figure 4.2

Restoring force Damping force


25 0.8
20 0.6

15 0.4
0.2
10
0
Fd,kN
Fr, kN

5 -0.004 -0.002 -0.2 0 0.002 0.004 0.006


0
-0.4
-0.004 -0.002 0 0.002 0.004 0.006
-5 -0.6
-10 -0.8
-15 -1

Displacement, m Displacement, m

Resistant force
25
20
15
10
Fr+Fd, kN

5
0
-0.004 -0.002 0 0.002 0.004 0.006
-5
-10
-15

Displacement, m

Figure 4.2- Resistant force components

In the Figure 4.3 the Matlab-based integration scheme for linear-elastic SDOF systems is
given.

52
Numerical example 4.2| For the SDOF model, given by the mass 𝑚 = 15 𝑘𝑁𝑠 /𝑚, stiffness
𝑘 = 40000 𝑘𝑁/𝑚 and damping ratio 𝜁 = 5%, subjected to a triangular blast load 𝐹(𝑡), the
dynamic response analysis is performed using the Linear Acceleration Method integration
routine developed in the Figure 4.3.
The time-history simplified blast triangular load model (Figure 4.4) can be expressed as

𝐹(𝑡) = 𝐹 1− (4.1)

where 𝐹 is the maximum force and 𝑡 is the positive phase blast loading duration.

The blast impulse is approximately the area under the curve i.e.
𝐼 = 0.5𝐹 𝑡 (4.2)
The equation of motion of an undamped linear-elastic SDOF system is therefore

𝑚𝑥̈ + 𝑘𝑥 = 𝐹 1− (4.3)

since the damping force is likely not to be activated due to blast positive phase small
duration 𝑡 . The closed-form displacement solution is given by (Biggs, 1964)

𝑥= (1 − 𝑐𝑜𝑠𝜔𝑡) + −𝑡 (4.4)

The response velocity is

𝑥̇ = 𝜔𝑠𝑖𝑛𝜔𝑡 + (𝑐𝑜𝑠𝜔𝑡 − 1) (4.5)

The displacement Dynamic Amplification Factor is

𝐷𝐴𝐹 = = = 𝑓(𝜔𝑡 ) = 𝑓( ) (4.6)


,

Three loading regimes are depicted as follows (Biggs, 1964):


 𝜔𝑡 < 0.4, impulsive loading regime
 𝜔𝑡 = 0.4, quasi-static loading regime
 0.4 < 𝜔𝑡 < 40, dynamic loading regime

The natural period yields 𝑇 = 0.12𝑠 and the damping coefficient 𝑐 = 𝑚𝜁 = 77.46𝑘𝑁/𝑚.
The 𝜔𝑡 parameter yields 10.32, that indicates a dynamic loading regime.
In the Figure 4.3, the Matlab-based integration scheme for linear-elastic SDOF systems is
given. The time-history response and peak response, are shown in the Figure 4.4. The
displacement Dynamic Amplification Factor is therefore 1.13.
 Task| Compare the closed-form “exact” solution with the numerical one given in the
Figure 4.4.

53
% Dynamic response analysis of linear-elastic SDOF systems
% coded by Mihail Iancovici, UTCB

clear all

% Input data
tf=10; % final time sequence, s
dt = 0.01; %the time step,s
t=0:dt:tf; % time interval
Npoints=unum(t);% number of discrete data points
durat = (Npoints-1)*dt; % time duration,s

% Input load
Fm=1000; % peak load, kN
td=0.2;% positive phase loading duration,s
f_t=1000*tripuls(t,2*td); %blast load generation

% Structural model input data


m=15;% mass kNs2/m
k=40000;% stiffness kN/m
damp=0.05;% critical damping ratio

%compute vibration eigenvalue and damping coefficient


omega=sqrt(k/m); % circular vibration frequency
T=2*pi/omega;% natural vibration period
f=1/T;% natural vibration frequency
c=2*damp*m*omega;% damping coefficient

% Dynamic response analysis


% Linear Acceleration Method integration scheme
kd=(m*6/dt^2)+(c*3/dt)+k; % dynamic stiffness
%set initial conditions (at t=0)
for i=1
x(i)=0; % displacement
xd(i)=0; % velocity
xdd(i)=0; % acceleration
dF(i)=0; % load increment
dFd(i)=0; % effective dynamic load
dx(i)=0; % displacement increment
end
for i=2:Npoints
dF(i)=f_t(i)-f_t(i-1); % load increment
dFd(i)=dF(i)+m*(3*xdd(i-1)+(6*xd(i-1)/dt))+c*(3*xd(i-1)+…
0.5*xdd(i-1)*dt); % effective dynamic load
dx(i)=inv(kd)*dFd(i); % displacement increment
x(i)=x(i-1)+dx(i); % displacement
xd(i)=(3*dx(i)/dt)-2*xd(i-1)-0.5*dt*xdd(i-1); % velocity
xdd(i)=(6*dx(i)/dt^2)-(6*xd(i-1)/dt)-2*xdd(i-1); % acceleration
end

%plots
subplot(2,2,1);plot(t,f_t,'linewidth',2);
xlabel('t,s','fontsize',20);
ylabel('Load,kN','fontsize',20); grid off;box off;
subplot(2,2,2);plot(t,x,'linewidth',2);
xlabel('t,s','fontsize',20);
ylabel('Disp,m','fontsize',20);grid off;box off;
subplot(2,2,3);plot(t,xd,'linewidth',2);
xlabel('t,s','fontsize',20);
ylabel('Vel,m/s','fontsize',20);grid off;box off;
subplot(2,2,4);plot(t,xdd,'linewidth',2);
xlabel('t,s','fontsize',20);
ylabel('Acc,m/s^2','fontsize',20);grid off;box off;

Figure 4.3- LAM integration routine for linear-elastic SDOF model under blast load

54
1000 0

800
-0.01
Load,kN

600

Disp,m
400
-0.02
200

0 -0.03
0 1 2 3 4 5 0 1 2 3 4 5
t,s t,s
0.028 at 0.24s
0.2 10

0.1 9 at 0.23s
5

Acc,m/s2
Vel,m/s

0
0
-0.1
-5
-0.2
0.023 at 0.07s
-0.3 -10
0 1 2 3 4 5 0 1 2 3 4 5
t,s t,s

Figure 4.4- Input blast load (0.2s duration) and linear-elastic SDOF time-history response on 5s

The LAM integration scheme can be further adapted and used for various load types and
SDOF models.

 Homework 4.1| Use the LAM routine to compute the time-history analysis for the
SDOF model above, subjected to a triangular periodic (sawtooth) function given in the
Figure 1.5.

 Task | Develop the LAM routine to obtain seismic response spectra of linear-elastic
systems.

Homework 4.2| Given the steel cantilever commercial panel subjected to an uniform
wind-induced pressure measured perpendicular to the panel (Figure 4.5 left) acting on 14
minutes duration (Figure 4.5 right, given in the form of data file).
4.00m
Trussed light panel

14
12
10
3.00m 8
6
p(t),kN/m2

4
2
0
Wind blowing -2 0 2 4 6 8 10 12 14
12.00m direction -4
-6
time, min

Figure 4.5 – Cantilever steel commercial panel (left figure) and


along-wind induced pressure (right figure)

55
Using a lateral sway SDOF linear-elastic model, the requirements are:
(i) Compute the time-histories and peak values of deflection, velocity and
acceleration of the panel;
(ii) Compute the time-histories and peak values of base shear force and bending
moment at the base, and
(iii) Estimate the duration to rest after wind stops blowing.
 Material: steel
Specific weight: 𝛾 = 7850 𝑘𝑔/𝑚
Young’s modulus: 𝐸 = 2.1𝑥10 𝑘𝑁/𝑚
 Sections
Column: ring section d=120 cm, t=10 cm
Damping ratio 𝜁 = 1%
 The SDOF- associated sway mass (column and trussed panel) is 15 kNs2/m

Answer

Peak deflection=___________(m)
Peak velocity=_____________(m/s)
Peak acceleration= _________(m/s2)
Peak damping force=__________(kN)
Peak restoring force=__________(kN)
Peak base shear force=_________(kN)
Peak base bending moment =_______(kNm)
Duration to rest=_________________(s)

Add your comment (if any):

Numerical example 4.3| For the nonlinear SDOF model, given by mass 𝑚 = 40 𝑘𝑁𝑠 /𝑚 ,
elastic stiffness 𝑘 = 4000 𝑘𝑁/𝑚 and damping ratio 𝜁 = 3%, subjected to an arbitrary
forcing function 𝐹(𝑡), the dynamic response analysis is performed using the integration
scheme given in the Table 1. The applied load 𝐹(𝑡) is chosen here ten times larger that in
the numerical example 4.1.

Recall the natural period 𝑇 = 0.63𝑠, damping coefficient 𝑐 = 𝑚𝜁 = 24𝑘𝑁/𝑚. The


restoring force-deflection relationship is chosen an elasto-plastic one (Figure 4.5).

56
𝑭

𝐹𝑦+

𝑘 𝑘
1 1
∆−
𝑢 ∆−
𝑦

∆+
𝑦 ∆+
𝑢 ∆

𝑘
1

𝐹𝑦−

Figure 4.5- Elasto-plastic force-deflection relationship

The constitutive parameters in the elasto-plastic curve are chosen as follows


 Yield restoring force: 𝐹 = 𝐹 = 0.4𝑚𝑔 = 160 𝑘𝑁

 Yield displacement: ∆ = ∆ = 0.04 𝑚

 Displacement ductility factor: 𝜇 = 5


Therefore the ultimate displacement (in both directions) yields
∆ = |∆ | = 𝜇∆ = 0.2 𝑚 . Reaching the ultimate displacement is associated to the failure.
The integration scheme, developed using the Microsoft Excel platform, is shown in the
Table 2 and the results are plotted in the Figure 4.6.

Table 2- LAM integration scheme (nonlinear model)


time, s 𝑭, 𝑘𝑁 ∆𝑭, 𝑘𝑁 𝒌, 𝑘𝑁/𝑚 𝒌𝒅 , 𝑘𝑁/𝑚 ∆𝑭𝒅 , 𝑘𝑁 ∆𝒙, 𝑚 𝒙,m/s 𝒙̇ ,m/s 𝒙̈ , 𝑚/𝑠
0.0 0 0 4000 28720 0 0 0 0 0
0.1 40 40 4000 28720 40.00000 0.00139 0.00139 0.04178 0.83565
0.2 60 20 4000 28720 224.56825 0.00782 0.00921 0.10923 0.51326
0.3 80 20 4000 28720 352.22089 0.01226 0.02148 0.12380 -0.22187
0.4 120 40 4000 28720 319.13811 0.01111 0.03259 0.09686 -0.31691
0.5 150 30 4000 28720 231.02486 0.00804 0.04063 0.06345 -0.35127
0.6 220 70 0 24720 184.27255 0.00745 0.04809 0.11430 1.36822
0.7 160 -60 0 24720 388.36934 0.01571 0.06380 0.17432 -0.16779
0.8 100 -60 0 24720 350.57700 0.01418 0.07798 0.08521 -1.61433
0.9 40 -60 0 24720 -45.01385 -0.00182 0.07616 -0.14434 -2.97660
1.0 20 -20 4000 28720 -737.56057 -0.02568 0.05048 -0.33293 -0.79534
1.1 10 -10 4000 28720 -929.40282 -0.03236 0.01812 -0.26519 2.15010
1.2 0 -10 4000 28720 -404.96650 -0.01410 0.00402 -0.00013 3.15112
1.3 0 0 4000 28720 381.58708 0.01329 0.01730 0.24130 1.67761
1.4 0 0 4000 28720 799.82844 0.02785 0.04515 0.26899 -1.12392
1.5 0 0 4000 28720 528.71917 0.01841 0.06356 0.07050 -2.84577
1.6 0 0 4000 28720 -170.62348 -0.00594 0.05762 -0.17695 -2.10321
1.7 0 0 4000 28720 -692.31930 -0.02411 0.03351 -0.26412 0.35968
1.8 0 0 4000 28720 -609.31786 -0.02122 0.01230 -0.12621 2.39851
1.9 0 0 4000 28720 -21.29879 -0.00074 0.01156 0.11025 2.33079
2.0 0 0 4000 28720 555.03587 0.01933 0.03088 0.24273 0.31873

57
For the sake of comparison with the linear-elastic model’s response, the integration
algorithm results are given in the Table 3.

Table 3- LAM integration scheme (linear model)


time, s 𝑭, 𝑘𝑁 ∆𝑭, 𝑘𝑁 𝒌, 𝑘𝑁/𝑚 𝒌𝒅 , 𝑘𝑁/𝑚 ∆𝑭𝒅 , 𝑘𝑁 ∆𝒙, 𝑚 𝒙,m/s 𝒙̇ ,m/s 𝒙̈ , 𝑚/𝑠
0.0 0 0 4000 28720 0 0 0 0 0
0.1 40 40 4000 28720 40.00000 0.00139 0.00139 0.04178 0.83565
0.2 60 20 4000 28720 224.56825 0.00782 0.00921 0.10923 0.51326
0.3 80 20 4000 28720 352.22089 0.01226 0.02148 0.12380 -0.22187
0.4 120 40 4000 28720 319.13811 0.01111 0.03259 0.09686 -0.31691
0.5 150 30 4000 28720 231.02486 0.00804 0.04063 0.06345 -0.35127
0.6 220 70 4000 28720 184.27255 0.00642 0.04705 0.08315 0.74529
0.7 160 -60 4000 28720 235.87617 0.00821 0.05526 0.04282 -1.55181
0.8 100 -60 4000 28720 -142.21859 -0.00495 0.05031 -0.15661 -2.43696
0.9 40 -60 4000 28720 -742.51033 -0.02585 0.02446 -0.34053 -1.24127
1.0 20 -20 4000 28720 -1012.22171 -0.03524 -0.01079 -0.31422 1.76740
1.1 10 -10 4000 28720 -572.54223 -0.01994 -0.03072 -0.05799 3.35719
1.2 0 -10 4000 28720 253.53958 0.00883 -0.02190 0.21296 2.06182
1.3 0 0 4000 28720 776.33073 0.02703 0.00514 0.28192 -0.68265
1.4 0 0 4000 28720 614.16517 0.02138 0.02652 0.11183 -2.71906
1.5 0 0 4000 28720 -53.09948 -0.00185 0.02467 -0.14318 -2.38117
1.6 0 0 4000 28720 -642.53514 -0.02237 0.00230 -0.26576 -0.07038
1.7 0 0 4000 28720 -665.47948 -0.02317 -0.02087 -0.16011 2.18336
1.8 0 0 4000 28720 -131.16242 -0.00457 -0.02544 0.07404 2.49956
1.9 0 0 4000 28720 485.97084 0.01692 -0.00852 0.23457 0.71114
2.0 0 0 4000 28720 666.05773 0.02319 0.01467 0.19104 -1.58188

Note that in the nonlinear integration scheme, the stiffness requires update at each time-
step.
In the Figure 4.6, the nonlinear response is compared with the linear-elastic SDOF model’s
response, for the same input load.
250
220
200

160
150 150
F(t),kN

120
100 100
80
60
50
40 40
20
10
0 0
0.0 0.5 1.0 1.5 2.0
time,s

0.1

0.08

0.06
Displacement, m

0.04

0.02

0
0.0 0.5 1.0 1.5 2.0
-0.02 linear
nonlinear
-0.04
time,s

58
0.4

0.3

0.2

Velocity, m/s
0.1

0
0.0 0.5 1.0 1.5 2.0
-0.1

-0.2

-0.3
linear
-0.4 nonlinear
time,s

2
Acceleration, m/s2

0
0.0 0.5 1.0 1.5 2.0
-1

-2

-3 linear
nonlinear
-4
time,s

Figure 4.6- Input load and time-history response functions


(linear-black dotted line and nonlinear-red line)

The resistant force components are shown in the Figure 4.7, for the linear-elastic and for
the nonlinear model, respectively.

Restoring force Damping force


250 8
200 6
150 4
100 2
50 0
Fd,kN
Fr, kN

0 -0.04 -0.02 -2 0 0.02 0.04 0.06 0.08 0.1


-0.04 -0.02 -50 0 0.02 0.04 0.06 0.08 0.1 -4
-100 -6
-150 linear -8
-200 nonlinear -10 linear
nonlinear
Displacement, m Displacement, m

Resistant force
250
200
150
100
50
Fr+Fd, kN

0
-0.04 -0.02 -50 0 0.02 0.04 0.06 0.08 0.1

-100
-150 linear
-200 nonlinear

Displacement, m

Figure 4.7- Resistant force components (linear and nonlinear model)


59
The effects of nonlinear behavior cannot be catched by a linear-elastic model. Moreover,
the nonlinear vibration phenomenon needs to be extended over the immediate range after
the load stops depending of load type. For instance, extending the time range for analysis,
in the Figure 4.7a, the linear and nonlinear responses are compared. While, due to the
presence of damping, the linear model’s displacement vanishes in time, in the case of
nonlinear model, the failure occurs at 2.6s time-instant, when the system’s reaches its
ultimate displacement given by the constitutive law, even if the load stopped (i.e.
impulsive load type).
0.25 250
0.22 200
0.2 linear
150
failure at <2.6s nonlinear
100
Displacement, m

0.15
failure at 0.2m
50

Fr, kN
0.1 0
-0.06 -0.01 0.04 0.09 0.14 0.19 0.24
0.05
Load stops -50
-100

0 -150 linear
0.0 5.0 10.0 15.0 -200 nonlinear
-0.05
time,s Displacement, m

Figure 4.7a- Response displacement (left figure) and restoring force characteristics (right figure)
for the linear and nonlinear models

The time-history up to the failure time-instant is represented in the Figure 4.7b, for both
linear and nonlinear models.
0.25

linear 0.22
0.2
nonlinear failure at <2.6s
0.15
Displacement, m

0.1

0.05

0
0.0 0.5 1.0 1.5 2.0 2.5

-0.05
time,s

0.6
0.5 linear
0.4 nonlinear
0.3
Velocity, m/s

0.2
0.1
0
-0.1 0.0 0.5 1.0 1.5 2.0 2.5

-0.2
-0.3
-0.4
time,s

60
4

Acceleration, m/s2
1

0
0.0 0.5 1.0 1.5 2.0 2.5
-1

-2

-3 linear
nonlinear
-4
time,s

Figure 4.7b- Time-history response functions (linear-black dotted line and nonlinear-red line)
up to the failure point

(i) Time-step size sensitivity analysis: Refining the integration time-step to half of the initial
one, the influence on the response is studied in this section. The nonlinear integration
scheme is shown in Table 4 and the results are plotted in the Figure 4.7c, altogether
with the linear model’s response and nonlinear model’s response, with initial step-size
of 0.1s.

Table 4- LAM integration scheme (nonlinear model, 0.05s time-step)

time, s 𝑭, 𝑘𝑁 ∆𝑭, 𝑘𝑁 𝒌, 𝑘𝑁/𝑚 𝒌𝒅 , 𝑘𝑁/𝑚 ∆𝑭𝒅 , 𝑘𝑁 ∆𝒙, 𝑚 𝒙,m/s 𝒙̇ ,m/s 𝒙̈ , 𝑚/𝑠


0.0 0 0 4000 101440 0.000 0.000 0.000 0.000 0.000
0.05 20 20 4000 101440 20.000 0.000 0.000 0.012 0.473
0.10 40 20 4000 101440 134.700 0.001 0.002 0.044 0.821
0.15 50 10 4000 101440 324.275 0.003 0.005 0.083 0.728
0.20 60 10 4000 101440 501.746 0.005 0.010 0.113 0.466
0.25 70 10 4000 101440 615.469 0.006 0.016 0.127 0.100
0.30 80 10 4000 101440 640.348 0.006 0.022 0.122 -0.278
0.35 100 20 4000 101440 583.018 0.006 0.028 0.107 -0.344
0.40 120 20 4000 101440 499.378 0.005 0.033 0.090 -0.326
0.45 135 15 4000 101440 414.963 0.004 0.037 0.073 -0.350
0.50 150 15 4000 101440 329.768 0.003 0.040 0.057 -0.290
0.55 185 35 0 97440 278.970 0.003 0.043 0.065 0.580
0.60 220 35 0 97440 419.290 0.004 0.047 0.115 1.425
0.65 190 -30 0 97440 700.462 0.007 0.054 0.166 0.644
0.70 160 -30 0 97440 858.324 0.009 0.063 0.180 -0.114
0.75 130 -30 0 97440 831.520 0.009 0.072 0.156 -0.849
0.80 100 -30 0 97440 625.508 0.006 0.078 0.095 -1.563
0.85 70 -30 0 97440 245.585 0.003 0.081 0.000 -2.256
0.90 40 -30 0 97440 -303.109 -0.003 0.078 -0.130 -2.928
0.95 30 -10 4000 101440 -995.585 -0.010 0.068 -0.256 -2.121
1.00 20 -10 4000 101440 -1513.206 -0.015 0.053 -0.330 -0.835

61
1.05 15 -5 4000 101440 -1713.116 -0.017 0.036 -0.333 0.731
1.10 10 -5 4000 101440 -1537.040 -0.015 0.021 -0.262 2.079
1.15 0 -10 4000 101440 -1037.310 -0.010 0.011 -0.141 2.778
1.20 0 0 4000 101440 -351.347 -0.003 0.007 0.005 3.037
1.25 0 0 4000 101440 388.267 0.004 0.011 0.145 2.571
1.30 0 0 4000 101440 1015.021 0.010 0.021 0.247 1.509
1.35 0 0 4000 101440 1383.834 0.014 0.035 0.287 0.120
1.40 0 0 4000 101440 1414.770 0.014 0.049 0.259 -1.257
1.45 0 0 0 97440 1110.103 0.011 0.060 0.197 -1.220
1.50 0 0 0 97440 812.793 0.008 0.068 0.137 -1.184
1.55 0 0 0 97440 524.270 0.005 0.074 0.079 -1.149
1.60 0 0 0 97440 244.274 0.003 0.076 0.022 -1.115
1.65 0 0 0 97440 -27.445 0.000 0.076 -0.033 -1.082
1.70 0 0 4000 101440 -291.134 -0.003 0.073 -0.079 -0.768
1.75 0 0 4000 101440 -478.487 -0.005 0.068 -0.105 -0.280
1.80 0 0 4000 101440 -547.315 -0.005 0.063 -0.106 0.260
1.85 0 0 4000 101440 -484.698 -0.005 0.058 -0.081 0.723
1.90 0 0 4000 101440 -309.182 -0.003 0.055 -0.038 1.002
1.95 0 0 4000 101440 -65.499 -0.001 0.054 0.013 1.036
2.00 0 0 0 97440 186.762 0.002 0.056 0.064 1.005
2.05 0 0 0 97440 431.643 0.004 0.061 0.113 0.975
2.10 0 0 0 97440 669.287 0.007 0.068 0.161 0.947
2.15 0 0 0 97440 899.907 0.009 0.077 0.208 0.919
2.20 0 0 0 97440 1123.710 0.012 0.088 0.253 0.891
2.25 0 0 0 97440 1340.899 0.014 0.102 0.297 0.865
2.30 0 0 0 97440 1551.668 0.016 0.118 0.340 0.840
2.35 0 0 0 97440 1756.207 0.018 0.136 0.381 0.815
2.40 0 0 0 97440 1954.701 0.020 0.156 0.421 0.791
2.45 0 0 0 97440 2147.329 0.022 0.178 0.460 0.767
2.50 0 0 0 97440 2334.262 0.024 0.202 0.498 0.745
2.55 0 0 0 97440 2515.671 0.026 0.228 0.535 0.723
2.60 0 0 0 97440 2691.718 0.028 0.256 0.570 0.701

The sensitivity analysis shows that there is no significant influence neither on the
nonlinear response nor on the displacement associated to failure (0.20m occurring at 2.5 s).

0.3
linear failure at <2.5s
0.25
nonlinear 0.1s
0.2 nonlinear 0.05s
Displacement, m

0.15

0.1

0.05

0
0.0 0.5 1.0 1.5 2.0 2.5
-0.05
time,s

62
0.8

0.6

0.4

Velocity, m/s
0.2

0
0.0 0.5 1.0 1.5 2.0 2.5
-0.2
linear
-0.4
nonlinear 0.1s
-0.6 nonlinear 0.05s
time,s

2
Acceleration, m/s2

0
0.0 0.5 1.0 1.5 2.0 2.5
-1

-2

-3 linear
nonlinear 0.1s
-4
nonlinear 0.05s time,s

Figure 4.7c- Time-history response functions (linear and nonlinear- 0.1s and 0.05s time-steps)
up to the failure point

 Homework 4.2| Derive the Matlab-based routine for the nonlinear response analysis,
based on the routine for the elastic model’s response (Figure 4.3). Perform then the
analysis for a bi-linear SDOF model, having 10% post-yield stiffness degrading ratio in
both directions and the nonlinear properties from numerical example 4.3. Compare then
the results obtained with both hysteretic curves.
 Homework 4.3| Based on the nonlinear LAM integration routine developed in the HW
4.2, perform the analysis of the bi-linear SDOF model in HW 4.2, for a blast load given in
the Figure 1.2b with 0.5 s duration. Compare then the elastic and the inelastic responses.
 Task|Based on the LAM integration routine for the elastic model in the numerical
example 4.1 (Figure 4.3), develop the energy balance analysis routine. Plot the time-
histories of energy components and find the peak values. Extend the energy balance
analysis for the blast load given in the HW 4.3. Plot the time-history of hysteretic energy,
altogether with the induced, potential, damping and elastic deformation energies. Find
then the peaks and their occurrence time-instants.

Numerical example 4.4| For the RC frame given below, analyzed in the numerical
example 2.1 , subjected to a random load (Figure 4.8) applied horizontally to the left joint,
the dynamic response analysis is performed using the integration scheme developed for
63
the elastic-linear MDOF model, in the Chapter 5 (Lecture Notes).
The matrix analysis-based routine is developed during the classes.

p(LL) = 24 kN/m
Young’s modulus E = 2.5x107 kN/m2

Specific weight 𝛾 =2500 kg/m3


4.00
Sections:

Columns|50x50 (cm)
7.50

Figure 4.8- RC plane frame and physical properties

The input load is provided in the file “inputdata.txt” andBeam|30x80(cm)


its time-history is shown in the
Figure 4.9.
The extreme input load values are 36.7 kN and -34.9 kN, respectively.

40

30

20

10
Load,kN

-10

-20

-30

-40
0 10 20 30 40 50 60
time,s

Figure 4.9- Time-history of the horizontal input random load at joint left joint

Recall the proposed discrete model consisting of 3 elements (fixed at both ends), their
stiffness and mass matrices being known (see Chapter 2). The full model (given in the
Figure 4.9) has therefore 12 DOFs, 6 DOFs effective (n-1-6) and 6 DOF-restricted (r- 7-12).
To be noticed that the body uniformly distributed load p will contribute beside the self-
weight of the structure to the overall mass matrix M.

2 5
Y 1 4
3 6

11 8
Z 12 10 9 7

0 X

Figure 4.10- Discrete (lumped) model and associated degrees of freedom


in the global reference system (XYZ)
64
The individual members and their local refference axes are recalled in the Figure 4.11.

0 e1: m1, k1 x
1 2
z
z y z y
0 0
e3: m3, k3 e2: m2, k2

x x

4 3

Figure 4.11- Discrete elements, joints and local reference axes (xyz)

The vibration modes are given in the Numerical example 2.1 of this textbook. The
damping model for this example, was chosen the Rayleigh model.
If the damping ratios of the first two modes are set to 5%, then 𝑎 = 0.7387 and 𝑎 =
0.0034. The damping ratios for the 3rd, 4th, 5th and the 6th mode are 8.64%, 23.16%, 31.98%
and 41.23%, respectively.
The Rayleigh main damping sub-matrix associated to the n-type degrees of freedom has
the form
2.8233 0 −0.1666 −2.6855 0 0
⎡ 0 5.6033 −0.4248 0 0.0705 0.0676 ⎤
⎢ ⎥
−0.1666 −0.4248 1.4398 0 −0.0676 −0.0282⎥
𝑪𝒏𝒏 = 10 ⎢ (4.7a)
⎢−2.6855 0 0 2.8233 0 −0.1666⎥
⎢ 0 0.0705 −0.0676 0 5.6033 0.4248 ⎥
⎣ 0 0.0676 −0.0282 −0.1666 0.4248 1.4398 ⎦
while the main damping sub-matrix associated to the r-type degrees of freedom has the
form is

0.0853 0 0.1666 0 0 0
⎡ 0 5.2799 0 0 0 0 ⎤
⎢ ⎥
0.1666 0 0.4411 0 0 0 ⎥
𝑪𝒓𝒓 = 10 ⎢ (4.7b)
⎢ 0 0 0 0.0853 0 0.1666⎥
⎢ 0 0 0 0 5.2799 0 ⎥
⎣ 0 0 0 0.1666 0 0.4411⎦
The secondary damping sub-matrices are as follows

0 0 0 −0.0815 0 −0.1640
⎡ 0 0 0 0 −5.2790 0 ⎤
⎢ ⎥
0 0 0 0.1640 0 0.2191 ⎥
𝑪𝒏𝒓 = 10 ⎢ (4.7c)
⎢ −0.0815 0 −0.1640 0 0 0 ⎥
⎢ 0 −5.2790 0 0 0 0 ⎥
⎣ 0.1640 0 0.2191 0 0 0 ⎦

65
0 0 0 −0.0815 0 0.1640
⎡ 0 0 0 0 −5.2790 0 ⎤
⎢ ⎥
0 0 0 −0.1640 0 0.2191⎥
𝑪𝒓𝒏 = 10 ⎢ (4.7d)
⎢−0.0815 0 0.1640 0 0 0 ⎥
⎢ 0 −5.2790 0 0 0 0 ⎥
⎣−0.1640 0 0.2191 0 0 0 ⎦
The sub-matrices are obtained by appropriate partition of the overall damping matrix 𝑪
(as shown in the Lecture Notes).
The corresponding stiffness sub-matrices are

0.0244 0 0.0488 0 0 0
⎡ 0 1.5625 0 0 0 0 ⎤
⎢ ⎥
0.0488 0 0.1302 0 0 0 ⎥
𝑲𝒓𝒓 = 10 ⎢ (4.8a)
⎢ 0 0 0 0.0244 0 0.0488⎥
⎢ 0 0 0 0 1.5625 0 ⎥
⎣ 0 0 0 0.0488 0 0.1302⎦

0 0 0 −0.0244 0 −0.0488
⎡ 0 0 0 0 −1.5625 0 ⎤
⎢ ⎥
0 0 0 0.0488 0 0.0651 ⎥
𝑲𝒏𝒓 = 10 ⎢ (4.8b)
⎢−0.0244 0 −0.0488 0 0 0 ⎥
⎢ 0 −1.5625 0 0 0 0 ⎥
⎣ 0.0488 0 0.0651 0 0 0 ⎦

0 0 0 −0.0244 0 0.0488
⎡ 0 0 0 0 −1.5625 0 ⎤
⎢ ⎥
0 0 0 −0.0488 0 0.0651⎥
𝑲𝒓𝒏 = 10 ⎢ (4.8c)
⎢ −0.0244 0 0.0488 0 0 0 ⎥
⎢ 0 −1.5625 0 0 0 0 ⎥
⎣ −0.0488 0.0651 0 0 0 ⎦

and further, the corresponding inertia matrices are

3.7862 0 2.1358 0 0 0
⎡ 0 0.8495 0 0 0 0 ⎤
⎢ ⎥
2.1358 0 1.5533 0 0 0 ⎥
𝑴𝒓𝒓 =⎢ (4.9a)
⎢ 0 0 0 3.7862 0 2.1358⎥
⎢ 0 0 0 0 0.8495 0 ⎥
⎣ 0 0 0 2.1358 0 1.5533⎦

0 0 0 1.3106 0 1.2621
⎡ 0 0 0 0 0.4247 0 ⎤
⎢ ⎥
0 0 0 −1.2621 0 −1.1650⎥
𝑴𝒏𝒓 =⎢ (4.9b)
⎢ 1.3106 0 1.2621 0 0 0 ⎥
⎢ 0 0.4247 0 0 0 0 ⎥
⎣−1.2621 0 −1.1650 0 0 0 ⎦

66
0 0 0 1.3106 0 −1.2621
⎡ 0 0 0 0 0.4247 0 ⎤
⎢ ⎥
0 0 0 1.2621 0 −1.1650⎥
𝑴𝒓𝒏 =⎢ (4.9c)
⎢ 1.3106 0 −1.2621 0 0 0 ⎥
⎢ 0 0.4247 0 0 0 0 ⎥
⎣ 1.2621 −1.1650 0 0 0 ⎦

𝑲𝒏𝒏 and 𝑴𝒏𝒏 are given in the numerical example 2.1.

By solving the reduced differential equations system given (2.42a) in the Lecture Notes
and then the secondary sub-system (2.42b), the displacements velocities and accelerations,
respectively, are determined by using the Linear Acceleration Integration scheme and
shown in the Figure 4.12.

Joint displacements (x10-3) Max. Min.


-3
x 10
Disp6,rad Disp5,m Disp4,m Disp3,rad Disp2,m Disp1,m

1
0
-1
0.7138 -0.6937
0 -6
10 20 30 40 50 60
x 10
5
0 0.0037 -0.0034
-5
0 -4
10 20 30 40 50 60
x 10
2
0
0.0978 -0.0970
-2
0 10 20 30 40 50 60
-3
x 10
1
0
-1 0.7129 -0.6909
0 10 20 30 40 50 60
-6
x 10
5
0
-5
0.0039 -0.0039
0 10 20 30 40 50 60
-4
x 10
2
0 0.0963 -0.0949
-2
0 10 20 30 40 50 60
time,s

Joint velocities Max. Min.

0.0115 -0.0115
Vel6,rad/s Vel5,m/s Vel4,m/s Vel3,rad/s Vel2,m/s Vel1,m/s

0.02
0
-0.02
0 10 20 30 40 50 60
-4
x 10
2
0 0.0001 -0.0001
-2
0 10 20 30 40 50 60
-3
x 10
2
0
0.0017 -0.0017
-2
0 10 20 30 40 50 60
0.02
0
-0.02 0.0115 -0.0115
0 10 20 30 40 50 60
-4
x 10
2
0 0.0002 -0.0002
-2
0 10 20 30 40 50 60
-3
x 10
2
0 0.0017 -0.0017
-2
0 10 20 30 40 50 60
time,s

67
Joint accelerations Max. Min.

Acc6,rad/s2Acc5,m/s2 Acc4,m/s2Acc3,rad/s2Acc2,m/s2 Acc1,m/s2


1
0.9675 -0.9044
0
-1
0 10 20 30 40 50 60
0.05
0 0.0221 -0.0189
-0.05
0 10 20 30 40 50 60
0.05
0 0.0430 -0.0381
-0.05
0 10 20 30 40 50 60
1
0
-1 0.9011 -0.7713
0 10 20 30 40 50 60
0.02
0
-0.02
0.0132 -0.0132
0 10 20 30 40 50 60
0.05
0 0.0345 -0.0343
-0.05
0 10 20 30 40 50 60
time,s

Figure 4.12- Time-histories of the effective joint displacements and extreme values (n=1,2,..,6 DOF)

The time-instant internal forces vector 𝒇( ) , which consists of member’s axial, shear and
bending moments is given by using the eq. (2.44) in the Lecture Notes i.e.
(𝒆)
𝒇( )
= 𝑻(𝒆) 𝒌𝑮 𝒙(𝒆) (4.10a)
(𝒆)
𝑻( ) is the transformation matrix of element’s (e) coordinates, 𝒌𝑮 is the stiffness matrix of
the element (e) in the global reference system (previously used to assemble the structure’s
stiffness matrix 𝑲 and 𝒙( ) are the joints displacements of element (e), properly selected
from the global displacements vector 𝒙).
Usually the effect of pre-existent gravity static loads is to be accounted for. The static
internal forces vector is
(𝒆)
𝒇( )
= 𝑻(𝒆) 𝒌𝑮 𝒙𝒔𝒕 (𝒆) (4.10b)

where 𝒙𝒔𝒕 (𝒆) is the static displacements vector of element (e) obtained by classical Direct
Displacements Method procedure of Structural Analysis (Iancovici, 2012).
The total load effect is the obtained by linear addition of static and dynamic effects on the
structure.
The time-histories and the peak values of the internal forces are given in the Figure 4.13
(origin joint-o and termination joint-t). The reaction forces are the time-histories of internal
forces in the end- joints for the elements 2 and 3.

68
Element 1-beam Max. Min.
o/t o/t

50

49.0408 1.6774
N,kN

0
-1.6774 -49.0408
-50
0 10 20 30 40 50 60

100

96.5225 83.4107
Q,kN

90
96.5893 83.4775
80
0 10 20 30 40 50 60

100
M,kNm

origin
0 -42.8849 -92.2030
termination
92.3049 43.2844
-100
0 10 20 30 40 50 60
time,s

Element 2-right column Max. Min.


o/t o/t

100

origin 96.0725 84.5809


N,kN

0
termination
-84.5809 -96.0725
-100
0 10 20 30 40 50 60

50

37.9924 13.0367
Q,kN

0
-13.0367 -37.9924
-50
0 10 20 30 40 50 60

0
M,kNm

-50 -43.2660 -89.9919


-5.8807 -61.9781
-100
0 10 20 30 40 50 60
time,s

Element 3-left column Max. Min.


o/t o/t
100

50

origin 95.2511 84.2261


N,kN

0
termination
-50
-84.2261 -95.2511
-100
0 10 20 30 40 50 60

50

-12.5632 -37.4768
Q,kN

0
37.4768 12.5632
-50
0 10 20 30 40 50 60

100
M,kNm

50 88.9042 45.3774
61.0030 4.8753
0
0 10 20 30 40 50 60
time,s

Figure 4.13- Time-histories of axial forces, shear forces and bending moments in members
and extreme values

69
 Homework 4.4|For the refined model developed in the HW 2.2, perform the time-
history analysis for a triangular impact load (Figure 1.16 in the Lecture Notes), acting
vertically at the middle of beam.

Numerical example 4.5| The five story RC frame structure from the numerical example
3.1, is subjected to a seismic support excitation given by the Vrancea 1977 NS component
acceleration time-series. By selecting the Rayleigh model to build the damping matrix, the
time-history analysis using the direct integration approach scheme developed in the
numerical example 4.4 for the support excitation case. The first two modes damping ratios
are set to 5%.
 Task| Modify the integration scheme developed for SDOF response analysis in the
Figure 4.4, for the case of MDOF models, subjected to support excitation.
The results, using the integration scheme developed above, are shown in the Figure 4.14a
and b.

Relative joint displacements, m Max. Min.


0.05
Disp5,m

0
0.0305 -0.0287
-0.05
0 5 10 15 20 25 30 35 40
0.05
Disp4,m

0
0.0272 -0.0256
-0.05
0 5 10 15 20 25 30 35 40
0.05
Disp3,m

-0.05 0.0219 -0.0206


0 5 10 15 20 25 30 35 40
0.02
Disp2,m

-0.02
0 5 10 15 20 25 30 35 40 0.0148 -0.0139
0.01
Disp1,m

-0.01
0 5 10 15 20 25 30 35 40 0.0065 -0.0061
time,s

Relative joint velocities, m/s Max. Min.


0.5
Vel5,m/s

0
0.3344 -0.3492
-0.5
0 5 10 15 20 25 30 35 40
0.5
Vel4,m/s

0
0.2961 -0.3079
-0.5
0 5 10 15 20 25 30 35 40
0.5
Vel3,m/s

-0.5 0.2346 -0.2425


0 5 10 15 20 25 30 35 40
0.2
Vel2,m/s

-0.2
0 5 10 15 20 25 30 35 40 0.1549 -0.1592
0.1
Vel1,m/s

-0.1
0 5 10 15 20 25 30 35 40 0.0657 -0.0676
time,s

70
Absolute joint accelerations, m/s2 Max. Min.

2
5

Acc1,m/s Acc2,m/s Acc3,m/s Acc4,m/s Acc5,m/s


0
4.1126 -3.2009
-5
0 5 10 15 20 25 30 35 40
2

5
3.6339 -3.7752
0

-5
0 5 10 15 20 25 30 35 40
2

0 3.0965 -3.3443
-5
0 5 10 15 20 25 30 35 40
2

0 2.6563 -2.8473
-2
0 5 10 15 20 25 30 35 40
2

0
2.1180 -2.3174
-1
0 5 10 15 20 25 30 35 40
time,s

Figure 4.14a- Time-histories of story relative displacements, relative velocities and absolute
accelerations and corresponding extreme values

Story seismic loads and base shear force, kN Max. Min.


500
321.1744 -314.4177
S5,kN

-500
0 5 10 15 20 25 30 35 40
500
S4,kN

0 278.8099 -268.3760
-500
0 5 10 15 20 25 30 35 40
500
S3,kN

0 246.9932 -228.6883
-500
0 5 10 15 20 25 30 35 40
500
S2,kN

0
210.2834 -196.1735
-500
0 5 10 15 20 25 30 35 40
200
S1,kN

0 171.1495 -156.4175
-200
0 5 10 15 20 25 30 35 40
2000
Fb,kN

-2000
1205.0518 -1126.3537
0 5 10 15 20 25 30 35 40
time,s

Figure 4.14b- Time-histories of story loads and base shear force


and corresponding extreme values

(i) By performing the analysis for each proportional damping model developed in the
numerical example 3.1, the peak story displacements and peak drift ratios,
straightforward obtained from the time-history analysis, are shown in the Table 5
and Figure 4.15.
Table 5- Peak story displacements, m
Story M prop. model K prop. model Rayleigh model
1 0.0065 0.0066 0.0065
2 0.0148 0.0149 0.0148
3 0.0219 0.0219 0.0219
4 0.0272 0.0272 0.0272
5 0.0305 0.0304 0.0305

71
5 5
Mprop
Kprop
4 4
Rayleigh

3 3
Story

Story
2 2

Mprop
1 Kprop 1
Rayleigh

0 0
0.00 0.01 0.02 0.03 0.04 0.00 0.05 0.10 0.15 0.20 0.25 0.30
Deflection, m Drift ratio, %

Figure 4.15- Peak story displacements (left figure) and peak story drift ratios (right figure)
for three different proportional damping models

It is to be noted that the response is not sensitive to damping model chosen.


The additional modal information is given in Table 6.

Table 6- Modal information

Vibration mode index Mode mass participation factors Mode equivalence factors
𝝓𝑻𝒊 𝑴𝟏 𝑚 , (∑ 𝑚 𝜙 ,)
𝛤= 𝑻 𝜆 = =
𝝓𝒊 𝑴𝝓𝒊 𝑚 ∑ 𝑚 (∑ 𝑚 𝜙 ,)
1 0.252 0.834
2 0.276 0.109
3 0.241 0.037
4 0.170 0.016
5 0.061 0.004

𝑚 , in Table 6 is the equivalent modal mass of the ith mode (eq.2.68a in the Lecture Notes).

Numerical example 4.6|Given the five story RC building structure in the Figure 4.16,
treated extensively in the numerical example 3.1, the dynamic response analysis to lateral
arbitrary (random) nodal loads is performed by using the Mode Superposition Method
approach.
The routine is developed during classes using the LAM integration scheme in the Figure
4.3.
The aerodynamic time-histories of wind loads (acting at left façade only) are provided in
the form of data file and represented along with point-in time maxima, in the Figure 4.17.
The sampling time- step is 0.01s and the loading duration, 11s.

72
p(LL)=55.5kN/m

Columns Beams
p(LL)=49.5kN/m
3.00

p(LL)=49.5kN/m 60 cm
3.00 90 cm

60 cm
p(LL)=49.5kN/m 3.00
40 cm

p(LL)= 49.5kN/m
3.00 Young’s modulus 𝐸 = 2x10 𝑘𝑁/𝑚

3.00 Specific weight 𝛾 = 2500 𝑘𝑔/𝑚

7.00 4.00

Figure 4.16- Five story RC frame structure under lateral aerodynamic wind loads

500 313 kN at 6.51s


Load 5,kN

-500
0 1 2 3 4 5 6 7 8 9 10

500 374 kN at 2.3s


Load 4,kN

-500
0 1 2 3 4 5 6 7 8 9 10

500 355 kN at 1.07s


Load 3,kN

-500
0 1 2 3 4 5 6 7 8 9 10

500 332 kN at 2.11s


Load 2,kN

-500
0 1 2 3 4 5 6 7 8 9 10

500
372 kN at 1.84s
Load 1,kN

-500
0 1 2 3 4 5 6 7 8 9 10
time,s

Figure 4.17- Time-histories of lateral nodal loads and peak values

The modal analysis was performed in the numerical example 3.1 as well as a sensitivity
analysis on damping model influence. The Mode Superposition Method analysis framework
is presented in the Chapter 4 of the accompanying Lecture Notes and further used.
 Task| Modify the integration scheme developed for MDOF response analysis in the
numerical example 4.5, for the Mode Superposition Method approach.
For this example, the mass proportional damping model is used. The fundamental mode
damping ratio is 5%. The higher modes damping ratios yield 1.61 %, 0.91%, 0.64% and
0.52% respectively.

73
Running the time-history analysis by the Linear Acceleration Method integration scheme,
both on (i) modal responses and (ii) total response, the story displacements are obtained
and represented in the Figure 4.18.

Mode 1 Mode 2 Mode 3 Mode 4 Mode 5 Total


-3 -3 -3 -4
x 10 x 10 x 10 x 10
0.01 5 2 1 5 0.02
at 9.33s
Disp51,m

Disp52,m

Disp53,m

Disp54,m

Disp55,m

Disp5,m
0.01
0 0 0 0 0
0

-0.01 -5 -2 -1 -5 -0.01
0 5 10 0 5 10 0 5 10 0 5 10 0 5 10 0 5 10
-4 -3 -3
at 9.33s
-3
x 10 x 10 x 10 x 10
0.01 5 2 2 1 0.01
Disp41,m

Disp42,m

Disp43,m

Disp44,m

Disp45,m

Disp4,m
0 0 0 0 0 0

-0.01 -5 -2 -2 -1 -0.01
0 5 10 0 5 10 0 5 10 0 5 10 0 5 10 0 5 10
-3 -3 -3 -3
x 10 x 10 x 10 x 10
0.01 2 2 1 2 0.01
Disp31,m

Disp32,m

Disp33,m

Disp34,m

Disp35,m

Disp3,m
0 0 0 0 0 0

-0.01 -2 -2 -1 -2 -0.01
0 5 10 0 5 10 0 5 10 0 5 10 0 5 10 0 5 10
-3 -3 -3 -3 -3
at 9.6s
x 10 x 10 x 10 x 10 x 10
5 5 1 1 2 0.01
Disp21,m

Disp22,m

Disp23,m

Disp24,m

Disp25,m

Disp2,m
0 0 0 0 0 0

-5 -5 -1 -1 -2 -0.01
0 5 10 0 5 10 0 5 10 0 5 10 0 5 10 0 5 10

x 10
-3
x 10
-3
x 10
-3
x 10
-3
x 10
-3
x 10
-3 at 7.43s
2 2 2 2 1 5
Disp11,m

Disp12,m

Disp13,m

Disp14,m

Disp15,m

Disp1,m
0 0 0 0 0 0

-2 -2 -2 -2 -1 -5
0 5 10 0 5 10 0 5 10 0 5 10 0 5 10 0 5 10
time,s time,s time,s time,s time,s time,s
at 3.36s
at 9.33s at 1.42s at 3.66s at 7.34s at 4.42s

Figure 4.18- Time-histories of modal, total story displacements and maxima point-in-time

The peaks of modal and total story displacements are given in Table 7. The total peak
response can be also obtained by various peak modal combination rules (e.g. ABSSUM,
SRSS, CQC etc.) since modal peaks are not occurring simultaneously. Among these, the
Square Root of Sum Squared (SRSS) rule is widely used.

Table 7 – Peak story modal and total displacement responses, m

Story Mode 1 Mode 2 Mode 3 Mode 4 Mode 5 Total Ratio


Max|Min Direct SRSS Direct/SRSS
5 0.00944 0.00226 0.00123 0.00061 0.00027 0.01051 0.00981 1.07
-0.00914 -0.00238 -0.00120 -0.00058 -0.00027
4 0.00835 0.00037 0.00114 0.00125 0.00082 0.00900 0.00858 1.05
-0.00809 -0.00039 -0.00117 -0.00130 -0.00082
3 0.00661 0.00180 0.00136 0.00062 0.00116 -0.00722 0.00711 1.02
-0.00640 -0.00171 -0.00139 -0.00059 -0.00116
2 0.00435 0.00268 0.00091 0.00074 0.00116 -0.00544 0.00537 1.01
-0.00422 -0.00254 -0.00089 -0.00071 -0.00116
1 0.00185 0.00163 0.00161 0.00125 0.00084 -0.00391 0.00331 1.04
-0.00180 -0.00154 -0.00158 -0.00131 -0.00084

74
One can notice a good approximation of peak story displacements, obtained
straightforward by direct time-instant addition by SRSS approach. Same analysis
framework approach is used for any other response parameter (i.e. velocities,
accelerations, internal forces, over-turning moments etc.).

Numerical example 4.7| The modal analysis of a single story tridimensional RC shear
model, having the plan dimensions 𝐿 = 25 𝑚 of five equal spans of 5 m and 𝐿 = 18 𝑚 of
three equal spans of 6 m, is performed.
The story height is ℎ = 3.7𝑚. The columns are of equal section of 50 by 50 cm, as well as
the beams section is 90 cm depth and 35 cm width. The floor thickness is 15 cm.
The Young’s modulus is E = 2.5x107 kN/m2 for all elements (floor plate, columns and
beams). The specific weight is 𝛾 =2500 kg/m3 .
The floor plate is homogenous and continuous, therefore the center of mass (CM) is same
with its geometric center (GC) as presented in the Chapter 5 of the accompanying Lecture
Notes.
The building is modeled as shear-type space frame with 3DOF with an additional floor
load of 5 kN/m2 from quasi-permanent-QPL and live loads-LL (Figure 4.19). The column’s
torque stiffness is negligible compared to the overall structure’s torque stiffness.

z z
CM 3
y y
2 1

x x

Figure 4.19- 3D structural view (left figure) and associated DOFs (right figure)

Two models are analyzed: (i) plan symmetric- with coincident mass and stiffness centers
and (ii) plan eccentric-with story stiffness eccentricity (Figure 4.20).
The symmetric structure is taken therefore as reference structure for analysis and
discussions.

75
Regular/symmetric Irregular/eccentric

y x(e) y x(e)
CS

y(e) ey y(e)
Ly z Ly z
x x

CM=CS CM ex

Lx Lx

Figure 4.21- Symmetric (left figure) and eccentric (right figure) models

The stiffness matrix, expressed in terms of DOFs, has the form of

𝑘 𝑘 𝑘
𝑲= 𝑘 𝑘 𝑘 (4.11)
𝑘 𝑘 𝑘

Its development is shown in the Figure 4.22.

1st column of K 2nd column of K 3rd column of K

ux=1 uθ=1

uy=1

k21 k22 k32

k23 k13
k31 k11 k33
k32

Figure 4.22- Stiffness matrix 𝑲 development

Using simple static equilibrium equations based on Figure 4.8, the stiffness matrix is
( ) ( ) ( )
⎡ ∑( ) 𝑘 0 − ∑( ) 𝑘 𝑦 ⎤
( ) ( ) ( )
𝑲=⎢ 0 ∑( ) 𝑘 ∑( ) 𝑘 𝑥 ⎥ (4.12a)
⎢ ( ) (

⎣− ∑( ) 𝑘 𝑦
) ∑( ) 𝑘 ( ) 𝑥 ( ) ∑( ) 𝑘 ( )
𝑦( )
+ ∑( ) 𝑘
( )
𝑥( )

where
76
( )
𝑘 = ∑( ) 𝑘 (4.12b)

and
( )
𝑘 = ∑( ) 𝑘 (4.12c)
( ) ( )
are the total stiffnesses of the frame in x- and y-directions, respectively. 𝑘 and 𝑘 is the
lateral stiffness of vertical element (e), in the x- and y-directions, respectively.
The term
( ) ( )
𝑘 = ∑( ) 𝑘 𝑦( )
+ ∑( ) 𝑘 𝑥( )
(4.12d)

is the torque stiffness of the frame; 𝑥 ( ) and 𝑦 ( )


are the distances from the Geometric
Center (GC) to the vertical element (e).
The inertia (mass) matrix has the form of
𝑚 0 0 𝑚 0 0 𝑚 0 0
𝑴= 0 𝑚 0 = 0 𝑚 0 = 0 𝑚 0 (4.13)
0 0 𝑚 0 0 𝐼 0 0 𝑚𝑟

where 𝑟 = is the radius of gyration of rectangular floor plate.

(i) Regular/symmetric structure

The total mass of the structure, including the plate, beams and half of the columns mass is
𝑚 = 264 𝑘𝑁𝑠 /𝑚. Therefore the inertia matrix is

𝑚 0 0 𝑚 0 0 264 0 0
𝑴= 0 𝑚 0 = 0 𝑚 0 = 0 264 0 (4.14a)
0 0 𝐼 0 0 𝑚𝑟 0 0 20875

The lateral sway stiffness for one column in both directions is 𝑘 = = 30847 𝑘𝑁/𝑚.
Therefore the stiffness matrix is
740331.27 0 0
𝑲= 0 740331.27 0 (4.14b)
0 0 87297396.00
Note that for no stiffness eccentricity, i.e. 𝑒 = 𝑒 = 0, the stiffness matrix is diagonal.
By running the eigenmodes analysis, the following results are obtained for the spectral
matrix

52.60 0 0 0.12 0 0 8.43 0 0


𝜴= 0 52.60 0 ;𝑻 = 0 0.12 0 ;𝒇= 0 8.43 0 (4.15)
0 0 64.67 0 0 0.10 0 0 10.92

The structure is torsional flexible.


The natural periods, for instance, can be easily determined, since the vibration modes are
fully decoupled as
77
𝑇 =𝑇 = 2𝜋 = 2𝜋 .
= 0.19 𝑠 ( 1st DOF) (4.16a)

𝑇 =𝑇 = 2𝜋 = 2𝜋 .
= 0.19 𝑠 ( 2nd DOF) (4.16b)

.
𝑇 =𝑇 = 2𝜋 = 2𝜋 .
= 0.10 𝑠 ( 3rd DOF) (4.16c)

The modal mass normalized eigenvectors are

0.0615 0 0
𝜱= 0 0.0615 0 (4.17)
0 0 0.0069

The vibration mode shapes are given in the Figure 4.23.

Mode 1 Mode 2 Mode 3


0.0615

0.0069

0.0615

Figure 4.23- Vibration mode shapes of regular structure (plan view)

The first two vibration modes are independent sways in x- and y- directions. The third
mode shape is pure torque. This can be also viewed in the mass mode participation factors
matrix given by
16.24 0 0
𝑷=𝜱 𝑴= 0𝑻
16.24 0 (4.18)
0 0 144.48

(ii) Irregular/eccentric structure

Let us suppose that for functional purposes, 2 columns are removed from the initial,
symmetric reference structure (Figure 4.21, right). The Center of Mass (CM) is therefore
considered in the Geometric Center (GC), since the influence of removed columns is
negligible, i.e. the mass story eccentricities 𝑒 = 𝑒 = 0.
The mass matrix therefore is nearly same as for the regular structure,

78
264 0 0
𝑴= 0 264 0 (4.19)
0 0 20875
The story stiffness eccentricities in the x-and y- directions yield
( ) ( )
∑( )
𝑒 = ( ) = 0.43 𝑚 (4.20a)
∑( )

( ) ( )
∑( )
𝑒 = ( ) = 0.26 𝑚 (4.20b)
∑( )

Therefore the total story eccentricities the x-and y- directions are


𝑒 =𝑒 +𝑒 = 0.43 𝑚 (4.21a)
𝑒 =𝑒 +𝑒 = 0.26 𝑚 (4.21b)
The stiffness matrix yields therefore
709484.14 0 −185082.82
𝑲= 0 709484.14 308471.36 (4.22)
−185082.81 308471.36 84814201.53
where the off-diagonal terms introduce the elastic coupling between DOFs.
By running the eigenmodes analysis, the following results are obtained for the spectral
matrix

51.68 0 0 0.12 0 0 8.23 0 0


𝜴= 0 51.84 0 or 𝑻 = 0 0.12 0 or 𝒇 = 0 8.25 0 (4.23a)
0 0 63.87 0 0 0.10 0 0 10.16

The normalized eigenvectors (4.23b) and the vibration mode shapes are given in the
Figure 4.24

0.0315 −0.0528 0.0035


𝜱 = −0.0525 −0.0317 −0.0058 (4.23b)
0.0008 0 −0.0069

Mode 1 Mode 2 Mode 3


0.0315 0.0528

0.0525 0.0035
0.0058

0.0317
0.0008

0.0069

Figure 4.24- Vibration mode shapes of irregular structure (plan view)

79
The coupling effect is observed in the mass mode participation factors matrix given by

8.31 13.85 15.82


𝑷 = 13.93 8.36 0 (4.24)
0.91 1.52 143.61
Thus, the first mode is dominated by sways and slight torque, the second one is close to
pure sway in both directions and the third mode is dominated by torque and slight sways.
In terms of eigenvalues the modal elastic coupling effect is less than 2%.

 Homework 4.5| Suppose that a slab hole is performed (Figure 4.25). Perform the modal
analysis for the case of inertia eccentricity.

Ly z
x

Hole 5m/6m
Lx

Figure 4.25- The inertial eccentric structure

Hint: Compute the coordinates of the Center of Mass (CM)


∑( ) ( ) ( )
𝑒 = ∑( ) ( )
= 0.38 𝑚 (4.25a)

∑( ) ( ) ( )
𝑒 = ∑( ) ( )
= 0.43 𝑚 (4.25b)

expressed with reference to the Geometric Center (GC). 𝐴( ) is the lth floor section area,
𝑥 ( ) and 𝑦 ( ) are the distances from the Geometric Center (GC) to the lth floor section mass
center. The Center of Stiffness (CS) is same, coincident with the GC.

Numerical example 4.8| For the regular single story tridimensional RC model, treated in
the numerical example 4.7, the time-history analysis is performed for the bi-directional
input ground motion induced by the El Centro EQ of May 5th 1940. The results of
accelerograms signal processing are given in Chapter 1 of the accompanying Lecture
Notes.
The damping matrix is a Rayleigh-based model, with 5% damping ratios set for the
fundamental modes, in both directions.

80
z

x
0

-1

-2

-3

-4
0 5 10 15 20 25 30 35 40

40
35
30
25
2

20
5
1.

15
5
0.

10
5
-0.

-1

5
5
-1.

-2

5
0
-2.

Figure 4.26- RC tridimensional model subjected to bi-directional input ground motion

The Rayleigh model-based damping matrix yields

1397.92 0 0
𝑪= 0 1397.92 0 (4.26)
0 0 137695.10

where 𝑎 = 2.6479 and 𝑎 = 0.0009. The damping ratio for the third mode (torque) yields
5.10%.
Running the dynamic analysis by Linear Acceleration Method integration scheme, the
time-history response is obtained. The relative floor displacements and their extreme
values are represented in the Figure 4.27.

Relative floor displacement, m Max. Min.


-3
x 10
2

0.0018 -0.0025
x-disp.,m

-1

-2

-3
0 5 10 15 20 25 30 35 40
time,s
-3
x 10
2

1
0.0011 -0.0009
y-disp.,m

-1

-2
0 5 10 15 20 25 30 35 40
time,s

Figure 4.27- Time-histories of relative floor displacements, in x- and y-directions

81
The bi-directional input ground motion accelerations and the floor relative displacements
are shown altogether in the Figure 4.28.

z
y

0
x
-1

-2

-3

-4
0 5 10 15 20 25 30 35 40
40
35
30
25
20
2

5
1.

15
0.5

10
5
-0.

-1

5
5
-1.

-2

5
0
-2.

Figure 4.28- Time-histories of acceleration input ground motion and relative floor response
displacements, in x- and y-directions

Since the structure is symmetric and there is no rotational ground motion component,
there will be consequently no torsional response. For wind-excited structures however,
torque components will be always present and consequently, torsional-induced response.
Any other response parameter (i.e. velocities, accelerations, internal forces, over-turning
moments etc.) can be directly addressed by analysis, as well as any input directionality can
be explicitly considered.
 Homework 4.6|Perform the time-history analysis for the elastic eccentric structure
given in the Figure 4.21 (right) and compare the results with those obtained in the
numerical example 4.8.
 Homework 4.7|Determine the time-history response and their peak values (relative
displacements, relative velocities, absolute accelerations) of the tridimensional RC shear
model (3DOFs/story) of 𝐿 = 36 𝑚 of five equal spans of 7.2 m and 𝐿 = 21 𝑚 of three
equal spans of 7 m (Figure 4.29). The story height is ℎ = 3 𝑚. The columns are of equal
section of 80 by 80 cm, as well as the beams section is 90 cm depth and 40 cm width. The
floor thickness is 15 cm. The Young’s modulus is E = 2x107 kN/m2 for all the elements
(floor plate, columns and beams). The specific weight is 𝛾 =2500 kg/m3 .
The input consist of a bi-directional ground motion given by VN77NS and VN77EW
accelerogram components, applied under 30o directivity with respect to x- and y- axes
(Figure 4.29).
82
z

y x(e)

y(e)
Ly z
x

y CM=CS

Lx

Figure 4.29- Six story RC frame structure: 3D view (left figure) and story plan view (right figure)

Compute then the mode correlation coefficients based on the Menun&Der Kiureghian’s
equation (eq. 5.61, in the accompanying Lecture Notes).

 Homework 4.8|For the structure given the Figure 4.29, develop the energy balance
analysis framework, using the theory developments given in the Chapter 5 of the
accompanying Lecture Notes.

 Homework 4.9|A 3 meters height space foundation frame structure having mounted 9
equipments (industrial machineries) on top is given (Figure 4.30). The rotating
machineries are running under the 2nd -type signal (Figure 1.18), generated in the
horizontal directions. The machines are half ton weight each. The machine starting order is
1-red, 2-blue and 3-yellow and shown in the Figure 4.31. The structural data are as follows

 Material: RC
Specific weight: 𝛾 = 2500 𝑘𝑔/𝑚
Young’s modulus: 𝐸 = 2𝑥10 𝑘𝑁/𝑚
 Loads:
- (i) Dead Load (implicitly assigned), (ii) Live Load - 3𝑘𝑁/𝑚 and (iii) rotating
machineries.
 Sections
Slab thickness: 25 cm
Beams section: 85(h) x 35(w), cm
Columns section: 60x60, cm
 Damping ratios for all vibration modes 𝜁 = 5%

83
y

2 2 2

3 1 3
4x3.50m

3 3 3

4x5.00m

Figure 4.30- 3D view (left figure) and machinery arrangement view (right figure)

Requirements:

(i) Compute the dynamic amplification factors for the dominant peaks of load using a
simplified model having 3DOF/floor, for the 1st machinery running only;
(ii) Compute the steady-state maximum floor deflection and acceleration
corresponding to the largest peak component;
(iii) Using a detailed FEM model perform the linear dynamic analysis for the first
rotating machinery only; represent graphically the time-histories of displacement,
velocity and acceleration on the machine’s acting point;
(iv) Apply the 2nd set of load with a 10 sec delay and the 3rd set of loads with 20 sec
delay; evaluate the effect of each load set on the time-history and peak dynamic
response; check the compliance with a functional limit acceleration of 50 mili-g. If
the floor acceleration limit is exceeded, propose a solution to redesign the structural
system in order to fulfill this requirement.

***

84
5. Dynamics of unconventional structures.
Vibration control

The energy dissipation devices are incorporated in conventional structures for the seismic-
, wind-, industrial- or ambient (micro-tremor) induced response reduction. This chapter is
devoted to the modeling and analysis of structures equipped with energy dissipation
devices (base isolation technique and Tuned Mass Damper- TMD technique). The overall
damping modeling is discussed and the time-history analysis framework is used to
determine the structural response to harmonic and random loads. This section allows
further extensions to non-proportional damping model, active controlled structures and
nonlinear response analysis of unconventional structures.

Numerical example 5.1|Given the one-story building structure (Figure 5.1), design the
columns having same rectangular sections so that the base shear force to be half of
maximum harmonic applied load of circular frequency 𝜃 = 40 𝑟𝑎𝑑/𝑠. The total weight is
𝑊 = 600𝑘𝑁, the Young’s modulus 𝐸 = 2x10 𝑘𝑁/𝑚 and damping ratio 𝜁 = 5%.

Actual structure SDOF model


W 1
F(t)

5.00

Figure 5.1- The actual structure (left figure) and the SDOF model (right)

The one-story frame can be simply modeled as a lateral sway SDOF. Using the
developments in Chapter 6 of the accompanying Lecture Notes, the Transmissibility Ratio
(TR) is given by
( )
𝑇𝑅 = ( ) ( )
(5.1)

By imposing the condition 𝑇𝑅 = 0.5 and solving the eq. (5.1) for the circular frequency
ratio, this yields 𝛽 = 𝜃⁄𝜔 = 1.73. Therefore 𝜔 = 23.12 𝑟𝑎𝑑/𝑠 and the lateral stiffness of
frame yields 𝑘 = 𝑚𝜔 = 32697 𝑘𝑁/𝑚.
On the other hand, the lateral stiffness can be expressed as a contribution of individual
columns i.e.

85
𝑘=3 (5.2a)

The equality yields

𝐼= = 56766 ∙ 10 𝑚 (5.2b)

The section’s moment of inertia of one column is

𝐼= (5.2c)

where b is the dimension of rectangular cross section, so that 𝑏 = √12𝐼 = 0.5108 𝑚 .


Therefore 𝑏 = 50 𝑐𝑚 can be chosen for the column dimension in both directions. For this
value the natural vibration period yields 𝑇 = 0.28 𝑠.
Notice that any increase of section over the limit provided will increase the
transmissibility ratio and vice-versa. On the other hand any force-based vibration
transmissibility analysis must be extended over the kinematic parameters as response
displacement, velocity and acceleration. In this regard any increase of section will push the
structure in the quasi-resonant region for the three parameters in the stationary response
regime (Figure 3.3 in the Lecture Notes). So that stiffening the structure for the lateral
sway is not beneficial for the load given. For random loads however, this approach
requires complete information on the frequency content.
 Task| Given the one-story frame in the Figure 5.1, modeled as SDOF system, subjected
to a quasi-harmonic ground acceleration having the predominant period 𝑇 = 0.5𝑠, pre-
design the columns section so that the seismic force to be 60% of inertia force induced by
the ground motion. Add your comments.

Numerical example 5.2|Given the base isolated RC shear frame structure (Figure 5.3),
perform (i) the modal analysis and (ii) the seismic response analysis, in terms of maximum
story deflections, story seismic forces and base shear force. The results are then compared
with the response of conventional structure (Figure 5.3, left).
Input ground motion data: The seismic load is represented by the 5% damping ratio,
normalized absolute acceleration elastic spectrum (Figure 5.2).

𝑆𝐴𝑒𝑙 /𝑃𝐺𝐴

2.5

1/T

0.4
𝑇, 𝑠

Figure 5.2- Normalized absolute acceleration response spectrum (5% damping ratio)
86
The Peak Ground Acceleration (PGA) is set to 0.3g and the following relationship is used
to obtain spectral accelerations corresponding to other damping level given 𝜁(%)

𝑆𝐴 % = 𝑆𝐴 % (5.3)

Conventional structure data: Weight 𝑊 = 600𝑘𝑁; Flexural stiffness 𝐸𝐼 = 82000𝑘𝑁𝑚 ;


Damping ratio 𝜁 = 2%; Story height ℎ = 3.50 𝑚

Isolation system data: Isolation system weight 𝑊 = 𝑊 = 400𝑘𝑁; Isolation system lateral
stiffness (shear stiffness) 𝑘 = 𝑘 /10; Isolation system damping ratio 𝜁 = 10%. The height
of isolators is negligible and the shear force-lateral deflection relationship is assumed to be
linear-elastic.

Ws Ws

3.50
ISOLATION SYSTEM Wi

CONVENTIONAL STRUCTURE ISOLATED STRUCTURE

Figure 5.3- Conventional structure (left figure) and corresponding isolated structure (right figure)

The coordinates associated to each model are given in the Figure 5.4.

1 2

ISOLATION SYSTEM
1

CONVENTIONAL STRUCTURE ISOLATED STRUCTURE

Figure 5.4- Degrees of freedom for the conventional structure (left figure) and
isolated structure (right figure)

(i) The conventional structure

The lateral stiffness is provided by columns i.e.

𝑘 =3 = 68851.31 𝑘𝑁/𝑚 (5.4)

Therefore the natural vibration period yields

87
𝑇 = 2𝜋 = 0.19 𝑠 (5.5)

(ii) The isolated structure


Simplified approach using the equivalent SDOF model
Since it is intended to provide a large decoupling degree of the super-structure from the
ground motion, we assume the isolated structure as an equivalent SDOF model, driven by
the stiffness of the isolation system and the total mass (𝑚 + 𝑚 ).
Thus, the fundamental natural period of isolated structure (in fact a 2 DOF model), yields

𝑇 = 2𝜋 = 0.75 𝑠 = 3.94𝑇 (5.6)

that is almost four times the natural period of conventional structure. This is the effect of
large flexibility brought into the system by the isolators. The target fundamental natural
period is used in practice for the initial design of isolators from based on the required
lateral stiffness.
Eigenmodes analysis of isolated structure
The mass matrix is given by
𝑚 0 40 0
𝑴= = (5.7a)
0 𝑚 0 60
and the stiffness matrix is

𝑘 +𝑘 −𝑘 75736.44 −68851.31
𝑲= = (5.7b)
−𝑘 𝑘 −68851.31 68851.31
Running the eigenmodes analysis, the following spectral matrix and modal matrix
(normalized with respect to the first ordinate) are obtained as
𝑇 0 𝟎. 𝟕𝟔 0 𝜔 0 8.15 0
𝑻= = ,𝑠 ; 𝜴 = = , 𝑟𝑎𝑑/𝑠 (5.8a)
0 𝑇 0 0.12 0 𝜔 0 54.54
1 1
𝜱= (5.8b)
1.06 −0.63
The mode shapes are given in the Figure 5.5.

Mode 1: 0.76 s- isolation mode Mode 2: 0.12 s-structural mode


1.06 0.63

1 1

Figure 5.5 -Mode shapes of the isolated structure


88
Since the fundamental eigenvector is almost constant, the deformation in the structure will
be negligibly small. Hence, the first mode is called isolation mode. The second mode is the
structural mode but it’s participation is negligibly small. The fundamental vibration period
is well estimated by the equivalent SDOF approach.

The modal parameters are summarized in the Table 1.

Table 1- Mass participation factors and mode equivalence factors

Vibration mode index Mode mass participation factors Mode equivalence factors
𝝓𝑻𝒊 𝑴𝟏 𝑚 , (∑ 𝑚 𝜙 ,)
𝛤= 𝜆 = =
𝝓𝑻𝒊 𝑴𝝓𝒊 𝑚 ∑ 𝑚 (∑ 𝑚 𝜙 ,)
1 0.9636 0.9992
2 0.0364 0.0008

Damping modeling

The damping coefficient of the conventional structure is


𝑐 = 2𝜁 𝑚 𝜔 = 79.36 𝑘𝑁/𝑚 (5.9)
while the damping coefficient of the isolation system is
𝑐 = 2𝜁 (𝑚 + 𝑚 )𝜔 = 165.35 𝑘𝑁/𝑚 (5.10)
The damping matrix of the global model is therefore
𝑐 +𝑐 −𝑐 244.71 −79.36
𝑪= −𝑐 𝑐 = (5.11)
−79.32 79.32
The generalized damping matrix yields
165.68 157.45
𝑪∗ = 𝜱𝑻 𝑪𝜱 = (5.12)
157.45 375.64
which is definitely not a diagonal matrix, the cross-diagonal terms are no more negligible
as in the case of a conventional structure.
By neglecting however these terms, i.e. assuming that the damping coupling is negligible
in the global model, we obtain
1.54 0 2𝜁 𝜔 0
(𝜱𝑻 𝑪𝜱)(𝜱𝑻 𝑴𝜱) 𝟏
= = (5.13)
0 5.89 0 2𝜁 𝜔
The modal damping ratios are therefore
.
𝜁 = ∙ .
= 0.094 ≈ 10% (5.14a)
.
𝜁 = ∙ .
= 0.054 ≈ 5.4% (5.14b)

While the damping ratio for the fundamental mode is driven by the isolation system, the
damping ratio of the second mode (structural) is higher than for the structure itself). In the
initial design stage however, the isolation system’s damping ratio can be attributed to the
89
fundamental mode, as well as super-structure damping ratio can be attributed to the
second mode.
However, the non-proportional damping model should be carefully studied and
incorporated in the analysis framework.

Response analysis by Mode Superposition Method (Response Spectrum Approach)

(i) Modal response


The mode mass participation factors are given in the Table 1. Therefore 96% of the ground
acceleration is carried-out by the isolation mode and only 4% by the structural mode.
Mode 1
Story relative deflections
( . , . %) 0.0468
𝒙𝟏 = 𝝓𝟏 𝑆𝐷(𝑇 = 0.76, 𝜁 = 9.4%)𝛤 = 𝝓𝟏 𝛤 = 𝑚 (5.15)
0.0496
where
𝑆𝐴(𝑇 = 0.19, 𝜁 = 9.4%) = 3.24 𝑚/𝑠

Story seismic forces


𝑆 , 124.42
𝑺𝟏 = = 𝑴𝝓𝟏 𝑆𝐴(𝑇 = 0.76, 𝜁 = 9.4%)𝛤 = 𝑘𝑁 (5.16a)
𝑆 , 197.82
Therefore the base shear force is 𝐹 , =𝑆 , +𝑆 , = 322.24 𝑘𝑁 (5.16b)

Mode 2
Story relative deflections
( . , . %) 0.000090
𝒙𝟐 = 𝝓𝟐 𝑆𝐷(𝑇 = 0.12, 𝜁 = 5.4%)𝛤 = 𝝓𝟐 𝛤 = 𝑚 (5.17)
0.000056
where
𝑆𝐴(𝑇 = 0.12, 𝜁 = 5.4%) = 7.35 𝑚/𝑠

Story seismic forces


𝑆 , 11.76
𝑺𝟐 = = 𝑴𝝓𝟐 𝑆𝐴(𝑇 = 0.12, 𝜁 = 5.4%)𝛤 = 𝑘𝑁 (5.18a)
𝑆 , −11.11
Therefore the base shear force is 𝐹 , =𝑆 , +𝑆 , = 0.65 𝑘𝑁 (5.18b)

(ii) Total response


The SRSS approach is used to combine the peak modal responses in order to obtain the
total peak response.

90
Story relative deflections

0.0468
𝒙 = √0.0468 + 0.000090 = 𝑚 (5.19)
√0.0496 + 0.000056 0.0496
where the contribution of second mode is obviously negligible.

Story seismic forces

√124.42 + 11.76 124.97


𝑺= = 𝑘𝑁 (5.20a)
197.82 + (−11.11) 198.13

and the total base shear force is 𝐹 = 323.1 𝑘𝑁.


The base shear force of the conventional structure is

𝐹 = 𝑚 𝑆𝐴(0.19,2%) = 60 ∙ 2.5 ∙ 0.3 ∙ 10 ∙ = 537.85 𝑘𝑁 (5.20b)

The base shear force of the isolated structure is therefore approximately 40% smaller than
the conventional structure’s base shear force.
The maximum deflection of the conventional structure yields

𝑥 = 𝑆𝐴(0.19,2%) = 0.8197 𝑐𝑚 (5.21)

When compared to the top deflection of 4.96 cm of the isolated structure, this is six times
smaller.
The peak relative deflection of the isolator must be limited to its deformation capacity
provided by producer. The absolute deflection includes however the ground
displacement; this is an important requirement for the displacement- controlled design
due to the presence of adjacent buildings.

Numerical example 5.3| Suppose that the isolated structure analyzed in the numerical
example 5.2, is subjected to a micro-tremor noise-like input acceleration time-series (Figure
5.6a). The input motion lasts for 30 minutes (Figure 5.6b) and the data file is given as
“microt.txt”. The dynamic analysis is performed by using the LAM integration scheme.

Ws

3.50
ISOLATION SYSTEM
Wi

Figure 5.6a- Base isolated structure subjected to micro-tremor ground motion

The micro-tremor input acceleration time-and frequency analysis results are given in the
Figure 5.6a and b. The peak ground acceleration is 0.1996 m/s2.
91
0.25
0.12
0.2

Fourier Amplitude, m/s


Ground acceleration, m/s/s

0.1
0.15
0.08
0.1
0.06
0.05
0.04
0
0.02
-0.05

-0.1
0
0 1 2 3 4 5 6 7 8 9 10 10 20 30 40 50
time,s Frequency,Hz

Figure 5.6b- Micro-tremor acceleration record (left figure; represented on 10 seconds only) and the
corresponding Fourier Amplitude Spectrum (right figure)

The frequency content is dominated by a multiple dominant components. The time-history


responses and their peak values are presented in the Figures 5.7, 5.8 and 5.9.

Relative displacements, m Max. Min.


-3
x 10
2

1
Disp. top,m

-1 0.015 -0.033
-2

-3

-4
0 200 400 600 800 1000 1200 1400 1600 1800
time,s
-3
x 10
2

1
Disp. isol.,m

-1

-2
0.014 -0.031
-3

-4
0 200 400 600 800 1000 1200 1400 1600 1800
time,s

Figure 5.7- Relative response displacements: top story and isolators

Relative velocities, m/s Max. Min.


0.03

0.02
Vel.top,m/s

0.01

0
0.0207 -0.0198
-0.01

-0.02
0 200 400 600 800 1000 1200 1400 1600 1800
time,s

0.02

0.01
Vel.isol,m/s

-0.01
0.0192 -0.0186
-0.02
0 200 400 600 800 1000 1200 1400 1600 1800
time,s

Figure 5.8- Relative response velocities: top story and isolators

92
Absolute accelerations, m/s2 Max. Min.
0.6

2
Acc.top,m/s
0.4

0.2

-0.2
0.2232 -0.1074
0 200 400 600 800 1000 1200 1400 1600 1800
time,s
0.3
2
Acc.isol,m/s

0.2

0.1

0
0.2120 -0.0955
-0.1
0 200 400 600 800 1000 1200 1400 1600 1800
time,s
0.3
2
Acc.base,m/s

0.2

0.1

0 0.1996 -0.0992
-0.1
0 200 400 600 800 1000 1200 1400 1600 1800
time,s

Figure 5.9- Absolute response accelerations: top story and isolators,


and input acceleration at the base

Story seismic forces and base shear force, kN Max. Min.


10
F2,kN

0
6.4422 -13.3938
-10

-20
0 200 400 600 800 1000 1200 1400 1600 1800
time,s
5
F1,kN

-5
3.8187 -8.4811
-10
0 200 400 600 800 1000 1200 1400 1600 1800
time,s
10

0
Fb,kN

-10

-20 9.8247 -21.8109


-30
0 200 400 600 800 1000 1200 1400 1600 1800
time,s

Figure 5.10- Story loads and base shear force

The results show that the input is filtered through the isolation system to the structure in a
way that the story drift is negligible and the distribution of absolute accelerations over the
height is nearly uniform. This is one of the objectives of the isolation technique. However,
the level of isolation is the objective of a detailed design process using the limitations
provided by the human vibration comfort criteria.

 Homework 5.1| Re-design the isolated structure in the above example so that the peak
horizontal acceleration to fulfill the Level 1 of the human vibration comfort limit (Table 1
in Chapter 6 of the Lecture Notes).
Hint: construct the absolute acceleration response spectrum corresponding to the isolation
system damping ratio. Impose then, a desired spectral acceleration limit and determine the
corresponding natural period. With this, find the effective lateral stiffness of the isolation
system. Run the time-history analysis and determine the floor peak absolute response
accelerations.

93
 Homework 5.2| For the irregular structure in the Figure 4.21 (right), perform the
design of a linear-elastic base isolation system, consisting of 20 isolators placed right
beneath the columns, so that the sway fundamental natural period of the isolated structure
to be 2.5 times the sway natural period of the regular conventional structure. With 15%
damping ratio attributed to the isolation system, perform the seismic response analysis
using El Centro bi-directional input motion and compare the results with those obtained
for the conventional structure, eccentric structure.

Numerical example 5.4|Given the conventional RC frame structure (Figure 5.11, left),
subjected to a lateral harmonic load, a TMD-system design and further, the dynamic
response analysis are performed. The results are then compared with the response of
conventional structure.
Conventional structure data: Weight 𝑊 = 1000 𝑘𝑁; Flexural stiffness 𝐸𝐼 = 84000𝑘𝑁𝑚 ;
Damping ratio 𝜁 = 2%; Story height ℎ = 4.00 𝑚
The primary structure is subjected to the horizontal harmonic load of the form 𝐹(𝑡) =
𝐹 sin(𝜃𝑡) where the load amplitude is 𝐹 = 50 𝑘𝑁, the circular frequency 𝜃 = 15 𝑟𝑎𝑑/𝑠
and the loading duration is 20 minutes.
The analysis routine will be developed during classes.
TMD
F(t) F(t) Wd
Ws Ws

4.00 4.00

CONVENTIONAL STRUCTURE TMD-EQUIPPED STRUCTURE

Figure 5.11- Conventional structure and TMD-equipped structure

The coordinates associated to each model are given in the Figure 5.12.

Wd 2
Ws 1 Ws 1

CONVENTIONAL STRUCTURE TMD-EQUIPPED STRUCTURE

Figure 5.12- Degrees of freedom of the conventional structure (SDOF) and


unconventional structure (2 DOF)
94
(i) The design of TMD system (damping neglected)

The lateral stiffness of conventional structure is

𝑘 =5 = 78750 𝑘𝑁/𝑚 (5.22)

Therefore its natural circular frequency is

𝜔 = = 28.06 𝑟𝑎𝑑/𝑠 (5.23)

and the vibration period


𝑇 = 2𝜋 = 0.22 𝑠 (5.24)

The damper’s weight is chosen 5% of structure’s weight i.e. 𝑊 = 50 𝑘𝑁.


The tuning condition given by the eq.(6.16) in the accompanying Lecture Notes, yields the
required stiffness of the mass damper as
𝑘 = 𝑚 𝜃 = 1125 𝑘𝑁/𝑚 (5.25)

(ii) Eigenmodes analysis of TMD-equipped structure (Figure 5.11, right)


The mass matrix of the TMD-equipped structure is
𝑚 0 100 0
𝑴= = (5.26a)
0 𝑚 0 5
and the stiffness matrix is
𝑘 +𝑘 −𝑘 79875 −1125
𝑲= = (5.26b)
−𝑘 𝑘 −1125 1125
Running the eigenmodes analysis, the following spectral and modal matrices are obtained
14.8533 0
𝜴= , 𝑟𝑎𝑑/𝑠 (5.27a)
0 28.3395
0.4230 0
𝑻= ,𝑠 (5.27b)
0 0.2217
1 1
𝜱= (5.27c)
51.3892 −0.3892

Thus, the first mode is called absorber mode and the second mode is called structural mode.
The modal parameters are summarized in the Table 2.

Table 2- Mass participation factors and mode equivalence factors

Vibration mode index Mode mass participation factors Mode equivalence factors
𝝓𝑻𝒊 𝑴𝟏 𝑚 , (∑ 𝑚 𝜙 ,)
𝛤= 𝜆 = =
𝝓𝑻𝒊 𝑴𝝓𝒊 𝑚 ∑ 𝑚 (∑ 𝑚 𝜙 ,)
1 0.0268 0.0912
2 0.9732 0.9088

95
The corresponding mode shapes are given in the Figure 5.13.

Mode 1: 0.42 s- damper controlled Mode 2: 0.22 s-structural controlled


51.39 0.39

1 1

Figure 5.13 -Mode shapes of TMD-equipped structure

The fundamental mode mass participation is negligible compared to the structural mode.
With these findings, the modal vibration periods, can be roughly estimated by

𝑇 = 2𝜋 = 0.4189 𝑠 (5.28a)

𝑇 = 2𝜋 = 0.2217 𝑠 (5.28b)

which are nearly same as those obtained from the modal analysis.

(iii) Damping modeling


The damping coefficient of the conventional structure is
𝑐 = 2𝜁 𝑚 𝜔 = 112.2497 𝑘𝑁/𝑚 (5.29a)
and the damping coefficient of the TMD system is assumed first
𝑐 =0 (5.29b)
The damping matrix of the global model is therefore
𝑐 +𝑐 −𝑐 112.2497 0
𝑪= −𝑐 𝑐 = (5.29c)
0 0
The generalized damping matrix yields
112.2497 112.2497
𝑪∗ = 𝜱𝑻 𝑪𝜱 = (5.29d)
112.2497 112.2497
which is not a diagonal matrix.
By neglecting however these terms, i.e. assuming the damping-associated coupling is
negligible in the global model, we obtain
0.0084 1.1141 2𝜁 𝜔 0
(𝜱𝑻 𝑪𝜱)(𝜱𝑻 𝑴𝜱) 𝟏
= = (5.30)
0.0084 1.1141 0 2𝜁 𝜔

96
The modal damping ratios are therefore
.
𝜁 = ∙ .
= 2.84 ∙ 10 ≈0 (5.31a)
.
𝜁 = .
= 0.0197 ≈ 2% (5.31b)

While the damping ratio for the fundamental mode is driven by the TMD system, the
second mode damping ratio is structural. In the initial design stage however, the TMD
system’s damping ratio can be attributed to the fundamental mode, as well as super-
structure damping ratio can be attributed to the second mode.
The time-histories of lateral displacements of the structure and TMD system, respectively,
is given in the Figure 5.14.

0.1
TMD disp,m

0.05

-0.05

-0.1
0 10 20 30 40 50 60
time,s
-3
x 10
2
Str disp,m

-1

-2
0 10 20 30 40 50 60
time,s

Figure 5.14- Response displacements: structure and TMD

The TMD-to-structure of steady-state peak displacements response ratio is 103. The


response of TMD is driven by a beating phenomenon, since the fundamental mode results
undamped and the harmonic load is quasi-resonant with the natural circular frequency of
TMD-equipped structure.
If the damping ratio of TMD system is selected now as 3%, then the generalized damping
matrix yields
2.4005 −0.0650
𝑪∗ = 𝜱𝑻 𝑪𝜱 = 10 (5.32a)
−0.0650 0.0029
which is still not a diagonal matrix but more closer to the diagonal assumption.
By neglecting however these terms, i.e. assuming the damping coupling is negligible in the
global model, we obtain
18.0434 −64.5387 2𝜁 𝜔 0
(𝜱𝑻 𝑪𝜱)(𝜱𝑻 𝑴𝜱) 𝟏
= = (5.32b)
−0.4888 2.9241 0 2𝜁 𝜔
The modal damping ratios are therefore

97
.
𝜁 = ∙ .
= 0.6074 = 60.74% (5.33a)
.
𝜁 = ∙ .
= 0.0516 = 5.16% (5.33b)

The time-histories of lateral displacements of the structure and damped TMD system,
respectively, is given in the Figure 5.15a.

-3
x 10
2
TMD disp,m

-1

-2
0 10 20 30 40 50 60
time,s
-3
x 10
2
Str disp,m

-1

-2
0 10 20 30 40 50 60
time,s

Figure 5.15a- Response displacements: structure (2% damping ratio) and TMD (3% damping ratio)

The TMD-to-structure of steady-state peak displacements response ratio reduces now to


1.32 and the peak steady-states response displacements of main structure increases by 21%
while the TMD steady-state displacement reduces by 65 times, as a result of adding
damping for the TMD system.
By providing only the optimum damping ratio to the damper (13% according to eq. 6.32b
in the Lecture Notes) the comparison of response displacements- initial and optimally
damped, are shown altogether in the Figure 5.15b. The reduction of peak structural
displacement is about 4 times.

1.50E-03 STRopt 1.00E-01


TMDopt

STRinit 8.00E-02 TMDinit


1.00E-03
6.00E-02

4.00E-02
5.00E-04
2.00E-02
Disp., m
Disp., m

0.00E+00 0.00E+00
0 10 20 30 40 50 60 0 10 20 30 40 50 60
-2.00E-02
-5.00E-04
-4.00E-02

-6.00E-02
-1.00E-03
-8.00E-02

-1.50E-03 time,s -1.00E-01 time,s

Figure 5.15b- Comparison on response displacements: initial and optimally damped

98
A schematic representation of harmonic input load and displacement responses for both,
structure and TMD, is given in the Figure 5.16.

0.04

0.03

0.02

0.01

-0.01
50

40
-0.02
30

20

10
-0.03
0

-10

-20 -0.04
0 x 10- 3
5 10 15
-30
1.5

0.5

-40
0

-0.5

-1

-50 -1.5
0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

Figure 5.16- Schematic representation of harmonic input load (black line, left joint) and TMD-
equipped structure’s response (blue lines, right joint and TMD)

Random loads require however, proper tuning with respect to a certain frequency
bandwidth. A more refined model needs a detailed modal treatment and careful optimum
design of TMD system using the developments given in the Lecture Notes.

 Homework 5.3| For the wind-excited one-story building in the Figure 5.11-left, design a
damped TMD system using the eq. (6.32a&b) for the optimum circular frequency and
optimum damping ratio, given in the Lecture Notes. The wind load is given in the file
“inputdata.txt” used in the numerical example 4.4.
Hint: perform the frequency-domain analysis of given load and determine the frequency
bandwidth information. Use then the tuning condition in order to pre-design the TMD
system. Run the analysis and check the validity of design.

 Homework 5.4| For the five story RC frame structure (Figure 5.17), analyzed in the
numerical example 4.6, subjected to lateral wind loads (the input loads are given in a form
of data file “5wdata.txt”), design a damped TMD system using the eq. (6.32a&b) for the
optimum circular frequency and optimum damping ratio, given in the Lecture Notes.
Check the validity of design by comparing the response of TMD-equipped structure with
the response of initial conventional structure.
Hint: perform the frequency-domain analysis of modal loads and determine their
frequency bandwidths information. Use then the tuning condition in order to pre-design
the TMD system based on modal approach.

99
TMD

3.0

3.0

3.0

3.0

3.0

7.0 4.0

Figure 5.17- Conventional structure (left figure) and TMD-equipped structure (right figure)

 Homework 5.5| For the structures analyzed in the HW 5.4, develop the energy balance
analysis framework and check the effectiveness of TMD system, in terms of energy
parameters.

100
ANNEX: Computational algebra using Matlab

Matlab platform and Simulink Toolboxes allow performing basic, standard and high level
dynamic analyses, as well. This section introduces the reader with basic computational
algebra using Matlab platform.
From the Matlab Editor window, the following basic operations allow performing*:

 Operation  Matlab syntax

4 −1 >> A=[4 -1;2 -6]


 Square matrix 𝑨 = A=
2 −6
4 -1
2 -6
5 >> b=[5;-3]
 Column vector 𝒃 = b=
−3
5
-3
>> b=[5 -3]
 Row vector 𝒖 = [ 5 −3]
b=
5 -3

 Identity matrix 𝑰𝟐 =
1 0 >> eye(2)
0 1 ans =
1 0
0 1
 Unit row vector 𝟏𝐫 = [ 1 1] >> ones(2,1)
ans =
1
1
 Unit column vector 𝟏𝐜 =
1 >> ones(1,2)
1 ans =
1 1
 Matrix trace 𝑡𝑟(𝑨) = −2 >> trace(A)
ans =
-2
 Vector transpose 𝒃𝑻 = [ 5 −3] >> b'
ans =
5 -3
 Matrix inversion 𝑩 = 𝑨 𝟏 >> inv(A)
ans =
0.2727 -0.0455
0.0909 -0.1818

101
 Matrix product 𝑨𝒃 =
23 >> A*b
28 ans =
23
28
 Matrix product 𝒃𝑻 𝑨 = [ 14 13] >> b'*A
ans =
14 13
>> det(A)
 Matrix determinant det (𝑨)
ans =
-22
>> D(1:2,1:2)
 Sub-matrix read
ans =
0.8147 0.6324 0.9575 0.9572 0.8147 0.6324
0.9058 0.0975 0.9649 0.4854
𝑫= 0.9058 0.0975
0.1270 0.2785 0.1576 0.8003
0.9134 0.5469 0.9706 0.1419

>> D(1:2,3:4)
 Sub-matrix read
ans =
0.8147 0.6324 0.9575 0.9572 0.9575 0.9572
0.9058 0.0975 0.9649 0.4854
𝑫= 0.9649 0.4854
0.1270 0.2785 0.1576 0.8003
0.9134 0.5469 0.9706 0.1419
>> [D(1:2,1:2); D(1:2,3:4)]
 Vertical concatenation
ans =
0.8147 0.6324 0.8147 0.6324
0.9058 0.0975
𝑩= 0.9058 0.0975
0.9575 0.9572
0.9649 0.4854 0.9575 0.9572
0.9649 0.4854
>> [D(1:2,1:2) D(1:2,3:4)]
 Horizontal concatenation
ans =
0.8147 0.6324 0.9575 0.9572 0.8147 0.6324 0.9575 0.9572
𝑩=
0.9058 0.0975 0.9649 0.4854
0.9058 0.0975 0.9649 0.4854
> u=rand(4,1)
 Random number generator
u=
𝒖 = [ 0.4218 0.9157 0.4218
0.7922 0.9595]
0.9157
0.7922
0.9595
>> sum(u)
 Sum of elements
ans =
𝑠𝑢𝑚(𝒖) = 3.0892
3.0892
>> max(u)
 Maximum value
𝑚𝑎𝑥(𝒖) = 0.9595 ans =
0.9595
>> min(u)
 Minimum value
ans =
𝑚𝑖𝑛(𝒖) = 0.4218
0.4218

*Additional operations used in practical applications are available in the Matlab Help
module.
102
REFERENCES

Akiyama, H., 1985, Earthquake-resistant limit-state design for building, the University of
Tokyo Press, Tokyo, Japan
Allemang, R.L., and Brown, D.,L., 1995, Experimental modal analysis, Shock and Vibration
Handbook, 4th edition, Chapter 18, Harris, C.M. eds., McGraw Hill, N.Y.
Baber, T., T., Noori, M., N., 1985, Random Vibration of Degrading Pinching Systems, Journal of
Engineering Mechanics, 111(8), p. 1010-1026
Bachmann, H., Ammann, W., 1987, Vibration of Structures induced by Man and Machines,
IABSE-AIPC-IVBH, Zürich 1987, ISBN 3-85748-052-X
Baker, W., E., 1973. Explosions in the Air, University of Texas Pr., Austin
Bathe, K., J., 1995, Finite Element Procedures, Prentice Hall, Upper Saddle River, NJ
Bathe, K., J., Wilson, E., L., 1976, Numerical Methods in Finite Element Analysis, Prentice Hall,
Inc., Englewood Cliffs, NJ
Biggs, J.M, Introduction to Structural Dynamics, McGraw-Hill, New York, 1964.
Biot, M., A., 1932, Vibrations of Buildings during Earthquake, Chapter II in Ph.D. Thesis No.
259 entitled Transient Oscillations in Elastic System, Aeronautics Department, Calif. Inst.
of Tech., Pasadena, California, U.S.A.
Bouc, R., 1967, Forced vibration of mechanical systems with hysteresis, Proc. of 4th conference
on Nonlinear oscillation, Prague, Czechoslovakia
Buchholdt, H., A., Moosavi Nejad, S., E., 2012, Structural Dynamics for Engineers, ICE
Publishing, 2nd Edition
Chapra, S., Canale, R., P., 1988, Numerical methods for engineers, McGraw-Hill, Inc., New
York, NY
Chopra, A., K., 2001, Dynamics of Structures: Theory and Applications to Earthquake
Engineering, 2nd ed., Prentice Hall, Englewood Cliffs, NJ
Clough, R., W., Penzien, J., 1993, Structural Dynamics, 2nd edition, McGraw-Hill, Inc., New
York, NY
Cooley, J., W., Tukey, J., W., 1965, An algorithm for the machine calculation of complex Fourier
series, Mathematics of Computation, 19 (1965), pp. 297-301.
Connor, J., 2003, Introduction to structural motion control, Prentice Hall Inc., 2003
Den Hartog, J., P., 1956, Mechanical Vibrations, 4th ed., McGraw-Hill, New York
Osgood, B., 2007, The Fourier transform and its applications. Lecture notes, Stanford University
Hatch, M., R., 2001, Vibration simulation using MATLAB and ANSYS, Chapman &
Hall/CRC, ISBN 1-58488-205-0
Holmes, D., J., 2001, Wind Loading of Structures, Spon Press, London
Housner, G., W., 1956, Limit Design of Structures to Resist Earthquakes, Proc. 1st World Conf.
Earthquake Engineering, 5-1 to 5-13, Berkeley, CA
Iancovici, M., 2012, Statica constructiilor II. Exemple numerice (2nd Ed. digital, in Romanian)
available at www.utcb.ro

103
Iancovici, M., Vezeanu, G., Bogdan, O., 2012, A time-domain approach for the performance-
based design of tall buildings, Proceedings of the 15WCEE (World Conference on
Earthquake Engineering), Lisbon, Portugal (paper no. 3013, on CD)
Ifrim, M., 1984, Dinamica structurilor şi Inginerie seismică, EDP (in Romanian)
Inman, D., 2007, Engineering vibration, Prentice Hall Inc., 3rd Ed.
Irwin, A., W., 1978, Human Response to Dynamics Motion of Structures. The Structural
Engineer London. Melbourne, Private Communication
Jeary, A., P., 1986, Damping in tall buildings – A mechanism and a predictor,Earthquake
Engineering and Structural Dynamics, 14, 733-750.
Kareem, A., 1982, Fluctuating wind loads on buildings, J. Engrg. Mech. Div., ASCE, 108(EM6),
p. 1086-1102
Kashima, T., 2001, Experimental testing, Lecture Notes for the Earthquake Engineering
Course, International Institute of Seismology and Earthquake Engineering (IISEE),
Japan
Kramer, S., L., 1996, Geotechnical Earthquake Engineering, Prentice Hall, Inc., NJ
Luiz, F., M., 2012, FTOOL Interactive, Graphical Program for Structural Analysis. Educational
Version 3.0, User Manual.
Lungu, D., Vacareanu, R., Aldea, A., Arion, C., 2000, Advanced Structural Analysis,
Conspress, Bucharest
Macavei, F., Poterasu, V., F., 1994, Complemente de Dinamica structurilor, Editura Virginia,
Iasi (in Romanian)
Mays, G., C., Smith, P., D., Hetherington, J., G., 2001, Blast effects on buildings, Thomas
Telford
Meirovitch, L., 1975, Elements of Vibration Analysis, McGraw-Hill, New York.
Mendis, P., Ngo, T., Haritos, N., Hira, A., Samali, B., Cheung, J., 2007, Wind Loading on Tall
Buildings, EJSE Special Issue: Loading on Structures
Mendis, P., A., Ngo, T., 2002, Assessment of tall buildings under blast loading and aircraft
impact, Toward a better Built Environment, Innovation, Sustainability and Information
Technology, International Association of Bridge and Structural Engineering, Australia,
pp. 375-376
Menun, C., Der Kiureghian, A., 1998, A Replacement for the 30 % Rule for Multi-component
Excitation, Earthquake Spectra, Vol. 13, Number 1, February 1998.
Mertins, A., 1999, Signal analysis: wavelets, filter banks, time-frequency transforms and
applications, John
Wiley & Sons Inc., ISBN: 978-0-471-98626-3
Newmark, N., M., 1959, A Method of Computation for Structural Dynamics, ASCE Journal of
the Engineering Mechanics Division, Vol. 85 No. EM3
Ormondroyd, J., Den Hartog, J., P., 1928, The theory of dynamic vibration absorber, Trans.
ASME, APM-50-7, pp.9-22
Otani, S., 1981, Hysteresis Model of Reinforced Concrete for Earthquake Response Analysis, J.
Faculty of Engineering, University of Tokyo, Series B, Vol. XXXVI-II, No. 2, p. 407-441
104
Park, Y., J., Ang, A., H., 1985, Mechanistic seismic damage model for reinforced concrete, J.
Struct. Eng., ASCE, 111, p. 740 - 757
Satake, N., Suda, K., Arakawa, T., Sasaki, A.,Tamura, Y., 2003, Damping Evaluation Using
Full-Scale Data of Buildings in Japan, Journal of Structural Engineering, 129, 4, 470-447,
2003
Simiu, E., Scanlan, R., H., 1996, Wind effects on structures, 3rd Ed., Wiley, New York
Simiu, E., 2011, Design of Buildings for Wind, 2nd Ed., Wiley.
Sivaselvan, M., V., Reinhorn, A., M., 2000, Hysteretic Models for Deteriorating Inelastic
Structures, Journal of Engineering Mechanics, ASCE, 126(6), p. 633-640
Smith, P.,D., Hetherington, J.,G., 1994. Blast and ballistic loading of structures, Butterworth
Heinemann Ltd.
Smith, S.,W., 1997, The Scientist and Engineer’s Guide to Digital Signal Processing, California
Technical Publishing
Soong, T.,T., Constantinou, M.,C., 1994, Passive and active structural vibration control in civil
engineering. New York: Springer
Soong, T., T., Dargush, G.,F., 1997, Passive energy dissipation systems in structural engineering,
Wiley, ISBN-13: 978-0471968214
Trifunac, M., D., 1971, Zero Baseline Correction of Strong Motion Accelerograms, Bull Seism.
Soc. Amer., 1971, 61, 1201-1211.
Turner, M., J. , Clough, R., W., Martin, H., C., Topp, L. J., 1956, Stiffness and Deflection
Analysis of Complex Structures, J. of Aero. Sci., 23 (9), Sept. 1956.
Vanmarcke, E., 2010, Random Fields. Analysis and Synthesis. World Scientific Publishing Co.
Pte. Ltd.
Wen, Y., K., 1976, Method for Random Vibration of Hysteretic Systems, Journal of Engineering
Mechanics, ASCE, 102 (2), p. 249-263
Wen, H., L., Chopra, A., 2002, Accidental torsion in buildings: analysis versus earthquake
motions, Earthquake Engineering Research Center University of California, Berkeley,
REPORT NO. 01
Wilson, E., L., Bathe, K., 1973, Stability and Accuracy Analysis of Direct Integration Methods,
Earthquake Engineering and Structural Dynamics, Vol. 1, p. 283-291
Zienkiewicz, O., C., Taylor, R., L., 1989, The Finite Element Method Vol I: Basic Formulation
and Linear Problems, 4th ed., McGraw-Hill, New York;

AIJ-GBV (1991), Guidelines for the evaluation of habitability to building vibration, Architectural
Institute of Japan, Maruzen
EN1991-1 Eurocode 1 (2002), Actions on structures - Part 1-4: Wind actions. European
Committee for Standardization
EN1993-1 Eurocode 3 (2005), Design of steel structures - Part 1-1: General rules and rules for
buildings. European Committee for Standardization
ISO 2631 (1997), Mechanical Vibration and Shock, Evaluation of human exposure to whole-body
vibration, International Organization for Standardization.

105
ISO 3010 (2001), Basis for design of structures– Seismic actions on structures, International
Organization for Standardization.
ISO 4354 (2009), Wind Actions on Structures, International Organization for
Standardization.
ISO 10137 (2007), Basis for design of structures– Serviceability of buildings and walkways against
vibrations, International Organization for Standardization.
NIST (National Institute of Standards and Technology), 2005. Federal Building and Fire
Safety Investigation of the World Trade Center Disaster: Final Report on the Collapse of the
World Trade Center Towers. NIST NCSTAR 1. Gaithersburg, MD.

*MATLAB, The Language of Technical Computing, Version 6.1 (Release 12.1), The Mathworks
Inc., http://www.mathworks.com

106

You might also like